Attorney Docket No.246-021PCT ATRIAL FIBRILLATION ANALYSIS METHODS, SYSTEMS AND APPARATUSES CROSS-REFERENCE TO RELATED APPLICATIONS [0001] This application claims priority to U.S. Provisional Application No. 63/546,857, entitled ATRIAL FIBRILLATION ANALYSIS METHODS, SYSTEMS AND APPARATUSES, filed on November 1, 2023, naming Nicholas SUNDERLAND as an inventor, the entire contents of that application being incorporated herein by reference in its entirety. BACKGROUND [0002] Atrial fibrillation (afib) is an irregular and often very rapid heart rhythm (arrhythmia). Atrial fibrillation increases the risk of stroke, heart failure and other heart-related complications. [0003] During atrial fibrillation, the heart's upper chambers (the atria) beat chaotically and irregularly — out of sync with the lower chambers (the ventricles) of the heart. For many people, atrial fibrillation may have no symptoms. However, A-fib may cause a fast, pounding heartbeat (palpitations), shortness of breath or weakness. [0004] Episodes of atrial fibrillation may come and go, or they may be persistent. Atrial fibrillation can be a serious medical condition that requires proper treatment to prevent stroke. [0005] Treatment for atrial fibrillation may include medications, therapy to reset the heart rhythm, and catheter procedures to block faulty heart signals. There can be waves of membrane potential change that propagate through cardiac muscle and trigger contraction. [0006] The typical heart has four chambers — two upper chambers (atria) and two lower chambers (ventricles). Within the upper right chamber of the heart (right atrium) is a group of cells called the sinus node. The sinus node is the heart's natural pacemaker. It produces the signal that starts each heartbeat. [0007] In a regular heart rhythm, the signal travels from the sinus node through the two upper heart chambers (atria), the signal passes through a pathway between the upper and lower chambers called the atrioventricular (AV) node and the movement of the signal causes your heart to squeeze (contract), sending blood to the heart and body. [0008] In atrial fibrillation, the signals in the upper chambers of the heart are chaotic. As a result, the upper chambers shake (quiver). The AV node is then bombarded with signals trying
Attorney Docket No.246-021PCT to get through to the lower heart chambers (ventricles). This causes a fast and irregular heart rhythm. [0009] The heart rate in atrial fibrillation may range from 100 to 170 or 180 or more beats per minute. In contrast, the normal range for a heart rate is 60 to 90ish beats a minute. [0010] Preventative concepts are often limited to choosing a healthy lifestyle believed to reduce the risk of heart disease and may prevent atrial fibrillation, such as managing stress, as intense stress and anger can cause heart rhythm problems. [0011] Also, there are medicines to control the heart’s rhythm and rate, blood-thinning medicine to prevent blood clots from forming and reduce stroke risk and medicine and healthy lifestyle changes to manage atrial fibrillation risk factors. More specifically, medications include beta blockers, blood thinners, calcium channel blockers, and heart rhythm medicines. SUMMARY [0012] In an embodiment, there is a method, comprising obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation and identifying regional atrial activation times based on data based on wavelet processing of data based on the first data. [0013] In an embodiment, there is a method, comprising obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation and reconstructing one or more unipolar EGMs based on data based on wavelet processing of data based on the first data. [0014] In an embodiment, there is a method, comprising obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation and obtaining second data based at least in part on wavelet processing of the obtained first data as part of a process to develop the second data, wherein the second data is data indicative of ventricular far-field artifact in the obtained first data. [0015] In an embodiment, there is a method, comprising obtaining first data based on electrical activity in a live human recorded in or on the live human who is one of afflicted with a condition, such as a heart condition, or who is not afflicted with a health condition or not afflicted with a heart ailment, and identifying activation times based on data based on wavelet processing of data based on the first data. In an embodiment, the identification includes
Attorney Docket No.246-021PCT identifying regional atrial activation times and/or ventricle activation times based on data based on wavelet processing of data based on the first data. [0016] In an embodiment, there is a method, comprising obtaining first data based on one or more unipolar recordings of electrical activity in a human, such as activity of an atrium and/or ventricle of a living human and/or in another body part of a human, such as a muscle group, who is afflicted with an ailment, such as atrial fibrillation, or who is not afflicted with a health ailment or not afflicted with a heart ailment (but could be afflicted with another ailment, such as muscle spasms), or who is otherwise healthy, or who is afflicted with a heart condition, such as palpitations, but not atrial fibrillation, and reconstructing one or more unipolar datasets based on data based on wavelet processing of data based on the first data. [0017] In an embodiment, there is a method, comprising obtaining first data based on one or more electrical phenomenon recordings of electrical phenomena in an atrium and/or other body part of a living human afflicted with atrial fibrillation or not afflicted with such but afflicted with another ailment (e.g., clogged arteries) or afflicted with that ailment and another ailment or who is not afflicted with an ailment or not afflicted with a heart ailment (but might be afflicted with another ailment unrelated to the heart, such as a wandering eye) and obtaining second data based at least in part on wavelet processing of the obtained first data as part of a process to develop the second data, wherein the second data is data indicative of ventricular far-field artifact in the obtained first data. BRIEF DESCRIPTION OF THE DRAWINGS [0018] FIGs.1-5 depict exemplary comparisons between normal electrical wave propagation in a normal functioning heart and electrical wave propagation in a heart afflicted by atrial fibrillation. [0019] FIGs.6 and 7 present exemplary flowcharts for exemplary methods according to an exemplary embodiment. [0020] FIGs.8a and 8b depict an exemplary catheter used in some embodiments. [0021] FIGs.9a is a schematic representation of a system embodiment showing a catheter in the left atrium. [0022] FIG.9b is a schematic representation of an atrial electrogram from one electrode. [0023] FIG.10 shows a schematic diagram of a catheter in a heart and additional recording, control and processing devices that are required for inverse endocardial mapping.
Attorney Docket No.246-021PCT [0024] FIGs.11-14 show exemplary potential maps over time for an exemplary scenario; [0025] FIG.15 presents pre-processing information according to an embodiment. [0026] FIG.16 presents potential to phase conversion information according to an embodiment. [0027] FIGs.17-20 show exemplary phase maps over time for an exemplary scenario. [0028] FIGs.21-22 show conceptual location teachings according to an embodiment. [0029] FIG.23 shows an exemplary phase gradient map in an exemplary scenario. [0030] FIG.24 shows actions of an embodiment of the method where (a) a representative potential distribution is sampled at internal points on a circle and (b) the potential distribution within the circle is reconstructed by a forward solution using potentials interpolated around the virtual inner circle from the sampled potentials. [0031] FIG.25 shows an exemplary algorithm according to an embodiment; [0032] FIG.26A shows raw atrial electrograms; [0033] FIGs.26B-D show wavelet related data; [0034] FIG.27 shows an electrograms with V far-field artifact subtracted; [0035] FIGs.28-32 show exemplary flowcharts for exemplary methods; [0036] FIG.33A shows an exemplary electrogram; [0037] FIGs.33B and 33C show wavelet related data; [0038] FIG.33D shows an exemplary electrogram; [0039] FIGs.34-38 show exemplary flowcharts for exemplary methods; [0040] FIG.39 shows an exemplary electrogram; [0041] FIG.40 shows an exemplary algorithm according to an embodiment; [0042] FIGs.41A-E and FIG.42 show graphs; [0043] FIG.43 shows an exemplary electrogram; [0044] FIGs.44-46 show exemplary flowcharts for exemplary methods; [0045] FIGs.47A-C and FIGs.48A-F and FIGs.49A-E show confirmational data; and [0046] FIG.50 shows an exemplary algorithm for an exemplary embodiment.
Attorney Docket No.246-021PCT DETAILED DESCRIPTION [0047] The teachings herein in part relate to identifying heart tissue / heart cells of interest in a living human, which tissue / cells have an association with the occurrence of atrial fibrillation. The teachings herein also relate to methods and procedures for altering the heart / implementing a surgical procedure on the heart to at least partially alleviate or otherwise reduce the occurrence and/or symptoms of atrial fibrillation. But as noted at the beginning of this paragraph, the teachings herein are only in part related to scenarios associated with atrial fibrillation and of the tissue/cells having an association with the occurrence of atrial fibrillation. In an exemplary embodiment, the teachings herein also relate to identifying heart tissue / heart cells of interest or other tissue or cells in a living human, where the tissue / cells do not have an association with the occurrence of atrial fibrillation. In an embodiment, the heart tissue / cells of interest or other cells / tissue relate to / are part of a heart or part of another organ or other body part of a human that does not suffer from atrial fibrillation and/or does not suffer from any other health condition or ailment or otherwise has a heart that functions regularly or otherwise in a statistically normal manner or otherwise is a heart of a living human in a state that corresponds to a statistically healthy heart, all other things being equal. In an exemplary embodiment, the heart is a heart that corresponds to that of a 15
th to 90
th percentile or a 25
th to 80
th percentile or a 35
th to 70
th percentile human factors engineering human male and/or female who is less than, greater than and/or equal to 1, 2, 3, 4, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95 or 100 years old or any range of values therebetween in one month increments, who was born in New Zealand, the United States, a country that was and/or became a member of the European Union, the State of Japan, the People’s Republic of China and/or the Commonwealth of Australia. To be clear, while the teachings detailed herein are often discussed and directed towards the identification of tissue/cells of a heart afflicted with atrial fibrillation or some other ailment, in alternate embodiments, the teachings detailed herein can be applicable and are applicable to analyzing and otherwise obtaining data and evaluating cells and or tissue of a heart that is not afflicted with atrial fibrillation or otherwise is a heart that suffers no other elements and otherwise is a healthy heart. Said, in an embodiment, the heart could be a heart that is afflicted with an ailment that is different than atrial tribulation. For example, the heart could be a heart that is diagnosed as in need of a pacemaker and otherwise could have a pacemaker, or clogged arteries, or has a hole therein. That is, the teachings herein can be applicable to a heart that has clogged passages or otherwise a heart of an individual that has clinically high cholesterol requiring some form of remediation. The heart could be a heart
Attorney Docket No.246-021PCT where there is an ascending aorta for example. Any condition of the heart to which the teachings detailed herein can be of utilitarian value can be a heart to which the teachings detailed herein can be applied or otherwise are applied unless otherwise noted, providing that the art enables such, all in the interest of textual economy. Thus, any disclosure herein of a method action and/or a device and/or system and/or a program product etc. having applicability to a heart afflicted with atrial tribulation corresponds to an alternate disclosure of any of these other scenarios, whether to a heart or to another organ or other body part, unless otherwise noted. [0048] Indeed, the teachings detailed herein can be utilized for diagnostic purposes or otherwise to evaluate a heart to determine the heart does not have any problems or otherwise is not afflicted with a condition, such as for example, atrial fibrillation, but is afflicted with another condition. Still, embodiments include detection, diagnosis and/or treatment of atrial fibrillation. Corollary to this is that embodiments include evaluating a heart and otherwise executing at least some of the methods detailed herein to determine that the heart is fine and otherwise is healthy or at least is not afflicted with atrial fibrillation. It could be that the diagnosis is, for example, that whatever is source of a phenomenon that caused the patient to seek treatment or otherwise caused a doctor to recommend the teachings detailed herein be utilized for a given person, the source is not atrial fibrillation. It could be simply age based, statistical demographic based or genetic based, etc., that caused the teachings herein to be executed, and the prognosis is that there is no atrial fibrillation, or at least not any effective atrial fibrillation. [0049] In an embodiment, electro-anatomic mapping is used to guide exemplary treatments of heart rhythm disturbances. This can involve the following actions: i) 3D heart surface geometry is reconstructed for the chamber (or chambers) of concern; ii) electrical signals (time varying electric potentials) are recorded at a number of registered points on the heart surface; iii) electrical activity throughout the region is rendered, in time and space; and iv) statistical analysis is implemented. Based on this information, likely sources of rhythm disturbance in the heart wall are then located and, in some embodiments, ablated. [0050] Embodiments can include the use of real time and near real time mapping and analysis of electrical activity in persistent and permanent atrial fibrillation using intracardiac catheters that record electrical activity simultaneously at multiple 3D locations.
Attorney Docket No.246-021PCT [0051] In some embodiments, acquisition, analysis and visualization processes can be completed within 30, 25, 20, 15, 10, 5, 4, 3, or 2 seconds, or any value or range of values therebetween in 0.1 second increments (e.g., 4.4, 3.9, 3.3 to 7.8 seconds, etc.) [0052] In some embodiments, there is the use of flexible multi-electrode basket catheters that make direct contact with the atrial surface or comes into close proximity with the heart surface(s). Electrical activity can be mapped throughout the cardiac cycle while the electrodes remain in contact with the chamber wall and/or are located in the chamber and their 3D position is known. The source of rhythm disturbances can also be identified while the electrodes are in the chamber, or within 20, 15, 10, 5, 4, 3, 2, or 1 minutes, or any value or range of values therebetween in 0.1 minute increments of the removal of the electrodes from the chamber (or movement of the electrodes to another portion of the chamber – embodiments include using standard catheters to read potentials at multiple regions within a chamber to harness the accuracy of a tightly spaced arrangement of electrodes while using conventional readily available electrode catheters (e.g., those with 64 or 128 electrodes)). [0053] In some embodiments, the Constellation catheter (Boston Scientific, Inc.) basket catheter with 64 electrodes to record potentials is used to obtain potential readings within the chamber / on the surface of the chamber. In embodiments, the catheters can be basket catheters. In embodiments, regional mapping can also be carried out using high-resolution grid catheters such as by way of example the Abbott Advisor with 16 electrodes at 3 mm centres or the BioSense Webster OPTRELL with 48 electrodes at 2.4 mm centres. In some embodiments, inverse mapping and/or wavelet-based signal processing can be applied with these catheters, and, in some embodiments, this can add considerable functional utility to the use of these systems. [0054] Conversely, some embodiments use noncontact mapping methods to obtain potentials within the heart. Here, electrical activity is measured on a surface adjacent to the inner or outer surface of the cardiac chamber of interest and is then mapped onto the heart surface in question using inverse problem techniques. [0055] In some embodiments, St. Jude Medical, Inc. catheters and mapping system intended for noncontact 3D electro-anatomic mapping are used to obtain the potentials within the chamber. The catheter has a 64-electrode array mounted on an inflatable balloon. [0056] In some embodiments, an Acutus Medical, Inc. mapping system based on an expandable basket catheter that contains 42 electrodes as well as ultrasound probes can be used
Attorney Docket No.246-021PCT to obtain data within a heart. With this approach, electrical activity recorded with a multi- electrode basket catheter in an atrial cavity is used to estimate an equivalent electrical dipole distribution within the atrial wall. In some embodiments, a Cardioinsight Technologies, Inc. system is used to map electrical activity measured on the body surface with a multi-electrode vest onto the epicardial surface of the heart using an inverse method. The approach is non- invasive, but it requires accurate 3D anatomic representations of body surface and epicardial geometry using computed tomography (CT) or magnetic resonance imaging (MRI). Note that while embodiments focus on measuring data using recording devices in the atrial cavity, in another embodiment, the recordings are taken outside the atrial cavity, providing that the art enables such. In an embodiment, the recordings are taken outside the human in fact (such as could be the case, for example, for non-heart related issues, such as muscle spasm, etc.) Any disclosure herein of recording in the atrial cavity corresponds to an alternate disclosure of recording outside the atrial cavity, providing the art enables such, unless otherwise noted, in the interests of textual economy. [0057] But embodiments are not limited to the above noted catheters or even the specific features associated therewith. Embodiments include using data from a device having less than or more than or equal to 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 325, 350, 375, 400, 450, 500, 550, 600, 700, 800, 900, 1000, 1250, 1500 or more or any value or range of values therebetween in 1 increment (e.g., 23, 38, 22-66, etc.) number of channels electrodes (at least read electrodes). Any device, system or method that can enable utilitarian data collection can be used in some embodiments providing that the art enables such. [0058] Embodiments include utilizing methods for determining physiological information for an internal body surface, such as an endocardial surface. [0059] FIGs.1-5 show general schematic representations of features of two human hearts 100. The heart on the left presents the sinus rhythm, and the heart on the right presents atrial fibrillation. The heart on the left is a normal beating heart, where regular impulses 110 produced by the sinus node 102 emanate therefrom as shown. The heart on the right is a heart beating under atrial fibrillation, where the impulses 120 are shown in a manner to represent by way of conceptual example the chaotic nature thereof. The heart on the right is meant to represent the most common heart rhythm disturbance scenario. The heart on the left shows the control heart, where the normally regular spread of electrical activation across the atria can be compared to the rapid chaotic rhythm with intermittent transmission of activation to the
Attorney Docket No.246-021PCT ventricles shown in the right (which replaces the normal regular spread on the left). This results in the irregular and often rapid heart rate that increases the risk of stroke and can limit heart function in some scenarios. [0060] As shown from figures 1 to 5, the impulses traveling through the heart in the atrial fibrillation scenario appear at first glance to be quite random. Indeed, most potential/current (electrical potential / electrical current) mapping systems and techniques result in data sets that are ambiguous at best. In some instances, some correlation can be deduced for short period of time, but often, the correlation is not replicated. The teachings herein go beyond the mere mapping of the electrical potentials within the heart, which teachings can provide a platform for identifying ambulation targets in a heart afflicted with atrial fibrillation. [0061] The teachings herein are directed to providing an interventional treatment of persistent and permanent atrial fibrillation, or at least providing an identification of heart cells / tissue that are causing or at least implicated in the atrial fibrillation. The teachings herein can be directed to proving the interventional treatment (or the identification) to sustained episodes of atrial fibrillation that do not spontaneously terminate within two weeks. Embodiments can be directed to episodes that do not spontaneously terminate in 1 week, or 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, or 15 days, or any value or range of values therebetween in 1 day increments. [0062] FIG.6 presents a high level flowchart for an exemplary method, method 600, according to an exemplary embodiment. Method 600 includes method action 610, which includes the action of obtaining electrical potentials inside a human, where in an embodiment, the potentials are inside a human heart while the human heart is beating and thus while the human associated with the human heart is alive. Corollary to this is that in at least some exemplary embodiments, at least those directed to treating atrial fibrillation, the obtained electrical potentials in method action 610 are obtained while the heart is experiencing a sustained episode of atrial fibrillation that has not spontaneously terminated within the past few days or weeks for that matter. In this exemplary method, the heart is a heart that is in a scenario of permanent atrial fibrillation. But it is briefly noted that at least some exemplary embodiments herein can utilize the teachings detailed herein to analyze healthy tissue or otherwise analyze parts that are not experiencing atrial fibrillation. [0063] Some exemplary embodiments include obtaining electrical potentials inside the heart utilizing a basket catheter. For example, FIG.8a shows a schematic representation of a multi- electrode mapping catheter 1. It includes multiple expandable splines 2 with sensors or
Attorney Docket No.246-021PCT electrodes 3 spaced evenly along the splines. The catheter is open in the sense that fluid such as blood for example, can pass freely between the splines. However, as shown in FIG.1b, in this exemplary embodiment, all electrodes lie on a continuous virtual surface 4 that is closed in the mathematical sense. [0064] FIG. 2a shows a schematic representation of the mapping problem in a heart 5. The catheter 1 is located in the left atrium (LA), and electrical potentials generated by electrical activity in the heart can be recorded by each of the multiple electrodes 3 simultaneously. An electrogram (potential as a function of time) at a typical electrode 3 is displayed for a single cardiac cycle in FIG. 2b. The potential distribution on the LA endocardial surface 6 at successive instants through the cardiac cycle can be reconstructed based on the corresponding potentials recorded at the multiple catheter electrodes. This can be executed using an inverse approach, or solving an inverse problem. The objective of the inverse problem in some embodiments is to reconstruct source information (e.g., atrial endocardial potentials) from the measured field (e.g., potentials recorded at the catheter electrodes) based on a priori information on the physical relationships between sources and measured field. In this setting, information is also required about the 3D geometry of the endocardial surface and the 3D location of each of the electrodes. This information can be obtained using a CT scan while the catheter electrodes are in the heart chamber or some other form of imaging technique, such as using radiopaque beads, by way of example, etc. In embodiments, if electrodes are opaque under radiation, the electrodes can be located in 3D using biplane cone fluoroscopy. Note that in some embodiments, use of intracardiac mapping systems supplied by Abbott (St Jude), BioSense Webster (Carto) and Medtronic provide instantaneous readout of 3D electrode locations using a hybrid magnetic/electrical impedance sensing system. [0065] Any device, system, and/or method of correlating the location of the catheter to locations on a chamber of a heart that can enable the teachings detailed herein can be used in at least some embodiments. The idea is to obtain a spatial relationship, whether it be for example in cartesian coordinates or polar coordinates or radial coordinates (and thus typically in three dimensions), between the electrodes and locations on the surface of the heart chamber so that the data obtained from the electrodes can be correlated to specific and discrete locations on the surface of the heart chamber. In an embodiment, the accuracy is within plus or minus 1 cm, 0.75, 0.5, 0.4, 0.3, 0.2, 0.1, 0.08, 0.06, 0.04, 0.02, or 0.01 cm, or any value or range of values therebetween in 0.01 cm increments.
Attorney Docket No.246-021PCT [0066] FIG.9a shows the four cardiac chambers: the left atrium (LA), right atrium (RA), right ventricle (RV) and left ventricle (LV). An endocardial surface 6 is typically at least part of the surface of one of the chambers of the heart. Where discussed herein the endocardial surface may be represented as a 2D surface, but it is understood that a user of the system would typically be investigating a 3D endocardial surface enclosing a chamber within. In some embodiments an endocardial surface may be only a portion of a chamber, that portion being of interest. [0067] Moreover, empirical research has shown that the further apart the electrodes are from each other, owing to, for example, expansion of the splines of the catheter, the less accurate the ultimate resulting potential map of the interior surface of the chamber. Embodiments include utilizing a catheter where the electrodes are no more than 5, 4.5, 4, 3.5, 3, 2.5, 2, 1.5, 1, 0.75, or 0.5 mm, or any value or range of values therebetween in 0.1 mm increments. Embodiments thus can include the controlled or otherwise limited expansion of the catheter splines to values that are lower than that which would otherwise be possible within a given chamber. By way of example only and not by way limitation, if the catheter was capable of expansion to the point where the electrodes would be 4 mm away from each other, the expansion might be limited to an expansion where the electrodes are only 2.5 mm away from each other, at most. This may not result in adequate data from the electrodes to map the entire surface of the chamber. However, it may not be necessary to map the entire chamber, and, alternatively, owing to the specific nature associated with implementing the teachings detailed herein, the catheter can be moved to another location within the chamber, and potentials can be obtained, and that particular region of the surface can then be mapped, and this can take place in a serial fashion for other locations within the chamber, and thus other locations on the surface. Put another way, embodiments can include obtaining the potentials within the chamber in one fell swoop for all locations on the surface of the chamber, and embodiments can include obtaining potentials within the chamber in a serial manner at different locations within the chamber for different regions of the surface of the chamber. This latter method can provide a more accurate data set, because the electrodes are closer to each other, which more accurate data set will provide more accurate potential mapping of the locations on the surface of the chamber. [0068] By way of example only and not by way of limitation, in an exemplary embodiment, the catheter can be inserted to be proximate a first region of the chamber, and can be controlled to expand the splines of the catheter to a point where the electrodes extend from each other but within the utilitarian distances detailed above her other utilitarian distances. The electrical
Attorney Docket No.246-021PCT potentials can be recorded for utilitarian time periods, such as for example, over at least and/or no more than 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 35, 40, 45, or 60 seconds or more, or any value or range of values therebetween in one second increments, where consecutive electrical potentials are recorded through a recording cycle that operates at at least 500 or 1000 or 1500 or 2000 or 3000 or 4000 or more hertz or any value or range of values therebetween in 1 Hz increments. The resulting data set obtained for the temporal period where the electrodes are at the given location within the chamber can be stored and/or manipulated or otherwise used to implement at least some of the teachings detailed herein, and then the catheter can be moved to a different location within the chamber, and, if the splines were contracted, the splines can be re-expanded to obtain utilitarian spacing of the electrodes, and then the data collection can be repeated at this new location within the chamber to obtain the data set that can be utilized to develop an accurate potential map for this new region of the surface of the chamber, and this movement / data collection series of actions can be repeated however many times needed to obtain accurate data and/or accurate potential mapping of the desired regions within the heart chamber. In an exemplary embodiment, the catheter is utilized with electrodes spaced at the aforementioned limits for example to map at least 50, 55, 60, 65, 70, 75, 80, 85, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, or 100% of the surface area of the chamber or any value or range of values therebetween in 1% increments in accurate manner. [0069] It is briefly noted that imaging of the catheter spline, or more specifically, the electrodes and the cardiac chamber surface, showing the electrodes and the surface of the chamber, and/or coordinate data indicating the relative locations between the electrodes and the surface of the chamber, will be obtained each time that the catheter is moved to a new location within the chamber to obtain data that will be used to develop a map of the potentials on the surface of the associated region of the chamber. [0070] To be clear, some exemplary embodiments include any device system and/or method that can enable electric potentials within a chamber of a heart, whether on a surface of the chamber or spaced away from the chamber, to be obtained in a utilitarian manner to implement the teachings detailed herein can be utilized in at least some exemplary embodiments providing that the art enable such. But briefly, FIG.10 shows an exemplary system that can be used to obtain the electrical potentials from within the heart chamber and/or develop at least some of the data herein. Additional details of this will be described below, but briefly, the utilization of the arrangement of figure 10 to execute one or more of the method actions detailed herein.
Attorney Docket No.246-021PCT Thus, in some embodiments, the arrangement of figure 10 is configured, such as via firmware and/or software and/or hardware, to implement one or more the method actions herein. FIG. 10 can have a control unit configured to implement one or more or all of the functionalities detailed herein and/or method actions detailed herein. Also, parts of FIG.10 can be bifurcated and/or trifurcated and spatially located remotely provided that such enables the teachings herein. [0071] Again with reference to figure 6, method 600 further includes method action 620, which includes the action of manipulating the obtained electrical potentials to obtain a utilitarian data set. In an exemplary embodiment, additional details of which will be described below, method action 620 is executed by implementing electrical potential mapping techniques further described below, but briefly, any one or more of the potential mapping teachings disclosed in United States Patent No. 10,610,112, issued on April 7, 2022, naming Bruce Smaill as an inventor, and naming Auckland UniServices Limited of New Zealand as the Applicant, can be used in some embodiments to obtain values for electrical potentials for locations on a surface of a heart chamber. By way of example, the inverse mapping techniques disclosed in the aforementioned patent can be utilized in some embodiments to develop a potential map of the surface or portions of a surface of the chamber of interest. But to be clear, and device system and/or method that will enable electrical potential mapping of potentials obtained from electrodes onto the surface of the chamber can be utilized in at least some exemplary embodiments, providing that the art enable such. [0072] Method action 620 further includes, at least in some embodiments, preprocessing of the data obtained from the potential mapping of the surface. Under the rubric of manipulating the obtained electrical potential to obtain a utilitarian data set, embodiments further include implementing cardiac tissue cell phase mapping techniques utilizing the data obtained from the execution of the potential mapping technique detailed above. Some additional features of this will be described below, but briefly, by way of example, a Hilbert transform can be utilized to develop instantaneous phase data from the potential data such as that disclosed by Pawel Kuklik et al in Reconstruction of Instantaneous Phase of Unipolar Atrial Contact Electrogram Using a Concept of Sinusoidal Decomposition and Hilbert Transform, published in IEEE Transactions On Biomedical Engineering, Vol.62, No.1, January 2015. But to be clear, any other transformation technique or any other device system and/or method that can enable these data to be developed from the electrode potential readings and/or from the surface potential data can be utilized in at least some exemplary embodiments providing that the art enable such.
Attorney Docket No.246-021PCT [0073] Method 600 further includes method action 630, which includes the action of statistically analyzing the utilitarian data set obtained in method action 620. In an embodiment where the utilitarian data set is a phase map or otherwise constitutes phase data of specific locations on the surface of the chamber, over a utilitarian time period, such as, for example, 10 or 15 or 20 seconds as noted above, which utilitarian time period can correspond to the timing of the readings of the electrical potentials utilizing the electrodes located in the chamber, the action of statistically analyzing the utilitarian data set can include time averaging maximum phase gradients between the different locations on the surface of the chamber. [0074] Method 600 further includes method action 640, which includes the action of analyzing the results of the statistical analysis executed in method action 630. In an exemplary embodiment, as will be described in greater detail below, chamber surface locations that have a statistically meaningful phase gradient can be considered locations where there exists heart tissue that is playing a role in causing atrial fibrillation of the heart, at least relative to other tissue of the heart. In at least some exemplary embodiments, at least some of these locations having the nonzero phase gradient can be considered for targeting in an ablation process. [0075] FIG.7 provides a flow variation of an exemplary method, method 700, according to an exemplary embodiment. This method does not specifically require the actor to obtain the electrical potentials from within the heart chamber. Instead, another actor could obtain the information and otherwise provide the information to another actor executing method 700. In this regard, method 700 could be executed remotely from the patient otherwise from the operating room where the electrical potentials are being recorded. By way of example only and not by way limitation, in an exemplary embodiment, an Internet connection or a telephone connection or some other form of relative high-speed data communication system can be utilized to transfer the role signal potentials and or the spatial location data associated with the electrodes relative to the surface from the operating room or whatever hospital or location where the human patient is being treated or otherwise where the human patient is located during the action of obtaining the electrical potential within the heart, to a remote location, such as where a server or a remote computer is located, which could be tens or hundreds or thousands of kilometers away, in this remote computer remote server could implement method 700. Accordingly, embodiments include methods of practicing remote treatments or remote analysis and/or devices and/or systems that enable such, such as by way of example only and not by way limitation, a laptop and or a desktop computer or some other type of computer system, such as a smart phone for that matter, located otherwise co-located with the patient, that can
Attorney Docket No.246-021PCT receive the data from the electrodes or otherwise receive the data based on the data from the electrodes, and transform this data into a communicator ball medium which can be communicated over the Internet or over a phone line etc. to the remote location, where method 700 could be executed. Corollary to this is that at least some exemplary embodiments include some form of computing system, such as one or more of the aforementioned systems, that can receive the transferred data and execute method action 700. Moreover, some embodiments include the ability to then send the results of method 700 back to the location where the patient is located so that an ablation treatment procedure for example can be implemented based on the result of method action 700 and/or any one or more the additional actions detailed herein. But of course, an embodiment includes a system that is located with the patient that can execute method 700. [0076] In at least some exemplary embodiments, all of the proceeding paragraphs, minus the actual treatment, can take place in at least some exemplary embodiments, within 30, 25, 20, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, or 2 minutes, or any value or range of values therebetween in one minute increments. And in some embodiments, the treatment will add no more than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 70, 80, or 90 minutes to those times. [0077] And returning to FIG. 7 and method 700, method 700 includes method action 710, which includes the action of obtaining data based on electrical potentials in a live human. In an embodiment, the potentials are within a live human heart. As just noted, method action 710 does not require per se the actual action of utilizing the electrodes located in the heart. Method action 710 can instead be executed by obtaining a data set or otherwise obtaining data based on those readings from the electrodes. For example, method action 710 can be executed by receiving over the internet a data package or a series of data sets or a single data set indicating the time based electrical potential values on one or more or all of the electrodes of the catheter and or the accompanying spatial relationships between the electrodes and the surface of the heart chamber. [0078] The dataset could be a set of raw electrical values, or could be data extrapolated from the raw electrical values (e.g., a normalized set of electrical potentials, or pre-processed electrical potentials, or electrical potentials where extraneous values are omitted or smoothed, etc.). As used herein, “data based on X” means X or data that is extrapolated from X or data that is extrapolated from data extrapolated from X.
Attorney Docket No.246-021PCT [0079] But still, with respect to embodiments where the system, such as a computer system, is implementing method 700, method action 710 could be executed by receiving the electrical values directly from the electrodes via leads extending from the catheter to the computer system utilized to execute method action 700. [0080] In an embodiment, the obtained data can be time based data (such as electrical potential readings) for ABC number of electrodes, where for respective electrodes, there are discrete values in time increments of at least 200, 500, 750, 1000, 1250, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000 Hz (the numbers need not be the same for each electrode), over at least, or equal to or no more than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, or 90 seconds (including consecutive seconds) or any value or range of values therebetween in 1 second increments. Thus, the data could be, for a 64 electrode catheter (by way of example only and not by way of limitation, a 64 channel Constellation
TM basket catheter available as of January 10, 2022, at the Royal Melbourne Hospital) and a system taking measurements at 2,000 Hz over a 13 second period, 1,664,000 time based values. And note that that might be for one location. But method action 710 could also be executed in one fell swoop, to cover multiple readings from multiple locations, and thus there could be potentially two or three or four or five or more times that number of values. Still, in at least some exemplary embodiments, it is envisioned that electrode readings for one location within the chamber will be obtained and transferred to the computer for the execution of method 700, or at least one or more method actions therein, such as the execution of method 720 and/or method 730 (introduced below), and then the catheter will be moved to obtain additional electrical potentials, which will then be sent to the computer, and so on. [0081] Method 700 further includes method action 720. This can include implementing a preprocessing of the obtained data to obtain second data. It is briefly noted that method action 720 is optional in some embodiments and/or is otherwise a method action that can be practiced in various extremes or lack thereof. FIG. 15 shows some conceptual data associated with addressing scenarios where the signals from the electrodes are noisy and contaminated by the asynchronous electrical activity of the ventricles. One or both of these aspects can be subtracted, using a suite of robust wavelet-based filtering applications that enable the recovery of the underlying atrial electrograms. In an embodiment, the filter applications can correspond to those presented in greater detail further below that are based on wavelet processing. Briefly, embodiments include utilizing a reconstructed electrograms obtained by executing and/or
Attorney Docket No.246-021PCT based on the wavelet filter applications detailed below. Any disclosure herein of the use of an electrogram and/or any action based on an electrogram corresponds to a disclosure of using the reconstructed electrogram obtained based on data based on wavelet filters detailed below unless otherwise noted, providing that the art enables such. To be clear, in an embodiment, any of the teachings herein can be based on the reconstructed electrograms based on the data based on wavelet filters. Moreover, any disclosure herein of the use of atrial activation timing and/or any action based on atrial activation timing corresponds to a disclosure of using the atrial activation timing developed / obtained based on data based on wavelet filters detailed below unless otherwise noted, providing that the art enables such. To be clear, in an embodiment, any of the teachings herein can be based on the atrial activation timing determined based on data based on the wavelet filters. Any disclosure herein of data where V artifact has been removed corresponds to a disclosure of using the wavelet filters detailed below to develop the V artifact and removing such from the electrogram and using data based thereon. [0082] In an embodiment, method 720 can be skipped, and instead, method action 710 can be such that the obtained data based on the electrical potentials within a live human heart includes the wavelet filtering detailed herein. Indeed, in an exemplary embodiment, the data based on the electrical potentials within a live human heart of method action 710 can be the reconstructed electrograms detailed below based on the wavelet processing. Alternatively, and/or in addition to this, the data based on the electrical potentials within a live human heart could be the atrial activation timings developed based on wavelet filtering and/or the ventricular artifact data developed based on wavelet filterings, etc. The second data is thus the reconstructed electrogram and/or atrial activation timing and/or V artifact data (which includes the atrial electrogram with the V artifact removed) all based on the data based on the wavelet filtering data. [0083] Further, in an embodiment, method action 620 can include executing any of the wavelet processing actions detailed herein and/or any of the associated actions detailed herein. In an embodiment, the result of method action 620 can be the reconstructed electrogram developed based on wavelet processing, the atrial activation timing developed based on wavelet processing, and/or the V artifact data developed based on wavelet processing (which includes the atrial electrogram with the V artifact removed). [0084] And note that this can be an example of where the data of method action 710 can be data based on electrical potentials within a live human heart. In this regard, in an exemplary embodiment, method action 720 could be executed by the hospital or another actor in a scenario
Attorney Docket No.246-021PCT where, for example, method 700 is executed by some remote facility located remotely from the patient (in which case for example method action 720 would not be part of method 700, and thus an abbreviated version of method 700 would be practiced). In any event, this is but one example of how data based on electrical potentials within a live human (or non-human) heart could come about, which data based on electrical potentials could be the data obtained in method action 710. [0085] Method 700 further includes method action 730, which can include executing potential mapping, such as forward mapping or inverse mapping, of the surface of the chamber in which the electrodes are or were located. By way example only and not by way of limitation, any one or more the techniques detailed in the aforementioned US patent noted above, U.S. Patent No. 10,610,112, can be utilized in at least some exemplary embodiments. Briefly, we will focus on inverse mapping techniques. That is, some embodiments use inverse potential mapping. This can be done even though at least some of the electrodes, including more than 20, 30, 40, 50, 60, 70, or 80% or more of the electrodes used to obtain the potential values are not in contact with the atrial wall. Inverse mapping can account for this and be used to reconstruct a time-varying potential field across the 3D surface, such as the left atrial chamber surface. [0086] Here, the second data obtained using the pre-processing action of method 720 is used in the potential mapping of method action 730. [0087] Embodiments include systems to enable reconstruction of panoramic electrical activity in a heart chamber from physiological information, most particularly, time-varying electrical potentials (which may also be referred to as electrical fields or simply, fields) recorded using an open catheter inside the chamber that contains multiple sensors which may comprise electrodes, some or all which are not in contact with the wall of the chamber. In some embodiments, a numerical approach can be used to estimate physiological information (such as electrical potentials, electrical fields, or fields) in the volume bounded by the electrodes from the recorded potentials. This provides the additional boundary conditions necessary for accurate inverse mapping of potentials onto the inner surface of the heart chamber. For instance, in inverse solution packages that employ Boundary Element Methods (BEMs), it is utilitarian to specify both potential and potential gradients at measurement points. [0088] But some embodiments utilize meshless methods to address the inverse problem represented in FIG. 2. There, a set of fictitious current sources (or sinks) is positioned on a boundary outside the cardiac surface and it is assumed that Laplace's law holds throughout the
Attorney Docket No.246-021PCT volume inside the boundary. Source magnitudes which give rise to potentials that best match potentials recorded on the catheter are then determined using standard inverse solution methods. Corresponding potentials on the cardiac surface are then estimated from these sources. Meshless methods can be efficient and robust in this setting, and can be much simpler computationally, are far less reliant on accurate 3D descriptions of the endocardial geometry of the heart cavity and provide solutions when measurement points lie on the endocardial surface relative to the numerical approach described above. [0089] This system can enable rapid reconstruction and visualization of electrical potentials on an internal body surface, particularly an internal surface bounded by a chamber such as the endocardial surface of a cardiac chamber, or region of that chamber. These potentials comprise physiological information and can be, in some embodiments as noted above, from electrical potentials which may be measured with an expandable multi-electrode basket catheter, in which either all or some of the electrodes are not in contact with the surface. Such a catheter is open in a sense that bodily fluid such as blood within the chamber passes freely through it, but in which the electrodes define a mathematically closed 3D surface. [0090] It is anticipated that the system or methods disclosed could be used as the basis of a complete mapping system or alternatively to enhance the performance of existing or proposed systems. [0091] Some exemplary methods obtain exemplary potential maps, will now be briefly described. [0092] An exemplary method of determining physiological information for an internal body surface using an open catheter comprising multiple electrodes bounding a volume within the catheter can include: [0093] a) obtaining a first set of electric potentials using a plurality of the electrodes, [0094] b) determining a first set of boundary conditions from the first set of electric potentials, [0095] c) using the first set of boundary conditions to perform a forward solution to a first set of differential equations to provide a second set of electric potentials within the volume, [0096] d) determining a second set of boundary conditions from the second set of electric potentials, [0097] e) using the second set of boundary conditions to perform an inverse solution to a second set of differential equations, and
Attorney Docket No.246-021PCT [0098] f) determining the physiological information for the internal body surface using the inverse solution. [0099] The method can also include interpolating the first set of electric potentials. The physiological can comprise electric potentials on the internal body surface. The internal body surface can comprise an endocardial surface. The endocardial surface can comprise, least in part, an atrium or ventricle. The physiological information can comprise an electro-anatomical mapping. The method can comprise using a numerical method to solve the first or second set of differential equations. The numerical method can comprise any one or more of a finite element method, a boundary element method, or a meshless method. The numerical method can be implemented using a processor. The open catheter can comprise a flexible basket. The method can comprise positioning the catheter within a chamber bounded by the internal body surface. The method can comprise positioning the catheter proximal to a region of the internal body surface. The method can comprise positioning the catheter in a plurality of positions within the chamber. The method can comprise locating the catheter in a first position to obtain a first set of physiological information for a first portion of the internal body surface, and at least one second position to obtain a second set of physiological information for a second portion of the internal body surface. [00100] The method can comprise introducing the catheter into the body using a percutaneous technique. [00101] The method can be a method of determining physiological information for an internal body surface of a human using an open catheter comprising multiple electrodes bounding a volume within the catheter, the method comprising: [00102] a) obtaining a set of electrode electric potentials using a plurality of the electrodes; [00103] b) determining a boundary that contains the internal body surface; [00104] c) determining a set of discrete fictitious sources on the boundary; [00105] d) using inverse solution methods, determining source magnitudes of discrete fictitious sources of the determined set of discrete fictitious sources that give rise to potentials that sufficiently match the obtained electric potentials; [00106] e) determining corresponding potentials on the internal body surface of the human from the determined source magnitudes; and
Attorney Docket No.246-021PCT [00107] f) using the determined corresponding potentials on the internal body surface, determining the physiological information for the internal body surface. [00108] In some embodiments, the inverse solution methods are standard inverse solution methods. [00109] In some embodiments, the internal body surface is an endocardial surface of a cardiac chamber. [00110] The method can comprise introducing the catheter into the human body using a percutaneous technique. [00111] The action of determining the boundary surface is executed before, during and/or after the action of obtaining the set of electrode electric potentials. [00112] In some methods, the inverse solution methods are standard inverse solution methods. [00113] In some methods, the methods include obtaining the geometry of the internal body surface, wherein the internal body surface is a heart chamber and obtaining data indicating the position of the catheter within the heart chamber. [00114] The method can include creating a visual representation of the heart chamber based on data indicative of the electrode electrical potentials, the source magnitudes of the discrete fictitious sources, the corresponding potentials on the internal body and the geometry of the heart chamber and position of the catheter in the heart chamber. The method further can include executing spatio-temporal processing of the determined corresponding potentials on the internal body surface. [00115] The method can include reconstructing position of the catheter relative to the internal body surface and accounting for potential error in the determined corresponding potentials on the internal body surface. The method can include displaying the determined physiological information on an image of the internal body surface. The method can include locating the catheter at a first position to ultimately obtain a global image of the internal body surface and locating the catheter at at least one second position to obtain a more accurate estimate of potentials on a portion of the internal body surface, the more accurate estimate of potentials corresponding to the determined corresponding potentials on the internal body surface. The action of determining corresponding potentials on the internal body surface can be executed by a numerical method. The action of determining corresponding potentials on the internal body surface can be executed by a meshless method.
Attorney Docket No.246-021PCT [00116] And there can be non-transitory storage medium having machine-readable instructions stored thereon, that when executed by a processor cause the processor to perform the following actions: determine a boundary that contains an internal body surface of a human; determine source magnitudes of discrete fictitious sources of a set of discrete fictitious sources that give rise to potentials that sufficiently match a set of electrode electric potentials obtained from a plurality of electrodes that were and/or are located in the internal body surface; and determine corresponding potentials on the internal body surface from the determined source magnitudes. [00117] FIG.24a illustrates how the inverse endocardial mapping problem is approached with meshless methods that use the MFS. In FIG. 24a a set of discrete fictitious sources 11 is positioned on a boundary 12 that contains the endocardial surface 6 of the cardiac chamber. The fictitious sources are each fundamental solutions of Laplace's equation (equation 1 above) within the source-free volume Ω contained in the cardiac chamber. The potential ϕ
i at any point (x,y,z) in 3 due to a source i located on the boundary 2 is
where r
i is the Euclidean distance between (x,y,z) and the source at (x
i,y
i,z
i). [00118] That is r
i=√((x−x
i)
2+(y−y
i)
2+(z−z
i)
2) [00119] The total potential Φ at any point in Ω contributed by a set of M fictitious sources is
[00120] Source densities β
i are selected to match potentials recorded at each of the electrodes on the catheter 4 in a utilitarian manner, but also potentials inside the catheter 5 estimated by solving the forward problem if desired. Source densities are determined by solving inverse
Attorney Docket No.246-021PCT problem with appropriate regularization and endocardial surface potentials are then mapped using equation 5. [00121] While optimal placement of fictitious sources is empirical to some extent, a number of relatively straightforward a priori rules apply. The number of independent fictitious sources is less than the degrees of freedom of the measured potential distributions as determined by singular value analysis. The distribution of sources on the fictitious boundary should be bound by this constraint and it should be adaptive—most dense where the measurements are closest to the cardiac surface and least dense where they are furthest from it. Where sources are sparse, the accuracy of inverse mapping is increased with displacement of the surface from the cardiac boundary and vice versa. Finally the displacement of the boundary should be sufficient to accommodate uncertainty in the representation of cardiac surface geometry. [00122] FIG. 24b represents a method of distributing fictitious sources that is consistent with these rules. The number of independent sources 11 is equal to the number of electrodes 3 on the catheter and their spacing reflects the position of the catheter with respect to the wall. It is possible to add additional sources without loss of generality by interpolating along the fictitious boundary between independent source points. [00123] The use of meshless methods in the setting above gives rise to computationally efficient inverse solutions that are marginally less accurate than BEM, but more robust in the presence of uncertainty about endocardial geometry and the relative position of measurement sites on the catheter with respect to the endocardial surface. Finally, the positioning of independent sources on the external boundaries provides direct feedback on the spatial resolution of potentials that are mapped onto the endocardial surface. [00124] By way of example only and not by way limitation, data techniques can be used to reconstruct electrical potentials (voltages) on the inner surface of the atrial cavity from signals recorded at the individual electrodes on a basket catheter where some or all electrodes are not touching the atrial surface. The 3D geometry of the inner surface of the heart chamber (such as the left atria, with reference to FIG.11) which can be specified and the positions of each of the electrodes with respect to this surface are also known. This information is information that can be obtained by using, for example, electrical mapping systems used to guide ablation in clinical electrophysiology laboratories, such as those at the Royal Melbourne Hospital on January 10, 2022.
Attorney Docket No.246-021PCT [00125] In an exemplary scenario, where a 64 channel catheter (Constellation
TM, Boston Scientific) was introduced into the left atrium (LA) by transeptal puncture of a person afflicted with permanent atrial fibrillation, method 730 can include the action of using software to execute inverse mapping based on the electrode readings. Method 730 can be executed in real- time while the basket catheter is within the atrial cavity, at least in some embodiments. [00126] FIGs. 11-14 show by image presentation data from a result of the merging of data obtained using an inverse mapping technique (e.g., via a meshless method such as that detailed above) with data from CT images for an exemplary left atrium (posterior on the left, anterior on the right), where the figures show potential values at instantaneous points in time at locations on the surface of the left atria, and collectively show changes with some exemplary time progressions at the various locations by way of example only. More specifically, the electrode potential values corresponding to the time periods (where the monitoring device operated at 2,000 Hz – collecting readings two-thousand times per second) for the respective electrodes was, using inverse mapping, used to obtain electrical potentials on the surface of the left atria (visually represented in FIGs.11-14). By way of example, the data from 64 electrodes can be used to develop potential values for 64, 100, 150, 200, 250, 300, 350, 400, 450, 500, 600, 700, 800, 900, 1000, 1250, 1500, 1750, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 6000, 7000, 8000 or more, or any value or range of values therebetween in 1 increment locations on the surface (or surface region – again, embodiments can include utilizing the catheter with a “reduced” total volume, where the electrodes are closer together relative to that which would otherwise be the case, to obtain electrical potentials at a sublocation or a subregion within the cavity, to obtain more accurate data, and then the catheter can be moved to another subregion in the process continued), and these developed potentials being present for respective time increments at the cycle of the monitoring system (here, 2000 Hz). FIGs.11-14 show the time varying potentials on the surface, or more accurately, at the various surface locations. [00127] But to be clear, embodiments need not provide this exemplary any exemplary imaging. It can be sufficient to simply obtain time varying data sets for the potentials at the locations on the surface of the cavity. [00128] It is briefly noted that there can be a point of diminishing returns with respect to the number of locations used in the inverse mapping technique. In this regard, the potentials on the surface that are calculated are limited in accuracy vis-à-vis the number of electrodes utilized in a given catheter. Thus, utilitarian value can potentially be obtained by utilizing less than the full number of locations in the inverse mapping the net which a given system could potentially
Attorney Docket No.246-021PCT develop otherwise implement. This can have utilitarian value with respect to improving computational times. That said, in some embodiments, there could be potentially more locations than those detailed above. Any number of locations resulting from the inverse mapping (or forward mapping for that matter in embodiments that utilize such) that can enable the teachings detailed herein can be utilized in at least some exemplary embodiments. Corollary to this is that any cycle time and/or total length of time that can enable the teachings detailed herein can be utilized in at least some exemplary embodiments providing that the art enable such. [00129] The present inventors have found that mere potential mapping, while utilitarian in some instances, can be better utilized (or more specifically the data of the mapping can be better utilized), at least in some embodiments, by for example developing heart tissue cell phase data therefrom. In this regard, the potential maps are not necessarily useful in the context of the exacting nature that is required to identify heart tissue that is a driver or a substrate of atrial fibrillation, or otherwise identify tissue that can be ablated to alleviate at least some of the effects of atrial fibrillation. Indeed, the sampling of images shown in FIGs.11-14 show clearly that the potential mapping result exhibits the chaotic activity that is typical in atrial fibrillation. The point is that it is very difficult to extrapolate what is occurring using even exacting time increment potential mapping, and on a per patient basis, typically less utilitarian than the teachings herein vis-à-vis tissue identification. [00130] Thus, embodiments include extracting useful information, or more useful information from the electrical signals or electrograms recoded at the electrodes and/or from the potential map developed for the surface from those electrical signals. [00131] And this leads to method action 740, which includes executing phase mapping on the results of the potential mapping. Again, as noted above, at least some of the techniques detailed in the Journal article published in IEEE Transactions On Biomedical Engineering, Vol.62, No. 1, JANUARY 2015, by Pawel Kuklik et al, entitled Reconstruction of Instantaneous Phase of Unipolar Atrial Contact Electrogram Using a Concept of Sinusoidal Recomposition and Hilbert Transform, are used to develop the phase data. But other method actions can be used. Any method that can enable phase data to be developed that is utilitarian can be used in some embodiments. [00132] Briefly, a Hilbert transform, which can be used to transform data from the potential realm to the phase realm, is a linear operator transforming a function u(t) into a function H(u)(t)
Attorney Docket No.246-021PCT
where P is the Cauchy principal value of the integral. Phase is defined as an angle between the original signal and the Hilbert transform of the signal. Exact formulations of the phase vary between studies. Here, for example only, we define instantaneous phase as follows:
where u∗ sets the origin of the phase plane with respect to which phase is computed. Instantaneous phase increases monotonically within consecutive cycles of oscillation, reverting to a base value after completion of each cycle (by way of example). This property results in a “sawtooth” appearance of the instantaneous phase plot (see FIG.16, more on this below). [00133] In order to assess effects of specific features of the unipolar deflection morphology, idealized signals can be constructed using analytic considerations. This approach can enable control the morphology of the electrograms (such as the amplitude of R and S waves and the level of noise) and assessment of their effect on the reconstructed phase. Signal processing can be conducted in MATLAB (version 7.12, Mathworks Inc., Natick, MA, USA). The Hilbert transform can be calculated using the “Hilbert” function. Phase can be calculated using (2) using four-quadrant inverse tangent function “atan2.” Methods can include a transformation of atrial unipolar electrograms that can be applied prior to application of the Hilbert transform. The transformation can be based on the following assumptions (by way of example): [00134] 1) Due to the mathematical properties of the Hilbert transform, phase reconstruction of the instantaneous phase performs best in case of a sinusoidal morphology of the signal. [00135] 2) Local activity in unipolar signals related with a beginning of new cycle is proportional to the signal slope. [00136] Based on these considerations, the following transformation can be used: [00137] 1) The transformed signal is a sum of sinusoidal waves of one period length (called “sinusoidal wavelets” below).
Attorney Docket No.246-021PCT [00138] 2) For each time point of the original signal, one sinusoidal wavelet is created. [00139] 3) The amplitude of the sinusoidal wavelet is proportional to the slope of the signal at a given time point. [00140] 4) A wavelet is generated only if a derivative of the signal is negative (since a negative slope in unipolar electrogram corresponds with the passing of a wave). [00141] 5) The period of the sinusoidal wavelet is equal to the mean cycle length of the electrogram derived from dominant frequency of given electrogram. [00142] These steps can be summarized in a following equation:
[00143] where w(t) is a transformed signal, v(t) is an original electrogram, and T is a mean cycle length of the original electrogram (derived from dominant frequency of the electrogram) and sign() is the signum function. [00144] Because a sinusoidal wavelets can be used to construct transformed signals, we term this transformation a “sinusoidal recomposition.” The phase of the recomposed signal can be calculated using (2). Because the recomposed signal is a sum of sinusoidal wavelets with a mean value equal to zero and wavelets of the greatest amplitude are clustered around the negative slope of the local deflection, the resultant recomposed signal also has a sinusoidal morphology oscillating around zero value. Based on this consideration, we set u∗ (origin of the phase space with respect to which phase is computed; see (2)) to zero. One can use a specific definition of phase presented in (2) in order to obtain timing of the phase inversion (time point at which phase changes value from maximum to minimum denoting a beginning of a new cycle) coinciding with the timing of the local deflection in the electrogram. Formulation in (2) results, in case of a sinusoidal signal, in phase inversion occurring at position of the maximum negative slope of sinusoid. Since during sinusoidal recomposition individual sinusoidal wavelets are triggered according to the timing of the negative slope in electrogram, this will result in timings of the phase
Attorney Docket No.246-021PCT inversions centered at timings of the local deflections in electrogram. (All of this by way of example to enable some exemplary embodiments.) [00145] With respect to embodiments that utilize phase, mapping can be used to analyze atrial fibrillation, in some embodiments, the electrograms at the various locations (points) on the surface of the chamber (developed in method action 730 for example) are transformed into a phase record that represents the “time-history” of recent activation at respective locations (points). In embodiments, the transformation removes the magnitude variation that complicates interpretation of potential maps. [00146] FIG. 16 shows a conceptual temporally consistent change of electrical potential for a given location converted into change of phase for that given location over time in a heart cell of a heart not afflicted by atrial fibrillation, but it is noted that this can also correspond to such for a heart cell in a heart afflicted by atrial fibrillation. While a Hilbert transform was used to obtain this data, other methods / regimes of transformation can be used providing that the result provides utilitarian data, as noted above. Corollary to this is that software packages can be utilized to transform the electrical potential data obtained in method action 730 to the phase data obtained in method action 740, and this software can be part of the systems detailed herein. [00147] FIG.17 shows an example of the phase of the various locations on the chamber surface at an instant in time (time 0.0010) represented in visual format. Blue indicates activation (depolarization), light blue/green shows that tissue is refractory (unable to be activated), yellow/gold (partially repolarized) indicates that activation may be able to occur although propagation is expected to be slow. Red shows fully repolarised regions that will support the normal spread of electrical activation. This information is rendered as a color map across the left atria surface that “predicts” the spread of activation. A moving wave of activation (blue) cannot enter refractory regions (green) and spreads via regions that are repolarized (red). [00148] From the phase maps, in at least some exemplary embodiments, more consistent patterns of electrical activity that appear to repeat over time can be recognized. For instance, activation that appears to be occurring in some regions more than others can be identified, or at least more easily identified, and in some embodiments, and this can be associated with patterns of phase difference. [00149] In any event, as a result of method action 740, in at least some embodiments, phase data for the various temporal locations associated with the potential data obtained in method action 730 for the respective locations can be developed. In an embodiment, where for example, there
Attorney Docket No.246-021PCT are 1,000 surface locations (in the potential map, and thus if one to one locations are used, the phase map), and the sample cycle was 2,000 Hz over a 13 second period, 26,000,000 time based values for phase can be obtained (each of the 1000 locations having 26,000 values in 0.0005 ms increments). But again, at least some exemplary embodiments include normalizing or otherwise removing extraneous data points, so the actual number of values will change or could change. But in any event, the ideas that while the visual images can be useful or otherwise appealing, the data obtained in method action 740 need not necessarily be in the form of a visual representation. It could be in a temporal based matrix or data set. [00150] But here, Figs.17-20 show by image presentation data from a result of the merging of data obtained using the phase mapping techqniue with data from CT images for a exempalry left atrium (posterior on the left, anterior on the right), where the figures show phase values at instanteous points in time at locations on the surface of the left atria, and collectively show changes with some exempalry time progressions at the various locations by way of example only. This data can be obtained using the tranformation techniques described above vis-a-vis transforming the potential data associated with FIGs. 11-14 to the phase realm for FIGs. 17- 20. Again by way of example, the data used to delelop the potential map based on the data from 64 physical electrodes can be used to develop phase values for 64, 100, 150, 200, 250, 300, 350, 400, 450, 500, 600, 700, 800, 900, 1000, 1250, 1500, 1750, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 6000, 7000, 8000 or more or any value or range of values therebetween in 1 increment locations on the surface (or surface region), and these developed phase values being present for respective time increments at the sampling rate of the monitoring system. [00151] Method 700 further includes method action 750, which includes the action of identifying maximum phase gradients over time for the phase data obtained in method action 740. In this regard, for the various locations on the surface of the chamber, temporally corresponding phase values at the locations are compared to one another. For example, by conceptual explanation only, FIG.21 shows a posterior view of the left atria with a grid of an exaggerated size superimposed thereon. The grid has five blocks each corresponding to a virtual location of the surface of the chamber, the virtual location corresponding to a location of the phase map (and the potential map). The blocks are labeled A through E for explanation purposes. It is noted that these grids are exaggerated in size and are presented for conceptual purposes only. The concept is that each block represents a location where the inverse mapping actions above developed potential values therefore, and, in method action 740, phase values were developed therefore. Accordingly, for the various temporal locations for which data
Attorney Docket No.246-021PCT exists, there are values of phase for locations A, B, C, D and E. In the embodiment of method action 750, for each spatial location the maximum phase difference is retained from the collection of phase differences computed from the difference in phase value from the location and a group of neighboring locations. This computation occurs for each temporal point. [00152] In an exemplary scenario, for explanatory purposes only, with reference to the time increments of figures 17 to 20, for location A, at time 0.0010, respective values for phase of A minus phase of B, phase of A minus phase of C, phase of A minus phase of D and phase of A minus phase of E, are obtained (4 values total). The maximum difference is recorded selected by the greatest absolute magnitude of the four numbers, and this recorded maximum difference could be a positive number or a negative number. Here, for example, the phase between A and C will be the maximum, and it is a positive number. At time 0.5035, for example only by way of presenting this concept, the values are again calculated (they can be calculated for each temporal period in between or portions thereof) and here, the maximum phase is between A and D, and this can be calculated for time 1.005, and the maximum phase difference will be calculated for time 1.500, and so on (the calculations for each block can be done for each measurement cycle, and thus (using the example above) for 13 seconds, there will be 26,000 values for each block in some embodiments). The point is that for each temporal location in at least some exemplary embodiments, there will be a maximum phase gradient for each location relative to locations surrounding that location, and that value will change over time and sometimes that value will be positive and other times that value will be negative, it is also possible for the value to be zero. [00153] The alternating positive / negative value can be because in atrial fibrillation, the electrical impulses are chaotic. Sometimes, the cells at location A will be activated before the cells at location B, because, for example, the electrical wavefront will move generally from South to North (pretending that up is north, down is south, etc.) during some temporal periods, and then the electrical wavefront will move (generally) from north to south in other temporal periods (because of the chaotic nature of atrial fibrillation), and thus the cells at location A will be activated after the cells at location B. And in some temporal periods, the electrical wave front will move from East to West, and thus the tissue cells at location D will be activated before the tissue cells at location A, but the cells at A and B will be activated at approximately the same time. The fronts can (and will) move at various angular directions relative to each location, concomitant with the chaotic nature of atrial fibrillation wavefronts, at least in healthy normal heart tissue.
Attorney Docket No.246-021PCT [00154] The point is that the nature of atrial fibrillation will cause the cells to be activated in a random manner relative to one another and thus the phase differences between adjacent locations will change overtime or otherwise be different relative to that which would be the case in a normally functioning heart, where the wavefronts move in direction in a generally consistent manner (e.g., location A will almost always be activated before cell D, if the wavefront moves from the Northwest to the south East, etc.). [00155] FIG. 22 presents an expanded grid system to conceptually explain the concept under explanation in some greater detail. In the exemplary embodiment where phase gradients are being calculated for each temporal location (or the desired temporal locations, which can be a subset of the total number of data collection cycles), there will also be phase difference calculations for location B, and thus there will be a phase difference calculation for B vs. G, B vs. F, B vs. A and B vs. H, and for location F, phase differences calculated at F vs. J, F vs. I, F vs. D, F vs, B, and so on. For each temporal location (in an embodiment), the maximum phase difference at each location is determined. For time 0.5000 for example, the maximum gradient for A could be with location E, and the maximum gradient for B could be with F, for example. These values are recorded for time 0.5000. And note that such values should exist for each location (A, B, C, D, E, F, G, H, I, J and not shown locations K, L, M and so on – again, embodiments can include thousands of locations) for each temporal period. Accordingly, method action 750 can result in tens of thousands of values for the maximum phase difference for each location over a 10 or 15 seconds of time period by way of example. [00156] It is briefly noted that in the embodiment under description, only the four 90 degree adjacent locations (nodes or points) of the phase map are utilized. This is for computational convenience. In some embodiments, the diagonal nodes could be utilized in addition to this or instead of the 90° adjacent locations. In some other embodiments, some other form of comparison selection regime can be utilized, such as, for example, all locations within X millimeters of a given location. In some other embodiments, the set of neighboring locations could be fewer than 4. Any set of locations the phase difference of which can be calculated that can enable the teachings detailed herein to be implemented in the utilitarian value can be utilized in at least some exemplary embodiments. [00157] In any event, upon completion of method action 750, there will be resulting data that includes the maximum phase gradients over time for the phase data obtained in method action 740 for each location or however many locations that are utilitarian. This then leads to method action 760, which includes executing time averaging to obtain regional fingerprints. In an
Attorney Docket No.246-021PCT embodiment, these are atrial fibrillation fingerprints. But again, in other embodiments, the teachings herein are applied to someone who is not afflicted with atrial fibrillation. In this regard, for a given location, the many maximum phase gradient values obtained in method action 750 are time averaged and resulting value for each location are obtained. [00158] In an embodiment, it can be sufficient to simply add the values of the maximum gradients together and evaluate the result. The normal tissue will result in a value of zero or close to zero. The tissue that is a driver and/or a substrate of afib will result in a non-zero value in a statistically significant manner. And of course there are other methods of evaluating the phase gradients to determine whether or not a given location is experiencing a nonrandom activation. In an exemplary embodiment, artificial intelligence or a neural network or a deep neural network or otherwise a trained neural network can be utilized to analyze the phase gradient data to identify the locations that are the drivers of afib / substrates of afib. [00159] In an embodiment, there is a method, such as method 2800, represented by the algorithm of FIG.50, which includes method action 2892, which includes obtaining data based on electrical potentials within a live human heart, which can correspond to any of the teachings herein. Method 2800 includes method action 2894, which includes executing wavelet processing on data based on the obtained data based on the electrical potentials. This can include any one or more of the actions detailed below with respect to wavelet processing / data based on wavelet processing. Method 2800 further includes method action 2896, which includes executing method action 730, based on the results of method action 2894. That is, the data obtained in method action 2894 is used in the potential mapping of method action 730.Method 2800 further includes method 2898, which includes executing method actions 740, 750 and 760. [00160] Figure 23 visually presents the results of the action of time averaging the maximum phase gradients at the various locations on the surface of the chamber. Here, it shows that for the vast majority of locations on the surface of the chamber, the normalized time average maximum phase gradient is zero or close to zero (note that normalization is used for convenience – embodiments may not normalize the data). That is, for the various thousands of maximum phase gradients at each location, over time, the maximum gradients cancel out each other. This is because at some temporal periods they are positive and at other temporal periods they are negative, and so on, because of the chaotic nature of the wavefronts in a heart afflicted by atrial fibrillation. But inspection of the features of figure 23 reveal that there are some areas on the surface of the chamber where the normalized time average maximum phase
Attorney Docket No.246-021PCT gradient is not zero. This is because the tissue at those locations is interfering in some consistent manner with the chaotic wavefronts. Put another way, there is something that is interfering with the randomness of the activation of the cells in these locations relative to that which would otherwise be the case. In at least some embodiments, the tissue that is interfering with the randomness is tissue that is exasperating or otherwise causing the symptoms of the atrial fibrillation in the first instance. This tissue can be a driver and/or a substrate of afib. [00161] More specifically, the teachings herein enable the testing of whether phase signatures do repeat in specific locations more regularly than would be expected for normal heart tissue subjected to afib. This is done by averaging maximum phase gradients over time to create a phase signature map. FIG.23 identifies the phase signature map for a specific patient at the current state of their disease condition based on maximum phase gradients (derived from electrical activation wave fronts) for all regions over the recording period of the procedure. The averaging executed per region over times from 5-35 seconds for example (or longer or shorter or any time that provides utilitarian values that enable the teachings herein – the time is typically the time used to collect the potentials using the electrodes). The resultant maps show that activation, in this atrial fibrillation scenario, was preferentially initiated in antero-lateral region (red and yellow) just below the LA roof and spreading toward the mitral valve, and that activation occurs with equal probability everywhere else (blue). Using other measures, one can identify other atrial fibrillation fingerprints. Time-averaged spatial phase gradient maps indicate regions in which electrical activation is repeatedly delayed with respect to adjacent tissue, while time-averaged entropy maps identify regions with complex electrical activity. [00162] It is briefly noted that in a normal heart without afib, the normalized time averaged maximum phase gradient over the chamber may not be blue or otherwise may not have time average to zero. This is because there will always be a phase difference between one location and at least one other location adjacent that location when wavefronts move consistently, and thus there is always a maximum phase difference with respect to a given location relative to at least one other location. The maximum phase differences cancel out owing to the randomness of afib, and that is the fact upon which at least some of the teachings herein rely. Conversely, instead of the ordered wave progression, activating heart cells as the wave progresses, in scenarios where the atrial fibrillation scenario causes activation of the cell, the activations will be random and thus the phase difference between adjacent cells will cancel each other out over time, at least for cells that do not influence the afib or are otherwise healthy or normal.
Attorney Docket No.246-021PCT [00163] And the maps are highly repeatable as the time interval increased from 5 to 35 sec, can be co-located and internally consistent and observed in different locations across patients. Using method 700, one can identify the substrates that drive afib and provide appropriate targets for isolation (e.g., via ablation). Embodiments include ablating those areas or some of those areas or otherwise areas relative to or based on those areas, or otherwise providing instructions to do so. [00164] Because the atrial fibrillation fingerprints are stable over time, embodiments include building the data over the surfaces of both atria by sequential region-of-interest mapping (e.g., by incrementally moving the catheter to different regions in the chamber, thus enabling the basket to be used in a state where the electrodes are closer to each other than that which would otherwise be the case. Moreover, it is noted that embodiments include obtaining the potential data with electrodes no more than 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29 or 30 mm or any value or range of values therebetween in 1 mm increments from the adjacent surface of the chamber. [00165] In view the above, it can be seen that in at least some exemplary embodiments, there is a method, comprising obtaining heart phase data for a plurality of activation cycles of a living human afflicted with atrial fibrillation and analyzing the heart phase data to identify specific heart tissue locations where there are repeated and consistent temporal discrepancies of electrical activation relative to other tissue locations. In an exemplary embodiment, the action of obtaining the heart phase data can be executed by executing one or more of the actions detailed herein, such as by implementing a Hilbert transformation, or by using other numerical techniques or by pure number crunching is such can be utilitarian. In at least some exemplary embodiments. The action of obtaining heart phase data can be executed by receiving from a remote processor or the like pre-calculated heart phase data. The action of identifying specific heart tissue locations with our repeated and consistent temporal discrepancies can be executed according to anyone where the teachings detailed above. In an embodiment, the obtained heart phase data is data based on data that was developed using the filter applications that correspond to those presented in greater detail further below that are based on wavelet processing. Briefly, embodiments include utilizing heart phase data that is based on reconstructed electrograms obtained by executing and/or based on the wavelet filter applications detailed below. In an embodiment, the heart phase data is based, directly or indirectly, on reconstructed electrogram(s) obtained based on data based on wavelet filters detailed below. The heart phase data can be based on any of the teachings below that are presented as being based on the
Attorney Docket No.246-021PCT reconstructed electrograms based on the data based on wavelet filters. In some embodiments, the phase data is also based on atrial activation timing developed / obtained based on data based on wavelet filters detailed below and/or the V artifact removal teachings detailed below. Also, with respect to “obtaining heart phase data,” this can include executing any one or more of the actions detailed herein with respect to wavelet filters / wavelet processing. All of this is presented in terms of textual economy. [00166] In an exemplary embodiment, this method can be executed for any one or more of the values of location herein for any one or more of the data collection numbers frequencies and/or for any one or more of the temporal periods detailed herein whether method action of analyzing the heart phase data is executed in a period of time within 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 70, 80, 90, 100, 120, 130, 140 or 150% or any value or range of guys therebetween in 1% increments of the temporal period that form the basis of the data collection. For example, if the data was collected over 10 seconds, the method action of analyzing the heart phase data can be executed within three seconds or 15 seconds. [00167] In an exemplary embodiment, the action of analyzing the heart phase data includes implementing a statistical analysis on the heart phase data. By way of example only and not by way limitation, this can correspond to the time averaging detailed above or otherwise implementing signal averaging techniques. Thus, the action of analyzing the heart phase data can include implementing time averaging analysis on the heart phase data. In an exemplary embodiment, the utilization of the time averaging permits the values to become stable within a very short period of time, such as any one or more of the aforementioned times detailed above. Time averaging supports the generation of a fingerprint image which becomes static after a few seconds of data. Without time averaging, in some embodiments, the phase map is continuously changing and the clinician is trying to identify areas which are not changing. Time averaging reduces the highly variable and active areas down to zero which shows the non-zero areas which are areas of interest, in some embodiments. [00168] In an embodiment the action of analyzing the heart phase data includes, for a plurality of spatial locations on an interior surface of the heart, which spatial locations include the identified specific heart tissue locations, identifying respective maximum phase gradients for respective locations of the plurality of spatial locations over a length of time, time averaging the respective maximum phase gradients for the respective locations an identifying corresponding locations where the time averaged results are statistically aberrant and/or are not statistically aberrant, wherein the identified corresponding locations of the time average results
Attorney Docket No.246-021PCT that are statistically aberrant are the identified specific heart tissue locations and/or the identified corresponding locations of the time average results that are not statistically aberrant are not the identified specific heart tissue locations. By way of example only and not by way of limitation, in an exemplary embodiment, the time average results that are statistically aberrant can be those that are substantially nonzero. By way of example only and not by way of limitation, in an exemplary embodiment, the time average results that are statistically aberrant can be those that are different from the majority of values. By way of example only and not by way limitation, in most parts, most of the tissue will be normal, at least in most parts they can be treated. In this regard, instead of operating on a binary time average zero versus time average nonzero, a more flexible approach can be utilized that statistically analyzes the entire data set for all of the locations or at least most of the locations, and then determines which of the locations indicate statistically aberrant time values or otherwise time averaged values. [00169] Corollary to this is that the time averaged results that are not statistically aberrant could be the majority of the results or otherwise could be the results that are closer to zero than a subset of other results. Statistical analysis can be implemented to identify the non-aberrant time averages. In some embodiments, the knowledge of one of or new skill in the art can be relied upon to identify the values that are statistically aberrant and/or not statistically aberrant. Indeed, the requirement that there be statistical aberrant and/or non-aberrance can be dispensed with in some embodiments. In this regard, a nonstatistical approach could be applied based on underlying knowledge. [00170] In an embodiment, the action of analyzing the heart phase data includes, for a plurality of spatial locations on an interior surface of the heart, which spatial locations include the identified specific heart tissue locations, identifying respective maximum phase gradients for respective locations of the plurality of spatial locations over a length of time, time averaging the respective maximum phase gradients for the respective locations and identifying corresponding locations where the time averaged results are non-zero and/or statistically zero, wherein the identified corresponding locations of the time average results that are non-zero are the identified specific heart tissue locations and/or the identified corresponding locations of the time average results that are statistically zero are not the identified specific heart tissue locations [00171] In an embodiment, the respective maximum phase gradients are the respective maximum phase gradients between the respective locations and a plurality (2, 3, 4, 5, 6, 7, 8,
Attorney Docket No.246-021PCT 9, 10, 11, 12, 13, 14, 15 or more or any value or range of values therebetween in 1 increment) of proximate locations on the surface of the heart. [00172] In an embodiment, the proximate locations are effectively North-South-East-West adjacent locations and/or can be the four closest nodes. The locations can be the 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 or more or any value or range of values therebetween in 1 increment closest nodes. The locations can be the locations immediately surrounding the respective location. The locations can be the locations that provide utilitarian value with respect to implementing the teachings herein. [00173] Some embodiments include obtaining respective plurality of temporally spaced electrical potentials for respective electrodes of at least 20, 30, 40, 50, 60, 70, 80, 90, 100, 120, 150, 200, 250, 300, 350, 400, 450 or 500 or more electrodes of a catheter located in a heart chamber at a first location in the heart chamber, converting the obtained respective plurality of temporally spaced electrical potentials to the heart phase data, thereby obtaining the heart phase data. In these embodiments, the actions of identifying respective maximum phase gradients, time averaging, identified corresponding locations where the time average results are statistically aberrant and/or not statistically aberrant, are based on the obtained respective plurality of temporally spaced electrical potentials for the catheter located at the first location. [00174] Note that with respect to “converting,” embodiments include implementing the filter applications that correspond to those presented in greater detail further below that are based on wavelet processing. Briefly, embodiments include executing actions of converting that include executing the wavelet filter applications detailed below. In an embodiment, the converting includes taking the obtained electrical potentials and implementing the wavelet filters detailed below to obtain reconstructed EGMs and/or atrial activation timing, or at least executing one or more of the actions associated therewith. Any of the teachings herein of using “normal”’ EGMs and/or timing based on “normal” EGMs (subjected to preprocessing or not) corresponds to a disclosure using the EGMs reconstructed based on wavelet filtering and/or using atrial activation timing based on wavelet filtering. [00175] In some embodiments, the method includes (with respect to repositioning the catheter to map another different region) obtaining respective second plurality of temporally spaced electrical potentials for respective electrodes of at least 20, 30, 40, 50, 60, 70, 80, 90, 100, 120, 150, 200, 250, 300, 350, 400, 450 or 500 or more or any value or range of values therebetween in 1 increment electrodes of the catheter located in the heart chamber at a second location in
Attorney Docket No.246-021PCT the heart chamber different from the first chamber, converting the obtained respective second plurality of temporally spaced electrical potentials to second heart phase data (again, this can entail using the wavelet processing herein), for a plurality of second spatial locations on the interior surface of the heart, which second spatial locations include the identified specific heart tissue locations, identifying respective second maximum phase gradients for respective second locations of the plurality of second spatial locations over second length of time, second time averaging the second respective maximum phase gradients for the second respective locations; and identifying corresponding second locations where the time averaged second results are statistically aberrant and/or not statistical aberrant, wherein the identified corresponding second locations of the second time average results that are statistically aberrant are included in the identified specific heart tissue locations and/or the identified corresponding second locations of the time second average results that are not statistically aberrant are not included in the identified specific heart tissue locations. [00176] In an embodiment, this obtaining of the temporally spaced electrical potentials can occur for a third plurality of temporally spaced electrical potentials, and 4
th plurality and a 5th plurality and so on for each of the locations where the catheter is moved to within the chamber, and the associated method actions there with can be repeated accordingly. The results can be combined to obtain a phase map or otherwise to evaluate the entire surface array much larger portion of the surface relative to that which exists with respect to the original temporally spaced electrical potentials that were obtained prior to the action of obtaining the respective second plurality of temporally spaced electrical potentials. [00177] In an embodiment, the action of obtaining heart phase data and analyzing is executed in real time vis-à-vis a catheter located in a heart chamber. [00178] In an embodiment, the action of obtaining heart phase data and analyzing is executed while the catheter is located in a heart chamber. In an embodiment, these actions are executed while the patient is in the reading room or otherwise before the patient leave the operating room in which the potentials were obtained from the catheter in the patient. In an exemplary embodiment, these actions are executed within one, two, three, four, five, six, seven, eight, nine, 10, 11, 12, 13, 14, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 2728, 29, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75 or 80 or 90, or 100, or 110 or 120 minutes or any value or range of values therebetween in one minute increments from the beginning and or middle and/or an end of any one or more of the above-noted or herein noted medical procedure(s) that is / are
Attorney Docket No.246-021PCT implemented on the heart of the patient (e.g., ablation) based on the results of the action of analyzing. [00179] Indeed, in an embodiment, the action of sensing the potentials with the electrodes, the actions of executing method 700 for example or the actions of analyzing, and the action of ablating to completion are executed within some of the above noted periods (which are not repeated here for purposes of textual economy). [00180] In an exemplary embodiment there is a method that includes the action of developing a time-varying electrical potential map of a surface of a cavity of a beating heart. This action can be accomplished by one or more of the methods otherwise the sub-methods detailed above vis-à-vis transforming the electrical potentials obtained from the electrodes to the surface on the heart chamber. By way of example only and not by way of limitation, this can be done by the inverse mapping method detailed above, using, for example, a meshless method. Any one or more of the method actions detailed in the above referenced ‘112 patent can be executed to execute this action. [00181] This method further includes the action of developing a time-varying phase map of the surface of the cavity based on the developed time-varying electrical potential map. The techniques associated with developing the time varying phase map according to some embodiments are detailed above. Any other technique that can result in a time varying phase map, or a data equivalent thereof, can be used providing that such enables the teachings detailed herein. And it is noted that the phase map need not have one to one correspondence with the potential map. For example, the potential map may have 1500 (by way of example only) locations for which respective potentials that are developed on a time varying basis. The developed phase map could have fewer locations by way of example only. It can be utilitarian to only analyze some locations or otherwise a subset of the locations, and thus only some of the phase data will be developed for only some of the locations. This could save computational time or otherwise be utilitarian with respect to eliminating extraneous locations based on the results of the creation of the potential map. The elimination of the extraneous locations would thus provide more accurate results with respect to the phase transformation or otherwise with respect to the data to be extracted from this phase transformation. In an embodiment, at least 50, 55, 60, 65, 70, 75, 80, 85, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99 or 100% or any value or range of values therebetween in 1% increments of the locations are subject to the transformation from the potential realm to the phase realm. In an embodiment, by way of example only, the action of identifying specific heart tissue locations on the basis of repeated
Attorney Docket No.246-021PCT and consistent temporal phase discrepancies can be executed by anyone using the teachings detailed above. In an embodiment, the obtained heart phase data is based on data developed using the filter applications that correspond to those presented in greater detail further below and are based on wavelet processing. Briefly, embodiments include utilizing heart phase data based on reconstructed electrograms obtained by executing the wavelet filter applications detailed below and/or applications based on them. In an embodiment, the heart phase data are based, directly or indirectly, on reconstructed electrogram(s) obtained using wavelet filter methods detailed below. The heart phase data can be based on any of the teachings below that are presented as being obtained from the reconstructed electrograms based on the data based on wavelet filters. In some embodiments, the phase data are also based on atrial activation timing developed / obtained from data based on wavelet filters detailed below and/or the V artifact removal teachings detailed below. Also, with respect to “obtaining heart phase data,” this can include executing any one or more of the actions detailed herein with respect to wavelet filters / wavelet processing. All of this is presented in terms of textual economy. In an embodiment, by way of example, the action of developing the time-varying electrical potential map is based on data that was developed using the filter applications that correspond to those presented in greater detail further below that are based on wavelet processing. Briefly, embodiments include using time-varying electrical potential map data that is based on reconstructed electrograms obtained by executing and/or based on the wavelet filter applications detailed below. In an embodiment, the potential map is based, directly or indirectly, on reconstructed electrogram(s) obtained based on data based on wavelet filters detailed below. The potential map can be based on any of the teachings below that are presented as being based on the reconstructed electrograms based on the data based on wavelet filters. In some embodiments, the time-varying electrical potential map is also based on atrial activation timing developed / obtained based on data based on wavelet filters detailed below and/or the V artifact removal teachings detailed below. Also, with respect to “developing a time-varying electrical potential map,” this can include executing any one or more of the actions detailed herein with respect to wavelet filters / wavelet processing. All of this is presented in terms of textual economy. [00182] Note that the actions of developing can instead be actions of obtaining, depending on the actor of a given method. [00183] The method under discussion can further include the action of identifying repeating phase signatures for respective locations on the surface of the atrial cavity from the time-
Attorney Docket No.246-021PCT varying phase map that repeat in a statistically aberrant manner relative to other phase signatures at other respective locations. As detailed, embodiment, by implementing time averaging of the maximum phase gradients at the various locations over various temporal periods detailed herein, or any other that can have utilitarian value, signatures that have aberrant occurrences can be identified. [00184] In an embodiment, the electrical potential map has at least 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1250, 1500, 2000, 3000, 4000, 5000 or more or any value or range of values therebetween in 1 increment electrical potential spatial locations and at least respective 500, 750, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 6000, 7000, 8000, 9000, 10000, 12K, 13K, 14K 15K 16K 17K 18K 19K, 20K, 25K, 30K, 35K or 40K or any value or range of values therebetween in 1 increment temporal potential values for the respective potential spatial locations, the phase map has at least any of the just noted values for the potential map phase spatial locations and the values can be different (we use the reference for textual economy) and at least the aforementioned number for the potential temporal values temporal phase values (and again they may not be the same – we reference for textual economy) for respective phase locations. In some embodiments, the respective electrical potential locations of the specified number of electrical potential locations have respective phase locations of the at least specified number of phase locations, but again, the number need not be the same (but can be). In some embodiments, the values are within 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24 or 25% or any value or range of values therebetween in 1% increments of the lower value (and this can be the case for the temporal locations as well). [00185] In some embodiments, the actions of developing a time-varying electrical potential map, developing the time-varying phase map, and identifying the repeating phase signatures are executed within a period of no more than 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes. Indeed, in some embodiments, the time averaging phase data reaches convergence within 40, 35, 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4, 3 or 2 seconds of the beginning of the calculations of the maximum phase gradients. (Convergence can mean that the statistically non-aberrant numbers (e.g., the zero values) average out to a level number. [00186] In some embodiments, the action of developing a time-varying electrical potential map of a surface of a cavity of a beating heart is based at least in part on time-varying readings from electrodes located in the cavity, and the action of identifying the repeating phase signatures is executed within 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes or any value or range of values
Attorney Docket No.246-021PCT therebetween in 1 minute increments of the electrodes being removed from the chamber (and this covers the electrodes being in the chamber). [00187] In some embodiments, the action of developing a time-varying electrical potential map of a surface of a cavity of a beating heart is based at least in part on invasive readings taken while a human in which the beating heart resides is in an operating room / before he or she leaves the OR, and the action of identifying the repeating phase signatures is executed while he or she is in the OR / before the human leaves the operating room. [00188] In some embodiments, the method include executing a medical procedure targeted at tissue of the heart corresponding to at least some of the respective locations identified as having the repeating phase signatures that repeat in the statistically aberrant manner before the human leaves the operating room / within 20, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3 or 2 minutes of the catheter first entering the cavity. [00189] In an embodiment, the method includes, after executing the medical procedure, repeating any one or more of the method actions detailed herein, to “validate” or determine whether or not the procedure was effective in whole or in part. For example, after ablating the tissues that were identified as having the aberrant activation times or otherwise ablating tissue based on the locations identified as having aberrant activation times, additional electrode readings are taken, which readings can be taken along any of the lines detailed above or differently providing such is utilitarian, and then the potential map is re-created utilizing these new readings, and in the phase map is re-created utilizing these new phase maps, and then the statistical analysis is reexecuted or another type of analysis is executed on the phase map, or more accurately, on the maximum phase gradients for the various locations, and locations with aberrant time average values or locations without aberrant time average times are identified, and compared to the original set of such locations. This new data is used to validate the efficacy of the medical treatment. If there are remaining tissue locations with aberrant readings, a second treatment can be executed, and then a new set of test can be run, and so on. All of this can be done while the patient is in the OR / before he or she leaves the OR, and this can be done within any of the total time periods detailed herein for such procedures providing that the art enables such. [00190] Thus, there is an exemplary method of developing second time-varying electrical potential map of the surface of the cavity of the beating heart, developing second time-varying phase map of the surface of the cavity based on the second developed time-varying electrical
Attorney Docket No.246-021PCT potential map and evaluating whether and/or how many repeating phase signatures for respective locations on the surface of the atrial cavity from the second time-varying phase map that repeat in a statistically aberrant manner relative to other phase signatures at other respective locations, and based on the evaluation, evaluation whether the medical procedure was successful, etc. [00191] In an embodiment, there is a method, comprising developing data including at least X spatial locations and at least Y respective phase gradients for the respective spatial locations of the X spatial locations, analyzing the developed data (according to any of the regimes herein), which analysis can be executed by statistical analysis or any other analysis that will provide utilitarian value, and identifying locations of the respective locations that are indicative of tissue influencing atrial fibrillation based on the statistical analysis, wherein X is at least any of the spatial values herein and Y is at least any of the temporal values herein. In an embodiment, the analysis is time averaging. [00192] In an embodiment, the analysis can entail adding all the phase gradients at each location and looking for resulting values that are aberrant. Indeed, even without time averaging, if indeed values will add to zero for the non aberrant tissue, the numerator will likely be zero in some embodiments, so the averaging will be the averaging of zero. [00193] In some embodiments, the action of identifying locations includes identifying locations where the statistical analysis of the developed data indicates non-random activation of respective heart tissue cells at the identified locations. In some embodiments, the action of identifying locations includes identifying locations where averaging of the maximum phase gradients yields a statistically meaningful non-zero value. In some embodiments, the action of identifying locations includes identifying locations where averaging of the maximum phase gradients yields a value that is at least 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1, 1.1, 1.2, 1.3, 1.4, 1.5 or 1.6 or more standard deviations from the mean average for all of the average phase gradients of the locations (these can be the aberrant locations). In some embodiments, the action of identifying locations includes identifying locations where averaging of the maximum phase gradients yields a value that is no more than 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1, 1.1, 1.2, 1.3, 1.4, 1.5 or 1.6 or more standard deviations from the mean average for all of the average phase gradients of the locations (these can be the non-aberrant locations). And with respect to these below standard deviation values, this can encompass time averages where the time average is not zero out for the healthier normal tissue, for whatever reason, based on, for example, the physiology of a given individual.
Attorney Docket No.246-021PCT [00194] In some embodiments, the action of identifying locations includes identifying other locations where averaging of the maximum phase gradients of at least a majority of the Y phase gradients yields a statistically zero value. In some embodiments, the action of identifying locations includes identifying locations where averaging of the maximum phase gradients of at least a majority of the Y phase gradients yields a statistically meaningful non-zero value and the action of identifying the locations includes further statistically analyzing the values of the non-zero values. Here, this can utilize to further vet the non zero values. For example, there could be statistically significant nonzero values that are still indicative of healthy tissue or tissue that is not a substrate for afib, or at least not a meaningful influencer of afib. These might not be targets for ablation, at least not at an initial pass. For example, only some of the locations that have the statistically significant nonzero value might be ablated, such as those that are above average of all of the locations that are statistically significantly nonzero, and then the potential data measurements could be re-taken and then the phase analysis could be rerun to see the new activation patterns, and then a percentage or some threshold amount of that tissue (e.g., the above average for the new non-zero locations) could be ablated, and so on. The point here is that by further analyzing the statistically significant nonzero locations, additional choices can be made as to what tissue should be targeted or not targeted. [00195] In some embodiments, the statistical analysis of the developed data identifies statistically consistent patterns of electrical activity that repeat in a statistically meaningful manner over time. [00196] As noted herein, embodiments, include non-transitory computer readable medium having recorded thereon, a computer program for executing at least a portion of a method, the computer program including any one or more all the actions detailed herein providing that the art enable such. In an exemplary embodiment, the computer readable medium includes code for statistically analyzing first data based on phase gradients for at least X locations (where X can be any of those detailed herein) on a surface of a chamber of a human heart and code for identifying a plurality of locations from the at least X locations, based on the statistical analysis of the first data, that should be targeted for treatment. In an embodiment, the medium has code for transforming respective electrograms for respective locations of the at least X locations to a phase record including the phase gradients. There can be code for creating the electrograms from data based on electrical potentials obtained from electrodes within the human heart, the number of electrodes within the human heart being less than X, such as at least 10, 20, 30, 40, 50, 60, 70, 80, 85, 86, 87, 88, 89, 909, 91, 92, 93, 94, 95, 96, 97, 98 or 99% or more or any
Attorney Docket No.246-021PCT value or range of values in 1% increments less than X. The code for creating the electrograms uses inverse solution methods. In an embodiment, the code for statistically analyzing the first data time averages respective maximum phase gradients for the at least X locations. In an embodiment, the code for statistically analyzing the first data time averages respective maximum phase gradients for respective locations of the at least X locations and the code for identifying the plurality of locations from the at least 150 location identifies respective locations where time averages of the respective maximum phase gradients are statistically significantly non-zero. [00197] In an exemplary embodiment, the non-transitory computer readable medium is such that the computer program can also include code for implementing any one or of the wavelet- based filtering related actions herein, or more accurately, code for implementing any one or more of the actions associated with developing the reconstructed atrial electrogram based on wavelet filtering techniques and/or the associated atrial timing identification and/or the V artifact removal, in the interest of textual economy. [00198] In an embodiment, the at least X locations includes at least 2,000 locations, respective locations of the at least 2,000 locations have at least 1000 respective maximum phase gradients and the medium creates the electrograms and identifies the plurality of locations from the at least 2,000 locations for the at least 1000 respective maximum phase gradients within 20, 15, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes when run on a Dell ™ laptop with an Intel Core i9 Microprocessor with at least a 2.8 GHz clock frequency, at least 16 by 1024 KB L2 cache, at least 22.00 MB L3 cash, a TDP of at least 160 W, a DMI 3.0 I/O bus and a 4 x DDR4-2666 memory and in some embodiments, the code for creating the electrograms uses at least 500 measurements from each electrode per second. [00199] FIG. 10 shows a diagram of an exemplary system that can implement one or more of the method actions herein. A catheter is placed inside a volume of interest, typically a heart chamber. Catheters are electrically connected to an interface 13, which is electrically isolated and may comprise a proprietary system or a set of such systems. Instantaneous potentials and the 3D positions are acquired from individual electrodes on one or more cardiac catheters. For instance potentials and 3D positions may be recorded simultaneously from multi-electrode basket catheters positioned in the RA and LA, or from a multi-electrode basket catheter and an ablation catheter in the same cardiac chamber. 3D electrode positions are recorded using impedance techniques, magnetic sensors, ultrasound sensors or combinations of these methods.
Attorney Docket No.246-021PCT Electrocardiograms (ECGs) are also acquired without position information for standard lead configurations. [00200] The processing unit 14 controls the acquisition and processing of data so that recorded potentials or information derived from them can be mapped onto the endocardial surface of a heart chamber or chambers in a form that is useful to the operator. The processing unit is an electronic computing device with a real-time operating system and it may include a field programmable gate array (FPGA) and appropriate memory. It may be connected to the interface 13 either directly or wirelessly. [00201] Alternative forms of the teachings herein can comprise a computer program or computer readable medium adapted to perform the method. The processor unit or means may comprise a microprocessor, FPGA, logic circuit or other form of processor. The processor means may be incorporated into a computer, for example a PC, mainframe, remote server or cloud-cluster. The processor means may be a single device or may interface with further processing means. Connections may be made between the processing means through a network or wired/wireless communications. The processor or processing means may receive information from a data storage device such as a removable storage disk, hard-disk, ROM, RAM or a network connection. [00202] The first processing step is to construct a computer representation of the 3D endocardial surface geometry of the heart chamber or chambers of interest. This may be derived from i) cardiac MR images ii) contrast-enhanced cardiac CT images or iii) surface coordinates mapped under fluoroscopic guidance using a contact catheter. Alternately, geometry created in iii) can be merged with endocardial surfaces segmented from i) or ii). Static 3D models can be integrated with cine-fluoroscopic imaging or ultrasound imaging to provide estimates of heart wall motion. Provision for the import of such video data is indicated in 15. [00203] Processing steps that will be carried out during the acquisition of data from a catheter or catheters include those described in relation to FIGS. 3, 4 and 5 above. For each time interval, a forward solution will be used to estimate potentials inside the catheter followed by an inverse solution that enables potentials to mapped onto the endocardial surface of the cardiac chamber or chambers. This will be repeated at successive intervals throughout a recorded data set. Endocardial potentials will be rendered on a computer representation of the 3D surface of the heart chamber or chambers presented on a screen or display device 16 in a form that can be
Attorney Docket No.246-021PCT manipulated interactively by the operator. The location of catheter or catheters with respect to the heart wall will also be displayed. Electrograms at selected endocardial points (or at selected catheter electrodes) and selected ECG leads will be presented simultaneously in a moving window with an adjustable time base. It will also be possible to display a projection of unique source points onto the endocardial surface providing intuitive information on the spatial resolution of the inverse maps presented. The processing and display of time-varying endocardial potentials can be completed in real-time or near real-time. These may include the following: o spatial and temporal filtering o activation time analysis o regional variability analysis o regional frequency analysis o phase analysis The results of the analysis steps listed will be mapped onto a representation of the 3D heart displayed in 11. [00204] The processing steps may be implemented in hardware or software or a combination of these. The steps or instructions may be stored in computer readable media or memory including a hard disk, random access memory (RAM), read only memory (ROM) a removable storage disk or device or other storage media. [00205] There are methods of operating a catheter in a sequence of steps guided by the information displayed in 11 above. Initially a global picture of electrical activity on the endocardial surface of a heart chamber will be acquired and displayed. This can use a catheter positioned in a first position, generally centrally with electrodes in contact with or adjacent to as much of the endocardial surface of the heart chamber as possible. The catheter may be moved to one or more second positions to investigate regions of interest. As noted above, a catheter can be used for global mapping can also be used for regional mapping, such as by reducing the dimensions of the catheter to move the electrodes closer to each other to obtain more accurate readings, for example, by adjusting the size of the catheter to map in specific regions of the chamber with greater precision. [00206] In a method of mapping of electrical activity, the activity is obtained over a short period of time (for instance continuous periods of at least 3 to 50, or 5 to 40 seconds are required in
Attorney Docket No.246-021PCT AF) before a user decides which areas require further investigation. Higher resolution mappings will be obtained in these regions-of-interest by moving multi-electrode arrays with smaller diameters into them (again in AF continuous periods of at least 10-20 seconds are required for region-of-interest mapping). This method will support more efficient high- resolution endocardial mapping of electrical activity because it utilizes potentials recorded at all electrodes whether they are in contact with the endocardial surface of the heart chamber or not. The operator will also receive direct feedback the accuracy of endocardial maps through visual comparison of maps and electrograms displayed as the catheter is moved closer to the surface and as some electrodes make contact with it. [00207] The mapping approach above could be carried out using combinations of catheters with different dimensions. In some embodiments, a single adjustable catheter can be used. Here the dimensions of the electrode array are altered by withdrawing or advancing the splines into or out of the catheter. In some embodiments, the catheter can meet the following general specifications: o The catheter will be steerable. o It will be possible to lock the dimensions of the electrode array in multiple dimensions between fully open and fully closed states. o Electrodes will be uniformly spaced as far as is possible in open and closed states and distributed evenly across the mathematically closed virtual surface that bounds them. o Inter-electrode spacing will be sufficient to characterize electrical activity appropriately within endocardial regions on the order of 10 mm diameter. [00208] It can be possible to introduce the catheter into atrial appendages and pulmonary vein in a closed state. [00209] In summary combination of the processes detailed herein can provide a regime for mapping endocardial electrical activity on a heart chamber or chambers. This is achieved, in part, by mapping potentials recorded on a catheter containing multiple electrodes back to the endocardial surface when some or all of the electrodes are not in contact with it. This can involve the solution of an inverse problem that is inherently ill-posed. This can involve executing a solution of the forward problem inside the catheter, which provides additional information that improves the conditioning of the inverse problem enabling more robust solutions, independent of the method used. Inverse mapping can be used to reconstruct global
Attorney Docket No.246-021PCT endocardial potentials. However, local or region or interest (regional) mapping can be executed to fully reconstruct endocardial surface potentials in regions where endocardial geometry is complex and/or there is rapid spatial variation of endocardial potentials. Meshless maps can be used to reduce the scale of the inverse problem enabling acceptably rapid solution in the presence of relative motion of the electrodes and the heart wall. By combining appropriate systems for acquisition, processing and display of endocardial mapping data (see FIG. 10), with improved multi-electrode catheters, it is possible to provide more accurate, more rapid and more comprehensive results than with existing methods. [00210] From the foregoing it will be seen that a method of determining physiological information of an endocardial surface is provided which improves the calculation of physiological information, for example, electrical potentials. [00211] Embodiments include executing treatment methods based on the obtained data with respect to the locations identified as being aberrant or otherwise having delayed phase gradients were otherwise based on the results of method 700. In an exemplary embodiment, the method includes ablating some or all of the tissue at the locations having the non-zero gradients otherwise having the time averaged regional afib fingerprints. In some exemplary embodiments, a level of the time averaged maximum phase gradient is set, above which the treatment method is executed to the tissue associated there with. With reference to figure 23 and the normalized time average maximum phase gradient map shown in that figure, tissue corresponding to a maximum phase gradient of above .65 by way of example could be targeted for ablation. In an embodiment, this can be a hard number based on empirical results over a statistically significant number of patients, while in other embodiments, this can be based on the general overall impression of a given result for a given patient. With respect to figure 23, it can be utilitarian to only ablate the area in red or around the red area and not treat the green areas even though the green areas may be as extensive as the red areas, but the green areas are not as intense as the red areas with respect to the time averaged maximum phase gradient values and thus it may be sufficient to only “process” the tissue that shows up as red. Conversely, the extensive nature of the areas in green with respect to spatial location could be tissue that serves as a greater block to the electrical signals then the tissue in red. In any event, methods include evaluating the data obtained from method 700 or whatever method is derived from the teachings detailed herein to obtain the data related to heart tissue that has statistically significant aberrant activation times and, based on that data, treating a portion of the heart via a medical procedure, such as for example, ablation therapy.
Attorney Docket No.246-021PCT [00212] In an exemplary embodiment, the medical therapy is executed to competition within GHI minutes from the time that method 700 or any other related method relating to evaluating the time average maximum phase gradient or any other data that is utilitarian relating to identifying the tissue that is aberrant is completed, where GHI is 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85 or 90 or any value or range of values there between in 1 increment. In an exemplary embodiment, the medical therapy is executed to completion within GHI minutes from the time that method 700 or any other related method relating to evaluating the time average maximum phase gradient or any other data that is utilitarian relating to identifying the tissue that is aberrant is first begun, and this need not be the same as the aforementioned period. In an exemplary embodiment, the medical therapy is executed to completion within GHI minutes from the time that the catheter first enters the chamber, and this need not be the same as the aforementioned period. [00213] In an embodiment, method 700 or a truncated method thereof is executed to completion within JKL minutes from the time that method 700 is commenced, where JKL can be 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55 or 60 or any value or range of values there between in 1 increment. In an embodiment, method 700 or a truncated method thereof is executed to completion within JKL minutes from the time that the catheter first enters the heart chamber. And this time need not be the same as the prior period. And method 700 can be executed to cover one or more regions within the chamber, and can be implemented for any of the location values detailed herein at any of the cycle values for any of the time periods detailed herein. [00214] Any control unit and/or test unit or the like disclosed herein can be a personal computer programs was to execute one or more or all of the functionalities associated there with are the other functionalities disclosed herein. In an exemplary embodiment, any control unit and/or test unit or the like can be a dedicated circuit assembly configured so as to execute one or more or all of the functionalities associated there with or the other functionalities disclosed therein. In an exemplary embodiment, and the control unit and/or test unit or the like disclosed herein can be a processor or the like or otherwise can be a programmed processor. The control unit can be a signal processor or the like or a personal computer or the like or a mainframe computer or the like etc., that is configured to receive signals from the electrodes or data based on data from the electrodes and implement method 700 for example or a related action. More particularly, the control unit can be configured with software the like to analyze the signals /
Attorney Docket No.246-021PCT the data based on the signals in real time and/or in near real time while the catheter is in the chamber. [00215] An exemplary system includes an exemplary device / devices that can enable the teachings detailed herein, which in at least some embodiments can utilize automation. That is, an exemplary embodiment includes executing one or more or all of the methods detailed herein and variations thereof, at least in part, in an automated or semiautomated manner using any of the teachings herein. Conversely, embodiments include devices and/or systems and/or methods where automation is specifically prohibited, either by lack of enablement of an automated feature or the complete absence of such capability in the first instance. [00216] [00217] Embodiments address processing of the raw signals from the electrodes. In this regard, unipolar electrograms directly obtained from the electrodes in the heart include the superposition of far-field components and noise. This confounds identification and analysis of local atrial activation. Embodiments include utilitarian methods of extracting near-field atrial activity from unipolar EGMs in AF. Embodiments can include wavelet-based filtering to extract detailed information from unipolar atrial EGMs in AF. Such exemplary embodiments and others described herein can aid in the enablement of the identification of atrial regions that drive persistent and permanent AF from aberrant patterns of electrical behavior. Such embodiments can enable the obtention of robust measures of local electrical activity from the extracellular electrograms (EGMs) recorded during electro-anatomic mapping of patients in AF. [00218] Embodiments can include implementing the teachings herein based on results of a 16 channel Abbott Advisor grid array, although other embodiments detailed herein utilize an array with more channels (less can be used in other embodiments). Some embodiments include utilizing wavelet based tools so that the fractionated electrical activity associated with atrial fibrillation can be utilitarianly characterized. [00219] Normally, the electrical activation which triggers contraction sweeps across the chambers of the heart as a uniform wave rapidly propagated (~1 m /sec) wavefront. This results in corresponding changes in potential on the surface of the heart as the activation wavefront moves through it (relatively large short lived electrograms with a simple deflection complex). Fractionated electrograms have lower than normal magnitudes with multiple activation complexes that occur over much longer time intervals. Hence the utilization of the positive
Attorney Docket No.246-021PCT power teachings herein. This is explained by nonuniform electrical propagation with substantial regional activation delays. Fractionated electrical activity occurs adjacent to heart regions where normal heart muscle cells have "died" and been replaced by collagen (scarring or replacement fibrosis). [00220] Fractionated electrograms are also observed in longstanding (persistent or permanent) AF and two mechanisms are involved. Embodiments can rely on nonuniform and disorganized activation of the atrial chambers, where fractionation occurs even if atrial muscle cells are entirely normal. Embodiments include identifying regions of the heart as ablation targets in interventions to reverse longstanding AF. [00221] A problem in linking recordings of fractionated electrical activity with regions of fibrosis is that the relationship between them is not unique. Fractionated electrical activity can be recorded some distance from a fibrotic region and it is difficult to determine whether recorded fractionated electrical activity is due to activation of a large distant region or a smaller but closer region. High resolution surface mapping resulting from the teachings herein can be used in some embodiments to provide more precise resolution, providing that the teachings can be implemented to have robust signal processing for classifying of fractionation. [00222] Embodiments go beyond simple Fourier transforms which use periodic basis functions with infinite extent as such is ill-suited for analyzing fibrillatory activity which is non stationary and stochastic. In embodiments, wavelet decomposition that uses time-scaled versions of a basis function with limited extent (the "mother” wavelet) are used to perform continuous decompositions with time. This is utilitarian for analysis of AF. A further utilitarian result is that a first-derivative (1
0) Gaussian mother wavelet can replicate key features of a unipolar cardiac electrogram. Based on this, embodiments include performing a wavelet decomposition that represents electrical activity in AF as a combination of electrogram-like components with different magnitudes across different time scales. In fractionated activity, this enables the separation of the effects of magnitude and distance. Local activity can be identified based on the relatively high amplitude in short time scales whereas these components are filtered with more distant activity and most of the energy is contained in longer time scale wavelet components. [00223] In normal electrical activation, decomposition with 1
0 derivative Gaussian wavelets produces components across time with a single complex in which the peak corresponds to the maximum negative rate of change of potential. In contrast, fractionated activity is associated
Attorney Docket No.246-021PCT with multiple such complexes spread over a much longer time interval, and it is this aspect that some embodiments rely upon for identification of the fractionated activity. Wavelet decomposition as used herein can be utilized to provide a quantitative classification of fractionation on the basis of the number, amplitude, timing and/or time-scale distribution of separate activation components. Embodiments include combining this with high resolution electrical mapping, for instance using a 2D grid array so as to enable identification of structural substrate that could contribute to the maintenance of AF, and thus provide targets for ablation. [00224] Embodiments include utilizing high-resolution phase mapping during fractionated activity so as to enable more detailed analysis of regional mechanisms that contribute to the maintenance persistent AF than has previously been possible. Some embodiments herein include, in the presence of prolonged low amplitude fractionation, the establishment of methods that identify the representative spread of activation across specific regions that are not confounded by fractionated local activity. Embodiments include doing this using the normalized positive wavelet power in longer time-scale wavelet components. This is done to provide for identifying and weighting dominant components of local activation so that time- averaged phase analysis can be employed. [00225] More specifically, embodiments include obtaining summations of potentials generated by current flows associated with propagating wavefronts of depolarization and repolarization in the heart (e.g., cardiac EGMs). From these summations, in an exemplary embodiment, activation times are identified based on, for example, the fact that activation has a higher frequency content than repolarization. Non near-field (electrotonic) contributions are also identified based on the fact that such undergoes frequency-dependent attenuation with distance. Embodiments can enable the separation of near-field components and uncorrelated activity from more distant atrial regions, while also removing or otherwise accounting for far-field ventricular artifact that would otherwise obscure local atrial activity. [00226] Embodiments can exclude or otherwise avoid the utilization of multi-polar spatial difference recordings (bipolar, omnipolar, Laplacian), at least with respect to implanting the obtention of atrial activation times and/or a reconstruction of an atrial EGM. Embodiments can include utilizing such to validate at least some aspects of the teachings herein. [00227] Briefly, in an embodiment, the intra-atrial unipolar EGMs acquired from patients with persistent AF (PeAF) using multi-channel basket catheters and high resolution grid catheters can, in some embodiments, be processed with first signal processing to reduce, including
Attorney Docket No.246-021PCT minimize (which includes eliminate if possible) noise, and second with continuous first and second order Gaussian wavelet transforms where, in an embodiment, this minimizes / eliminates non near-field signal components. With respect to the transforms, this can be done to (1) estimate and subtract beat-to-beat ventricular (V) artifact (2) identify and classify local atrial activation, and 3) reconstruct near-field atrial EGMs using wavelet-based matched filters. More particularly, wavelet-based estimation of V artifact using QT ensemble averaging and current V amplitude can be implemented in some embodiments. Subtraction residues across the V QT interval can be substantially better than fixed template subtraction (in some embodiments, by way of example only, p<1.0x10
-5, n=50). Embodiments can include wavelet- based methods for identification and classification of local atrial activation that are well- constrained. These can provide more (including much more) robustness in the presence of noise than standard detection algorithms employing bipolar and Laplacian EGMs. Embodiments also include recovering realistic atrial unipolar EGM morphology using wavelet filters based on a priori knowledge of the temporal variation in frequency content for near- field extracellular potentials following activation. [00228] As mentioned above, embodiments include wavelet transformation techniques. Individual unipolar atrial EGMs f(t) can be decomposed using the continuous wavelet transformation (CWT). By way of example, these actions can be based on the following equation:
where ^^^
^^
^^^ are the wavelet coefficients, while ^^^^ and ^^^^ are time shift and time scale dilation
parameters, respectively. Ψ ∗( ^^^^ ) is the complex conjugate of the decomposition function Ψ ( ^^^^ ) also referred to as the mother wavelet. In this embodiment, wavelets will be calculated at 10 different time scales with first and second derivative Gaussian mother wavelets using the Fast Fourier transform (FFT). In an embodiment, the wavelets can be calculated at 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, or more (or less than), or any value or range of values therebetween in 1 increments different scales. Any number of scales that can have utilitarian value can be used in some embodiments providing that the art enables such. That said, embodiments include executing one or more of the actions / obtaining one or more results herein in real time or near real time, so there are limits on the scales in some embodiments, which can be in some instances, correlated to the processing power of the computer systems used to implement some of the teachings herein.
Attorney Docket No.246-021PCT [00229] Atrial EGMs can be reconstructed following "filtering" across time scales in CWTs calculated with second derivative Gaussian mother wavelets (sometimes, only), which can be, in some embodiments, evaluated as ^
^^^ = 1 10 +∞ 1
^ � ^^^ ^^^^ ^^^^ ^^^^ ^^^^ ^^^^− ^^^^ ^^^^ ^^^^
where ^^^^( ^^^^) is t reconstructed atrial EGM, ^^^^
^^^^( ^^^^, ^^^^
^^^^) are "filtered" wavelet coefficients, ^^^^
^^^^ represents the separate time scales that can be / are considered in some embodiments and the constant C can ensure or otherwise drive the implementation towards energy conservation. [00230] As noted above, in some embodiments, ventricular far-field artifact subtraction can be achieved. We now present an exemplary embodiment of wavelet-based subtraction of ventricular far-field artifact. For example, far-field artifact generated in individual atrial EGM channels by ventricular activation and repolarization can be estimated beat-to-beat and subtracted using the workflow summarized in Fig.25. FIG.25 shows an exemplary processing pipeline in which estimated far-field artifact in each QT window is subtracted for each wavelet scale during QT window and inverse wavelet transformation is then used to recover. [00231] Initially, by way of example, as shown in FIG.26A, we start with exemplary raw atrial EGMs for a channel of the basket catheter where ventricular (V) activation timing is identified from coronary sinus (CS) EGMs is indicated by the broken red lines superimposed on this figure. This is by example representative unprocessed atrial EGMs, where V activation is indicated by broken red lines. In this regard, a 10 second period from a representative unprocessed atrial EGM is presented in Fig.26A while the continuous wavelet decomposition over the first 3 seconds of this signal is shown in Fig.26B. That is, FIG.26B shows wavelet decomposition of initial segment of (26A) with a 2nd order Gaussian wavelet in 10 levels. Wavelet coefficients in a specific QT window (indicated in blue – this is a window encompassing a V activation) are enlarged in Fig. 26C. Put another way, FIG.26C shows a magnified wavelet decomposition in QT window. That is, the grey shading of FIG.26A is an example of the temporal length of one QT window. [00232] In an embodiment, for one or more or all of the channels of the catheter, a CWT is computed across time windows from -100 to +400 ms by way of example with respect to respective ventricular activation times (hereafter referred to as QT windows) using a 2nd order Gaussian mother wavelet at for example 10 time scales. In an embodiment, the base raw signal extends (in some embodiments contiguously) for less than, greater than, and/or equal to 20, 21,
Attorney Docket No.246-021PCT 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, or 900 ms or more, or any value or range of values therebetween in 1 ms increments. In an embodiment, the CWT can be computed across time windows from plus and/or minus less than and/or greater than and/or equal to 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 325, 350, 375, 400, 425, 450, 475, 500, 525, 550, 575, 600, 625, 650, 675, 700, 725, 750, 775, 800 ms, or more, or any value or range of values therebetween in 1 ms increments from the V activation timing (and the minus need not be the same absolute value as the positive, and the numbers need not be the same in each window, although in other embodiments, the numbers will be the same in each window). In an embodiment, the CWT can be computed continuously and/or for time windows that form a contiguous set of time windows for the given data from the channel, where in some embodiments, each window has a V activation. In an embodiment, there are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, or more, or any value or range of values therebetween in 1 increment time windows. [00233] In an embodiment, ventricular far-field artifact is estimated in each QT window from wavelet components time-averaged across repeated QT windows. Fig. 26D shows an exemplary result of such (right column). That is, far-field artifact in QT window estimated from ensemble averages of wavelet coefficients across successive QT windows are shown in the right column. The left column shows the data of the right column scaled to match current V amplitude. [00234] In an embodiment, because these windows are synchronized with V activation, averaging identifies mean wavelet coefficients associated with the V artifact and reduces atrial contributions in AF effectively to zero (which includes to zero in some embodiments). This can be because the atrial contributions are not correlated with V activation. Embodiments include accounting for beat-to-beat artifact variation during V activation by scaling ensemble- averaged wavelet components to match recorded V amplitude (the left column of FIG.26D). In an embodiment, the results of scaling are tapered such with, for example, Hanning windows HW
i(τ) centered on V activation time for example. In an embodiment, the windows are centered with a maximum value of 1.0 and widths from 15 to 150 msec for wavelet scales i = 1,10. In an embodiment, the windows have widths of less than, greater than, and/or equal to 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55,
Attorney Docket No.246-021PCT 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 275, 300, 325, 350, 375, 400, 425, 450 or 500ms, or more (or less potentially), or any value or range of values therebetween in 1 increment, and as can be seen, can be different for the different wavelet scales. The scaled components can be added to fixed ensemble-averaged wavelet components tapered by complementary windows 1- HW
i(τ) to estimate the wavelet decomposition of each V artifact. (This provides the data outside the window for the remainder of the time average. That is, the white / non-greyed areas are combined to create the Estimated V artifact in the QT windows. [00235] Next, in an embodiment, this wavelet decomposition of estimated far-field artifact can be subtracted from the CWT of the raw EGM and an inverse transform can be performed. As shown in Fig. 27, this processing approach extracts atrial activation complexes that overlap high amplitude ventricular artifact. Note that the start and finish of the QT interval could also be varied in a patient-specific manner if such is utilitarian. [00236] It is noted that in some embodiments, there are no set constraints, at least in general, to the implementations of the teachings herein with respect to the aforementioned scales and/or timings. In an exemplary embodiment, the system that implements the teachings herein can for example compute CWTs at five scales, for example, and if this is not sufficient, compute additional scales. The window widths could start with -75 ms to +250 ms relative to the V activation time, and if this is not sufficient, the windows can be expanded. This is all by way of example. The point is that in an embodiment, an adaptive arrangement can be implemented where more limited number crunching is initially executed, and if more is utilitarian or otherwise if more is needed, more is done in an adaptive manner. That said, after a sufficient number of implementations on different patients, a historical database can be developed to identify scales and timings that provide utilitarian value over a statistically significant number of patients. In this regard, in an exemplary embodiment, a set of data constraints can be utilized for different patients having common attributes where the constraints have shown to be utilitarian in past implementations. It is found that these constraints do not provide suitable data for a given patient, the constraints can be changed. Accordingly, computational efficiency can be implemented in most instances, and for the occasional occurrence where those predetermined constraints do not provide utilitarian value, the constraints can be expanded at the potential cost of computational efficiency. Put another way, for example, if 9 out of 10 implementations are as efficient as possible, the 10
th implementation could be relatively inefficient (because perhaps the calculations would have to be redone, perhaps completely from
Attorney Docket No.246-021PCT the beginning, at the greater scales and/or longer time frames) and that would be acceptable overall. [00237] Embodiments use averaging to identify the mean wavelet coefficients associated with the V artifact and to reduce atrial contributions to zero (at least sufficiently so). In an embodiment, as a result, the atrial contributions on a magnitude basis that remain is no more than 8, 7, 6, 5, 4, 3, 2, 1, 0.75, 0.5, 0.4, 0.3, 0.2 or 0.1 % or any value or range of values therebetween in 0.01% increments that which was the case prior and/or relative to the V artifact. As noted above, the window widths can be different in some embodiments, and the widths can be different for different scales, as seen above. In an embodiment, the widths increase linearly with scale. In an embodiment, the widths increase non-linearly with scale. In an embodiment, the width of a given scale can be higher than a preceding scale by less than, greater than and/or equal to 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 175, 200, 250, 300, 350, 400, 500, 600, 700, 800, 900, 1000, 1250, 1500, 1750 or 2000% or more or any value or range of values therebetween in 0.1% increments. Thus, for example, scale 1 and scale 2 could have the same widths, or have different widths (and it could be that scale 2 has a more narrow width, however unlikely). In an embodiment, scale 1 could have a width of 25 ms and scale 2 could have a width of 40 ms (60% greater) and scale 3 could have a width of 60 ms (140% greater than 1, and 50% greater than 2). The windows can have any of the above-noted timings. [00238] And note while Hann / Hanning (sometimes called Hamming) windows are used here, other types of windows / functions can be used. A Parzen window can be used in some embodiments. A Blackman window can be used in some embodiments. A Nuttal window can be used in some embodiments. A Blackman-Nuttall window can be used in some embodiments. A Welch window can be used in some embodiments. A Blackman-Harris window can be used in some embodiments. A sine window can be used in some embodiments. A Cosine-sum windows could be used. Gaussian and/or Rife-Vincent windows can be used in some embodiments. B-spline window can be used in some embodiments. Confined Gaussian windows and/or Approximate confined Gaussian windows can be used in some embodiments. A Tukey window can be used in some embodiments. A Generalized normal window can be used in some embodiments. A Planck-taper window can be used in some embodiments. A DPSS window can be used in some embodiments. A Slepian window can be used in some embodiments. In an embodiment, a function that provides / uses a smooth taper is used.
Attorney Docket No.246-021PCT [00239] A Kaiser window can be used in some embodiments. A Dolph-Chebyshev window can be used in some embodiments. An Ultraspherical window can be used in some embodiments. [00240] Embodiments also include using hybrid windows, such as for example in some embodiments, a Barlett-Hann window, a Planck-Bessel window, a generalized adaptive polynomial (GAP) window and/or Lanczos window could be used. [00241] In view of the above, any window functioning an/or apodization function and/or tapering function used herein can be based on one or more of the just noted window arrangements (by “based on,” it is meant that it does not have to be exactly that window, but can be based on that window, so one might take the results of such and then modify those results, such as by applying another window function thereto and/or by adjusting a portion thereof (such as the portions that are a certain percentage of time away from the V activation time on one or both sides of the function, etc.). [00242] In an embodiment, the product of a trained neural network, trained on a sufficient number of examples, can be used to select the window function and/or the parameters of the window function, as detailed herein (e.g., window length). [00243] In an embodiment, the complementary of any given window (in simple terms, the 1- the window function) can be used in a manner concomitant with the teachings above with the complementary windows 1- HW
i(τ)), and the addition of the results can be obtained to estimate the wavelet decompositions of each V artifact so as to, for example, provide the data outside the window for the remainer of the time average or otherwise to combine the results to create an estimated V artifact for the QT windows, where this result is subtracted from the CWT for the raw EGM and an inverse transform is executed in accordance with the teachings above so as to extract the atrial activation complexes that overlap high amplitude ventricular artifact. [00244] In an exemplary embodiment, a plurality of window functions can be applied, and the most utilitarian window function can be selected from the results, or more accurately, the results that are most utilitarian can be selected for use. Note also that in an exemplary embodiment, different window functions that are utilized in the results are compared for empirical purposes, and the most utilitarian function is selected for use when applied to patients for treatment or otherwise diagnostic purposes in the clinical scenarios. In an exemplary embodiment, again as noted above, machine learning can be used to select the “best” window function and/or the parameters thereof (length of time for example). In an exemplary embodiment, different window functions could be utilized for different types of people based
Attorney Docket No.246-021PCT on demographics. In an exemplary embodiment, depending on the resulting wavelet transforms, machine learning, or more accurately, the results of machine learning, could be utilized to identify the window functions that will be utilized. That is, for example, in the clinical condition, upon the wavelet decomposition of the different scales, the results of a trained neural network by way of example could be utilized to analyze the results of the decompositions and select the window function that will be applied accordingly. And then of course the system could apply that window function in accordance with the teachings detailed herein. And note that in an exemplary embodiment, if an applied window function is deemed to results in less than utilitarian results, another window function can be used. There could be a primacy of window functions that could be used where if the first window function which is deemed the “go to” window function in normal circumstances is not providing sufficient results or otherwise providing results that may be less than utilitarian, the next window on the list could be utilized, and so on. [00245] In view of the above, with respect to FIG. 28, there is an exemplary method, method 1030, which includes method action 1032, which includes the action of obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation. In an exemplary embodiment, this can entail the entire procedure of placing the lumen catheter in the human and snaking such to the human’s heart, and then utilizing the electrodes on the catheter to record potentials at one or the electrodes. In an embodiment, this can entail obtaining data from a third party that is executing the recordings. Indeed, in an exemplary embodiment, this could be done remotely, where a surgeon or healthcare professional performs the actual physical operation / procedure on the human, and the actor of method 1030 is located in another country or another continent, where the first data is obtained via the Internet by way of example only and not by way of limitation. It could be obtained as an attachment to an email, or even could be obtained utilizing facsimile or some other regime. Any device system and/or method that can enable method action 1032 can be utilized in at least some exemplary embodiments. Conversely, in an embodiment, in a variation of method 1030, jumping ahead a bit, with reference to FIG. 28A, there is a method 1031, which includes method action 1033, instead of method action 1032, instead entails obtaining first data based on one or more electrical phenomena in an organ and/or a portion of an organ in a living human. In an exemplary embodiment, the first data is based on one or more unipolar EGMs which represent the electrical phenomena. In an embodiment, the first data is based on one or more unipolar EGMs which correspond to the electrical phenomena, the one or more unipolar EGMs
Attorney Docket No.246-021PCT being recorded in an atrium of the living human. In an embodiment, the electrical phenomena are phenomena in an atrium of the living human. The phenomena can be in another part of the living human. In an embodiment, the identified regional activation times are regional atrial activation times. The organ and/or a portion of an organ is a heart and/or a portion of a heart atrium. In the interests of textual economy, in an embodiment, method action 1032 is replaced with method action 1033, and the method is executed accordingly. [00246] As noted above, embodiments include applying some of or all of the teachings detailed herein some of the teachings detailed herein to a heart that afflicted with atrial fibrillation. The heart could be a perfectly healthy heart. But in other embodiments, the to be a heart with other issues, such as by way of example only and not by way of limitation, a heart with an ascending aorta. The heart could be a heart with arteries that are clogged by a certain amount, such as by way of example only and not by way of limitation, by at least 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, or 60% or more rainy value range by therebetween in 1% increments. [00247] In an embodiment, the electrical phenomenon is disorganized electrical activity, concomitant with the teachings above. But in an embodiment, the electrical phenomenon is organized electrical activity. [00248] By data based on one or more unipolar EGMs, this can be data based on one or more unipolar EGMs, this can be data directly from the channels of the basket catheter, or can be data from data from the channels of the basket catheter, etc. With respect to the former, this could be the raw signals from one or more electrodes of the basket catheter, and with respect to the latter, this could be processed signals and/or a data set that is developed from those signals. Indeed, in an exemplary embodiment, where the EGM’s are obtained at an earlier temporal location, such as a day or two or even a week or more before method 1030 is executed, the first data can correspond to a dataset where the original EGM data has been weighted for example, or subjected to smoothing or manipulation where clear extraneous data has been removed from the data set. [00249] It is briefly noted that by EGM, it is meant a recording of electrical activity over time. This does not require a graph format, although the EGM can certainly be in a graph format. Any data format from which monitored changes in electrical potential over time can be extracted can correspond to EGMs. In an embodiment, the first data includes EGMs for less than, greater than, and/or equal to 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, or 300, or more, or any
Attorney Docket No.246-021PCT value or range of values therebetween in 1 increment channels of an electrode assembly monitoring electrical activity of the heart (e.g., a basket catheter). In an embodiment, one or more or all of the EGMs span a period of time less than, greater than, and/or equal to 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 350, 375, 400, 450, 500, 550, or 600, or more, or any value or range of values therebetween in 1 increment seconds, and the EGMs need not be the same period of time as the others. [00250] In an embodiment, two or more time intervals can be utilized for the same scale, and the timing that has more utilitarian value can be utilized for further calculations. This might result in additional computational times, but if the computational system utilized is sufficiently advanced, this could be de minimis or otherwise this can be an acceptable delay on the overall processing. In an exemplary embodiment, two or three or four or five or six or seven or eight or nine or 10 or more time intervals are developed and the best time interval is utilized for further processing. In an embodiment, an average of the results can be utilized. The results can also be weighted. [00251] Any EGM data that can enable the teachings herein can be used in some embodiments. Here, in method 1030, the EGM must be a unipolar EGM. [00252] Method 1030 and method 1031 include method action 1034, which includes the action of obtaining second data based on wavelet processing of the obtained first data as part of a process to develop the second data. In an embodiment, this can entail executing wavelet processing of the obtained first data as part of a process to develop second data. This can be done according to any of the implementations detailed herein associated with wavelet processing. The actor executing method action 1034 can be the actor implementing the wavelet processing. Alternatively, and/or in addition to this, method action 1034 can be executed by receiving the second data, where the wavelet processing was executed remote from the actor executing method 1030. In an exemplary embodiment, the second data can be obtained over the Internet or by email, etc. Still, embodiments contemplate executing wavelet processing under the control of the actor implementing method 1030. In an exemplary embodiment, and one site computer, such as a mainframe computer in a hospital or the like, where the procedure is being implemented, can execute method action 1034. That said, in an alternate embodiment, a laptop computer with the appropriate software can be utilized to execute method action 1034. Accordingly, embodiments include non-transitory computer readable mediums that have code
Attorney Docket No.246-021PCT thereon for implementing or otherwise executing one or more of the method actions detailed herein. More on this below. [00253] In this exemplary embodiment, concomitant with the teachings above associated with FIGs.25-27, the second data is data indicative of a ventricle far field artifact obtained first data. In an embodiment, the ventricular far-field artifact is a result of ventricular activation and repolarization. [00254] In an exemplary embodiment, such as where the wavelet processing is executed by the actor of method 1030, the action of executing wavelet processing includes decomposing an individual unipolar atrial EGM of the obtained first data using CWT. Thus, in an exemplary embodiment, the wavelet processing includes decomposition of an individual unipolar atrial EGM of the obtained first data using CWT (again, the actor of method 1030 need not do the decomposing). [00255] In an embodiment, the wavelet processing includes calculating wavelets at three (3) to fifteen (15) different time scales (e.g., 10 different time scales) with second derivative Gaussian mother wavelets. Any number that can have utilitarian value can be used, as noted above. In an embodiment, the processing includes computing a CWT across a plurality of time windows from respective first times before respective beat-to-beat ventricular activation times to respective second times after beat-to-beat ventricular activation times. The timeframes can be as detailed above, and need not be the same for each time window, again as noted above. Corollary to this is that in an embodiment, there is the method 1040 of FIG.29, which includes method action 1042, which includes obtaining data based on an estimate of beat-to-beat ventricular activation times. Method 1040 also includes method action 1044, which includes the action of executing method 1030 or method 1031 (or executing method 1030 with method action 1033 instead of method action 1032), including executing the wavelet processing in this exemplary embodiment. In this exemplary method action, the action of executing wavelet processing includes computing a CWT across a plurality of time windows from respective first times before respective beat-to-beat ventricular activation times to respective second times after beat-to-beat ventricular activation times. [00256] Note that when implementing method 1040, method action 1042 can be executed while executing method action 1044, or at least while executing portions of method action 1044. In this regard, the first data based on one or more unipolar EGM’s recorded in an atrium of a living human can be obtained while also obtaining the estimate of the ventricular activation
Attorney Docket No.246-021PCT times. In an embodiment, the ventricular activation times can be obtained afterwards, depending on who and/or where the various parts of the method are executed. The point is, unless otherwise noted, there is no temporal sequence associated with the presentation of the method actions detailed herein providing that the art enables such. When there is an order that is to be implemented, such will be noted herein. [00257] In an embodiment, the second times are between and inclusive of 1 to 10 times the first times. For example, if the time before activation is 100 ms, the time after can be 100, 101, 102, 103, 103.1, 104, 110, 120, 150, 200, 250, 300, 777, or 1000 ms, etc., all by way of example. That said, some of these times are less than utilitarian. In an embodiment, the second times are less than, greater than, and/or equal to 1, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2, 2.1, 2.2, 2.3, 2.4, 2.5, 2.6, 2.7, 2.8, 2.9, 3, 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7, 3.8, 3.9, 4, 4.1, 4.2, 4.3, 4.4, 4.5.4.6.4.7, 4.8, 4.9, 5, 5.25, 5.5, 5.75, 6, 6.5, 7, 7.5, 8.5, 9, 9.5, 10, 11, or 12 times, or any value or range of values therebetween in 0.01 increments between and inclusive the first times. (In an embodiment, if the first time is treated as a negative value, the second time can be the absolute value of the first time times whatever multiplier is desired.) [00258] In an embodiment, the wavelet processing includes developing pluralities of respective wavelet coefficients at different scales for respective periods of time extending from before respective ventricular activation to after ventricular activation. In an embodiment, the pluralities amount to any of those noted above. In an embodiment, this amounts to less than, greater than, and/or equal to 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, or 100, or more, or any value or range of values therebetween in 1 increment, where the different scales can be the same or different for respective periods of time. In an embodiment, the pluralities are for each ventricular activation within a timeframe of less than, greater than, and/or equal to 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 125, 150, 175, 200, 250, or 300 seconds, or more, or any value or range of values therebetween in 1 second increment. In an embodiment, there are coefficients for each time period associated with each activation within the time just noted. In an embodiment, it can be every other activation, or every three or every four, etc. In an embodiment, less than, greater than (limited by the math – we cannot have greater than 100%) and/or equal to 100, 99, 98, 97, 96, 95, 94, 93, 92, 91,
Attorney Docket No.246-021PCT 90, 85, 80, 75, 70, 65, 60, 55, 50, 45, 40, 35 or 30% or any value or range of values therebetween in 1% increments of the times associated with ventricular activations can have the coefficients. [00259] In an embodiment, the obtained second data is based on respective time-averaged wavelet coefficients for the respective different scales. This, by way of example, as explained above with respect to figure 26D. In an embodiment, the wavelet processing includes developing CWTs of the first data and the second data is based on time averaging of respective portions of the developed CWTs and implementing a window function on the time averaged respective portions. In an embodiment where the actor of method 1030 executes the processing, the action of executing wavelet processing includes developing CWTs of the first data and the method includes time averaging respective portions of the developed CWTs and implementing a window function on the time averaged respective portions. [00260] Note that while the time averaging and the developments of the pluralities of respective wavelet coefficients at different scales was just presented in terms of something that existed within the obtained second data, in an embodiment, the actor of method 1030 executes the wavelet processing, and as part of that wavelet processing, the actor executes the action of developing respective pluralities of wavelet coefficients at different scales as noted above. Further, in an exemplary embodiment, the method further includes as part of the process to develop second data, time averaging the respective pluralities of wavelet coefficients to develop respective time-averaged wavelet coefficients for the respective different scales, wherein the second data is based on the developed respective time-averaged wavelet coefficients. Accordingly, in the interests of completeness, figure 30 shows an exemplary method, method 1050, which includes method action 1052, which includes executing method 1030 or method 1031 (or executing method 1030 with method action 1033 instead of method action 1032), including executing the wavelet processing. The reader is referred to the statements in method 1050 which correspond to those just detailed, albeit presented in an algorithmic format. [00261] Again, consistent with the teachings above, in an embodiment, the time-averaging identifies mean wavelet coefficients associated with the ventricular artifact and reduces atrial contributions in atrial fibrillation to at least effectively zero, which includes zero. In an embodiment, the contributions are reduced by at least and/or equal to 20, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4.5, 4, 3.5, 3, 2.5, 2, 1.9., 1.8, 1.7, 1.6, 1.5, 1.4, 1.3, 1.2, 1.1, 1.0, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.09, 0.08, 0.07, 0.06, 0.05, 0.04, 0.03, 0.02, 0.01%,
Attorney Docket No.246-021PCT or less, or any value or range of values therebetween in 0.005% increments of the amount that existed before the reduction. [00262] In an embodiment, any one or more of the methods herein, such as method 1050, further includes as part of the process to develop second data: (1) scaling ensemble time-averaged wavelet components of the developed time-averaged wavelet coefficients for the respective different scales; (2) tapering the scaled components at least towards zero value outside an interval relative to the ventricular activation time for the respective different scales; (3) tapering fixed ensemble time-averaged wavelet components towards zero value inside an interval relative to the ventricular activation time for the respective different scales; and (4) adding the tapered scaled components to the tapered fixed components to obtain an estimate of a wavelet decomposition of each ventricular artifact. [00263] Accordingly, in an embodiment, the second data is based on an estimate of a wavelet decomposition of each ventricular artifact; the estimate is based on scaled ensemble time- averaged wavelet components of the respective time-averaged wavelet coefficients for the respective different scales; the scaled components are tapered at least towards zero value outside an interval relative to the ventricular activation time for the respective different scales; the fixed ensemble time-averaged wavelet components are tapered towards zero value inside an interval relative to the ventricular activation time for the respective different scales; and the tapered scaled components are added to the tapered fixed components, resulting in an estimate of a wavelet decomposition of each ventricular artifact. [00264] In embodiments, as noted above, the scaling scales to 1 (the maximum amplitude is 1, and all other amplitudes are scaled based thereon). That said, embodiments can utilize different scales providing that such has utilitarian value. In an embodiment, the scaling is executed to match recorded V amplitude. While tapering with Hanning windows centered on the activation time was presented in the embodiments above, other window functions that can enable the teachings detailed herein that have utilitarian value can be utilized. As noted above, the mathematical function of the window function is zero valued outside of the above-noted time intervals, which are different for different scales in the embodiment noted above, with increasing time interval for the coarser scales, and is symmetric about the V activation time. In an embodiment, the time intervals need not be different for every scale and/or need not increase or otherwise change according to a linear trajectory with increasing scale. In an embodiment, two or more time intervals can be utilized for the same scale, and the timing that has more utilitarian value can be utilized for further calculations. This might result in additional
Attorney Docket No.246-021PCT computational times, but if the computational system utilized is sufficiently advanced, this could be de minimis or otherwise this can be an acceptable delay on the overall processing. In an exemplary embodiment, two or three or four or five or six or seven or eight or nine or 10 or more time intervals are developed and the best time interval is utilized for further processing. In an embodiment, an average of the results can be utilized. The results can also be weighted. [00265] This is also the case with respect to the fixed ensemble time-averaged wavelet components which are tapered towards zero value inside an interval relative to the ventricular activation time. This interval can be the same as the interval utilized for the tapering towards zero outside the interval. [00266] As noted above, in an exemplary embodiment, the fixed ensemble averaged wavelet components tapered by complementary windows 1- HW
i(τ), and the tapered scaled components are added thereto. [00267] In an exemplary embodiment, the goal of the scaling and tapering is to estimate the wavelet decompositions of each V artifact. The amplitude scaled data is utilized within the window, and outside the window the fixed ensemble averaged wavelet components are utilized. The tapering can provide seamless transition from inside the window to outside the window. [00268] Any window function that can enable the teachings herein can be used in some embodiments. Some embodiments do not use window functions per se. Other types of data manipulation can be used in some embodiments. Indeed, as noted below, a trained neural network can be used. [00269] While the above is presented in terms of the obtained second data being previously processed, in an embodiment, where the actor of method 1030 executes this processing, there is a method that includes any one or more of the actions herein, that further includes as part of the process to develop second data (1) scaling ensemble time-averaged wavelet components of the developed time-averaged wavelet coefficients for the respective different scales; (2) tapering the scaled components at least towards zero value outside an interval relative to the ventricular activation time for the respective different scales; (3) tapering fixed ensemble time- averaged wavelet components towards zero value inside an interval relative to the ventricular activation time for the respective different scales; and (4) adding the tapered scaled components to the tapered fixed components to obtain an estimate of a wavelet decomposition of each ventricular artifact.
Attorney Docket No.246-021PCT [00270] In an embodiment, the wavelet processing includes developing respective pluralities of wavelet coefficients at different scales for respective periods of time extending from before respective ventricular activation to after ventricular activation. On the scales, the above noted features / variations can apply. Further, the wavelet processing can include developing wavelet decompositions of estimated far field artifacts at the different scales for the respective periods of time. Here, in an embodiment, the second data is also based on third data developed by: (1) subtracting the wavelet decompositions from the respective pluralities of wavelet coefficients for the different scales to develop third data and (2) manipulating the third data to extract atrial activation complexes from the first data that overlap high amplitude ventricular artifact. Accordingly, in an embodiment, the second data is the data of FIG. 27 (by way of example only). [00271] Concomitant with the actor of method 1030 executing the processing, in an embodiment there is a method that includes one or more of the method actions herein, further comprising: (1) executing the wavelet processing; (2) developing respective pluralities of wavelet coefficients at different scales for respective periods of time extending from before respective ventricular activation to after ventricular activation; (3) developing wavelet decompositions of estimated far field artifacts at the different scales for the respective periods of time; (4) the method further includes developing third data by subtracting the wavelet decompositions from the respective pluralities of wavelet coefficients for the different scales; and (5) the method further includes manipulating the third data to extract atrial activation complexes from the first data that overlap high amplitude ventricular artifact. [00272] Still with respect to the above-noted process flow of FIG. 25, in an embodiment, the second data is further based on a subtraction of data resulting from the wavelet processing of the obtained first data from CWTs of the first data and performing an inverse transform on the result to remove ventricular far-field artifacts from the first data. And where the actor of method 1030 is executing the computations, in an embodiment, there is the method 1060 presented in FIG. 31, which includes method action 1062, which includes the action of executing method 1030 or method 1031 (or executing method 1030 with method action 1033 instead of method action 1032), including executing the wavelet processing. Method 1060 includes method action 1064, which includes the action of subtracting the second data from CWTs of the first data and performing an inverse transform on the result to remove ventricular far-field artifacts from the first data. Thus, the second data is different from that disclosed in
Attorney Docket No.246-021PCT the prior paragraph. Effectively, third data is obtained from method action 1064. This is the same as the second data noted immediately above. [00273] To put things more simply, in an embodiment, there is the method of 1070, as represented by the flow chart of FIG.32, which includes method action 1062 and method action 1074, where action 1074 includes subtracting results of tapering from CWTs developed in the wavelet processing to remove ventricular far-field artifacts from the first data. [00274] In an embodiment, to investigate the accuracy of the above detailed methods, one can remove V artifact from two adjacent unipolar channels recorded with an HD grid catheter and construct bipolar EGMs from the processed unipolar signals and compare such with the corresponding bipolar output from the mapping system. In addition, atrial recordings from which V artifact had been removed can be treated as “ground-truth” signals and combined with time-shifted versions of the estimated V artifact. The performance of subsequent wavelet-based artifact removal and subtraction of both fixed mean and median QT templates can then be evaluated with respect to the "ground truth" data. [00275] Embodiments include utilizing one or more of the above teachings to identify / determine atrial activation. Embodiments can include starting with the data developed using the atrial activation complexes developed according to the teachings above (e.g., data corresponding to FIG. 27 for example). Embodiments can include executing one or more or all of the method actions detailed above to obtain data analogous to / that of FIG.27 (it need not be in graphical form). Embodiments can include developing first derivative Gaussian wavelets to identify atrial activation. In an embodiment, wavelets having a function that has a form that is effectively similar enough to the unipolar atrial activation complex obtained above can be used in some embodiments. [00276] Embodiments can be directed to detecting periods when power due to depolarization (positive deflections in 1st derivative Gaussian wavelet decomposition for example) is maintained at a level greater (significantly in some embodiments) than background. In an embodiment, the power that is maintained is at least 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 125, 150, 175, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750 or 800 or 850% or more or any value or range of values therebetween in 1% increments of the background, by way of example. [00277] Within this window, maxima in this "positive" power can be used to identify the timing and reflect the magnitude of local depolarization. This can enable robust classification of
Attorney Docket No.246-021PCT normal and fractionated activation complexes. Embodiments can include setting a wavelet scale (or scales) that best or at least utilitarianly matches local atrial electrical activity and minimizes confounding effects of noise. Embodiment can include identifying the scale(s) based on inspection visually of the resulting graphs, or based on a metric of predetermined requirements or thresholds, or utilizing the results of a trained neural network by way of example only and not by way of limitation, where, with respect to this latter arrangement, a number of wavelet scales for a number of patients are selected by a trained professional, and the neural network is trained accordingly, and the results thereof are utilized to pick the scales going forward when the teachings herein are utilized in a clinical setting. [00278] Some exemplary details of an exemplary method that can enable the identification of atrial activation will now be described by way of example. [00279] FIG.33A shows an exemplary atrial activation complex for a person with PeAF (Atrial Fibrillation). That is, for the purposes of discussion, this is representative data of atrial EGMs (with V artifact subtracted) from a patient with PeAF. Two-timeframes a and b are highlighted, which frames overlap. In an embodiment, a CWT can be computed at 5 or 7 or 10 scales using a 1st derivative Gaussian mother wavelet. FIG. 33B shows five scales out of the 10 scales computed. (Again, more or less can be computed.) In an embodiment, instantaneous measure of the corresponding power at each wavelet scale is determined / estimated. In this exemplary embodiment, the wavelet decomposition is rectified to extract positive deflections (due to depolarization) and wavelet coefficients are squared to obtain the instantaneous measure of corresponding power by way of exemplary action, by way of example. In an embodiment, other functions can be used, such as cubing for example. Here, we have used power because it has a physical meaning, but from the point of view of mathematics, the transformation that is most effective can be used. [00280] Embodiments can include detecting local activation as an elevation, such as by way of example with the qualifier a sustained elevation, in "positive" power across selected scales. In an embodiment, this can be qualified that it must be with respect to a threshold (which can be an adaptive threshold) greater, including substantially greater than estimated background noise and non near-field artifact (for example, and not by way of limitation, 10x median in a 500 msec running lag window delayed by 20 msec – in some embodiments, greater than and/or equal to 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 175, 200, 225, 250, 275 or 300 ms or more or any value or range of values therebetween in 1 ms increments of the window length can be used, and in some embodiments, the window delay is less than, greater
Attorney Docket No.246-021PCT than and/or equal to 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 45, 50, 55 or 60 ms or more or any value or range of values therebetween in 1ms increments can be used – less than, greater than and/or equal to 7x, 8x, 9x, 10x, 11x, 12x, 13x, 14x, 15x, 16x, 17x, 18x, 19x, 20x, 21x, 22x, 23x, 24x, 25x, 26x, 27x, 28x, 29x, 30x, 35x or 40x or more or any vale or range of values therebetween in 1x increments can be used). In an embodiment, a timing of activation events is estimated by identifying maxima within this window across scales while quantifying a duration as the time for which this threshold is exceeded. More specifically, referring to FIGs. 33B and 33C, presented are time-series records of wavelet decomposition with 1
st derivative Gaussian wavelets with corresponding positive power at selected scales for normal (a) and low amplitude fractionated EGM segments (b), respectively. With regard to the latter, this is an example of a signal resulting from a temporal period where the fractionated activity associated with atrial fibrillation is occurring. Briefly, as can be seen from figure 33A, this does not occur in every segment that follows atrial activation. In many instances, there are atrial activations where this fragmented activity does not exist. The skilled person in the art will be able to identify the fractionated activity from the extracted atrial activation complexes often by inspection. More on this in a moment, but briefly, the use of "positive" power to identify activation events has proved sufficiently robust for a single threshold to be used for a particular patient across all scales in all channels. In an embodiment, this could be set by the operator / technician or whoever is running the testing / data processing. In an embodiment, there is utilitarian value in basing this on signal statistics across a substantial running lag window over sufficient time to enable noise and non near-field artifact to be segregated. [00281] In an embodiment, the temporal period of the wavelet decompositions and/or the positive power evaluation is computed across time windows from -125 to +125 ms by way of example with respect to the maxima of the EGMs. In an embodiment, the base raw signal extends (in some embodiments contiguously) for less than, greater than, and/or equal to 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, or 900 ms, or more, or any value or range of values therebetween in 1 ms increments for respective windows, and the time need not be the same for each window. In an embodiment, the CWT can be computed across time windows from plus and/or minus less than and/or greater than and/or equal to 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 325, 350, 375, 400, 425, 450, 475, 500, 525, 550, 575, 600, 625, 650, 675, 700, 725, 750, 775,
Attorney Docket No.246-021PCT 800 ms, or more, or any value or range of values therebetween in 1 ms increments from the maxima in the EMG with V artifact removed (and the minus need not be the same absolute value as the positive, and the numbers need not be the same in each window, although in other embodiments, the numbers will be the same in each window), and this can be different for different windows. In an embodiment, the CWT can be computed continuously and/or for time windows that form a contiguous set of time windows for the given data from the channel, where in some embodiments, each window has a maxima. In an embodiment, there are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, or more, or any value or range of values therebetween in 1 increment time windows. [00282] Note that the shaded area “b” of FIG.33A does not have the pronounced maxima as in the other areas, due to the fractionated activity. An embodiment can include centering the “b” window on an area where there “should” be a maxima based on the preceding maxima (and/or the following maxima). For example, if the normalized mode between maxima occurs within X ms of each other, half of that from the last maxima would be where the window is centered. The data can be evaluated to remove the fractionated areas / the areas without maxima, and then a mean and/or a median average time between the remaining maxima can be obtained, and then the “b” timeframe can be set to half the average from the last maxima. That said, inspection by a trained technician and/or a healthcare professional can identify such. One long continuous window could be used. [00283] Wavelet decomposition and power segments are presented horizontally for same scales, with scale becoming more course from top to bottom. The data is also normalized. Maxima in both decompositions and power segments identify contributions of different wavelet scales to local depolarization. In some embodiments, positive power is relied upon, as it can provide robust detection with fractionated EGMs in the presence of noise. [00284] Briefly, utilitarian value can be obtained using the wavelet approach in that in some embodiments, this approach increases the meaningful parameters that can be adjusted. In an embodiment, the increase is less than, greater than and/or equal to 50, 75, 100, 125, 150, 200, 250, 300, 350, 400, 450, 500, 600, 700, 800, 900 or 1000% or more or any value or range of values therebetween in 1% increments relative to that which would otherwise be the case. The techniques detailed herein can provide for the indication of the activation times or otherwise the onset of the fractionated activity, which is not apparent from the extracted complexes after the V artifact is removed. Put another way, in some embodiments, because the skilled person
Attorney Docket No.246-021PCT will be able to evaluate the extracted complexes, the person can “zoom in” on the pertinent timeframes, such as timeframe segment b. Alternatively, and/or in addition to this, based on the wavelet decompositions and/or the positive power data, the skilled person will be able to identify the fractionated activity by inspection, indeed, in some instances, more easily than that which would be the case from the extracted complexes. For example, the wavelet decompositions of figures 33B and 33C are clearly different from each other in most all the scales, and certainly sufficiently different in all the scales to identify which time periods have the fractionated activity. In an exemplary embodiment, an algorithm can be implemented that identifies the time frames of the fractionated activity. By way of example only and not by way of limitation, if there are a relatively higher number of maximas and minimas within a certain period of time, this could be an indication of fractionated activity relative to other activity. With respect to the wavelet decompositions, if there are minimas that have an absolute value in the neighborhood of that of the maximas, that could also be an indication of fractionated activity. By way of example only and not by way of limitation, if the wavelet decomposition provides for minimas that have absolute values that are within 5, 10, 15, 20, 25, 30, 35 or 40% or any value or range of values therebetween in 1% increments of the value of the maximas, such as the immediate minima following a maxima, that can be an indication of the fractionated activity. This can also be a basis from which to choose a scale for activation time evaluation as will be explained below. [00285] That is, embodiments include choosing a temporal resolution with which local activation complexes are identified. As is seen, the resolution is reduced as wavelet scale is increased. Embodiments include balancing resolution with clarity. By way of exemplary scenario, in this case, scale 5 is deemed to be the resolution that can provide utilitarian value for detection. Note that in an embodiment, scale 7 could have been used with respect to the electrogram during timing b, and 3 could have been used with respect to the electrogram during timing a. But when comparing timing b and timing a, it can be seen that scale 5 has maxima in both timings that can be identified and otherwise easily utilized. Scale 3 is somewhat choppy with respect to timeframe b, and this choppiness can skew the actual activation time with respect to the maximum. Corollary to this is that there would be between four and five maxima depending on the threshold that is utilized (more on the threshold in a moment). Conversely, with scale 5, there are only three maxima at play, and all three fall within a range that exceeds a threshold in a distinct matter.
Attorney Docket No.246-021PCT [00286] Embodiments can include identifying the times of fractionated activity from the extracted atrial activation complexes and/or from the wavelet decomposition of the complexes and/or from the positive power evaluation thereof. Indeed, each can be utilized to supplement the other. Any of the techniques noted herein or any others that can enable the teachings herein can be utilized in at least some exemplary embodiments. Embodiments include the product of a machine learning algorithm that analyzes any one or more or all of the just detailed data sets to identify the fractionated activity. [00287] Still further, embodiments include selecting the wavelet scale that provides the most utilitarian value or otherwise that provide sufficient utilitarian value to implement the teachings detailed herein. In this regard, embodiments include obtaining the wavelet transformations for various scales, and evaluating the data of the individual scales to determine which of the scales will be utilized for atrial activation timing. As noted above, in this exemplary scenario, scale 5 is deemed to be the most utilitarian scale, or at least usable to implement the teachings herein. [00288] Embodiments include trial and error implementation where over time, as the processes are implemented, the scale to be selected becomes more evident based on the end results. In an embodiment, again, a machine learning algorithm can be utilized to develop a fixed algorithm that can be utilized to select the wavelet scale to be utilized. In an embodiment, the end results of the selected scales can be evaluated by the machine learning algorithm to develop a product from machine learning that can be placed on a computer chip or otherwise implemented in a computer system by way of example, to pick the wavelet scale. On the opposite side of the spectrum, a trained technician or otherwise a trained healthcare provider implementing the teachings detailed herein can evaluate the results of the wavelet decompositions and the results of the positive power analysis based on his or her experience to pick the scale that has utilitarian value. [00289] In an exemplary embodiment, the scale that has maxima that on average are closest to a threshold (e.g., as represented by the horizontal dashed line in figures 33B and 33C) can be utilized, whether that be for the positive power calculations or for the wavelet decomposition calculations. The threshold can be predetermined. The threshold can be adaptive. The scale that has a mode of maxima that are above the threshold but within a certain percentage of each other (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, or 30%, or any value or range of values therebetween in 0.1% increments of the amplitude of the highest or lowest of the group for example.) In an exemplary embodiment, the scale that has maxima that are relatively consistent with respect to “distance” from a threshold can be utilized, or otherwise the scales
Attorney Docket No.246-021PCT that have maxima that are more consistent than others can be utilized. In an exemplary embodiment, there will be a single large maxima, such as seen in all of the wavelet scales in figure 33C, and there will also be one or more smaller maximas. Embodiments can include selecting the scales that have the smaller maximas that are within a certain percentage of each other with respect to amplitude, such as by way of example only and not by way of limitation, if the positive power analysis includes “secondary” maximas that are within 5, 10, 15, 20, 25, 30, 35 or 40% or any value or range of values therebetween in 1% increments of each other (using the smallest amplitude or largest amplitude), that can be the scale to use. In an embodiment, the maximas used must be over a threshold. In an embodiment, the threshold is 5, 10, 15, 20, 25, 30, 35 or 40% or any value or range of values therebetween in 1% increments of the main maxima. As noted above, there must be maximas that are within a certain distance of the scale. [00290] In an embodiment, the scale selected has maximas above the threshold where the signal goes below the threshold before reaching another maxima. This would rule out scale 3 for example and scale 1 for example. In an embodiment, the selected scale has maximas where the signal goes to zero before another maxima. This would rule out scales 1 and 3. In an embodiment, the scale used has secondary maximas that are similar to each other within a certain time. [00291] Embodiments can include selecting the scale where there are a certain number of maximas that are within 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, or 200 ms, or any value or range of values therebetween in 1 ms increments of each other. For example, in a given range, there are at least and/or equal to and/or no more than 2, 3, 4, 5 or 6 or any value or range of values therebetween. In an embodiment, there is a temporal separation of a certain amount between them, such as for example, at least 10, 25, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, or 150 ms, or any value or range of values therebetween in 1 ms increments. In an embodiment, there are no more than and/or equal to than 2, 3, 4, 5 or 6 or any value or range of values in that range. [00292] Embodiments combine the various “rules.” By way of example, if the positive power analysis includes “secondary” maximas that are within 5, 10, 15, 20, 25, 30, 35, or 40%, or any value or range of values therebetween in 1% increments of each other (using the smallest amplitude or largest amplitude), and within 20, 25, 30, 35, 40, 45, 50, 55, 60, 70, 80, 90, 100, 110, 120, 130, 140 or 150 ms or any value or range of values therebetween in 1 ms increments, that could be the scale to use. Depending on the numbers, this would rule out all but scale 5.
Attorney Docket No.246-021PCT [00293] An embodiment can include choosing the scales that have the fewest maxima above the threshold as long as there are at least 2, 3, 4, 5, 6, or 7, or any value or range of values therebetween in 1 increment. [00294] In an embodiment, the scale selected will have a certain number of secondary maxima before the primary maxima, in some embodiments, at least that exceed the threshold, and a certain number of secondary maxima after the primary maxima, and in some embodiments, that exceed the threshold. In an embodiment, there are at least 1, 2, 3, 4 or 5 or any value or range of values therebetween in 1 increment before and/or after the primary maxima (and the two need not be the same), where in some embodiments, all are above the threshold and in some embodiments there is no threshold applied for this determination. [00295] Embodiments include choosing a scale based on any one or more features present in scale 5 and/or in scale 7 and/or in scale 3 and/or that are not present in scales 1 and/or 3 and/or 7 and/or 9. [00296] But again, embodiments will typically include the skilled technician or healthcare professional choosing a scale based on the graphical data and/or the numerical data associated with given scales, and embodiments include keeping a log of such choices so that neural network can be trained based on those choices, where the results of such trained neural network can be utilized to choose the given scales in the future. [00297] But note also that depending on the computational times in play, more than one scale can be utilized in the results of the specific scales can be compared to one another, and the results that appear more utilitarian can then be chosen. That is, the choice of scales need not be based on the data that is specific to the positive power and/or the wavelet decomposition, but can be chosen based on how those scales are utilized downstream. [00298] Also, in some embodiments, the scale(s) will be predetermined based on empirical evaluation of exemplary results. That is, in an embodiment, it could be that a given scale is the scale that is always used (e.g., it could be that scale 5 is the one that is used always in the clinical setting). In an embodiment, the scale used can be demographically linked (certain types of people will use one scale, and certain types will use another scale). In some embodiments, forecasting, statistical or AI or otherwise, can be used to select the given scale. Still, as noted, a predetermined threshold and/or set of requirements can be applied to select the given scale for use, and these can be demographically based as well (different requirements
Attorney Docket No.246-021PCT / thresholds for different types of people, or at least for different types of starting data for example). [00299] In an embodiment, the threshold is determined by taking a value that is less than, greater than and/or equal to 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65 or 70% or more or any value or range of values therebetween in 1% increments of a value extracted from the positive power information. For example by way of example, taking the largest amplitude from the positive power (for fractionated and/or unfractionated activity – the same threshold can be used) and setting the threshold at a value that is less than, greater than and/or equal to 13% (or any of the appropriate percentages just noted. In an exemplary embodiment, the threshold is set at one or more of the aforementioned percentage values of the average amplitude of the positive power. In an exemplary embodiment, this averaging can exclude the zero values. In an exemplary embodiment, this can be the mean, median and/or mode average. In an exemplary embodiment, the average can include the zero values (hence why the “larger” percentages might be used, to accommodate the expansive zero values). And because the length of the window can vary, the averages can change depending on the length of the window, at least where the zero values are taking into account. Accordingly, a lower percentage would be applicable for a window that has a more narrow length, because there would be fewer zero values as compared to a window that has a longer length, where there would be more zero values, and hence a higher percentage. [00300] In an embodiment, the threshold can be based on empirical analysis. In an embodiment, a significant number of results can be evaluated, and based on the evaluation, a threshold can be set, which threshold will be utilized in a clinical setting. And again, thresholds can be demographically based, where some threshold will be used for some people and others will be used for other people. The threshold can be data based, where certain threshold will be used for different types of data or data that includes different values. [00301] In an embodiment, the threshold can be the one that excludes for example at least and/or equal to and/or no more than 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85 or 90% or more or any value or range of values therebetween in 1% increments of the local maxima (amplitude) in the positive power. In this regard, with respect to scale five, as can be seen, the threshold excludes at least five maxima out of the eight total maxima as seen. In this regard, separately or in combination with this arrangement, the threshold can be set at the average of all maxima, whether such is the mean, median and/or mode. In an embodiment, this can be weighted, such as, for example, 33% greater than the average of all maxima. In an exemplary
Attorney Docket No.246-021PCT embodiment, the threshold can be set so that the threshold excludes maxima that are separated by a certain amount from each other, and only include maxima that are close to each other, such as by way of example only and not by way limitation, the three maxima having the circle shown in the wavelet scale five. [00302] Note also that the features associated herein with selecting the threshold can also be utilized in a variation thereof to select the wavelet scale. For example, if the average maxima (amplitude), mean/median and/or mode is X over the timescale, the one that has less than, greater than and/or equal to for example 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25 or more or any value or range of values therebetween in 1 increment maxima over that average are excluded or included (depending on the number, and it might be “only” 3 or only 4 or only 2 or only 1 above that threshold for example). In an embodiment, the average, or more accurately the threshold, can be weighted (say 30% above or 40% below the average). In an embodiment, the threshold is less than, greater than and/or equal to 10, 20, 30, 40, 50, 60, 70, 80 or 90% or more or any value or range of values less than or is less than, greater than and/or equal to 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 125, 150, 175, 200, 225, 250, 275, 300, 325 or 350% or greater than the raw average or the raw value used, or whatever data is used for the threshold (e.g., 80% less than the greatest maxima is the threshold, and the scale that has 3 maxima over that threshold is the scale used). Again, in some embodiments, zero values can be included in the averages, and in some the zero values are excluded. Any of the teachings related to setting a threshold detailed herein for one data can be applicable for setting a threshold for another data, or at least based thereon, providing that the art enables such, if such results in utilitarian value. [00303] Note that in an embodiment, the drawings showing the graphs with the timelines are to scale, and can be used for at least plus or minus 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1% or any value or range of values therebetween in 0.1% increments accuracy (e.g., the 250mSec scale can be plus 11.3% of that or minus 8.4% of that, etc. [00304] And note that while positive power is used herein in some embodiments, raw power or other data based on massaged scale data can be used in some embodiments, providing that utilitarian results can be extracted from the operation. [00305] Any device, system, and/or method that can enable the choice of scales to be determined so that the teachings detailed herein can be implemented can be utilized in at least some exemplary embodiments providing that the art enables such unless otherwise noted.
Attorney Docket No.246-021PCT [00306] Returning back to what to do once the scale is selected, having selected scale 5, again, in an embodiment, a timing of activation events is estimated by identifying maxima within the window across scales while quantifying a duration as the time for which a threshold is exceeded. In FIGs.33B and 33C, positive power maxima (circles) are identified with respect to a preset threshold (broken horizontal lines) and these times (broken vertical lines) are overlaid on EGMs and corresponding wavelet decompositions for both a and b. From this, local activation times can be determined along with fractionated activity. An example of this is in FIG.33D, which shows the EGM record with local activation marked by filled circles and fractionated activity indicated by the star. [00307] In this embodiment, the first maxima of scale 5 that exceeds the threshold of timeframe b is used as the onset of the fractionated activity. In an embodiment, the timing of largest maxima can be used (here, the third). In an embodiment, the average timing of the maxima (from a set reference time, such as the beginning of the window for example, or from the beginning of the first maxima or from some other utilitarian temporal beginning point) can be used. In an embodiment, the mode can be utilized instead of the median, and where there is an even number of maxima, the median between the mode can be utilized, or a default can utilize to utilize one versus the other, such as the first or the last or the one that is larger or smaller in amplitude by way of example. In an embodiment, the timing of the last maxima can be used. In an embodiment, each maxima can signify a timing that has utility for the teachings herein. Also, in some embodiments, the timing is weighted, so say an average timing would be skewed towards the timing of the highest amplitude, because the average would be weighted based on amplitude in some embodiments. Any regime of selection of timing that can have utilitarian value can be used in some embodiments. In an embodiment, the product of a trained neural network can be utilized to select the timing in this regard, if a sufficient number of examples are evaluated and this is provided to a neural network (a sufficient number of examples for training), the neural network can be trained thereon and can identify the timing. In this regard, a human can evaluate the data beforehand and identify the timing based on the knowledge and/or experience of the human or the plurality of humans, and then the evaluations established by the human(s) can be utilized to train the system. In an embodiment, the activation timing that is identified is not necessarily the time with the greatest amplitude or otherwise is not always such. This can be a result of data that comes from other locations / originates from outside the area of interest vis-à-vis a given channel that has been captured by a given channel (given electrode) and in some embodiments, at a given frequency. In this regard, there can be
Attorney Docket No.246-021PCT sampling between electrodes. If there is fractionation, such can occur at the electrode and/or within 0.25, 0.5, 0.75, 1, 1.25 or 1.5 mm or so of that electrode, and such can be considered “for” that electrode and not for other electrodes, but there can also be or instead only (for a given temporal period) fractionation that should be associated with another electrode / another channel and/or that originates outside of that space. In some embodiments, this is not wanted in the channel of interest, or more specifically, the influence thereof should not be taken into account for the activation timing. Hence the evaluation of the power data beyond taking the maximum amplitude. [00308] In an embodiment, inverse mapping can allow and thus is used for the identification of the fractionated activity and where the activity originates from / from where the activity comes. Here, one can evaluate the fractionated activity and, in some embodiments, use a temporal data context, and determine that a maxima is not associated with the fractionated activity that should be associated with the electrode / channel of interest, but instead is associated with another electrode / another channel, and thus should be discounted (that maxima should not be used for the activation timing. By way of example, the data from other electrodes as well as the electrode for the channel of interest can be considered within a given temporal framework. A comparison of data from other channels / electrodes can be made at given times (which can account for travel for example or such can be de minimus and ignored in other embodiments) and from such, the maxima within the power data that is utilitarian or otherwise appropriate for activation timing for that channel can be selected, and the other data disregarded. Thus, in some embodiments, there can be a method (which can be automated for example, and can be implemented using any of the computational techniques / devices / systems and/or methods detailed herein by way of example or otherwise modified versions thereof) which refers back to an underlying source of given data, and otherwise implements / relies upon spatial localization to identify the maxima that is to be used and/or the maxima that should not be used. In an embodiment, the maps detailed herein can put a current source within 2, 3, 4, 5, 6 or 7 or 8 or 9 or 10 or more electrodes. One will be greatest due to a given source and typically such will related to proximity. Typically, the highest will be at one electrode and the surrounding electrodes will have lower values (temporally based). That can be used to identify the location of the source between electrodes for example, using mapping techniques. And this is not the maximum maxima selection – it could be that the maxima at a given time for the correct source is lower than another maxima at another time for the wrong source – this is why temporal based comparisons can be useful. Thus, one can obtain in real time or near real time
Attorney Docket No.246-021PCT in some embodiments a clear recognition of from where the activity is coming / originates and chose whether or not it is originating from near the electrode / channel of interest or further away, and if the latter, likely discounted. This can be revealed from subsampling and is such in some embodiments. [00309] In this embodiment, only one maxima is used, which maxima corresponds to the originating activation that is near(est) the electrode, again based on map techniques to determine such in this exemplary embodiment. In an embodiment, the data that comes from outside the area of interest (that is identified as originating in tissue outside the area of interest for a given channel) is removed or not used or otherwise discounted from the data used to develop the activation timing / in the method actions to develop the activation timing. All of this said, weighting can be used, which weighting can be based on a relative magnitude of the potentials. In an embodiment, a combination of weighted averaging with other information from other electrodes, such as an inverse solution, can be used. [00310] Indeed, the propagation of electrical activation in cardiac muscle cells is driven by diffusion currents generated in surrounding tissue by cell membrane depolarization. This electrotonic current flow is effectively instantaneous and alters potential distributions in adjacent inactive muscle cells bringing them to the threshold at which depolarisation occurs. Unipolar electrograms (EGMs) are records of potential changes associated with these processes across an electrode in the extracellular domain. As the distance between a unipolar electrode and a distributed current source increase the resultant potential is attenuated and blurred, both as a result of diffusion. Again, atrial fibrillation (AF) is associated with regions where EGMs have multiple low-amplitude components which spread over a much longer time than normal. These fractionated EGMs reflect slow heterogeneous spread of electrical activation within the region addressed by the electrode or electrodes at which they are recorded. (Fractionated EGMs can occur intermittently in normal atrial tissue in AF due to spatial variation in activation time- history where some regions are refactory and activation cannot be initiated while others are partially repolarised and able to support slow, heterogenous electrical propagation. However, it also occurs more regularly in regions where structural remodeling, for instance due to atrial muscle cell death and subsequent infiltration scar tissue (replacement fibrosis) causes conduction slowing and conduction block.) Under these circumstances, it is difficult using the state of the art to identify activation times accurately at an individual extracellular electrode. This is because EGM deflections caused by activation a few millimeters away from the electrode may not be immediately distinguishable from those caused by depolarisation of
Attorney Docket No.246-021PCT smaller tissue volumes immediately adjacent to it. It is utilitarian to identify the location of sources of fractionated activity more precisely to improve the classification and mapping of this behavior. [00311] Embodiments thus include devices for, systems for and/or methods of determining an origin of energy or data, etc., or, at least determining or identifying energy or data or portions of the signal or signals that does not originate in a certain area, such as the tissue that is the area of interest or otherwise applicable to a certain channel, or otherwise that the energy or data signal originates from outside a specific area of interest or otherwise is more applicable to be utilized in activation time development for another channel. Embodiments thus include any one or more of the methods detailed herein along with such action. Corollary to this is that these devices, systems and/or methods can be utilized for into respectively, identify which maxima from a plurality of maxima can be utilized or otherwise should be utilized in the activation timing identification. [00312] Wavelet transformation with a ten derivative Gaussian mother wavelet is useful here. Fractionated unipolar atrial EGMs are decomposed into a set of time-varying components at different wavelet scales which carry information about the distribution of frequencies in EGM deflections associated with activation. In embodiments, activation sources adjacent to the electrode typically carry relatively more "power" in low-order (short-duration) wavelet scales whereas more distant sources should carry relatively more power in higher-order (longer duration) wavelet scales. Hence the different scales as taught herein to differentiate between the durations / identify the durations and determine the origin thereof so that the ones that are applicable to tissue associated with the channel can be utilized and the ones that are not can be removed (or at least weighted so that they impact the evaluation less than that which would otherwise be the case / reduce so that they are effectively negligible with respect to implementing the teachings herein). Comparison of wavelet decompositions for adjacent electrodes over time is also utilitarian to infer the location of sources from fractionated EGMs because this enables separation of passive non-near field components (which can occur on adjacent electrodes at the same time) from propagating electrical activity (where there are internally consistent time delays between activation detected at adjacent clusters of electrodes). Embodiments can include identifying the location of non-near field components (or identifying that such is not a near field component) in a fractionated EMG at less than grid spacing dimensions by, for example, "triangulation" (based on the attenuation of simultaneous wavelet components across adjacent clusters of electrodes) or by for example, increasing the
Attorney Docket No.246-021PCT resolution of our inverse mapping methods. Embodiments can use these techniques to do so, by way of example only and not by way of limitation. These rules can be utilized to identify wavelet components in a complex fractionated EGM that are associated with activation spread. Where more than one such component is identified at an electrode within the accepted absolute refractory period (this can happen if there is more than one local activation pathway), a weighted average of the activation time selected can be used. The process can be automated and could be made more adaptive using artificial intelligence (AI). By way of example only and not by way of limitation, any of the machine learning implementations detailed herein or otherwise the machine learning teachings where the products of machine learning detailed herein can be utilized in at least some exemplary embodiments to implement the weighted average or otherwise to identify the weights and to apply such. [00313] Accordingly, teachings herein include utilizing one or more of the techniques thus detailed to identify the particular maxima in the fractionated activity to utilize as the activation timing. The techniques herein provide for the development (extraction) of the maxima, and then the identification of the one maxima out of the various extracted maxima (if there are more than one). And note that it could be that there is no maxima that is used – that is, the maxima are not for the tissue associated with the channel. Any device for, system for and/or method that can enable utilitarian selection of a maxima to implement the teachings herein can be used providing that the art enables such and such can support the teachings herein, unless otherwise noted. [00314] In the above, the timing examples provide additional grounds for selecting or excluding a scale. Because in at least some exemplary embodiments, the purpose of choosing a scale is to choose the scale that can provide a utilitarian timing, if the scale does not provide a utilitarian timing according to the given algorithm that is utilized to identify the timing, that scale could be eliminated or otherwise not used in the evaluation of the positive power to determine timing. By way of example, if there is no clean mode of maxima with respect to timing (e.g. four or six maxima as opposed to three or five maxima, the latter being a clean mode because it would be the second maxima or the third maxima respectively) where the timing is based on the mode of the maxima, that scale might be dropped in lieu of another scale that does provide a clean maxima. In this underscored the fact that the scale that is selected is the scale that can provide the utilitarian value to select the timing. And this is where, for example, the trained neural network, or more accurately, the product of the trained neural network, can come into play in at least some exemplary embodiments. Still, a predetermined hard coded algorithm /
Attorney Docket No.246-021PCT predetermined chip and/or predetermined firmware can be utilized as well. Note also that big data could be used by way of example only and not by way of limitation, as will be described in greater detail below. [00315] FIG. 34 presents an exemplary algorithm for an exemplary method, method 1090, which includes method action 1092, which entails obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation. This can be executed according to any of the teachings herein. Accordingly, in an exemplary embodiment the data based on the first data is the obtained one or more unipolar EGMs. Conversely, in another exemplary embodiment, the data based on the first data is based on one or more unipolar EGMs with ventricular far-field artifact subtracted, concomitant with the teachings associated with method 1030, etc. But in an alternate embodiment of method 1090, as represented in FIG. 34A by method 1091, method action 1092 instead includes obtaining first data based on one or more unipolar EGMs of electrical properties in a living human (this is method action 1093 of FIG. 34A). Concomitant with the teachings herein, the electrical properties are electrical properties in an atrium of the living human. So, in an embodiment, the electrical properties are electrical properties in part of the living human. In some instances, the one or more unipolar EGMs are recorded in an atrium of a heart of the human. But in other embodiments, the one or more unipolar EGMs are recorded outside an atrium of a heart of the human. Any device or system that can enable the acquisition of, and any method of acquiring the data, can be used in some embodiments providing that the art enables such. Again, the method action 1093 can be executed in a human that has a normally functioning heart (method action 1092 can be replaced with method action 1093). Alternatively and/or in addition to this, the human is suspected of having atrial fibrillation but is not afflicted by atrial fibrillation. And while the teachings detailed herein can be utilized to diagnose or otherwise detect atrial fibrillation, the teachings detailed herein can also be utilized to determine that the person is in fact not afflicted with atrial fibrillation. [00316] To be clear, in some embodiments, the person is afflicted with atrial fibrillation, in an embodiment, the electrical activity is electrical activity in a heart not affiliated with atrial fibrillation, wherein the human has a pacemaker for his/her heart. In an embodiment, the electrical properties are properties of disorganized electrical activity, while in other embodiments, the electrical properties are properties of organized electrical activity. This may or may not be known beforehand. Method 1090 and method 1091 includes method action 1094, which entails identifying regional atrial activation times based on wavelet processing of
Attorney Docket No.246-021PCT data based on the first data. This can be done according to, for example, the teachings just detailed. The actor of method action 1094 can be the actor executing the processing, etc., or can be the actor that receives the processed data and from that data, identifies the regional atrial activation times therefrom. Note also that method action 1094 can be executed by evaluating data that already indicates where the activation times are present. [00317] Still, embodiments include a single actor implementing the processing, and identifying the activation. Thus, FIG.35 shows an exemplary algorithm for an exemplary method, method 1110, which includes method action 1112, which entails executing method action 1092 or method action 1093 (the text of method action 1112 in FIG.35 is an amalgamation of the two in the interests of textual economy) or a variation of one or both method actions. Method 1110 further includes method action 1114, which includes wavelet processing data based on the first data to obtain second data, and method action 1116, which includes identifying regional atrial activation times based on data based on the second data. This can be directly from the second data, or a result of manipulation of the second data (e.g., such as developing positive power). Again, any of the above teachings can be utilized to implement this method. [00318] As noted above, the wavelet processing can include developing first derivative Gaussian wavelets. Embodiments can include identifying, from wavelets produced by the wavelet processing, period(s) when power due to depolarization is maintained at a level sufficiently the greater than background power. In an exemplary embodiment, the power is at least 100, 125, 150, 175, 200, 250, 300, 350, 400, 450, 500, 600 or 700% or more or any value or range of values therebetween in 1% increments greater than the background power. [00319] In an embodiment, there is a method 2010, as seen in the algorithm presented in FIG. 36, which includes method action 2012, which includes executing method 1110. Method 2010 includes method action 2014, which includes processing the second data to obtain positive power data. Alternatively, such as where method 1090 is executed, the action of identifying regional atrial activation times is based on positive power data that is obtained from the results of the wavelet processing of the data based on the first data. In an exemplary embodiment with respect to method 2010, the positive power data includes data in a plurality of scales. Further, in an exemplary embodiment, method 2010 includes identifying data in at least one scale from the plurality of scales and using the data in the at least one scale for detection of positive power maxima, wherein the action of identifying regional activation times is based on the detected positive power maxima. And again, such as where method 1090 is executed, the action of identifying regional activation times is based on detected positive power maxima based on the
Attorney Docket No.246-021PCT results of the wavelet processing of the data based on the first data. In an embodiment, there is a method 2110, as seen in the algorithm presented in FIG.37, which includes method action 2112, which includes executing method 1110. Method 2110 includes method action 2114, which includes manipulating and/or processing the second data to develop the data based on the second data. By way of example, method action 2114 can include the action of selecting one or more wavelet scales of the second data that better matches local atrial electrical activity than other scales of the second data, wherein the selected one or more wavelet scales minimize confounding effects of noise more than the other scales of the second data. This can be done in accordance with the teachings above by way of example only and not by way of limitation, such as based on the number of maxima above a threshold value where all of the maxima are between zero power values, etc. [00320] In an embodiment, the wavelet processing of data results in a CWT at a plurality of different scales. In an embodiment, the data based on second data is rectified wavelet decompositions of the CWT. Thus, in an embodiment, the methods detailed herein can further include rectifying wavelet decomposition of the CWT. This can be done to extract positive deflections due to depolarization. Accordingly, in an embodiment, the data based on second data is based on positive deflections due to depolarization. Moreover, as noted above, only some scales of the CWT are used (sometimes only 1). Thus, in an embodiment, the method further includes rectifying wavelet decomposition of the CWT to extract positive deflections due to depolarization, the method further includes selecting one or more scales that are a subset of the total number of different scales (e.g., scale 5 from the 10 scales). Further with respect to methods that include rectifying the decompositions, in an embodiment, the method includes obtaining instantaneous measures of corresponding power at at least one scale of the plurality of different scales, and the data based on the second data includes maxima of the instantaneous measures. In an embodiment with respect to the rectification embodiments, concomitant with the teachings above associated with FIG. 33B and FIG. 33C, the method includes obtaining instantaneous measures of corresponding power at a plurality of scales of the plurality of different scales, as well as identifying one or more scales of the plurality of scales that provide a utilitarian balance between temporal resolution maxima identification during time periods of fractionated activity and the data based on the second data includes maxima (one or more) of the instantaneous measures from the identified one or more scales. Also, embodiments can include executing one or more of the method actions detailed above, such as those associated with method 1110, etc., and also classifying at least one of normal or fractionated activation
Attorney Docket No.246-021PCT complexes based on the data based on the second data to identify the regional atrial activation times. Corollary to this is that embodiments can include executing one or more of the method actions detailed above, such as those associated with method 1090, etc., and also classifying at least one of normal or fractionated activation complexes based on the wavelet processing of data based on the first data to identify the regional atrial activation times. [00321] To round things out, the action of executing wavelet processing can include decomposing the data based on the first data using CWT. Also, in an embodiment, the action of executing wavelet processing includes calculating wavelets at three (3) to fifteen (15) different time scales with first derivative Gaussian mother wavelets. And in the interest of textual economy, it is noted that any one or more of the teachings detailed above with respect to the embodiments associated with removing the artifact can apply to the embodiments of determining atrial activation times, etc., and vice versa, providing that the art enables such, unless otherwise noted. Accordingly, by way of example, the number of scales developed with respect to the embodiments related to removing the artifact can apply to the embodiments of determining atrial activation times, the times to which CWT is implemented detailed above can correspond in part or in whole to the times to which CWT is implemented with respect to the embodiments associated with the atrial activation, etc. And of course all this is in reverse as well, all in the interest of textual economy. [00322] And embodiments include combined methods. In an embodiment, any one or more of the method actions associated with the V artifact removal embodiments can be combined in a method that includes any one or more of the method actions associated with the atrial activation timing embodiment. Indeed, as would be understood from the above, the methods associated with V artifact removal are a precursor to the methods associated with atrial activation timing. [00323] Accordingly, in an exemplary embodiment, such as with respect to method 1110, in an exemplary embodiment, the first data obtained in method action 1112 can correspond to the results of the methods to remove the V artifact. In an exemplary embodiment, the first data can be the extracted atrial activation complexes thereof. Thus, with reference to FIG.38, there is a method, method 2310, which includes method action 2312, which includes executing method 1030 or method 1031 or a method or a method action associated therewith (such as method 1040 or method 1050 or method 1060 or method 1070 or the other methods actions detailed above to remove V artifact). Method 2310 also includes method action 2314, which includes executing method 1090 and/or method 1110 or a method or method action associated therewith, wherein the first data of method 1090 and/or 1110 is the second data of method 1030
Attorney Docket No.246-021PCT or the manipulated third data that results in the extracted atrial activation complexes. As will be understood, method 2310 is a broad method that encompasses combining any one or more of the teachings associated with V artifact removal with one or more of the teachings of atrial activation timing determination, all in the interests of textual economy. [00324] In an embodiment, the accuracy of atrial activation times identified in unipolar signals can be validated by comparing with corresponding Laplacian difference signals. Unipolar recordings can be acquired from, for example, 5 adjacent electrodes in an HD grid™ catheter. The Laplacian difference signal V
L(t) can be estimated as: ^^^^
^^^ ^ ^
, ^^ ^
^ ^
^^ ( ^^^^) = ^^^^
^^^^, ^^^^ − 0.25 ∗� ^^^^
^^^^, ^^^^+1 ( ^^^^) + ^^^^
^^^^, ^^^^−1 ( ^^^^) + ^^^^
^^^^+1, ^^^^ ( ^^^^) + ^^^^
^^^^−1, ^^^^ ( ^^^^)� (6) where ^^^^
^^^^, ^^^^+1 ( ^^^^
) is the voltage recorded at the centre electrode and the four remaining voltages are recorded on the electrodes surrounding it. Activation times for Laplacian signals can be identified as maxima in the time series using a threshold of 1.5x standard deviation. [00325] Embodiments also include reconstruction of near-field atrial electrograms. Here, atrial EGMs can be reconstructed using a wavelet-based filter that is synchronized with local activation and matches the known temporal variation of electrical activation, based on an occurrence where the highest instantaneous frequencies occur during depolarization whereas subsequent activity during repolarization is characterized by lower frequencies. FIG.40 shows an exemplary flowchart for an exemplary method. To start with, there is FIG. 39, which presents representative atrial EGMs (with V artifact subtracted) acquired from a patient with persistent AF, where the broken red lines / broken vertical lines are activation times determined in accordance with the teachings above." That is, FIG. 39 shows, a wavelet-based reconstruction of near-field atria EGMs, where the vertical lines indicate previously detected activation times. [00326] After identification of atrial activation times in unipolar recordings (V artifact subtracted – FIG. 39), a CWT can be computed at various scales (e.g., 10 scales, but other numbers can be used – we refer to the above statements about the applicability of the teachings of one embodiment being applicable to the other and not being repeated in the interests of textual economy – that applies for this reconstruction embodiment as well, and vis-a-versa) using a 2nd derivative Gaussian mother wavelet. That is, wavelet decomposition can be performed, here, in 10 scales, using 2
nd derivative Gaussian wavelets, but again, more or fewer scales can be used (any of the numbers detailed herein can be used, and the number and/or scales can be different from each other (e.g., the scales for V subtraction can be different in
Attorney Docket No.246-021PCT type and/or number than those for activation time and/or for the reconstruction of the atrial electrogram, and visa-versa (the scales for the reconstruction can be different from the V subtraction and/or the timing, and the timing can be different from the V subtraction and the reconstruction). [00327] The following will focus on a processing workflow for an atrial signal segment (300 msec – the grey area of FIG.39, expanded in FIG.41A for clarity) encompassing a complete activation cycle. FIG. 41B shows the wavelet decomposition for each scale, with the lowest scale at the top and the highest (coarsest) scale at the bottom, some of which are labeled (11701 is scale 1 and 11710 is scale 10, 11709 is scale 9 and 11704 is scale 4, and the scales that are not labeled would follow the pattern in 1 integer increments). [00328] In an embodiment, the temporal period of the wavelet decompositions computed across time windows from -150 to +150 ms by way of example with respect to the maxima of the EGMs and/or the activation times. In an embodiment, the base raw signal extends (in some embodiments contiguously) for less than, greater than, and/or equal to 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, or 900 ms, or more or any value or range of values therebetween in 1 ms increments for respective windows, and the time need not be the same for each window. It is to be understood that some of these values may start to overlap with the next window, depending on the spread of the maxima and/or activation timing, and thus the window could end at the beginning of the next window. In an embodiment, the CWT can be computed across time windows from plus and/or minus less than and/or greater than and/or equal to 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 325, 350, 375, 400, 425, 450, 475, 500, 525, 550, 575, 600, 625, 650, 675, 700, 725, 750, 775, 800 ms, or more, or any value or range of values therebetween in 1 ms increments from the atrial activation time (and the minus need not be the same absolute value as the positive, and the numbers need not be the same in each window, although in other embodiments, the numbers will be the same in each window), and this can be different for different windows, and again, this can be truncated upon running up onto the next window). In an embodiment, the CWT can be computed continuously and/or for time windows that form a contiguous set of time windows for the given data from the channel, where in some embodiments, each window has a maxima and/or activation times or two and/or three or more of such (whatever basis for window length that can have utilitarian value can be used – indeed, window length can be an artificial construct – a single window can exist where certain times
Attorney Docket No.246-021PCT therein are the basis from which to extract the data that has utilitarian value). In an embodiment, the temporal limitation is to prevent interference from coming through or otherwise unwanted data from coming through / being included in the calculations. In an embodiment, there are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350 or 400 or more, or any value or range of values therebetween in 1 increment time windows for each scale (this can be the case for all the windows herein for the other embodiments in the interest of textual economy). [00329] In an embodiment, each window (wavelet window) can include 1 or 2 or 3 or more (and that includes only 1 or 2 or 3 – any disclosure of values herein corresponds to an alternate disclosure of having only those values) activation times. [00330] In the figures, wavelet decomposition and power segments are presented horizontally for same scales, with scale becoming more coarse from top to bottom. The data is also normalized. Maxima in both decompositions and power segments identify contributions of different wavelet scales. [00331] Next, in this embodiment, the wavelet decomposition for each scale is tapered with window function, here a Tukey window aligned with activation time and centered on the activation time for finer wavelet scales. This is seen in FIG. 41C. Note that in some other embodiments, other types of window functions can be used, such as those disclosed above, providing that such can enable the teachings herein and provide utilitarian results. Tukey is simply an exemplary window function for this embodiment. And as noted above, windows based on any of those herein can be utilized. [00332] Again, in an embodiment, the product of a trained neural network, trained on a sufficient number of examples, can be used to select the window function and/or the parameters of the window function (length of time, centering / off-centering, which decompositions will be off centered and which will be centered, for example). [00333] Window length and the period of cosine tapering for one or more or all scales can be set to preserve components associated with activation. For remaining scales, the onset and duration of the initial cosine taper, which are often consistent with finer scales, the window length can be prolonged so that final cosine taper and can be synchronized with the next atrial activation, as seen in FIG.41C and FIG.42 (where FIG.42 is a larger view of FIG.41C, with the scales labeled – as seen scales 11710 and 11709 are synchronized with the next atrial
Attorney Docket No.246-021PCT activity / the window is prolonged – additional scales (scale 8 and/or scale 7 by way of example) can also be synchronized with the next atrial activity / the windows therefore can be prolonged). To be clear, in an embodiment, for each scale, the decomposition can be tapered with Tukey windows aligned as indicated. The finest scales can be centered on the activation time, while window lengths for the coarsest scales can be extended to coincide with the start of the next activation cycle. [00334] In an embodiment, the symmetric portion of the window for the relatively high frequencies and/or lower scales is 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 260, 27, 280, 290, 300, 310, 320, 330, 340 or 350 ms, or more, or any value or range of values therebetween in 1 ms increments (e.g., in the embodiment of FIG. 41C, 140 ms). Concomitant with the CWT windows noted above, some of this might result in the window overlapping with the next window and if the window starts abutting the next window, depending on the activation timing, that would truncate the window. Outside of this, the data is zeroed. Conversely, the non-symmetric portion of the window for the relatively low frequencies and/or higher scales is 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650 ms, or more, or any value or range of values therebetween in 1 ms increments (but again, if the window starts abutting the next window, depending on the activation timing, that would truncate the window). That said, in embodiments, the non-symmetrical portion (relative to the activation times – the window is symmetrical – this can be more accurately explained as window with a tail that is extended past what would be the end based on symmetry regarding activation time) of the window can be extended to the beginning of the symmetrical portion (symmetrical relative to the activation time) of the next window. [00335] Note that in at least the embodiment depicted here, it is the symmetric portion of the windowing that drives the tail of the non-symmetric / adjusted window. In this regard, here, the tail is moved to the right until the beginning of the next window for the next activation time. In an embodiment, the tail can be stopped before that time, such as, for example, less than, greater than and/or equal to 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95 or 100 ms or more or any value or range of values therebetween in 1 ms increments before the beginning of the next window, and thereagain, if the end starts running onto the symmetric window, it could end there.
Attorney Docket No.246-021PCT [00336] In an embodiment, as can be the case with respect to determining which scale to use in the embodiment to identify atrial activation times, the person of ordinary skill in the art can perform an inspection of the wavelet decomposition and determine which scales should have the prolonged window length. That said, in embodiments, the highest scales can be prolonged as a matter of course. In an embodiment, the highest 2, 3, 4, 5, 10, 15, 20, 25, 30, 35, or 40%, or more, or any value or range of values therebetween in 1% increments scales can be automatically prolonged by default, while the remaining scales are subjected to the Tukey windows centered on the activation time. That said, curve fitting techniques and/or statistical data analysis techniques can be utilized to identify the scales where onset and/or duration of the initial cosine taper should be prolonged. In an embodiment, a judgment call must be made as to which frequencies should be subjected to the strict Tukey windowing in which frequencies are more appropriate to be expanded outside the more limited window. In an embodiment, this can be predetermined. In an embodiment, a product of a trained neural network can be utilized to identify which scales are to be manipulated in which manner, based on observation or otherwise analysis of data where a technician made the judgment as to which scales should be modified in which manner. In an embodiment, this can depend on the mother wavelet time scale and/or the rate of fibrillation. In some embodiments, the judgement call can be whether to use the longer duration window for scale 10 only or to extend to scale 9, scale 9 and 8 (or potentially just 8), scale 9, 8 and 7 (or one or two of them and not 9 and/or not 8), scale 9, 8, 7 and 6, and so on. In an embodiment, a trial and error approach can be used and the results can be evaluated, which evaluation can be based on a predetermined set of requirements or can be based on knowledge of one of skill based on visual inspection. In an embodiment, the judgement(s) will be related to what underlying frequency is intended to be captured / desired to be captured. Shorter timescales can use lower scales, which capture faster activation. In an embodiment, the shorter timescales can capture up to 80, 90, 100, 110, 120, 130, 140, 150 Hz or more or any value or range of values therebetween in 1 Hz increments. In the embodiment presented in the figures, it is desired to capture up to 60 or 70 Hz. 5 to 10 Hz would be considered a low frequency component, which components can be present for repolarization. Accordingly, the scales that are extended can be the scales that provide fair indication of repolarization. Thus, in an embodiment, the scales are adjusted to the frequency one seeks to address. One of skill will make a tradeoff between the increased scale and the costs associated therewith, such as, for example, the fact that longer scales can give the ability to get rid of the slowly varying background noise (which is useful in some embodiments). This can be something that is not obtained as cleanly with a standard continuous filter.
Attorney Docket No.246-021PCT [00337] In an embodiment, the frequencies / scales that are subjected to “strict windowing” (symmetric tapering about the activation time) are those that have a maximum amplitude that is less than and/or equal to 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97 or 98% or more or any value or range of values therebetween in 1% increments of the average amplitude (mean, median and/or mode, depending on the embodiment) within the window and/or the average (mean, median and/or mode) of the positive and/or negative amplitude (for the positive, the negative is ignored (rectification) and vis-a-versa) - the maximum amplitude would be the maximum positive if positive average and negative if average negative for example). In an embodiment, the frequencies / scales that are subjected to the “adjusted windowing” (non-symmetric tapering about the activation time) are those that have a maximum amplitude that is equal to and/or greater than 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98 or 99% or more or any value or range of values therebetween in 1% increments of the average amplitude (mean, median and/or mode, depending on the embodiment) within the window and/or the average (mean, median and/or mode) of the positive and/or negative amplitude. [00338] In an embodiment, the frequencies / scales that are subjected to “strict windowing” are those that have a variation from zero or from a predetermined range of no more than a certain amount. In an embodiment, the frequencies / scales that are subjected to “adjusted windowing” are those that have a variation from zero or from a predetermined range of more than a certain amount. In an embodiment, the frequencies / scales that are subjected to “strict windowing” are those that have more than a certain number of maxima and/or minima, such as for example, more than or equal to for example 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 or 20 or more or any value or range of values therebetween in 1 increment maxima and/or minima (collectively or singularly depending on the embodiment) within the timescale of the given window and the opposite would be the case for the adjusted windowing (those that have less than or equal to 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 or more or any value or range of values therebetween in 1 increment maxima and/or minima (collectively or singularly depending on the embodiment). [00339] In an embodiment, at least 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98 or 99% or more or any value or range of values therebetween in 1% increments of the values over the time are within 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85 or 90% or any value or range of values therebetween in
Attorney Docket No.246-021PCT 1% increments of the average (mean median and/or mode) or of a predetermined value / range of values for the scale to be subjected to strict windowing, and at least 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60 or 65% or more or any value or range of values therebetween in 1% increments of the values over the time are outside of 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85 or 90% or any value or range of values therebetween in 1% increments of the average (mean median and/or mode) or of a predetermined value / range of values for the scale to be subjected to adjusted windowing. [00340] The timing noted above can be the timing of the taper windows. Conversely, the timing could be a timing before and/or after the activation timing in the embodiments above. In an embodiment, such can be used to establish the windows for the other embodiments, all by way of example. [00341] FIG.41D shows the results of the wavelet decomposition modified with the results of the Tukey windows (where scales 1-8 show the result of the symmetrical window and where scales 9 and 10 show the results of the non-symmetrical window / the elongate window). FIG. 41E shows the reconstructed electrogram (upon the application of the inverse transform). The result of this processing is shown in FIG. 43 where high frequency electrical activity in the period after local activation is attenuated but low frequency variation is preserved. FIG. 43 shows reconstructed electrograms from the processed wavelet components. [00342] Window lengths and cosine tapering periods for each scale can be adjusted but a single set of default values can be used (and is used for all results presented here, all by way of example). [00343] In view of the above, there is a method, method 1190, as seen in FIG.44, which includes method action 1192, which includes obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation. The method further includes method action 1194, which includes reconstructing one or more unipolar EGMs based on data based on wavelet processing of data based on the first data. The wavelet processing can be any of the processing detailed above / herein. In an embodiment, the actor of method 1190 executes the wavelet processing, while in another embodiment, the actor obtains the data based on wavelet processing of data based on the first data, and identifies the regional activation times accordingly. In an embodiment, the action of reconstructing is executed by filtering across time scales in CWTs calculated with second derivative Gaussian mother wavelets. Corollary to this is that the data based on wavelet processing is data based on results of filtering
Attorney Docket No.246-021PCT across time scales in CWTs calculated with second derivative Gaussian mother wavelets. In an embodiment, the action of reconstructing is executed by filtering across time scales in CWTs calculated with only second derivative Gaussian mother wavelets. Corollary to this is that the data based on wavelet processing is data based on results of filtering across time scales in CWTs calculated only with second derivative Gaussian mother wavelets. [00344] In an embodiment, there is an alternate version of method action 1192, which includes obtaining first data based on electrical activity in a living human. In an embodiment, the first data is based on one or more unipolar EGMs, which correspond to the electrical activity. In embodiment, the first data is based on one or more unipolar EGMs, which represent the electrical activity, and the one or more unipolar EGMs are recorded in a heart of the living human, and in embodiment, the electrical activity is electrical activity in an atrium of the living human. Again, in an embodiment, the electrical activity etc. can be in another part of the living human. In an embodiment, the electrical activity is electrical activity in an organ of the living human. The electrical activity is electrical activity in a heart of the living human. In an embodiment, electrical activity and/or electrical phenomenon detailed herein, etc., is electrical activity in a muscle of the human. In an embodiment, the electrical activity and/or electrical phenomenon detailed here, etc., is electrical activity in an arm muscle and/or a leg muscle and/or a finger muscle and/or an eye muscle and/or a muscle of the face, etc. Consistent with the teachings detailed herein, human may or may not be a human afflicted with atrial fibrillation, and the human may or may not be a healthy human or may or may not have a heart that is healthy. The heart could have a heart with a hole therein and/or the living human has at least one heart stent. Again, the human could have one or more ailments but not atrial fibrillation. Again, the electrical activity can be organized or disorganized. And as noted above, the human may not be afflicted with atrial fibrillation. Thus, there is method 1191 as represented in FIG.44A, which has method action 1193 instead of method action 1192, which includes obtaining first data based on one or more unipolar EGMs of activity in a living human. In an embodiment, method action 1193 is substituted for method action 1192. [00345] In an embodiment, the action of reconstructing is executed by using a wavelet-based filter that is synchronized with local activation and that matches known temporal variation of electrical activation, and thus the data based on wavelet processing is synchronized with local activation and that matches known temporal variations of electrical activation. Embodiments can include the action of filtering for high instantaneous frequency components proximate atrial activation times and filtering for lower frequency components away from the atrial
Attorney Docket No.246-021PCT activation times. By way of example, the Tukey windows detailed above can be used to implement this filtering, where the lower frequency components are in the scale(s) that have the extended windows. Thus, in an embodiment, the data based on wavelet processing is data where the data based on wavelet processing has been filtered accordingly. [00346] Note that the method 1190 is a method that includes expanded actions, such as any of the actions associated with V artifact removal and/or the activation time identification. In this regard, there is method 2190, as represented in FIG.45, which includes method action 2192, which includes executing method action 1192. Further, method 2190 includes method action 2194, which includes identifying atrial activation times based on data based on the first data. In an expanded embodiment, there are some actions associated with method action 2194, such as any one or more of the method actions detailed above with respect to the embodiment of identifying atrial activation times. The method also includes action 2196, which includes executing method action 1194, where the action of reconstructing includes filtering out components based on the identified activation times. Figure 46 presents an exemplary flowchart for an exemplary algorithm for an exemplary method, method 2199, which includes the actions of method 2190 detailed above, with the caveat that follows, with the additional action of method action 2193, which entails removing V artifact from data based on the first data. And note that this data based on the first data could be different data based on the first data in that for example, the data based on the first data utilized in method 1194A could have the Tukey windows applied thereto. With respect to the caveat, method action 2194A parallels method action 2194, except that this action requires that the action based on the first data is data resulting from the removal of V artifact from the first data. And any of the actions detailed above associated with removal of the V artifact can be undertaken in method action 2193. [00347] In an embodiment, the wavelet processing of methods 1190, 2190 and/or 2199 includes obtaining wavelet decompositions for a plurality of scales and the method further includes tapering the obtained wavelet decompositions based on atrial activation times. In this exemplary embodiment, the atrial activation times can be based on data based on the decompositions for the plurality of scales. Note that these decompositions could also be utilized to execute method action 1194 by way of example. Alternatively, the decompositions could be reexecuted or otherwise can be different in some other embodiments. In this regard, it is noted that in some embodiments, the scales that are utilized across the various embodiments can be the same while in other embodiments, the scales that are utilized across the various embodiments could be different. Any scale of use that can enable the teachings
Attorney Docket No.246-021PCT detailed herein can be utilized in at least some exemplary embodiments, providing that the art enables such, unless otherwise noted. Note further that in some embodiments, more scales can be utilized for, for example, the V artifact removal process than that which is utilized for the atrial activation time identification and/or the reconstruction of the atrial EGM. In an embodiment, more utilized for the former than the latter to and in another exemplary embodiment, more utilized for the latter than the former to, in an exemplary embodiment, more utilized for the second than the first, and more utilized for the third than the second, and in an embodiment, more utilized for the second than are utilized for the first and third etc. [00348] Corollary to the above is that in an exemplary embodiment, the data based on wavelet processing of data based on the first data in method action 1194 can be the results of the tapering of the obtained wavelet decompositions based on atrial activation times. [00349] But to bring things back a step or two, in an exemplary embodiment, the tapering includes tapering for at least some finer wavelet scales, which tapering for the at least some finer wavelet scales is centered on the atrial activation times. Consistent with the teachings above with respect to the coarser scales, in an embodiment, the tapering includes tapering for at least one coarser wavelet scale(s), which tapering for at least one coarser wavelet scale(s) is off centered from the atrial activation times. [00350] While embodiments above have been described in terms of the Tukey window, any window function and/or tapering function that can enable the teachings detailed herein and otherwise provide utilitarian result can be utilized in at least some exemplary embodiments, providing that the art enables such, unless otherwise noted. Is also the case with the Hann windows detailed above. Any arrangement for implementation of a mathematical function that is zero valued outside some chosen interval, whether symmetric or not symmetric around the middle of the interval, which may or may not approach a maximum in the middle or may or may not approach a maximum offset from the middle, and may or may not be tapered away from the middle or wherever the offset is determined to be, can be utilized in some embodiments providing that the art enables such. [00351] In an embodiment, the wavelet processing includes obtaining wavelet decompositions for a plurality of scales. In an embodiment, the methods detailed herein such as methods 2199, 2190, and/or 1190, further include the action of window function processing the obtained wavelet decompositions based on atrial activation times. In this exemplary embodiment, window length and a period of tapering for at least some scales are set to preserve components
Attorney Docket No.246-021PCT associated with atrial activation. In an embodiment, onset and duration of initial taper for at least some other scales different from the at least some scales (e.g., scale 10 in FIG. 41B) is consistent with that of the at least some scales and window length for the at least some other scales is prolonged relative to the window length of the at least some scales. In an embodiment, window length and a period of tapering for at least some scales are set to preserve components associated with atrial activation and window length for at least some other scales is prolonged relative to the window length of the at least some scales so that final taper is synchronized with next atrial activations. [00352] In view of the above, it can be seen that in some embodiments, the wavelet processing attenuates higher frequency electrical activity in a period after local atrial activation and low frequency variation is preserved. [00353] In an embodiment, correspondence between ground-truth (GT) and processed (P) EGMs was quantified by evaluating the correlation coefficient (CC) and normalized root- mean-squared error (nRMSE). ^^^^
where N is the number of data points compared, ^^^^
^^^^ are potentials at time i and ^^^^ are mean values. Percent activation detection accuracy can be quantified as 100
where ^^^^
^^^^ is the number of activation times identified correctly (±5 msec) and ^^^^
^^^^ the total number detected. Activation time differences between ground-truth (GT) and processed (P) recordings can be estimated as ∆T=|AT
GT - AT
P| (9) Signal processing functions can be written in the MATLAB programming language (The Mathworks, Natick, Massachusetts). Distributions can be plotted as box plots and the significance of differences between them was estimated using the non-parametric Mann- Whitney U test. Data can be expressed as mean±SD when normally distributed but can
Attorney Docket No.246-021PCT otherwise represented as median [interquartile range]. Below we explain an exemplary scenario based on this noted data analysis regime. [00354] The frequency content of the activation complex in atrial EGMs for patients in longstanding (persistent of permanent AF) is from 10-100Hz whereas the frequency content of the repolarization complex ranges from ~2-10 Hz, all by way of example. Wavelet scale 1 equates roughly with 80 Hz and scale 9 equates roughly with 8 Hz in this embodiment and scale 10 equates to roughly 5 or 6 Hz in this embodiment. In an embodiment, the scale 1 is any value or range of values between 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 105, 110, 115, 120, 125, or 130 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 2 is any value or range of values between 35, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 105, 110 or 115 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 3 is any value or range of values between 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100 or 105 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 4 is any value or range of values between 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, or 85 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 5 is any value or range of values between 20, 30, 35, 40, 45, 50, 55, 60, 65 or 70 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 6 is any value or range of values between 15, 20, 30, 35, 40, 45 or 50 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 7 is any value or range of values between 10, 15, 20, 30, 35, 40 or 45 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 8 is any value or range of values between 5, 10, 15, 20, 30 or 35 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 9 is any value or range of values between 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 15, 17, 18, 19, 20, 21, 22, 23, 24 or 25 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 10 is any value or range of values between 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 15, 17, 18, 19 or 20 Hz or any value or range of values therebetween in 1 Hz increments. All of these are by example only. In an embodiment, any one or more scales can have these values increased or decreased by 5, 10, 15, 20, 25, 30, 35, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 175, 200, 250, 300, 350 or 400% or more or any value or range of values therebetween in 1% increments, and such need not be the same across all scales. [00355] In an exemplary embodiment, the teachings above can be validated vis-à-vis ventricular far-field artifact subtraction. In this regard, Fig. 47 presents a representative
Attorney Docket No.246-021PCT example of wavelet-based V subtraction from unipolar EGMs recorded from adjacent HD grid electrodes in a patient with PeAF. Overlaid records before and after processing in Figs 47A & 47B indicate that this subtraction algorithm removes V artifact while preserving near synchronous atrial EGMs. In Fig 47C, bipolar EGMs constructed from the processed unipolar records are near identical to corresponding bipolar signals output from the mapping system showing that V subtraction has had minimal impact on atrial components in the two unipolar channels. Further, it is noted that FIG. 47 provides validation of wavelet-based V artifact subtraction. FIG. 47A provides unipolar EGMs recorded at adjacent electrodes of Advisor™ HD grid catheter before (grey) and after (blue) V artifact subtraction. FIG.47B provides bipolar EGMs for electrodes in (A) recorded from mapping system (red) and constructed from unipolar EGMs in (A) after V artifact subtraction. FIG. 47C provides for, following addition of a V signal component to "gold standard" unipolar atrial EGMs, artifact was removed with wavelet- based method and by subtracting mean and median QT templates synchronized with V activation. The correspondence of all three approaches with the "gold standard" atrial signal is indicated by the box-plots of correlation coefficient and normalized RMS error above. P value <1.0 x10
-5 indicated by ***. [00356] Systematic comparison of our wavelet-based artifact cancellation with subtraction of fixed mean or median QT templates demonstrates that this technique has utilitarian value beyond either alternative. The effects of V subtraction on synchronous atrial EGMs can be evaluated as outlined herein across 50 successive QT intervals. Figure 47C shows that with wavelet-based correction CC (0.94 [0.01]) was significantly better (p <1.0x10
-5) than both mean or median QT template subtraction (0.64 [0.11] and (0.65 [0.13], respectively) and that these were not different from each other (p=0.873). Corresponding results for nRMSE were 0.08 [0.01] compared with 0.22[0.06] and 0.22 [0.07] with the wavelet-based correction again a highly significant (p <1.0x10
-5) improvement on mean or median QT template subtraction, which were not different from each other (p=0.764). Embodiments include methods, devices, and/or systems disclosed herein where the results are at least 50%, 55%, 60, 65, 70, 75, 80, 85, 90, 95, or 100%, or more as good as these results. [00357] With respect to the identification of atrial activation, the wavelet-based method of identifying atrial activation is based on detection of elevated signal power across characteristic combinations of time and frequency. Representative results as noted above with respect to FIGs. 33A and 33B indicate that thresholds for the duration and magnitudes of power in individual wavelet scales can be set to segment different classes of local electrical activity. In
Attorney Docket No.246-021PCT cycle a, power exceeds threshold for 18ms with a single peak during this period in wavelet decompositions for all scales. In cycle b, however, lower levels of power are sustained for much longer (78ms) with multiple lower amplitude peaks in the wavelet decomposition during this period, both supporting classification as a fractionated EGM. Three fractionated EGMs identified on this basis are shown in the figures. Again, in an embodiment, there are methods of identifying the fractionated activity based on identifying the presence of two or three or four or five or six or more lower amplitude peaks relative to the single high peak in the non- fractionated activity over a sufficiently longer time, such as 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, 6, 6.5, 7, 7.5, 8, 9, or 10, or more, or any value or range of values therebetween in 0.1 increments times the time of the single peak that exceeds the threshold, such as the average thereof with respect to the nonfractionated activities. In an embodiment, the prerequisite is that the occurrence of the peaks must occur by at least and/or equal to the aforementioned times over the given period that is specified, and in some embodiments, there must be a certain amount of time between the peaks, such as at least 5, 10, 15, 20, 25 or 30% or more or any value or range of values therebetween in 1% increments of the total time to which the analysis is limited (e.g., five times the time that the single peak was above the threshold). [00358] In an embodiment, the methods used here for detection and classification of atrial activation are not affected by the magnitude of EGM components. As shown in FIG. 33B, thresholds used to detect the beginning and end of activation can be set for individual scales based on the previous power in that scale. Likewise, activation times can be estimated from maxima during activation in decompositions at different wavelet scales. The independence of identification on magnitude across wavelet scales is underscored by the fact that time series records for wavelet decomposition and associated power for the 5 scales presented in 33B and 33C are normalized. In some embodiments, there is a high level of consistency between activation characteristics detected at different scales and in this case we have selected scale 5 as best matching the EGM components that we are seeking to identify. Finally, a second possible activation is evident at scales 1 and 3 in cycle a, 37ms after initial detection of activation. This can be judged as being nonlocal because similar synchronous deflections were recorded at adjacent electrodes and markedly attenuated in associated Laplacian EGMs. [00359] In Fig. 48, an exemplary comparison of the performance of the teachings herein with Laplacian difference EGMs in the presence of wide-band non common-mode noise. Briefly, this shows a comparison of wavelet-based and Laplacian atrial activation detection in presence of uncorrelated wide-band noise. FIG. 48A is a Schematic of Advisor™ HD grid catheter
Attorney Docket No.246-021PCT showing 5 electrodes used to construct Laplacian difference signal B3. FIG. 48B shows a unipolar signal recorded at B3 with V artifact removed. Green dots represent activation times detected using wavelet-based method. FIG.48C shows Laplacian signal at B3 without added noise. FIG.48D shows Unipolar signal with white noise added to reduce SNR to -10dB. Red dots indicate activation times detected using wavelet-based approach. FIG. 48E shows Laplacian signal at B3 with white noise with the same distribution as at B3 also added at A3,B2,C3 and B4. FIG. 48F shows comparison of detection accuracy and timing error for activation times detected in unipolar signal in B) using wavelet-based method (red) and from corresponding Laplacian signal with progressive addition of wide-band white noise. Noise and SNR are referred to the unipolar signal B3 but the same noise power is also applied to the 4 other electrodes in the Laplacian signal. [00360] More specifically, unipolar recordings acquired at 5 adjacent sites with an HD grid catheter (Fig.48A) are processed as outlined herein and a Laplacian difference signal is formed from these data. While the Laplacian difference signal attenuates components that are common across electrodes, it also sharpens atrial activation complexes and amplifies uncorrelated noise (compare Figs. 48B and 48C). Addition of bandlimited Gaussian white noise sufficient to reduce SNR in unipolar B3 to -10dB lowered wavelet-based detection accuracy in Fig.48D to 88%, with the same number of activation complexes detected but incorrectly identified in 3 cases as preceding the actual time by 46 to 52 msec. However, the same noise has a substantially greater impact on Laplacian EGMs; SNR can be reduced to -14dB and detection accuracy dropped to 10.4% with a timing error of 43.2±25.0 msec. The robustness of the wavelet-based detection in the presence of increasing levels of uncorrelated noise is demonstrated in Figure 48F where accuracy and timing error are plotted as functions of decreasing SNR for unipolar electrode E5. Detection accuracy for the former is >80% with minimal timing error until SNR is < -10dB. In contrast, both are rapidly degraded with Laplacian difference signals when SNR drops below +10dB. Embodiments include methods, devices, and/or systems disclosed herein where the results are at least 50%, 55%, 60, 65, 70, 75, 80, 85, 90, 95, or 100%, or more, or any value or range of values therebetween in 1% increments of those presented. [00361] With respect to near-field atrial signal reconstruction, representative results obtained with wavelet-based reconstruction of unipolar atrial EGMs are presented in Fig. 49. Application of our time-dependent filter synchronized with atrial activation time preserved signal components associated with local atrial activation and this is verified by the consistency
Attorney Docket No.246-021PCT of timing and magnitude between reconstructed unipolar and corresponding Laplacian EGMs (compare FIG.49B and 49C). As expected though, there are clear differences in morphology; Laplacian EGMs are shorter in duration with peak amplitudes overlapping the maximum negative slope of unipolar EGMs (Fig. 49D). Attenuation of fine wavelet scales following activation resulted in stable electrical activity during this period, enabling estimation of an activation recovery interval of -220.8±13.5 msec in Fig.49B at a base cycle length of 363.2±7.0 msec. Wavelet-based EGM reconstruction proved extremely robust in the presence of noise. Addition of white Gaussian noise to the unipolar EGMs in Fig.49A sufficient to reduce SNR by > 20 dB had surprisingly little effect on the EGMs recovered. CC and nRMSE were preserved with progressive degradation occurring only when SNR was reduced below -10 dB (see Fig.49E). That is, performance of wavelet-based reconstruction of atrial unipolar EGMs can be shown by way of example by FIG.49. FIG.49A shows unipolar EGMs acquired using Advisor™ HD grid catheter with V artifact subtracted and FIG.49B shows corresponding near- field EGMs signals reconstructed as outlined above. Reconstruction is synchronized with activation times detected from FIG.49A, dashed red lines indicate activation times identified. FIG. 49C shows Laplacian EGM corresponding to unipolar EGM in FIG.49A. Voltage and time scales in upper panels apply to FIGs.49 (A), (B) and (C). FIG.49D shows representative reconstructed unipolar EGM cycle (blue) with corresponding Laplace EGM (red) overlaid. Timing within record indicated by grey shading in (B) and (C). FIG. 49E shows correspondence between reconstructed EGM in (B) with progressive addition of band-limited white Gaussian noise sufficient to reduce SNR to < -20dB. Embodiments include methods, devices, and/or systems disclosed herein where the results are at least 50%, 55%, 60, 65, 70, 75, 80, 85, 90, 95 or 100% or more as good as these results. [00362] Spatial difference EGMs that can provide a streamlined way of removing common mode noise recorded with multi-electrode arrays (including ventricular far-field artifacts) and enable robust identification of local activation times. However, important information is lost and difference signals cannot be used for inverse intracardiac potential mapping. The teachings herein present novel wavelet-based methods for extracting near-field atrial EGMs recorded during AF that are fast (2, 3, 4, 5, 6, 7, 8, 9 or 10 times faster than alternative methods mentioned above), enable more reliable estimation of atrial activation times in the presence of noise than difference-based methods such as the Laplacian filter (where reliability can be increased by 10, 20, 30, 40, 50, 75, 100, 125, 150, 175, 200, 250, 300, 350 or 400% or more or any value or range of values therebetween in 1% increments over the other methods, for at least
Attorney Docket No.246-021PCT 50, 55, 60, 65, 70, 75, 80, 85 or 90% of patients out of 100 patients tested), and provide robust direction-independent information on local activation and recovery that cannot be obtained with spatial difference methods. In addition to the standard signal processing workflow (bandpass filtering and application of a notch filter to remove power line noise), the teachings herein can include an approach involves 3 key steps: 1) estimation and subtraction of artifact due to ventricular activation using a method that accounts for beat-to-beat variation of this far- field signal component 2) identification and classification of local atrial activation from distributions of instantaneous power following removal of V artifact, and 3) reconstruction of near-field atrial unipolar EGMs using a "matched" filter synchronised with atrial activation times. The teachings herein can be based on the fact that both V artifact and atrial near-field EGMs have characteristic time-varying frequency “fingerprints” that can be identified and separated using wavelet-based methods. The teachings herein contrast from the variety of methods have been used to subtract V artifact from unipolar EGMs in AF ( Slocum et al., 1985; Shkurovich et al., 1998; Castells et al., 2005; Salinet et al., 2010; Ahmad et al., 2011; Salinet et al, 2013). The teachings herein can include ensemble-averaging of unipolar EGMs synchronized with ECG timing accurately characterizes the mean V artifact in AF, while accounting for beat-to-beat variation in this far-field signal component. As a result, removal of mean or median electrical activity within fixed QT templates aligned with V activation reduces artifact, and in embodiments, there is little to no subtraction residues. The teachings herein can be implemented without principal component analysis to account for variation in QRST morphology (Castells et al., 2005) and/or in a computationally efficient manner. [00363] The teachings herein can have near real-time cancellation achieved by continuous estimation of a QT subtraction template weighted heavily by the R peak of the current V activation in some embodiments. The teachings can have wavelet based methods instead to scale the QRS complex in the average QT template so that it matches current V activation amplitude. The analysis presented in Figure 47 indicates that atrial activation complexes are recovered from overlying V artifact with excellent fidelity although some residue remains with nRMSE across the QT window reduced to ~8% from ~22% for fixed QT template subtraction. [00364] Precise identification and classification of local activation can be utilitarian for effective analysis of factors responsible for atrial rhythm disturbances, and the teachings herein can provide such. The teachings herein provide methods that do not magnify error introduced by non common-mode noise. In AF, classification accuracy is also compromised by the complexity of EGMs which are often fractionated and this has been addressed using template
Attorney Docket No.246-021PCT matching schemes based on data sets of “characteristic” unipolar atrial activation patterns. The wavelet decomposition used here in some embodiments is a generalization of a “matched filter” approach. Embodiments use the morphology of 1st-differential Gaussian wavelets closely resembling that of the unipolar activation complex, and corresponding wavelet decompositions effectively reliably identify such events in the presence of wide-band noise because the probability of random occurrence of such characteristics across scales is very low. Furthermore, wavelet based methods provide a sensitive means of identifying activation that is not biased by the magnitude of signal components in unipolar EGMs as is the case with conventional gradient measures. Embodiments include recording from 2D (or 3D) arrays with appropriate resolution. [00365] Duration of activation and number of peaks detected provide a straightforward basis for segmenting and further classifying normal and fractionated activation. [00366] As seen above, a wavelet-based filter can also be used to reconstruct atrial electrograms. In this case, scale-weighted windows synchronized with atrial activation time can be used to attenuate fine scale wavelet components not associated with the current activation but to pass coarse scale components throughout most of the cycle. The filter according to an exemplary embodiment passes only high frequency signal components only during activation, and captures the subsequent low frequency recovery of potential and then restores the baseline immediately prior to the next activation. As with wavelet-based identification of activation time, these reconstruction methods can be extremely robust in the presence of noise and for very similar reasons. This can be a matched filter, in this case based on a priori knowledge that electrical activity is characterized by high frequency components during activation but lower frequencies following this. The recovery in potential following activation may not be linked with the delayed attenuation of coarse-scale wavelet components. In an embodiment, this is not interpreted as local atrial repolarization. [00367] The three wavelet-based filters outlined above can have a relatively large number of parameters which specify wavelet time scales, scale-based attenuation characteristics etc. and adjustment of some of these is utilitarian to optimize performance. For example, the wavelet- scale at which activation time is identified impacts the accuracy with which it can be estimated and the most appropriate setting can vary between studies. That said, the methods described are relatively consistent across scales and the default values tabulated are used for all examples presented here (with respect to the figures – other examples are presented). Despite the apparent complexity of the signal processing methods used, in at least some embodiments, they
Attorney Docket No.246-021PCT are computationally efficient and can be completed in near real-time, in part because the Fast Fourier Transform is used to compute CWT decompositions. In this study, most emphasis is placed on estimation of atrial activation times and recovery of near-field electrical activity from individual unipolar recordings. However, the results demonstrate that these methods are particularly powerful for analysing signals acquired with high resolution multi-electrode catheter arrays where signals recorded on adjacent electrodes can be used to validate and extend the findings. In this setting, these wavelet-based signal processing tools provide a platform for more sophisticated 2D (or 3D) analysis, for instance using phase analysis to investigate underlying local mechanisms responsible for rhythm disturbance. [00368] The above thus presents a rich set of novel, computationally efficient, wavelet-based signal processing tools that can be used to extract detailed information about local electrical activity from unipolar atrial EGMs in AF. The teachings herein are extremely robust in the presence of noise because it can employ matched filtering informed by a priori knowledge of the temporal variation of frequency content in cardiac extracellular potentials generated by electrical activation. At least some of these embodiments are such that the methods are readily generalizable to clinical mapping applications with high-resolution multi-electrode catheter arrays. [00369] Embodiments include spatial mapping technologies, or more accurately, the implementation of the spatial mapping technologies detailed herein. The spatial mapping technologies may or may not be associated with atrial fibrillation. The spatial mapping technologies may be implemented for a heart that is not afflicted with atrial fibrillation. The teachings detailed herein do have utilitarian value with respect to mapping and evaluating a heart where there is irregular activity. But the teachings detailed herein are also applicable to mapping a heart where there is regular activity (whether known beforehand or not, or at least suspected beforehand or not), which regular activity falls within any of the human factors engineering values detailed herein unless otherwise noted. [00370] In an embodiment, there are methods that include using the teachings herein to identify re-entrant pathways in atrial tachycardia and/or ventricular tachycardia. Exemplary embodiment, these are regular rhythms that can be some instances be traced utilizing existing mapping techniques. In this regard, it is noted that in at least some exemplary scenarios, those existing mapping techniques cannot be utilized to trace the irregular rhythms of atrial fibrillation. Conversely, the teachings detailed herein can be utilized to trace the irregular rhythms of atrial fibrillation, which is thus an example of a utilitarian value of the teachings
Attorney Docket No.246-021PCT detailed herein. So, embodiments of the teachings detailed herein can be distinguished from such prior mapping techniques because such prior mapping techniques cannot be used trace such rhythms, at least not in a manner that has efficacy, or at least medical efficacy, and certainly not in the manner providing the accuracy and/or specificity that results from the teachings detailed herein. [00371] Still, referring back to the reentrant pathways in atrial tachycardia and ventricle tachycardia, the techniques detailed herein can be utilized to identify such. Accordingly, the teachings detailed herein are not limited to irregular rhythms, but include the regular rhythms of such. Any disclosure herein of tracing and/or mapping irregular rhythms or otherwise applying the teachings detailed herein disclosed in relationship to working with a heart that is afflicted with atrial tribulation corresponds to an alternate disclosure of implementing any one or more the teachings detailed herein to a heart that has regular rhythms, and otherwise implementing any one or more of the teachings detailed herein identifying reentrant pathways in atrial tachycardia and or ventricle tachycardia providing that the art enable such, unless otherwise noted, all in the interest of textual economy. [00372] Embodiments can also include implementing the teachings detailed herein with respect to an analysis that does not involve all of the specifics of the sophisticated (relative) spatial mapping technologies detailed herein. In an exemplary embodiment, there is its analysis that involves by way of example only and not by way of limitation, assessing a single lead electrogram and extracting atrial components in the presence of ventricular components. In an exemplary embodiment, the ventricular components are likely to appear to be dominant. The teachings detailed herein can be applied to such in some embodiments by way of example. In this regard, the teachings detailed herein can be utilized to overcome or otherwise account for the fact that ventricular components are often dominant in the resulting data collection, thus the techniques detailed herein can be utilized to assess the atrial components even in the presence of the dominant ventricular components. Embodiments include devices and systems that enable such and/or methods of doing so. To be clear, embodiments can include extracting atrial data, which can be valuable atrial data, from signals that appear to be dominated by noise. Typically, the noises from ventricle activity. But to be clear, the noise could come from other activity or other sources. Ventricle activity is disclosed herein as but one example of activity or otherwise a phenomenon that can cause noise where the teachings detailed herein can be utilized to extract the valuable data from the signal that is dominated by data that is not wanted
Attorney Docket No.246-021PCT (which data could also be valuable, it is just not wanted, or at least the underlying hidden data is wanted more than the dominant data). [00373] Embodiments include extracting atrial data, which atrial data can be utilitarian, from signals that appear to be dominated by noise, again which noise can be but is not limited to ventricle activity. This atrial data may or may not be associated with atrial fibrillation, as noted above. Again, the heart from which the signals are obtained for otherwise the body tissue of the human from which the signals are obtained may be one that is not afflicted by atrial fibrillation or otherwise any ailment. [00374] Also, there is utilitarian value with respect to avoiding, in part or in totality, reliance on coupling improved atrial signals into a modeling environment for assessing spatial atrial activity. Embodiments include devices and systems that when utilized, avoid such, in part or in total, and methods that also avoid such, again in part or in total. Embodiments can and do include utilizing such (such can provide utilitarian value with respect to improving the accuracy of the spatial analysis). Embodiments may also not use such. In this regard, there can be utilitarian value or otherwise just a desirable goal of simply having atrial signals which are clean or otherwise more clean than that which is the case with respect to implementing other techniques, and or are not swamped or otherwise camouflaged by the ventricular activity, or by way of example only and not by way of limitation. [00375] Note that in some embodiments, the potentials measured using the basket catheter for example, the potentials can be between the respective electrodes of the one or more electrodes and a common reference electrode or a plurality of respective common reference electrodes - in some embodiments, a plurality of electrodes but not all the electrodes could use one reference electrode, and a plurality of electrodes but not all the electrodes could use another reference electrode, etc.). Note that in principle, the unipolar potential is the potential measured at the electrode. In practice, one can at least usually only measure potential difference and thus in an embodiment the reference electrode is set to zero volts, and here, this can be done by forming a Wilson central terminal or otherwise averaging voltages around closed "triangle" on the body surface as far from the heart as possible. Conservation should make this as close to zero as possible. Embodiment can use catheters with a far-field reference which might be argued to allow V artifact to be subtracted. While such certainly minimizes V artifact, there are differences in the V artifact measured at different electrodes and this results in subtraction residues.
Attorney Docket No.246-021PCT [00376] Embodiments can include techniques that are at least effectively agnostic, or otherwise not affected, by the direction in which activation wavefronts propagate with respect to the recording electrodes. Embodiments can also be implemented where useful data is obtained when activation spreads perpendicular to the electrode array (that is, the data obtained by the arrays can be effectively used in such a scenario, including used effectively). [00377] Embodiments can include obtaining one or more of the utilitarian data herein without interpolation across the interval from the onset to the end of ventricular depolarization. Embodiments include obtaining data where at least 80, 85, 90, 95, or 100%, or any value or range of values therebetween in 1% increments of the near-field atrial activity (as measured based on energy level) that overlaps ventricular activation is not lost / is preserved and is otherwise not distorted. Embodiments can avoid the use of subtraction of time-averaged estimates of far-field activity during ventricular activation and repolarization to obtain the data detailed herein. Adaptive algorithms can also be avoided, at least with respect to such that are designed to account for beat-to-beat variation in ventricular (V) artifact. [00378] Embodiments include utilizing the results of one or more of the method actions detailed above with respect to V artifact removal, atrial activation and/or atrial EGM reconstruction. [00379] In some embodiments, the actions of executing any one or more or all of: (1) the wavelet processing and the implementation of the window functioning and the Fourier transformation and the subtraction of the results from the initial EGM to develop the EGM without the V artifact, starting with the obtained first data based on one or unipolar EGMs recorded in an atrium (2) the wavelet processing and the positive power development (and in some embodiments, the power maxima identification and/or the selection of scale and/or the activation timing) starting with the obtained first data based on one or unipolar EGMs recorded in an atrium; and/or (3) the wavelet processing and the application of the window functioning and the Fourier transformation and the reconstructing the atrial EGM are executed within a period of no more than 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes or within 50, 40, 35, 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4 or 3 seconds of the beginning of the calculations or any value or range of values therebetween in one second increments. Note also that in some embodiments, the actions of developing a time-varying electrical potential map (based on the wavelet filtering, with in some embodiments and without another embodiments actually performing the calculations therefore (as opposed to starting with data that is already wavelet processed, such as for example, starting with a reconstructed atrial EGM), developing the time- varying phase map, and identifying the repeating phase signatures are executed within a period
Attorney Docket No.246-021PCT of no more than 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes. Indeed, in some embodiments, the time averaging phase data reaches convergence within 40, 35, 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4, or 3 seconds of the beginning of the calculations of the maximum phase gradients. (Convergence can mean that the statistically non-aberrant numbers (e.g., the zero values) average out to a level number. [00380] In the interest of textual economy, any temporal disclosure herein of executing any one or more or all of the method actions detailed herein corresponds to a disclosure where any one or more or all of the method actions associated with the wavelet filtering embodiments to develop the V artifact data and/or the EGM with the V artifact removed and/or the atrial activation timing based on wavelet filtering and/or the reconstructed EGM based on wavelet filtering are executed within that temporal period. Embodiments also include scenarios where the temporal disclosures herein do not include one or more or all of those actions. For example, the time varying electrical potential map could start with the reconstructed atrial electrograms based on wavelet filtering. Still, the temporal features detailed herein can apply to embodiments where, for example, the action of developing a time-varying electrical potential map is based on and includes actually taking invasive readings taken while a human in which the beating heart resides is in an operating room and where the method includes actually taking those invasive readings from the start of potential recording for example (potential recording of data that is actually used to implement the teachings herein, as opposed to potential recordings for the purposes of calibration and/or for verification that the system is functioning properly to enable the recordation of potentials that are usable). [00381] As noted herein, embodiments, include non-transitory computer readable medium having recorded thereon, a computer program for executing at least a portion of a method, the computer program including any one or more all the actions detailed herein providing that the art enables such. In an exemplary embodiment, the computer readable medium includes code for executing one or more or all of the method actions associated with methods 1030, 1040, 1050, 1060, 1070, 1090, 1110, 2010, 2110, 2310, 1190, 2190 and/or 2199, and the variations thereof and the various permutations detailed above. More specifically, in at least some exemplary embodiments, the computer readable medium includes code for executing one or more of the method actions detailed herein associated with the wavelet processing and the implementation of the window functioning and the Fourier transformation and the subtraction of the results from the initial EGM to develop the EGM without the V artifact, starting with the obtained first data based on one or unipolar EGMs recorded in an atrium. In at least some
Attorney Docket No.246-021PCT exemplary embodiments, the computer readable medium includes code for executing one or more of the method actions detailed herein associated with the wavelet processing and the positive power development (and in some embodiments, the power maxima identification and/or the selection of scale and/or the activation timing) starting with the obtained first data based on one or unipolar EGMs recorded in an atrium. In at least some exemplary embodiments, the computer readable medium includes code for executing one or more of the method actions detailed herein associated with the wavelet processing and the application of the window functioning and the Fourier transformation and the reconstructing the atrial EGM. Any computer readable medium disclosed herein corresponds to a disclosure in the interests of textual economy of a medium that executes any one or all of the method actions associated with the wavelet processing and the wavelet filtering based embodiments disclosed herein unless otherwise noted, providing that the art enables such. [00382] Embodiments include a model based on results from or that is a product of machine learning to execute and/or otherwise develop any one or more of the results of any one or more of the method actions detailed herein associated with the wavelet filtering embodiments. Embodiments also include utilizing that model to develop or otherwise obtain the results of those method actions / execute any one or more of those method actions. Embodiments disclosed herein have briefly addressed this feature with respect to certain aspects of the teachings herein. It is noted that this does not indicate a limitation on the utilization of the machine learning techniques. In an embodiment, those results are developed or otherwise obtained and/or the method actions are executed utilizing by way of example, a neural network, such as a deep neural network, that is trained. Corollary to this is that at least some exemplary embodiments include training the machine learning system, such as training the neural network. Embodiments include executing one or more of the actions herein a certain number of times, such as at least and/or equal to 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 70, 80, 90, 100, 125 or 150 or any value or range of values therebetween in one increment manually, and providing the results thereof to a neural network to train the neural network. In an embodiment, the resulting product of the machine learning can be based on the just detailed actions. In an embodiment, the selection of the scale(s) to utilize with respect to the positive power can be an example where the initial selection of scales are developed based on the judgment of the skilled technician and/or the skilled healthcare professional, and those selections are provided to the neural network which upon training, provides for a product that automatically selects the scale.
Attorney Docket No.246-021PCT [00383] To be clear, in an exemplary embodiment, any action of analyzing or evaluating or calculating or identifying, etc. can correspond to an action of utilizing the results of machine learning to obtain the results thereof. By way of example only and not by way of limitation, in an exemplary embodiment, the atrial activation timing can be identified by the results of machine learning. Corollary to this is that any disclosure herein with respect to identifying or choosing, etc., something vs. something else corresponds to a disclosure of doing so utilizing the results of machine learning, and also a disclosure where such choices and/or identification are provided to a machine learning system to develop the results. An embodiment, the machine learning develops the algorithms utilized by internally identifying features, such as important features, from the data provided. [00384] In an exemplary embodiment, there is an exemplary method that includes analyzing any one or more of the data presented herein using a computer chip or a logic circuit or electronics or software to develop the resulting data. In an embodiment, the computer chip or logic circuit or electronics or software is based on a statistically significant population of persons who were afflicted with afib. [00385] Corollary to the above is that in an exemplary embodiment, the code detailed herein for the computer readable media is code that results from machine learning. In an exemplary embodiment, the teachings detailed herein associated with machine learning can be implemented to produce the code, or at least a portion of the code that is for executing one or more of the actions detailed herein. [00386] In view of the above, in at least some exemplary embodiments, there is a system that includes an input subsystem configured to receive input regarding information about potentials or based on potentials, and output subsystem that provides output, such as any one or more of the results of any one or more of the method actions detailed herein. In an embodiment, there is an artificial intelligence subsystem interposed between the input subsystem and the output subsystem, wherein the system is configured to determine with the use of the artificial intelligence subsystem, any one or of the results herein. In an embodiment, the artificial intelligence subsystem is a neural network that is at least partially taught. In an exemplary embodiment, it can be a fully taught neural network. Note also that embodiments include not only training, but retraining a neural network or otherwise continuously training the neural network. In this regard, the results of the neural networks decisions can be evaluated and ratified or corrected in this ratification and/or corrective data can be provided to the neural network for retraining purposes. In this regard, the neural network is a partially taught neural
Attorney Docket No.246-021PCT network that is trainable with feedback provided through, for example, the input subsystem or another subsystem of the system. Any one or more or all of the variables detailed herein can be identified by using a product of machine learning / NN / DNN. Embodiments include providing a sufficient number of examples and/or solutions to a NN / DNN or otherwise a machine learning algorithm or device and training such accordingly, which examples / solutions can include any one or more or all of the variables herein for a sufficient number of individual patients or otherwise individual examples to train the system and otherwise obtain a product that results therefrom, which can be used to identify one or more of the solutions and/or results presented herein. [00387] Corollary to all of this is that in embodiments where, for example, there is the data, a database can be utilized instead of or in addition to the artificial intelligence arrangements detailed above. Here, in an exemplary embodiment, straightforward algorithms can be utilized to arrive at certain results based on a statistically significant amount of prior data. In an exemplary embodiment, there are methods that include the action of consulting a database that may include at least 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 900, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 5000, 7000, 10,000 or more or any value or range of values therebetween in one increment set of data respectively associated with individuals suffering from atrial fibrillation, where this database includes any one or more of the data disclosed herein. By way of example only and not by way of limitation, for a given positive power plot over different scales, the database can be consulted to identify a statistically significant similar plot that was utilized, and thus provide the identification of the scale to be utilized. Here, a non-AI algorithm and/or an algorithm that is not based on the results of machine learning can be utilized. This is because the amount of data collected in the database is sufficient to utilize statistical techniques outside of the realm of artificial intelligence to select a given scale. In the interest of textual economy, this concept can be applied to any of the other selection / determination actions detailed herein unless otherwise noted, providing that the art enables such. This is the case with any of the machine learning teachings as well, unless otherwise noted, providing that the art enables such. Of course, embodiments include developing the actual statistical analysis algorithm or other algorithm to make the determination or otherwise to select a given scale. [00388] Embodiments include dedicated electronic circuitry configured to analyze any one or more of the data provided to the system detailed herein or otherwise any one or the data
Attorney Docket No.246-021PCT obtained that is disclosed herein that is needed or otherwise as a basis to move forward on any one or more of the method actions detailed herein. [00389] In an embodiment, there is a device and/or a system, comprising an input terminal including one of a monitor, keyboard and mouse combination or a server and/or a USB port or an electronic signal reception port, the input terminal being configured to receive input relating to any one or more of the base data detailed herein (e.g., the raw EKG signals, the processed EKG signals, etc.). The device and/or system includes dedicated electronics circuitry configured to analyze data based on input into the input terminal, wherein the electronics circuitry includes at least one of a model that is used by the device/system to automatically execute one or more of the method actions herein based on the input into the input suite. An embodiment includes a computer system, comprising an input subsystem configured to receive input regarding data based on EKG information, and an output subsystem and an artificial intelligence subsystem interposed between the input subsystem and the output subsystem, wherein the system is configured to execute one or more of the method actions detailed herein based on input into the input subsystem and output the results using the output subsystem. [00390] In an embodiment, there is a method, comprising obtaining heart phase data for a plurality of activation cycles of a living human afflicted with atrial fibrillation, and analyzing the heart phase data to identify specific heart tissue locations where there are repeated and consistent temporal discrepancies of electrical activation relative to other tissue locations, wherein the obtained heart phase data can be based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of near field atrial EGMs, and this embodiment can include obtaining such based on any one or more of the techniques described herein. [00391] In an embodiment, the action of analyzing the heart phase data includes implementing a statistical analysis on the heart phase data and/or the action of analyzing the heart phase data includes implementing time averaging analysis on the heart phase data and/or the action of analyzing the heart phase data includes: for a plurality of spatial locations on an interior surface of the heart, which spatial locations include the identified specific heart tissue locations, identifying respective maximum phase gradients for respective locations of the plurality of spatial locations over a length of time;
Attorney Docket No.246-021PCT time averaging the respective maximum phase gradients for the respective locations; and identifying corresponding locations where the time averaged results are statistically aberrant and/or are not statistically aberrant. [00392] In an embodiment, the identified corresponding locations of the time average results that are statistically aberrant are the identified specific heart tissue locations and/or the identified corresponding locations of the time average results that are not statistically aberrant are not the identified specific heart tissue locations and/or the respective maximum phase gradients are the respective maximum phase gradients between the respective locations and a plurality of proximate locations on the surface of the heart and/or the proximate locations are effectively North-South-East-West adjacent locations and/or wherein the proximate locations are the locations immediately surrounding the respective location. [00393] In an embodiment, any one or more of the method actions herein and/or methods include abating a surface of the heart based at least on the identified specific heart tissue locations and/or identifying the surface of the heart to be ablated based on at least the identified specific heart tissue locations. Note that any disclosure herein of ablating includes a corresponding disclosure of identifying what to ablate. [00394] In an embodiment, the plurality of spatial locations includes at least 24 locations; and there are at least 36 temporally spaced respective maximum phase gradients for the respective locations. [00395] In an embodiment, the method further comprises: obtaining respective plurality of temporally spaced electrical potentials for respective electrodes of at least 24 electrodes of a catheter located in a heart chamber at a first location in the heart chamber; converting the obtained respective plurality of temporally spaced electrical potentials to the heart phase data based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein, wherein the converting results in the obtention of heart phase data, wherein the actions of identifying respective maximum phase gradients, time averaging, identified corresponding locations where the time average results
Attorney Docket No.246-021PCT are statistically aberrant and/or not statistically aberrant, are based on the obtained respective plurality of temporally spaced electrical potentials for the catheter located at the first location. [00396] In an embodiment the method further includes: [00397] obtaining respective second plurality of temporally spaced electrical potentials for respective electrodes of at least 64 electrodes of the catheter located in the heart chamber at a second location in the heart chamber different from the first chamber; [00398] converting the obtained respective second plurality of temporally spaced electrical potentials to second heart phase data based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein, for a plurality of second spatial locations on the interior surface of the heart, which second spatial locations include the identified specific heart tissue locations, identifying respective second maximum phase gradients for respective second locations of the plurality of second spatial locations over second length of time; second time averaging the second respective maximum phase gradients for the second respective locations; and identifying corresponding second locations where the time averaged second results are statistically aberrant and/or not statistical aberrant, wherein the identified corresponding second locations of the second time average results that are statistically aberrant are included in the identified specific heart tissue locations and/or the identified corresponding second locations of the time second average results that are not statistically aberrant are not included in the identified specific heart tissue locations. [00399] In an embodiment, the action of obtaining heart phase data and analyzing is executed in real time vis-à-vis a catheter located in a heart chamber. [00400] In an embodiment the action of analyzing the heart phase data includes: for a plurality of spatial locations on an interior surface of the heart, which spatial locations include the identified specific heart tissue locations, identifying respective maximum phase gradients for respective locations of the plurality of spatial locations over a length of time; time averaging the respective maximum phase gradients for the respective locations; and
Attorney Docket No.246-021PCT identifying corresponding locations where the time averaged results are non- zero and/or statistically zero, wherein the identified corresponding locations of the time average results that are non-zero are the identified specific heart tissue locations and/or the identified corresponding locations of the time average results that are statistically zero are not the identified specific heart tissue locations. [00401] In an embodiment, there is a method, comprising: developing a time-varying electrical potential map of a surface of a cavity of a beating heart; developing a time-varying phase map of the surface of the cavity based on the developed time-varying electrical potential map; and identifying repeating phase signatures for respective locations on the surface of the atrial cavity from the time-varying phase map that repeat in a statistically aberrant manner relative to other phase signatures at other respective locations, wherein in an embodiment, the developed phase map and/or the developed potential map is based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein. [00402] In an embodiment, the electrical potential map has at least 500 electrical potential spatial locations and at least respective 5,000 temporal potential values for the respective potential spatial locations, the phase map has at least 400 phase spatial locations and at least respective 4,000 temporal phase values for respective phase locations. [00403] In an embodiment, the respective electrical potential locations of the at least 500 electrical potential locations have respective phase locations of the at least 500 phase locations. In an embodiment, the electrical potential map has at least 500 electrical potential locations and at least respective 20,000 temporal locations for respective potential locations of the 500 potential locations, the phase map has at least 500 phase locations and at least respective 20,000 temporal locations for respective phase locations of the 500 phase locations. In an embodiment, the actions of developing a time-varying electrical potential map, developing the time-varying phase map, and identifying the repeating phase signatures are executed within a period of no more than 20 minutes. In an embodiment, the action of developing a time-varying electrical potential map of a surface of a cavity of a beating heart is based at least in part on
Attorney Docket No.246-021PCT time-varying readings from electrodes located in the cavity; and the action of identifying the repeating phase signatures is executed within 20 minutes of the electrodes being removed from the chamber. [00404] In an embodiment, the action of developing a time-varying electrical potential map of a surface of a cavity of a beating heart is based at least in part on invasive readings taken while a human in which the beating heart resides is in an operating room; and the action of identifying the repeating phase signatures is executed before the human leaves the operating room. [00405] In an embodiment, the method further includes executing a medical procedure targeted at tissue of the heart corresponding to at least some of the respective locations identified as having the repeating phase signatures that repeat in the statistically aberrant manner before the human leaves the operating room. [00406] In an embodiment, the method further includes: after executing the medical procedure, developing second time-varying electrical potential map of the surface of the cavity of the beating heart; developing second time-varying phase map of the surface of the cavity based on the second developed time-varying electrical potential map; and evaluating whether and/or how many repeating phase signatures for respective locations on the surface of the atrial cavity from the second time-varying phase map that repeat in a statistically aberrant manner relative to other phase signatures at other respective locations, and based on the evaluation, evaluation whether the medical procedure was successful, wherein one or both of the obtained maps is / can be based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein. [00407] In an embodiment, the medical procedure is ablation of the targeted tissue. [00408] In an embodiment, there is a method, comprising: developing data including at least X spatial locations and at least Y respective phase gradients for the respective spatial locations of the X spatial locations; statistically analyzing the developed data; and
Attorney Docket No.246-021PCT identifying locations of the respective locations that are indicative of tissue influencing atrial fibrillation based on the statistical analysis, wherein X is at least 64 and Y is at least 50, and wherein the phase gradients is based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein. [00409] In an embodiment, the statistical analysis is time averaging and/or the action of identifying locations includes identifying locations where the statistical analysis of the developed data indicates non-random activation of respective heart tissue cells at the identified locations and/or the action of identifying locations includes identifying locations where averaging of the maximum phase gradients yields a statistically meaningful non-zero value. In an embodiment, the action of identifying locations includes identifying other locations where averaging of the maximum phase gradients of at least a majority of the Y phase gradients yields a statistically zero value. [00410] In an embodiment, the action of identifying locations includes identifying locations where averaging of the maximum phase gradients of at least a majority of the Y phase gradients yields a statistically meaningful non-zero value and the action of identifying the locations includes further statistically analyzing the values of the non-zero values. In an embodiment, X is at least 300 and Y is at least 75. In an embodiment, X is at least 1,000 and Y is between 60 and 1,000, inclusive. [00411] In an embodiment, the statistical analysis of the developed data identifies statistically consistent patterns of electrical activity that repeat in a statistically meaningful manner over time. [00412] In an embodiment, there is a non-transitory computer readable medium having recorded thereon, a computer program for executing at least a portion of a method, the computer program including: code for statistically analyzing first data based on phase gradients for at least 150 locations on a surface of a chamber of a human heart; and code for identifying a plurality of locations from the at least 150 locations, based on the statistical analysis of the first data, that should be targeted for treatment; and code for obtaining / developing data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the
Attorney Docket No.246-021PCT reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein, wherein such is used to develop the phrase gradients in some embodiments. [00413] The medium can further include code for transforming respective electrograms for respective locations of the at least 150 locations to a phase record including the phase gradients and/or code for creating the electrograms from data based on electrical potentials obtained from electrodes within the human heart, the number of electrodes within the human heart being less than 150. In an embodiment, the code for creating the electrograms uses inverse solution methods and/or the code for statistically analyzing the first data time averages respective maximum phase gradients for the at least 150 locations and/or the code for statistically analyzing the first data time averages respective maximum phase gradients for respective locations of the at least 150 locations and/or the code for identifying the plurality of locations from the at least 150 location identifies respective locations where time averages of the respective maximum phase gradients are statistically significantly non-zero. In an embodiment, the at least 150 locations includes at least 2,000 locations, respective locations of the at least 2,000 locations have at least 1000 respective maximum phase gradients and/or the medium creates the electrograms and identifies the plurality of locations from the at least 2,000 locations for the at least 1000 respective maximum phase gradients within 20 minutes when run on a Dell ™ laptop with an Intel Core i9 Microprocessor with at least a 2.8 GHz clock frequency, at least 16 by 1024 KB L2 cache, at least 22.00 MB L3 cash, a TDP of at least 160 W, a DMI 3.0 I/O bus and a 4 x DDR4-2666 memory. In an embodiment, the code for creating the electrograms uses at least 500 measurements from each electrode per second. [00414] In an embodiment, there is code for developing data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein and this is done within 30, 29, 28, 27, 26, 25, 24, 23, 22, 21, 20, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes when run on a Dell ™ laptop with an Intel Core i9 Microprocessor with at least a 2.8 GHz clock frequency, at least 16 by 1024 KB L2 cache, at least 22.00 MB L3 cash, a TDP of at least 160 W, a DMI 3.0 I/O bus and a 4 x DDR4-2666 memory. This code can execute any one or more or all of the method actions associated with the V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM disclosed herein.
Attorney Docket No.246-021PCT [00415] Embodiments can include a system and/or simply an embodiment that includes a non- transitory computer readable medium having recorded thereon, a computer program for executing at least a portion of a method, the computer program including code for executing any one or more of the method actions and/or functionalities detailed herein. Thus, any disclosure herein of a method action or functionality corresponds to a disclosure of a non- transitory computer readable medium having programed thereon code to execute one or more of those actions and also a product to execute one or more of those actions. [00416] Embodiments include any functionality disclosed herein and/or method action disclosed herein being executed by a computer chip, a processor, software, logic circuitry and/or electronics, and all are not mutually exclusive. Any circuit that can enable the teachings herein can be used providing that the art enables such. Thus, in the interests of textual economy, and disclosure herein of a functionality of an article of manufacture corresponds to any one or more of the aforementioned structures being configured to execute such and otherwise for such, and the same is for any method action disclosed herein, where any such action corresponds to a disclosure of any one or more of the aforementioned structures being configured to execute such and otherwise for such. [00417] In some embodiments, a neural network, such as a DNN, is used to directly interface to the input into the systems / devices detailed above, and process this input via its neural net, and determine the information detailed above. The network can be, in some embodiments, either a standard pre-trained network where weights have been previously determined (e.g., optimized) and loaded onto the network, or alternatively, the network can be initially a standard network, but is then trained to improve specific recipient results based on outcome oriented reinforcement learning techniques. [00418] Any disclosure herein of a processor corresponds to a disclosure in an embodiment of a non-processor device or a combined processor-non-processor device where the non-processor is a result of machine learning. Embodiments can include a link from the cloud to a clinic to pass information back and forth, enabling the remote processing noted above and/or enabling the obtaining of additional data for retraining purposes. Information can be uploaded to the cloud to the clinic, where the information can be analyzed. Another exemplary system includes a smart device, such as a smart phone or tablet, etc., that is running a purpose built application to implement some of the teachings detailed herein. This can be used by the clinician, and can contain at least the front end portions of the systems and devices detailed herein, or otherwise provide the interface portal to the back end. Any disclosure herein of a processor corresponds
Attorney Docket No.246-021PCT to a disclosure of a non-processing device, or includes non-processing devices, such as a chip or the like that is a result of a machine learning algorithm or machine learning system, etc. [00419] In an exemplary embodiment, the smart device can be configured to present the windows for the interface that will be used by the user. [00420] Any method action and/or functionality disclosed herein where the art enables such corresponds to a disclosure of a code from a machine learning algorithm and/or a code of a machine learning algorithm and/or a product of machine learning for execution of such. Still as noted above, in an exemplary embodiment, the code need not necessarily be from a machine learning algorithm, and in some embodiments, the code is not from a machine learning algorithm or the like. That is, in some embodiments, the code results from traditional programming. Still, in this regard, the code can correspond to a trained neural network. In an embodiment, the trained neural network can be utilized to provide (or extract therefrom) an algorithm that can be utilized separately from the trainable neural network. In one embodiment, there is a path of training that constitutes a machine learning algorithm starting off untrained, and then the machine learning algorithm is trained and “graduates,” or matures into a usable code – code of trained machine learning algorithm. With respect to another path, the code from a trained machine learning algorithm is the “offspring” of the trained machine learning algorithm (or some variant thereof, or predecessor thereof), which could be considered a mutant offspring or a clone thereof. That is, with respect to this second path, in at least some exemplary embodiments, the features of the machine learning algorithm that enabled the machine learning algorithm to learn may not be utilized in the practice some of the method actions, and thus are not present the ultimate system. Instead, only the resulting product of the learning is used. [00421] And to be clear, in an exemplary embodiment, there are products of machine learning algorithms (e.g., the code from the trained machine learning algorithm) that are included in any one or more of the systems / subsystems detailed herein, that can be utilized to analyze any of the data obtained or otherwise available disclosed above that can be utilized or otherwise is utilized to evaluate the data obtained herein. This can be embodied in software code and/or in computer chip(s) that are included in the system(s). [00422] An exemplary system includes an exemplary device / devices that can enable the teachings detailed herein, which in at least some embodiments can utilize automation. That is, an exemplary embodiment includes executing one or more or all of the methods and/or
Attorney Docket No.246-021PCT functionalities detailed herein and variations thereof, at least in part, in an automated or semiautomated manner using any of the teachings herein. Conversely, embodiments include devices and/or systems and/or methods where automation is specifically prohibited, either by lack of enablement of an automated feature or the complete absence of such capability in the first instance. [00423] Automated actions can be executed by an algorithm where circuitry receives the input (embodied in an analogue or a digital signal), where the input suite converts the “physical” input into electronic signals using analog to digital converters for example, or in the case of the input suite corresponding to an Internet server, receives the digital signal from a remote location, and the digital data is stored in a memory and/or received by the electronics. The electronics, which is a result of the machine learning by way of example, takes the digital signal and deconstructs the digital signal to evaluate properties, and then, using its “knowledge” from its training, provides an output based on the knowledge. [00424] Database herein can be a database such as Microsoft ™ Access, where the computer automatically matches the data instead of the human matching the data. The results of machine learning and/or a product thereof can be used to perform the automatic matching. [00425] In an exemplary embodiment, the some or all of the teachings herein are implanted on a computer chip and/or a computer circuit. There are comparators based on big data noted above. In an exemplary embodiment, comparison can be represented by an algorithm where circuitry receives the input (embodied in an analogue or a digital signal), where the input suite converts the “physical” input into electronic signals using analog to digital converters for example, or in the case of the input suite corresponding to an Internet server, receives the digital signal from a remote location, and the digital data is stored in a memory and/or received by the electronics. The electronics takes the digital data and “looks” for certain strings of zeros and ones that correspond to a match with signatures / identifiers linked to prestored data regarding performance capabilities. The data linked to the signatures / identifiers is the cohort identified. [00426] It is further noted that any disclosure of a device and/or system detailed herein also corresponds to a disclosure of otherwise providing that device and/or system and/or utilizing that device and/or system. [00427] The systems detailed herein can be configured to transform input into numerical form, and the artificial intelligence subsystem can be configured to, using the numerical form, produce an estimated outcomes measure, produce one or more of the results detailed herein.
Attorney Docket No.246-021PCT More specifically, in an embodiment, the system or device is configured to automatically transform input into numerical form. This can be executed using a computer chip or a logic circuit or electronics or software or a processor, that is programmed to take the input and transform the input. In some embodiments, the input subsystem is configured to execute this functionality. Thus, the input subsystem can be more than just a mouse and computer screen and keyboard, etc. Embodiments include an input subsystem that includes a processor and/or software and/or firmware and/or hardware and/or a computer chip or a logic circuit otherwise electronics that is specifically designed and configured to execute one or more of the functionalities of the input subsystem detailed herein. In an embodiment, the artificial intelligence subsystem is configured to, using the numerical form, automatically produce results based on this input, as transformed. [00428] As can be seen, embodiments of the output subsystem can be more than just a “dumb” computer screen or the like. Embodiments include an output subsystem that includes a processor and/or software and/or firmware and/or hardware and/or a computer chip (herein a computer chip also corresponds to a plurality of such, interconnected with a motherboard, etc.) or otherwise electronics that is specifically designed and configured to execute one or more of the functionalities of the output subsystem detailed herein. Any disclosure herein of software corresponds to an alternate disclosure of a computer chip or a logic circuit or electronics. [00429] It is also noted that any disclosure herein of any process of manufacturing or providing a device corresponds to a disclosure of a device and/or system that results therefrom. Is also noted that any disclosure herein of any device and/or system corresponds to a disclosure of a method of producing or otherwise providing or otherwise making such. [00430] An exemplary system includes an exemplary device / devices that can enable the teachings detailed herein, which in at least some embodiments can utilize automation, as will now be described in the context of an automated system. That is, an exemplary embodiment includes executing one or more or all of the methods detailed herein and variations thereof, at least in part, in an automated or semiautomated manner using any of the teachings herein. [00431] Any embodiment or any feature disclosed herein can be combined with any one or more or other embodiments and/or other features disclosed herein, unless explicitly indicated and/or unless the art does not enable such. Any embodiment or any feature disclosed herein can be explicitly excluded from use with any one or more other embodiments and/or other features
Attorney Docket No.246-021PCT disclosed herein, unless explicitly indicated that such is combined and/or unless the art does not enable such exclusion. [00432] Any function or method action detailed herein corresponds to a disclosure of doing so an automated or semi-automated manner. [00433] It is further noted that any disclosure of a device and/or system detailed herein also corresponds to a disclosure of otherwise providing that device and/or system and/or utilizing that device and/or system. [00434] It is also noted that any disclosure herein of any process of manufacturing other providing a device corresponds to a disclosure of a device and/or system that results there from. Is also noted that any disclosure herein of any device and/or system corresponds to a disclosure of a method of producing or otherwise providing or otherwise making such. [00435] It is further noted that any disclosure of a device and/or system detailed herein also corresponds to a disclosure of otherwise providing that device and/or system and/or utilizing that device and/or system. [00436] It is also noted that any disclosure herein of any process of manufacturing other providing a device corresponds to a disclosure of a device and/or system that results there from. Is also noted that any disclosure herein of any device and/or system corresponds to a disclosure of a method of producing or otherwise providing or otherwise making such. [00437] Any function or method action detailed herein corresponds to a disclosure of doing so an automated or semi-automated manner. [00438] While various embodiments of the present invention have been described above, it should be understood that they have been presented by way of example only, and not limitation. It will be apparent to persons skilled in the relevant art that various changes in form and detail can be made therein without departing from the spirit and scope of the invention.
where ri is the Euclidean distance between (x,y,z) and the source at (xi,yi,zi). [00118] That is ri=√((x−xi)2+(y−yi)2+(z−zi)2) [00119] The total potential Φ at any point in Ω contributed by a set of M fictitious sources is
[00120] Source densities βi are selected to match potentials recorded at each of the electrodes on the catheter 4 in a utilitarian manner, but also potentials inside the catheter 5 estimated by solving the forward problem if desired. Source densities are determined by solving inverse
Attorney Docket No.246-021PCT problem with appropriate regularization and endocardial surface potentials are then mapped using equation 5. [00121] While optimal placement of fictitious sources is empirical to some extent, a number of relatively straightforward a priori rules apply. The number of independent fictitious sources is less than the degrees of freedom of the measured potential distributions as determined by singular value analysis. The distribution of sources on the fictitious boundary should be bound by this constraint and it should be adaptive—most dense where the measurements are closest to the cardiac surface and least dense where they are furthest from it. Where sources are sparse, the accuracy of inverse mapping is increased with displacement of the surface from the cardiac boundary and vice versa. Finally the displacement of the boundary should be sufficient to accommodate uncertainty in the representation of cardiac surface geometry. [00122] FIG. 24b represents a method of distributing fictitious sources that is consistent with these rules. The number of independent sources 11 is equal to the number of electrodes 3 on the catheter and their spacing reflects the position of the catheter with respect to the wall. It is possible to add additional sources without loss of generality by interpolating along the fictitious boundary between independent source points. [00123] The use of meshless methods in the setting above gives rise to computationally efficient inverse solutions that are marginally less accurate than BEM, but more robust in the presence of uncertainty about endocardial geometry and the relative position of measurement sites on the catheter with respect to the endocardial surface. Finally, the positioning of independent sources on the external boundaries provides direct feedback on the spatial resolution of potentials that are mapped onto the endocardial surface. [00124] By way of example only and not by way limitation, data techniques can be used to reconstruct electrical potentials (voltages) on the inner surface of the atrial cavity from signals recorded at the individual electrodes on a basket catheter where some or all electrodes are not touching the atrial surface. The 3D geometry of the inner surface of the heart chamber (such as the left atria, with reference to FIG.11) which can be specified and the positions of each of the electrodes with respect to this surface are also known. This information is information that can be obtained by using, for example, electrical mapping systems used to guide ablation in clinical electrophysiology laboratories, such as those at the Royal Melbourne Hospital on January 10, 2022.
Attorney Docket No.246-021PCT [00125] In an exemplary scenario, where a 64 channel catheter (ConstellationTM, Boston Scientific) was introduced into the left atrium (LA) by transeptal puncture of a person afflicted with permanent atrial fibrillation, method 730 can include the action of using software to execute inverse mapping based on the electrode readings. Method 730 can be executed in real- time while the basket catheter is within the atrial cavity, at least in some embodiments. [00126] FIGs. 11-14 show by image presentation data from a result of the merging of data obtained using an inverse mapping technique (e.g., via a meshless method such as that detailed above) with data from CT images for an exemplary left atrium (posterior on the left, anterior on the right), where the figures show potential values at instantaneous points in time at locations on the surface of the left atria, and collectively show changes with some exemplary time progressions at the various locations by way of example only. More specifically, the electrode potential values corresponding to the time periods (where the monitoring device operated at 2,000 Hz – collecting readings two-thousand times per second) for the respective electrodes was, using inverse mapping, used to obtain electrical potentials on the surface of the left atria (visually represented in FIGs.11-14). By way of example, the data from 64 electrodes can be used to develop potential values for 64, 100, 150, 200, 250, 300, 350, 400, 450, 500, 600, 700, 800, 900, 1000, 1250, 1500, 1750, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 6000, 7000, 8000 or more, or any value or range of values therebetween in 1 increment locations on the surface (or surface region – again, embodiments can include utilizing the catheter with a “reduced” total volume, where the electrodes are closer together relative to that which would otherwise be the case, to obtain electrical potentials at a sublocation or a subregion within the cavity, to obtain more accurate data, and then the catheter can be moved to another subregion in the process continued), and these developed potentials being present for respective time increments at the cycle of the monitoring system (here, 2000 Hz). FIGs.11-14 show the time varying potentials on the surface, or more accurately, at the various surface locations. [00127] But to be clear, embodiments need not provide this exemplary any exemplary imaging. It can be sufficient to simply obtain time varying data sets for the potentials at the locations on the surface of the cavity. [00128] It is briefly noted that there can be a point of diminishing returns with respect to the number of locations used in the inverse mapping technique. In this regard, the potentials on the surface that are calculated are limited in accuracy vis-à-vis the number of electrodes utilized in a given catheter. Thus, utilitarian value can potentially be obtained by utilizing less than the full number of locations in the inverse mapping the net which a given system could potentially
Attorney Docket No.246-021PCT develop otherwise implement. This can have utilitarian value with respect to improving computational times. That said, in some embodiments, there could be potentially more locations than those detailed above. Any number of locations resulting from the inverse mapping (or forward mapping for that matter in embodiments that utilize such) that can enable the teachings detailed herein can be utilized in at least some exemplary embodiments. Corollary to this is that any cycle time and/or total length of time that can enable the teachings detailed herein can be utilized in at least some exemplary embodiments providing that the art enable such. [00129] The present inventors have found that mere potential mapping, while utilitarian in some instances, can be better utilized (or more specifically the data of the mapping can be better utilized), at least in some embodiments, by for example developing heart tissue cell phase data therefrom. In this regard, the potential maps are not necessarily useful in the context of the exacting nature that is required to identify heart tissue that is a driver or a substrate of atrial fibrillation, or otherwise identify tissue that can be ablated to alleviate at least some of the effects of atrial fibrillation. Indeed, the sampling of images shown in FIGs.11-14 show clearly that the potential mapping result exhibits the chaotic activity that is typical in atrial fibrillation. The point is that it is very difficult to extrapolate what is occurring using even exacting time increment potential mapping, and on a per patient basis, typically less utilitarian than the teachings herein vis-à-vis tissue identification. [00130] Thus, embodiments include extracting useful information, or more useful information from the electrical signals or electrograms recoded at the electrodes and/or from the potential map developed for the surface from those electrical signals. [00131] And this leads to method action 740, which includes executing phase mapping on the results of the potential mapping. Again, as noted above, at least some of the techniques detailed in the Journal article published in IEEE Transactions On Biomedical Engineering, Vol.62, No. 1, JANUARY 2015, by Pawel Kuklik et al, entitled Reconstruction of Instantaneous Phase of Unipolar Atrial Contact Electrogram Using a Concept of Sinusoidal Recomposition and Hilbert Transform, are used to develop the phase data. But other method actions can be used. Any method that can enable phase data to be developed that is utilitarian can be used in some embodiments. [00132] Briefly, a Hilbert transform, which can be used to transform data from the potential realm to the phase realm, is a linear operator transforming a function u(t) into a function H(u)(t)
Attorney Docket No.246-021PCT
where ^^^ ^^^^^ are the wavelet coefficients, while ^^^^ and ^^^^ are time shift and time scale dilation parameters, respectively. Ψ ∗( ^^^^ ) is the complex conjugate of the decomposition function Ψ ( ^^^^ ) also referred to as the mother wavelet. In this embodiment, wavelets will be calculated at 10 different time scales with first and second derivative Gaussian mother wavelets using the Fast Fourier transform (FFT). In an embodiment, the wavelets can be calculated at 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, or more (or less than), or any value or range of values therebetween in 1 increments different scales. Any number of scales that can have utilitarian value can be used in some embodiments providing that the art enables such. That said, embodiments include executing one or more of the actions / obtaining one or more results herein in real time or near real time, so there are limits on the scales in some embodiments, which can be in some instances, correlated to the processing power of the computer systems used to implement some of the teachings herein.
Attorney Docket No.246-021PCT [00229] Atrial EGMs can be reconstructed following "filtering" across time scales in CWTs calculated with second derivative Gaussian mother wavelets (sometimes, only), which can be, in some embodiments, evaluated as ^^^^ = 1 10 +∞ 1 ^ � ^^^ ^^^^ ^^^^ ^^^^ ^^^^ ^^^^− ^^^^ ^^^^ ^^^^
where ^^^^( ^^^^) is t reconstructed atrial EGM, ^^^^ ^^^^( ^^^^, ^^^^ ^^^^) are "filtered" wavelet coefficients, ^^^^ ^^^^ represents the separate time scales that can be / are considered in some embodiments and the constant C can ensure or otherwise drive the implementation towards energy conservation. [00230] As noted above, in some embodiments, ventricular far-field artifact subtraction can be achieved. We now present an exemplary embodiment of wavelet-based subtraction of ventricular far-field artifact. For example, far-field artifact generated in individual atrial EGM channels by ventricular activation and repolarization can be estimated beat-to-beat and subtracted using the workflow summarized in Fig.25. FIG.25 shows an exemplary processing pipeline in which estimated far-field artifact in each QT window is subtracted for each wavelet scale during QT window and inverse wavelet transformation is then used to recover. [00231] Initially, by way of example, as shown in FIG.26A, we start with exemplary raw atrial EGMs for a channel of the basket catheter where ventricular (V) activation timing is identified from coronary sinus (CS) EGMs is indicated by the broken red lines superimposed on this figure. This is by example representative unprocessed atrial EGMs, where V activation is indicated by broken red lines. In this regard, a 10 second period from a representative unprocessed atrial EGM is presented in Fig.26A while the continuous wavelet decomposition over the first 3 seconds of this signal is shown in Fig.26B. That is, FIG.26B shows wavelet decomposition of initial segment of (26A) with a 2nd order Gaussian wavelet in 10 levels. Wavelet coefficients in a specific QT window (indicated in blue – this is a window encompassing a V activation) are enlarged in Fig. 26C. Put another way, FIG.26C shows a magnified wavelet decomposition in QT window. That is, the grey shading of FIG.26A is an example of the temporal length of one QT window. [00232] In an embodiment, for one or more or all of the channels of the catheter, a CWT is computed across time windows from -100 to +400 ms by way of example with respect to respective ventricular activation times (hereafter referred to as QT windows) using a 2nd order Gaussian mother wavelet at for example 10 time scales. In an embodiment, the base raw signal extends (in some embodiments contiguously) for less than, greater than, and/or equal to 20, 21,
Attorney Docket No.246-021PCT 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, or 900 ms or more, or any value or range of values therebetween in 1 ms increments. In an embodiment, the CWT can be computed across time windows from plus and/or minus less than and/or greater than and/or equal to 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 325, 350, 375, 400, 425, 450, 475, 500, 525, 550, 575, 600, 625, 650, 675, 700, 725, 750, 775, 800 ms, or more, or any value or range of values therebetween in 1 ms increments from the V activation timing (and the minus need not be the same absolute value as the positive, and the numbers need not be the same in each window, although in other embodiments, the numbers will be the same in each window). In an embodiment, the CWT can be computed continuously and/or for time windows that form a contiguous set of time windows for the given data from the channel, where in some embodiments, each window has a V activation. In an embodiment, there are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, or more, or any value or range of values therebetween in 1 increment time windows. [00233] In an embodiment, ventricular far-field artifact is estimated in each QT window from wavelet components time-averaged across repeated QT windows. Fig. 26D shows an exemplary result of such (right column). That is, far-field artifact in QT window estimated from ensemble averages of wavelet coefficients across successive QT windows are shown in the right column. The left column shows the data of the right column scaled to match current V amplitude. [00234] In an embodiment, because these windows are synchronized with V activation, averaging identifies mean wavelet coefficients associated with the V artifact and reduces atrial contributions in AF effectively to zero (which includes to zero in some embodiments). This can be because the atrial contributions are not correlated with V activation. Embodiments include accounting for beat-to-beat artifact variation during V activation by scaling ensemble- averaged wavelet components to match recorded V amplitude (the left column of FIG.26D). In an embodiment, the results of scaling are tapered such with, for example, Hanning windows HWi(τ) centered on V activation time for example. In an embodiment, the windows are centered with a maximum value of 1.0 and widths from 15 to 150 msec for wavelet scales i = 1,10. In an embodiment, the windows have widths of less than, greater than, and/or equal to 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55,
Attorney Docket No.246-021PCT 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 275, 300, 325, 350, 375, 400, 425, 450 or 500ms, or more (or less potentially), or any value or range of values therebetween in 1 increment, and as can be seen, can be different for the different wavelet scales. The scaled components can be added to fixed ensemble-averaged wavelet components tapered by complementary windows 1- HWi(τ) to estimate the wavelet decomposition of each V artifact. (This provides the data outside the window for the remainder of the time average. That is, the white / non-greyed areas are combined to create the Estimated V artifact in the QT windows. [00235] Next, in an embodiment, this wavelet decomposition of estimated far-field artifact can be subtracted from the CWT of the raw EGM and an inverse transform can be performed. As shown in Fig. 27, this processing approach extracts atrial activation complexes that overlap high amplitude ventricular artifact. Note that the start and finish of the QT interval could also be varied in a patient-specific manner if such is utilitarian. [00236] It is noted that in some embodiments, there are no set constraints, at least in general, to the implementations of the teachings herein with respect to the aforementioned scales and/or timings. In an exemplary embodiment, the system that implements the teachings herein can for example compute CWTs at five scales, for example, and if this is not sufficient, compute additional scales. The window widths could start with -75 ms to +250 ms relative to the V activation time, and if this is not sufficient, the windows can be expanded. This is all by way of example. The point is that in an embodiment, an adaptive arrangement can be implemented where more limited number crunching is initially executed, and if more is utilitarian or otherwise if more is needed, more is done in an adaptive manner. That said, after a sufficient number of implementations on different patients, a historical database can be developed to identify scales and timings that provide utilitarian value over a statistically significant number of patients. In this regard, in an exemplary embodiment, a set of data constraints can be utilized for different patients having common attributes where the constraints have shown to be utilitarian in past implementations. It is found that these constraints do not provide suitable data for a given patient, the constraints can be changed. Accordingly, computational efficiency can be implemented in most instances, and for the occasional occurrence where those predetermined constraints do not provide utilitarian value, the constraints can be expanded at the potential cost of computational efficiency. Put another way, for example, if 9 out of 10 implementations are as efficient as possible, the 10th implementation could be relatively inefficient (because perhaps the calculations would have to be redone, perhaps completely from
Attorney Docket No.246-021PCT the beginning, at the greater scales and/or longer time frames) and that would be acceptable overall. [00237] Embodiments use averaging to identify the mean wavelet coefficients associated with the V artifact and to reduce atrial contributions to zero (at least sufficiently so). In an embodiment, as a result, the atrial contributions on a magnitude basis that remain is no more than 8, 7, 6, 5, 4, 3, 2, 1, 0.75, 0.5, 0.4, 0.3, 0.2 or 0.1 % or any value or range of values therebetween in 0.01% increments that which was the case prior and/or relative to the V artifact. As noted above, the window widths can be different in some embodiments, and the widths can be different for different scales, as seen above. In an embodiment, the widths increase linearly with scale. In an embodiment, the widths increase non-linearly with scale. In an embodiment, the width of a given scale can be higher than a preceding scale by less than, greater than and/or equal to 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 175, 200, 250, 300, 350, 400, 500, 600, 700, 800, 900, 1000, 1250, 1500, 1750 or 2000% or more or any value or range of values therebetween in 0.1% increments. Thus, for example, scale 1 and scale 2 could have the same widths, or have different widths (and it could be that scale 2 has a more narrow width, however unlikely). In an embodiment, scale 1 could have a width of 25 ms and scale 2 could have a width of 40 ms (60% greater) and scale 3 could have a width of 60 ms (140% greater than 1, and 50% greater than 2). The windows can have any of the above-noted timings. [00238] And note while Hann / Hanning (sometimes called Hamming) windows are used here, other types of windows / functions can be used. A Parzen window can be used in some embodiments. A Blackman window can be used in some embodiments. A Nuttal window can be used in some embodiments. A Blackman-Nuttall window can be used in some embodiments. A Welch window can be used in some embodiments. A Blackman-Harris window can be used in some embodiments. A sine window can be used in some embodiments. A Cosine-sum windows could be used. Gaussian and/or Rife-Vincent windows can be used in some embodiments. B-spline window can be used in some embodiments. Confined Gaussian windows and/or Approximate confined Gaussian windows can be used in some embodiments. A Tukey window can be used in some embodiments. A Generalized normal window can be used in some embodiments. A Planck-taper window can be used in some embodiments. A DPSS window can be used in some embodiments. A Slepian window can be used in some embodiments. In an embodiment, a function that provides / uses a smooth taper is used.
Attorney Docket No.246-021PCT [00239] A Kaiser window can be used in some embodiments. A Dolph-Chebyshev window can be used in some embodiments. An Ultraspherical window can be used in some embodiments. [00240] Embodiments also include using hybrid windows, such as for example in some embodiments, a Barlett-Hann window, a Planck-Bessel window, a generalized adaptive polynomial (GAP) window and/or Lanczos window could be used. [00241] In view of the above, any window functioning an/or apodization function and/or tapering function used herein can be based on one or more of the just noted window arrangements (by “based on,” it is meant that it does not have to be exactly that window, but can be based on that window, so one might take the results of such and then modify those results, such as by applying another window function thereto and/or by adjusting a portion thereof (such as the portions that are a certain percentage of time away from the V activation time on one or both sides of the function, etc.). [00242] In an embodiment, the product of a trained neural network, trained on a sufficient number of examples, can be used to select the window function and/or the parameters of the window function, as detailed herein (e.g., window length). [00243] In an embodiment, the complementary of any given window (in simple terms, the 1- the window function) can be used in a manner concomitant with the teachings above with the complementary windows 1- HWi(τ)), and the addition of the results can be obtained to estimate the wavelet decompositions of each V artifact so as to, for example, provide the data outside the window for the remainer of the time average or otherwise to combine the results to create an estimated V artifact for the QT windows, where this result is subtracted from the CWT for the raw EGM and an inverse transform is executed in accordance with the teachings above so as to extract the atrial activation complexes that overlap high amplitude ventricular artifact. [00244] In an exemplary embodiment, a plurality of window functions can be applied, and the most utilitarian window function can be selected from the results, or more accurately, the results that are most utilitarian can be selected for use. Note also that in an exemplary embodiment, different window functions that are utilized in the results are compared for empirical purposes, and the most utilitarian function is selected for use when applied to patients for treatment or otherwise diagnostic purposes in the clinical scenarios. In an exemplary embodiment, again as noted above, machine learning can be used to select the “best” window function and/or the parameters thereof (length of time for example). In an exemplary embodiment, different window functions could be utilized for different types of people based
Attorney Docket No.246-021PCT on demographics. In an exemplary embodiment, depending on the resulting wavelet transforms, machine learning, or more accurately, the results of machine learning, could be utilized to identify the window functions that will be utilized. That is, for example, in the clinical condition, upon the wavelet decomposition of the different scales, the results of a trained neural network by way of example could be utilized to analyze the results of the decompositions and select the window function that will be applied accordingly. And then of course the system could apply that window function in accordance with the teachings detailed herein. And note that in an exemplary embodiment, if an applied window function is deemed to results in less than utilitarian results, another window function can be used. There could be a primacy of window functions that could be used where if the first window function which is deemed the “go to” window function in normal circumstances is not providing sufficient results or otherwise providing results that may be less than utilitarian, the next window on the list could be utilized, and so on. [00245] In view of the above, with respect to FIG. 28, there is an exemplary method, method 1030, which includes method action 1032, which includes the action of obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation. In an exemplary embodiment, this can entail the entire procedure of placing the lumen catheter in the human and snaking such to the human’s heart, and then utilizing the electrodes on the catheter to record potentials at one or the electrodes. In an embodiment, this can entail obtaining data from a third party that is executing the recordings. Indeed, in an exemplary embodiment, this could be done remotely, where a surgeon or healthcare professional performs the actual physical operation / procedure on the human, and the actor of method 1030 is located in another country or another continent, where the first data is obtained via the Internet by way of example only and not by way of limitation. It could be obtained as an attachment to an email, or even could be obtained utilizing facsimile or some other regime. Any device system and/or method that can enable method action 1032 can be utilized in at least some exemplary embodiments. Conversely, in an embodiment, in a variation of method 1030, jumping ahead a bit, with reference to FIG. 28A, there is a method 1031, which includes method action 1033, instead of method action 1032, instead entails obtaining first data based on one or more electrical phenomena in an organ and/or a portion of an organ in a living human. In an exemplary embodiment, the first data is based on one or more unipolar EGMs which represent the electrical phenomena. In an embodiment, the first data is based on one or more unipolar EGMs which correspond to the electrical phenomena, the one or more unipolar EGMs
Attorney Docket No.246-021PCT being recorded in an atrium of the living human. In an embodiment, the electrical phenomena are phenomena in an atrium of the living human. The phenomena can be in another part of the living human. In an embodiment, the identified regional activation times are regional atrial activation times. The organ and/or a portion of an organ is a heart and/or a portion of a heart atrium. In the interests of textual economy, in an embodiment, method action 1032 is replaced with method action 1033, and the method is executed accordingly. [00246] As noted above, embodiments include applying some of or all of the teachings detailed herein some of the teachings detailed herein to a heart that afflicted with atrial fibrillation. The heart could be a perfectly healthy heart. But in other embodiments, the to be a heart with other issues, such as by way of example only and not by way of limitation, a heart with an ascending aorta. The heart could be a heart with arteries that are clogged by a certain amount, such as by way of example only and not by way of limitation, by at least 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, or 60% or more rainy value range by therebetween in 1% increments. [00247] In an embodiment, the electrical phenomenon is disorganized electrical activity, concomitant with the teachings above. But in an embodiment, the electrical phenomenon is organized electrical activity. [00248] By data based on one or more unipolar EGMs, this can be data based on one or more unipolar EGMs, this can be data directly from the channels of the basket catheter, or can be data from data from the channels of the basket catheter, etc. With respect to the former, this could be the raw signals from one or more electrodes of the basket catheter, and with respect to the latter, this could be processed signals and/or a data set that is developed from those signals. Indeed, in an exemplary embodiment, where the EGM’s are obtained at an earlier temporal location, such as a day or two or even a week or more before method 1030 is executed, the first data can correspond to a dataset where the original EGM data has been weighted for example, or subjected to smoothing or manipulation where clear extraneous data has been removed from the data set. [00249] It is briefly noted that by EGM, it is meant a recording of electrical activity over time. This does not require a graph format, although the EGM can certainly be in a graph format. Any data format from which monitored changes in electrical potential over time can be extracted can correspond to EGMs. In an embodiment, the first data includes EGMs for less than, greater than, and/or equal to 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, or 300, or more, or any
Attorney Docket No.246-021PCT value or range of values therebetween in 1 increment channels of an electrode assembly monitoring electrical activity of the heart (e.g., a basket catheter). In an embodiment, one or more or all of the EGMs span a period of time less than, greater than, and/or equal to 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 350, 375, 400, 450, 500, 550, or 600, or more, or any value or range of values therebetween in 1 increment seconds, and the EGMs need not be the same period of time as the others. [00250] In an embodiment, two or more time intervals can be utilized for the same scale, and the timing that has more utilitarian value can be utilized for further calculations. This might result in additional computational times, but if the computational system utilized is sufficiently advanced, this could be de minimis or otherwise this can be an acceptable delay on the overall processing. In an exemplary embodiment, two or three or four or five or six or seven or eight or nine or 10 or more time intervals are developed and the best time interval is utilized for further processing. In an embodiment, an average of the results can be utilized. The results can also be weighted. [00251] Any EGM data that can enable the teachings herein can be used in some embodiments. Here, in method 1030, the EGM must be a unipolar EGM. [00252] Method 1030 and method 1031 include method action 1034, which includes the action of obtaining second data based on wavelet processing of the obtained first data as part of a process to develop the second data. In an embodiment, this can entail executing wavelet processing of the obtained first data as part of a process to develop second data. This can be done according to any of the implementations detailed herein associated with wavelet processing. The actor executing method action 1034 can be the actor implementing the wavelet processing. Alternatively, and/or in addition to this, method action 1034 can be executed by receiving the second data, where the wavelet processing was executed remote from the actor executing method 1030. In an exemplary embodiment, the second data can be obtained over the Internet or by email, etc. Still, embodiments contemplate executing wavelet processing under the control of the actor implementing method 1030. In an exemplary embodiment, and one site computer, such as a mainframe computer in a hospital or the like, where the procedure is being implemented, can execute method action 1034. That said, in an alternate embodiment, a laptop computer with the appropriate software can be utilized to execute method action 1034. Accordingly, embodiments include non-transitory computer readable mediums that have code
Attorney Docket No.246-021PCT thereon for implementing or otherwise executing one or more of the method actions detailed herein. More on this below. [00253] In this exemplary embodiment, concomitant with the teachings above associated with FIGs.25-27, the second data is data indicative of a ventricle far field artifact obtained first data. In an embodiment, the ventricular far-field artifact is a result of ventricular activation and repolarization. [00254] In an exemplary embodiment, such as where the wavelet processing is executed by the actor of method 1030, the action of executing wavelet processing includes decomposing an individual unipolar atrial EGM of the obtained first data using CWT. Thus, in an exemplary embodiment, the wavelet processing includes decomposition of an individual unipolar atrial EGM of the obtained first data using CWT (again, the actor of method 1030 need not do the decomposing). [00255] In an embodiment, the wavelet processing includes calculating wavelets at three (3) to fifteen (15) different time scales (e.g., 10 different time scales) with second derivative Gaussian mother wavelets. Any number that can have utilitarian value can be used, as noted above. In an embodiment, the processing includes computing a CWT across a plurality of time windows from respective first times before respective beat-to-beat ventricular activation times to respective second times after beat-to-beat ventricular activation times. The timeframes can be as detailed above, and need not be the same for each time window, again as noted above. Corollary to this is that in an embodiment, there is the method 1040 of FIG.29, which includes method action 1042, which includes obtaining data based on an estimate of beat-to-beat ventricular activation times. Method 1040 also includes method action 1044, which includes the action of executing method 1030 or method 1031 (or executing method 1030 with method action 1033 instead of method action 1032), including executing the wavelet processing in this exemplary embodiment. In this exemplary method action, the action of executing wavelet processing includes computing a CWT across a plurality of time windows from respective first times before respective beat-to-beat ventricular activation times to respective second times after beat-to-beat ventricular activation times. [00256] Note that when implementing method 1040, method action 1042 can be executed while executing method action 1044, or at least while executing portions of method action 1044. In this regard, the first data based on one or more unipolar EGM’s recorded in an atrium of a living human can be obtained while also obtaining the estimate of the ventricular activation
Attorney Docket No.246-021PCT times. In an embodiment, the ventricular activation times can be obtained afterwards, depending on who and/or where the various parts of the method are executed. The point is, unless otherwise noted, there is no temporal sequence associated with the presentation of the method actions detailed herein providing that the art enables such. When there is an order that is to be implemented, such will be noted herein. [00257] In an embodiment, the second times are between and inclusive of 1 to 10 times the first times. For example, if the time before activation is 100 ms, the time after can be 100, 101, 102, 103, 103.1, 104, 110, 120, 150, 200, 250, 300, 777, or 1000 ms, etc., all by way of example. That said, some of these times are less than utilitarian. In an embodiment, the second times are less than, greater than, and/or equal to 1, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2, 2.1, 2.2, 2.3, 2.4, 2.5, 2.6, 2.7, 2.8, 2.9, 3, 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7, 3.8, 3.9, 4, 4.1, 4.2, 4.3, 4.4, 4.5.4.6.4.7, 4.8, 4.9, 5, 5.25, 5.5, 5.75, 6, 6.5, 7, 7.5, 8.5, 9, 9.5, 10, 11, or 12 times, or any value or range of values therebetween in 0.01 increments between and inclusive the first times. (In an embodiment, if the first time is treated as a negative value, the second time can be the absolute value of the first time times whatever multiplier is desired.) [00258] In an embodiment, the wavelet processing includes developing pluralities of respective wavelet coefficients at different scales for respective periods of time extending from before respective ventricular activation to after ventricular activation. In an embodiment, the pluralities amount to any of those noted above. In an embodiment, this amounts to less than, greater than, and/or equal to 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, or 100, or more, or any value or range of values therebetween in 1 increment, where the different scales can be the same or different for respective periods of time. In an embodiment, the pluralities are for each ventricular activation within a timeframe of less than, greater than, and/or equal to 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 125, 150, 175, 200, 250, or 300 seconds, or more, or any value or range of values therebetween in 1 second increment. In an embodiment, there are coefficients for each time period associated with each activation within the time just noted. In an embodiment, it can be every other activation, or every three or every four, etc. In an embodiment, less than, greater than (limited by the math – we cannot have greater than 100%) and/or equal to 100, 99, 98, 97, 96, 95, 94, 93, 92, 91,
Attorney Docket No.246-021PCT 90, 85, 80, 75, 70, 65, 60, 55, 50, 45, 40, 35 or 30% or any value or range of values therebetween in 1% increments of the times associated with ventricular activations can have the coefficients. [00259] In an embodiment, the obtained second data is based on respective time-averaged wavelet coefficients for the respective different scales. This, by way of example, as explained above with respect to figure 26D. In an embodiment, the wavelet processing includes developing CWTs of the first data and the second data is based on time averaging of respective portions of the developed CWTs and implementing a window function on the time averaged respective portions. In an embodiment where the actor of method 1030 executes the processing, the action of executing wavelet processing includes developing CWTs of the first data and the method includes time averaging respective portions of the developed CWTs and implementing a window function on the time averaged respective portions. [00260] Note that while the time averaging and the developments of the pluralities of respective wavelet coefficients at different scales was just presented in terms of something that existed within the obtained second data, in an embodiment, the actor of method 1030 executes the wavelet processing, and as part of that wavelet processing, the actor executes the action of developing respective pluralities of wavelet coefficients at different scales as noted above. Further, in an exemplary embodiment, the method further includes as part of the process to develop second data, time averaging the respective pluralities of wavelet coefficients to develop respective time-averaged wavelet coefficients for the respective different scales, wherein the second data is based on the developed respective time-averaged wavelet coefficients. Accordingly, in the interests of completeness, figure 30 shows an exemplary method, method 1050, which includes method action 1052, which includes executing method 1030 or method 1031 (or executing method 1030 with method action 1033 instead of method action 1032), including executing the wavelet processing. The reader is referred to the statements in method 1050 which correspond to those just detailed, albeit presented in an algorithmic format. [00261] Again, consistent with the teachings above, in an embodiment, the time-averaging identifies mean wavelet coefficients associated with the ventricular artifact and reduces atrial contributions in atrial fibrillation to at least effectively zero, which includes zero. In an embodiment, the contributions are reduced by at least and/or equal to 20, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4.5, 4, 3.5, 3, 2.5, 2, 1.9., 1.8, 1.7, 1.6, 1.5, 1.4, 1.3, 1.2, 1.1, 1.0, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.09, 0.08, 0.07, 0.06, 0.05, 0.04, 0.03, 0.02, 0.01%,
Attorney Docket No.246-021PCT or less, or any value or range of values therebetween in 0.005% increments of the amount that existed before the reduction. [00262] In an embodiment, any one or more of the methods herein, such as method 1050, further includes as part of the process to develop second data: (1) scaling ensemble time-averaged wavelet components of the developed time-averaged wavelet coefficients for the respective different scales; (2) tapering the scaled components at least towards zero value outside an interval relative to the ventricular activation time for the respective different scales; (3) tapering fixed ensemble time-averaged wavelet components towards zero value inside an interval relative to the ventricular activation time for the respective different scales; and (4) adding the tapered scaled components to the tapered fixed components to obtain an estimate of a wavelet decomposition of each ventricular artifact. [00263] Accordingly, in an embodiment, the second data is based on an estimate of a wavelet decomposition of each ventricular artifact; the estimate is based on scaled ensemble time- averaged wavelet components of the respective time-averaged wavelet coefficients for the respective different scales; the scaled components are tapered at least towards zero value outside an interval relative to the ventricular activation time for the respective different scales; the fixed ensemble time-averaged wavelet components are tapered towards zero value inside an interval relative to the ventricular activation time for the respective different scales; and the tapered scaled components are added to the tapered fixed components, resulting in an estimate of a wavelet decomposition of each ventricular artifact. [00264] In embodiments, as noted above, the scaling scales to 1 (the maximum amplitude is 1, and all other amplitudes are scaled based thereon). That said, embodiments can utilize different scales providing that such has utilitarian value. In an embodiment, the scaling is executed to match recorded V amplitude. While tapering with Hanning windows centered on the activation time was presented in the embodiments above, other window functions that can enable the teachings detailed herein that have utilitarian value can be utilized. As noted above, the mathematical function of the window function is zero valued outside of the above-noted time intervals, which are different for different scales in the embodiment noted above, with increasing time interval for the coarser scales, and is symmetric about the V activation time. In an embodiment, the time intervals need not be different for every scale and/or need not increase or otherwise change according to a linear trajectory with increasing scale. In an embodiment, two or more time intervals can be utilized for the same scale, and the timing that has more utilitarian value can be utilized for further calculations. This might result in additional
Attorney Docket No.246-021PCT computational times, but if the computational system utilized is sufficiently advanced, this could be de minimis or otherwise this can be an acceptable delay on the overall processing. In an exemplary embodiment, two or three or four or five or six or seven or eight or nine or 10 or more time intervals are developed and the best time interval is utilized for further processing. In an embodiment, an average of the results can be utilized. The results can also be weighted. [00265] This is also the case with respect to the fixed ensemble time-averaged wavelet components which are tapered towards zero value inside an interval relative to the ventricular activation time. This interval can be the same as the interval utilized for the tapering towards zero outside the interval. [00266] As noted above, in an exemplary embodiment, the fixed ensemble averaged wavelet components tapered by complementary windows 1- HWi(τ), and the tapered scaled components are added thereto. [00267] In an exemplary embodiment, the goal of the scaling and tapering is to estimate the wavelet decompositions of each V artifact. The amplitude scaled data is utilized within the window, and outside the window the fixed ensemble averaged wavelet components are utilized. The tapering can provide seamless transition from inside the window to outside the window. [00268] Any window function that can enable the teachings herein can be used in some embodiments. Some embodiments do not use window functions per se. Other types of data manipulation can be used in some embodiments. Indeed, as noted below, a trained neural network can be used. [00269] While the above is presented in terms of the obtained second data being previously processed, in an embodiment, where the actor of method 1030 executes this processing, there is a method that includes any one or more of the actions herein, that further includes as part of the process to develop second data (1) scaling ensemble time-averaged wavelet components of the developed time-averaged wavelet coefficients for the respective different scales; (2) tapering the scaled components at least towards zero value outside an interval relative to the ventricular activation time for the respective different scales; (3) tapering fixed ensemble time- averaged wavelet components towards zero value inside an interval relative to the ventricular activation time for the respective different scales; and (4) adding the tapered scaled components to the tapered fixed components to obtain an estimate of a wavelet decomposition of each ventricular artifact.
Attorney Docket No.246-021PCT [00270] In an embodiment, the wavelet processing includes developing respective pluralities of wavelet coefficients at different scales for respective periods of time extending from before respective ventricular activation to after ventricular activation. On the scales, the above noted features / variations can apply. Further, the wavelet processing can include developing wavelet decompositions of estimated far field artifacts at the different scales for the respective periods of time. Here, in an embodiment, the second data is also based on third data developed by: (1) subtracting the wavelet decompositions from the respective pluralities of wavelet coefficients for the different scales to develop third data and (2) manipulating the third data to extract atrial activation complexes from the first data that overlap high amplitude ventricular artifact. Accordingly, in an embodiment, the second data is the data of FIG. 27 (by way of example only). [00271] Concomitant with the actor of method 1030 executing the processing, in an embodiment there is a method that includes one or more of the method actions herein, further comprising: (1) executing the wavelet processing; (2) developing respective pluralities of wavelet coefficients at different scales for respective periods of time extending from before respective ventricular activation to after ventricular activation; (3) developing wavelet decompositions of estimated far field artifacts at the different scales for the respective periods of time; (4) the method further includes developing third data by subtracting the wavelet decompositions from the respective pluralities of wavelet coefficients for the different scales; and (5) the method further includes manipulating the third data to extract atrial activation complexes from the first data that overlap high amplitude ventricular artifact. [00272] Still with respect to the above-noted process flow of FIG. 25, in an embodiment, the second data is further based on a subtraction of data resulting from the wavelet processing of the obtained first data from CWTs of the first data and performing an inverse transform on the result to remove ventricular far-field artifacts from the first data. And where the actor of method 1030 is executing the computations, in an embodiment, there is the method 1060 presented in FIG. 31, which includes method action 1062, which includes the action of executing method 1030 or method 1031 (or executing method 1030 with method action 1033 instead of method action 1032), including executing the wavelet processing. Method 1060 includes method action 1064, which includes the action of subtracting the second data from CWTs of the first data and performing an inverse transform on the result to remove ventricular far-field artifacts from the first data. Thus, the second data is different from that disclosed in
Attorney Docket No.246-021PCT the prior paragraph. Effectively, third data is obtained from method action 1064. This is the same as the second data noted immediately above. [00273] To put things more simply, in an embodiment, there is the method of 1070, as represented by the flow chart of FIG.32, which includes method action 1062 and method action 1074, where action 1074 includes subtracting results of tapering from CWTs developed in the wavelet processing to remove ventricular far-field artifacts from the first data. [00274] In an embodiment, to investigate the accuracy of the above detailed methods, one can remove V artifact from two adjacent unipolar channels recorded with an HD grid catheter and construct bipolar EGMs from the processed unipolar signals and compare such with the corresponding bipolar output from the mapping system. In addition, atrial recordings from which V artifact had been removed can be treated as “ground-truth” signals and combined with time-shifted versions of the estimated V artifact. The performance of subsequent wavelet-based artifact removal and subtraction of both fixed mean and median QT templates can then be evaluated with respect to the "ground truth" data. [00275] Embodiments include utilizing one or more of the above teachings to identify / determine atrial activation. Embodiments can include starting with the data developed using the atrial activation complexes developed according to the teachings above (e.g., data corresponding to FIG. 27 for example). Embodiments can include executing one or more or all of the method actions detailed above to obtain data analogous to / that of FIG.27 (it need not be in graphical form). Embodiments can include developing first derivative Gaussian wavelets to identify atrial activation. In an embodiment, wavelets having a function that has a form that is effectively similar enough to the unipolar atrial activation complex obtained above can be used in some embodiments. [00276] Embodiments can be directed to detecting periods when power due to depolarization (positive deflections in 1st derivative Gaussian wavelet decomposition for example) is maintained at a level greater (significantly in some embodiments) than background. In an embodiment, the power that is maintained is at least 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 125, 150, 175, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750 or 800 or 850% or more or any value or range of values therebetween in 1% increments of the background, by way of example. [00277] Within this window, maxima in this "positive" power can be used to identify the timing and reflect the magnitude of local depolarization. This can enable robust classification of
Attorney Docket No.246-021PCT normal and fractionated activation complexes. Embodiments can include setting a wavelet scale (or scales) that best or at least utilitarianly matches local atrial electrical activity and minimizes confounding effects of noise. Embodiment can include identifying the scale(s) based on inspection visually of the resulting graphs, or based on a metric of predetermined requirements or thresholds, or utilizing the results of a trained neural network by way of example only and not by way of limitation, where, with respect to this latter arrangement, a number of wavelet scales for a number of patients are selected by a trained professional, and the neural network is trained accordingly, and the results thereof are utilized to pick the scales going forward when the teachings herein are utilized in a clinical setting. [00278] Some exemplary details of an exemplary method that can enable the identification of atrial activation will now be described by way of example. [00279] FIG.33A shows an exemplary atrial activation complex for a person with PeAF (Atrial Fibrillation). That is, for the purposes of discussion, this is representative data of atrial EGMs (with V artifact subtracted) from a patient with PeAF. Two-timeframes a and b are highlighted, which frames overlap. In an embodiment, a CWT can be computed at 5 or 7 or 10 scales using a 1st derivative Gaussian mother wavelet. FIG. 33B shows five scales out of the 10 scales computed. (Again, more or less can be computed.) In an embodiment, instantaneous measure of the corresponding power at each wavelet scale is determined / estimated. In this exemplary embodiment, the wavelet decomposition is rectified to extract positive deflections (due to depolarization) and wavelet coefficients are squared to obtain the instantaneous measure of corresponding power by way of exemplary action, by way of example. In an embodiment, other functions can be used, such as cubing for example. Here, we have used power because it has a physical meaning, but from the point of view of mathematics, the transformation that is most effective can be used. [00280] Embodiments can include detecting local activation as an elevation, such as by way of example with the qualifier a sustained elevation, in "positive" power across selected scales. In an embodiment, this can be qualified that it must be with respect to a threshold (which can be an adaptive threshold) greater, including substantially greater than estimated background noise and non near-field artifact (for example, and not by way of limitation, 10x median in a 500 msec running lag window delayed by 20 msec – in some embodiments, greater than and/or equal to 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 175, 200, 225, 250, 275 or 300 ms or more or any value or range of values therebetween in 1 ms increments of the window length can be used, and in some embodiments, the window delay is less than, greater
Attorney Docket No.246-021PCT than and/or equal to 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 45, 50, 55 or 60 ms or more or any value or range of values therebetween in 1ms increments can be used – less than, greater than and/or equal to 7x, 8x, 9x, 10x, 11x, 12x, 13x, 14x, 15x, 16x, 17x, 18x, 19x, 20x, 21x, 22x, 23x, 24x, 25x, 26x, 27x, 28x, 29x, 30x, 35x or 40x or more or any vale or range of values therebetween in 1x increments can be used). In an embodiment, a timing of activation events is estimated by identifying maxima within this window across scales while quantifying a duration as the time for which this threshold is exceeded. More specifically, referring to FIGs. 33B and 33C, presented are time-series records of wavelet decomposition with 1st derivative Gaussian wavelets with corresponding positive power at selected scales for normal (a) and low amplitude fractionated EGM segments (b), respectively. With regard to the latter, this is an example of a signal resulting from a temporal period where the fractionated activity associated with atrial fibrillation is occurring. Briefly, as can be seen from figure 33A, this does not occur in every segment that follows atrial activation. In many instances, there are atrial activations where this fragmented activity does not exist. The skilled person in the art will be able to identify the fractionated activity from the extracted atrial activation complexes often by inspection. More on this in a moment, but briefly, the use of "positive" power to identify activation events has proved sufficiently robust for a single threshold to be used for a particular patient across all scales in all channels. In an embodiment, this could be set by the operator / technician or whoever is running the testing / data processing. In an embodiment, there is utilitarian value in basing this on signal statistics across a substantial running lag window over sufficient time to enable noise and non near-field artifact to be segregated. [00281] In an embodiment, the temporal period of the wavelet decompositions and/or the positive power evaluation is computed across time windows from -125 to +125 ms by way of example with respect to the maxima of the EGMs. In an embodiment, the base raw signal extends (in some embodiments contiguously) for less than, greater than, and/or equal to 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, or 900 ms, or more, or any value or range of values therebetween in 1 ms increments for respective windows, and the time need not be the same for each window. In an embodiment, the CWT can be computed across time windows from plus and/or minus less than and/or greater than and/or equal to 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 325, 350, 375, 400, 425, 450, 475, 500, 525, 550, 575, 600, 625, 650, 675, 700, 725, 750, 775,
Attorney Docket No.246-021PCT 800 ms, or more, or any value or range of values therebetween in 1 ms increments from the maxima in the EMG with V artifact removed (and the minus need not be the same absolute value as the positive, and the numbers need not be the same in each window, although in other embodiments, the numbers will be the same in each window), and this can be different for different windows. In an embodiment, the CWT can be computed continuously and/or for time windows that form a contiguous set of time windows for the given data from the channel, where in some embodiments, each window has a maxima. In an embodiment, there are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, or more, or any value or range of values therebetween in 1 increment time windows. [00282] Note that the shaded area “b” of FIG.33A does not have the pronounced maxima as in the other areas, due to the fractionated activity. An embodiment can include centering the “b” window on an area where there “should” be a maxima based on the preceding maxima (and/or the following maxima). For example, if the normalized mode between maxima occurs within X ms of each other, half of that from the last maxima would be where the window is centered. The data can be evaluated to remove the fractionated areas / the areas without maxima, and then a mean and/or a median average time between the remaining maxima can be obtained, and then the “b” timeframe can be set to half the average from the last maxima. That said, inspection by a trained technician and/or a healthcare professional can identify such. One long continuous window could be used. [00283] Wavelet decomposition and power segments are presented horizontally for same scales, with scale becoming more course from top to bottom. The data is also normalized. Maxima in both decompositions and power segments identify contributions of different wavelet scales to local depolarization. In some embodiments, positive power is relied upon, as it can provide robust detection with fractionated EGMs in the presence of noise. [00284] Briefly, utilitarian value can be obtained using the wavelet approach in that in some embodiments, this approach increases the meaningful parameters that can be adjusted. In an embodiment, the increase is less than, greater than and/or equal to 50, 75, 100, 125, 150, 200, 250, 300, 350, 400, 450, 500, 600, 700, 800, 900 or 1000% or more or any value or range of values therebetween in 1% increments relative to that which would otherwise be the case. The techniques detailed herein can provide for the indication of the activation times or otherwise the onset of the fractionated activity, which is not apparent from the extracted complexes after the V artifact is removed. Put another way, in some embodiments, because the skilled person
Attorney Docket No.246-021PCT will be able to evaluate the extracted complexes, the person can “zoom in” on the pertinent timeframes, such as timeframe segment b. Alternatively, and/or in addition to this, based on the wavelet decompositions and/or the positive power data, the skilled person will be able to identify the fractionated activity by inspection, indeed, in some instances, more easily than that which would be the case from the extracted complexes. For example, the wavelet decompositions of figures 33B and 33C are clearly different from each other in most all the scales, and certainly sufficiently different in all the scales to identify which time periods have the fractionated activity. In an exemplary embodiment, an algorithm can be implemented that identifies the time frames of the fractionated activity. By way of example only and not by way of limitation, if there are a relatively higher number of maximas and minimas within a certain period of time, this could be an indication of fractionated activity relative to other activity. With respect to the wavelet decompositions, if there are minimas that have an absolute value in the neighborhood of that of the maximas, that could also be an indication of fractionated activity. By way of example only and not by way of limitation, if the wavelet decomposition provides for minimas that have absolute values that are within 5, 10, 15, 20, 25, 30, 35 or 40% or any value or range of values therebetween in 1% increments of the value of the maximas, such as the immediate minima following a maxima, that can be an indication of the fractionated activity. This can also be a basis from which to choose a scale for activation time evaluation as will be explained below. [00285] That is, embodiments include choosing a temporal resolution with which local activation complexes are identified. As is seen, the resolution is reduced as wavelet scale is increased. Embodiments include balancing resolution with clarity. By way of exemplary scenario, in this case, scale 5 is deemed to be the resolution that can provide utilitarian value for detection. Note that in an embodiment, scale 7 could have been used with respect to the electrogram during timing b, and 3 could have been used with respect to the electrogram during timing a. But when comparing timing b and timing a, it can be seen that scale 5 has maxima in both timings that can be identified and otherwise easily utilized. Scale 3 is somewhat choppy with respect to timeframe b, and this choppiness can skew the actual activation time with respect to the maximum. Corollary to this is that there would be between four and five maxima depending on the threshold that is utilized (more on the threshold in a moment). Conversely, with scale 5, there are only three maxima at play, and all three fall within a range that exceeds a threshold in a distinct matter.
Attorney Docket No.246-021PCT [00286] Embodiments can include identifying the times of fractionated activity from the extracted atrial activation complexes and/or from the wavelet decomposition of the complexes and/or from the positive power evaluation thereof. Indeed, each can be utilized to supplement the other. Any of the techniques noted herein or any others that can enable the teachings herein can be utilized in at least some exemplary embodiments. Embodiments include the product of a machine learning algorithm that analyzes any one or more or all of the just detailed data sets to identify the fractionated activity. [00287] Still further, embodiments include selecting the wavelet scale that provides the most utilitarian value or otherwise that provide sufficient utilitarian value to implement the teachings detailed herein. In this regard, embodiments include obtaining the wavelet transformations for various scales, and evaluating the data of the individual scales to determine which of the scales will be utilized for atrial activation timing. As noted above, in this exemplary scenario, scale 5 is deemed to be the most utilitarian scale, or at least usable to implement the teachings herein. [00288] Embodiments include trial and error implementation where over time, as the processes are implemented, the scale to be selected becomes more evident based on the end results. In an embodiment, again, a machine learning algorithm can be utilized to develop a fixed algorithm that can be utilized to select the wavelet scale to be utilized. In an embodiment, the end results of the selected scales can be evaluated by the machine learning algorithm to develop a product from machine learning that can be placed on a computer chip or otherwise implemented in a computer system by way of example, to pick the wavelet scale. On the opposite side of the spectrum, a trained technician or otherwise a trained healthcare provider implementing the teachings detailed herein can evaluate the results of the wavelet decompositions and the results of the positive power analysis based on his or her experience to pick the scale that has utilitarian value. [00289] In an exemplary embodiment, the scale that has maxima that on average are closest to a threshold (e.g., as represented by the horizontal dashed line in figures 33B and 33C) can be utilized, whether that be for the positive power calculations or for the wavelet decomposition calculations. The threshold can be predetermined. The threshold can be adaptive. The scale that has a mode of maxima that are above the threshold but within a certain percentage of each other (e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, or 30%, or any value or range of values therebetween in 0.1% increments of the amplitude of the highest or lowest of the group for example.) In an exemplary embodiment, the scale that has maxima that are relatively consistent with respect to “distance” from a threshold can be utilized, or otherwise the scales
Attorney Docket No.246-021PCT that have maxima that are more consistent than others can be utilized. In an exemplary embodiment, there will be a single large maxima, such as seen in all of the wavelet scales in figure 33C, and there will also be one or more smaller maximas. Embodiments can include selecting the scales that have the smaller maximas that are within a certain percentage of each other with respect to amplitude, such as by way of example only and not by way of limitation, if the positive power analysis includes “secondary” maximas that are within 5, 10, 15, 20, 25, 30, 35 or 40% or any value or range of values therebetween in 1% increments of each other (using the smallest amplitude or largest amplitude), that can be the scale to use. In an embodiment, the maximas used must be over a threshold. In an embodiment, the threshold is 5, 10, 15, 20, 25, 30, 35 or 40% or any value or range of values therebetween in 1% increments of the main maxima. As noted above, there must be maximas that are within a certain distance of the scale. [00290] In an embodiment, the scale selected has maximas above the threshold where the signal goes below the threshold before reaching another maxima. This would rule out scale 3 for example and scale 1 for example. In an embodiment, the selected scale has maximas where the signal goes to zero before another maxima. This would rule out scales 1 and 3. In an embodiment, the scale used has secondary maximas that are similar to each other within a certain time. [00291] Embodiments can include selecting the scale where there are a certain number of maximas that are within 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, or 200 ms, or any value or range of values therebetween in 1 ms increments of each other. For example, in a given range, there are at least and/or equal to and/or no more than 2, 3, 4, 5 or 6 or any value or range of values therebetween. In an embodiment, there is a temporal separation of a certain amount between them, such as for example, at least 10, 25, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, or 150 ms, or any value or range of values therebetween in 1 ms increments. In an embodiment, there are no more than and/or equal to than 2, 3, 4, 5 or 6 or any value or range of values in that range. [00292] Embodiments combine the various “rules.” By way of example, if the positive power analysis includes “secondary” maximas that are within 5, 10, 15, 20, 25, 30, 35, or 40%, or any value or range of values therebetween in 1% increments of each other (using the smallest amplitude or largest amplitude), and within 20, 25, 30, 35, 40, 45, 50, 55, 60, 70, 80, 90, 100, 110, 120, 130, 140 or 150 ms or any value or range of values therebetween in 1 ms increments, that could be the scale to use. Depending on the numbers, this would rule out all but scale 5.
Attorney Docket No.246-021PCT [00293] An embodiment can include choosing the scales that have the fewest maxima above the threshold as long as there are at least 2, 3, 4, 5, 6, or 7, or any value or range of values therebetween in 1 increment. [00294] In an embodiment, the scale selected will have a certain number of secondary maxima before the primary maxima, in some embodiments, at least that exceed the threshold, and a certain number of secondary maxima after the primary maxima, and in some embodiments, that exceed the threshold. In an embodiment, there are at least 1, 2, 3, 4 or 5 or any value or range of values therebetween in 1 increment before and/or after the primary maxima (and the two need not be the same), where in some embodiments, all are above the threshold and in some embodiments there is no threshold applied for this determination. [00295] Embodiments include choosing a scale based on any one or more features present in scale 5 and/or in scale 7 and/or in scale 3 and/or that are not present in scales 1 and/or 3 and/or 7 and/or 9. [00296] But again, embodiments will typically include the skilled technician or healthcare professional choosing a scale based on the graphical data and/or the numerical data associated with given scales, and embodiments include keeping a log of such choices so that neural network can be trained based on those choices, where the results of such trained neural network can be utilized to choose the given scales in the future. [00297] But note also that depending on the computational times in play, more than one scale can be utilized in the results of the specific scales can be compared to one another, and the results that appear more utilitarian can then be chosen. That is, the choice of scales need not be based on the data that is specific to the positive power and/or the wavelet decomposition, but can be chosen based on how those scales are utilized downstream. [00298] Also, in some embodiments, the scale(s) will be predetermined based on empirical evaluation of exemplary results. That is, in an embodiment, it could be that a given scale is the scale that is always used (e.g., it could be that scale 5 is the one that is used always in the clinical setting). In an embodiment, the scale used can be demographically linked (certain types of people will use one scale, and certain types will use another scale). In some embodiments, forecasting, statistical or AI or otherwise, can be used to select the given scale. Still, as noted, a predetermined threshold and/or set of requirements can be applied to select the given scale for use, and these can be demographically based as well (different requirements
Attorney Docket No.246-021PCT / thresholds for different types of people, or at least for different types of starting data for example). [00299] In an embodiment, the threshold is determined by taking a value that is less than, greater than and/or equal to 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65 or 70% or more or any value or range of values therebetween in 1% increments of a value extracted from the positive power information. For example by way of example, taking the largest amplitude from the positive power (for fractionated and/or unfractionated activity – the same threshold can be used) and setting the threshold at a value that is less than, greater than and/or equal to 13% (or any of the appropriate percentages just noted. In an exemplary embodiment, the threshold is set at one or more of the aforementioned percentage values of the average amplitude of the positive power. In an exemplary embodiment, this averaging can exclude the zero values. In an exemplary embodiment, this can be the mean, median and/or mode average. In an exemplary embodiment, the average can include the zero values (hence why the “larger” percentages might be used, to accommodate the expansive zero values). And because the length of the window can vary, the averages can change depending on the length of the window, at least where the zero values are taking into account. Accordingly, a lower percentage would be applicable for a window that has a more narrow length, because there would be fewer zero values as compared to a window that has a longer length, where there would be more zero values, and hence a higher percentage. [00300] In an embodiment, the threshold can be based on empirical analysis. In an embodiment, a significant number of results can be evaluated, and based on the evaluation, a threshold can be set, which threshold will be utilized in a clinical setting. And again, thresholds can be demographically based, where some threshold will be used for some people and others will be used for other people. The threshold can be data based, where certain threshold will be used for different types of data or data that includes different values. [00301] In an embodiment, the threshold can be the one that excludes for example at least and/or equal to and/or no more than 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85 or 90% or more or any value or range of values therebetween in 1% increments of the local maxima (amplitude) in the positive power. In this regard, with respect to scale five, as can be seen, the threshold excludes at least five maxima out of the eight total maxima as seen. In this regard, separately or in combination with this arrangement, the threshold can be set at the average of all maxima, whether such is the mean, median and/or mode. In an embodiment, this can be weighted, such as, for example, 33% greater than the average of all maxima. In an exemplary
Attorney Docket No.246-021PCT embodiment, the threshold can be set so that the threshold excludes maxima that are separated by a certain amount from each other, and only include maxima that are close to each other, such as by way of example only and not by way limitation, the three maxima having the circle shown in the wavelet scale five. [00302] Note also that the features associated herein with selecting the threshold can also be utilized in a variation thereof to select the wavelet scale. For example, if the average maxima (amplitude), mean/median and/or mode is X over the timescale, the one that has less than, greater than and/or equal to for example 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25 or more or any value or range of values therebetween in 1 increment maxima over that average are excluded or included (depending on the number, and it might be “only” 3 or only 4 or only 2 or only 1 above that threshold for example). In an embodiment, the average, or more accurately the threshold, can be weighted (say 30% above or 40% below the average). In an embodiment, the threshold is less than, greater than and/or equal to 10, 20, 30, 40, 50, 60, 70, 80 or 90% or more or any value or range of values less than or is less than, greater than and/or equal to 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 125, 150, 175, 200, 225, 250, 275, 300, 325 or 350% or greater than the raw average or the raw value used, or whatever data is used for the threshold (e.g., 80% less than the greatest maxima is the threshold, and the scale that has 3 maxima over that threshold is the scale used). Again, in some embodiments, zero values can be included in the averages, and in some the zero values are excluded. Any of the teachings related to setting a threshold detailed herein for one data can be applicable for setting a threshold for another data, or at least based thereon, providing that the art enables such, if such results in utilitarian value. [00303] Note that in an embodiment, the drawings showing the graphs with the timelines are to scale, and can be used for at least plus or minus 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1% or any value or range of values therebetween in 0.1% increments accuracy (e.g., the 250mSec scale can be plus 11.3% of that or minus 8.4% of that, etc. [00304] And note that while positive power is used herein in some embodiments, raw power or other data based on massaged scale data can be used in some embodiments, providing that utilitarian results can be extracted from the operation. [00305] Any device, system, and/or method that can enable the choice of scales to be determined so that the teachings detailed herein can be implemented can be utilized in at least some exemplary embodiments providing that the art enables such unless otherwise noted.
Attorney Docket No.246-021PCT [00306] Returning back to what to do once the scale is selected, having selected scale 5, again, in an embodiment, a timing of activation events is estimated by identifying maxima within the window across scales while quantifying a duration as the time for which a threshold is exceeded. In FIGs.33B and 33C, positive power maxima (circles) are identified with respect to a preset threshold (broken horizontal lines) and these times (broken vertical lines) are overlaid on EGMs and corresponding wavelet decompositions for both a and b. From this, local activation times can be determined along with fractionated activity. An example of this is in FIG.33D, which shows the EGM record with local activation marked by filled circles and fractionated activity indicated by the star. [00307] In this embodiment, the first maxima of scale 5 that exceeds the threshold of timeframe b is used as the onset of the fractionated activity. In an embodiment, the timing of largest maxima can be used (here, the third). In an embodiment, the average timing of the maxima (from a set reference time, such as the beginning of the window for example, or from the beginning of the first maxima or from some other utilitarian temporal beginning point) can be used. In an embodiment, the mode can be utilized instead of the median, and where there is an even number of maxima, the median between the mode can be utilized, or a default can utilize to utilize one versus the other, such as the first or the last or the one that is larger or smaller in amplitude by way of example. In an embodiment, the timing of the last maxima can be used. In an embodiment, each maxima can signify a timing that has utility for the teachings herein. Also, in some embodiments, the timing is weighted, so say an average timing would be skewed towards the timing of the highest amplitude, because the average would be weighted based on amplitude in some embodiments. Any regime of selection of timing that can have utilitarian value can be used in some embodiments. In an embodiment, the product of a trained neural network can be utilized to select the timing in this regard, if a sufficient number of examples are evaluated and this is provided to a neural network (a sufficient number of examples for training), the neural network can be trained thereon and can identify the timing. In this regard, a human can evaluate the data beforehand and identify the timing based on the knowledge and/or experience of the human or the plurality of humans, and then the evaluations established by the human(s) can be utilized to train the system. In an embodiment, the activation timing that is identified is not necessarily the time with the greatest amplitude or otherwise is not always such. This can be a result of data that comes from other locations / originates from outside the area of interest vis-à-vis a given channel that has been captured by a given channel (given electrode) and in some embodiments, at a given frequency. In this regard, there can be
Attorney Docket No.246-021PCT sampling between electrodes. If there is fractionation, such can occur at the electrode and/or within 0.25, 0.5, 0.75, 1, 1.25 or 1.5 mm or so of that electrode, and such can be considered “for” that electrode and not for other electrodes, but there can also be or instead only (for a given temporal period) fractionation that should be associated with another electrode / another channel and/or that originates outside of that space. In some embodiments, this is not wanted in the channel of interest, or more specifically, the influence thereof should not be taken into account for the activation timing. Hence the evaluation of the power data beyond taking the maximum amplitude. [00308] In an embodiment, inverse mapping can allow and thus is used for the identification of the fractionated activity and where the activity originates from / from where the activity comes. Here, one can evaluate the fractionated activity and, in some embodiments, use a temporal data context, and determine that a maxima is not associated with the fractionated activity that should be associated with the electrode / channel of interest, but instead is associated with another electrode / another channel, and thus should be discounted (that maxima should not be used for the activation timing. By way of example, the data from other electrodes as well as the electrode for the channel of interest can be considered within a given temporal framework. A comparison of data from other channels / electrodes can be made at given times (which can account for travel for example or such can be de minimus and ignored in other embodiments) and from such, the maxima within the power data that is utilitarian or otherwise appropriate for activation timing for that channel can be selected, and the other data disregarded. Thus, in some embodiments, there can be a method (which can be automated for example, and can be implemented using any of the computational techniques / devices / systems and/or methods detailed herein by way of example or otherwise modified versions thereof) which refers back to an underlying source of given data, and otherwise implements / relies upon spatial localization to identify the maxima that is to be used and/or the maxima that should not be used. In an embodiment, the maps detailed herein can put a current source within 2, 3, 4, 5, 6 or 7 or 8 or 9 or 10 or more electrodes. One will be greatest due to a given source and typically such will related to proximity. Typically, the highest will be at one electrode and the surrounding electrodes will have lower values (temporally based). That can be used to identify the location of the source between electrodes for example, using mapping techniques. And this is not the maximum maxima selection – it could be that the maxima at a given time for the correct source is lower than another maxima at another time for the wrong source – this is why temporal based comparisons can be useful. Thus, one can obtain in real time or near real time
Attorney Docket No.246-021PCT in some embodiments a clear recognition of from where the activity is coming / originates and chose whether or not it is originating from near the electrode / channel of interest or further away, and if the latter, likely discounted. This can be revealed from subsampling and is such in some embodiments. [00309] In this embodiment, only one maxima is used, which maxima corresponds to the originating activation that is near(est) the electrode, again based on map techniques to determine such in this exemplary embodiment. In an embodiment, the data that comes from outside the area of interest (that is identified as originating in tissue outside the area of interest for a given channel) is removed or not used or otherwise discounted from the data used to develop the activation timing / in the method actions to develop the activation timing. All of this said, weighting can be used, which weighting can be based on a relative magnitude of the potentials. In an embodiment, a combination of weighted averaging with other information from other electrodes, such as an inverse solution, can be used. [00310] Indeed, the propagation of electrical activation in cardiac muscle cells is driven by diffusion currents generated in surrounding tissue by cell membrane depolarization. This electrotonic current flow is effectively instantaneous and alters potential distributions in adjacent inactive muscle cells bringing them to the threshold at which depolarisation occurs. Unipolar electrograms (EGMs) are records of potential changes associated with these processes across an electrode in the extracellular domain. As the distance between a unipolar electrode and a distributed current source increase the resultant potential is attenuated and blurred, both as a result of diffusion. Again, atrial fibrillation (AF) is associated with regions where EGMs have multiple low-amplitude components which spread over a much longer time than normal. These fractionated EGMs reflect slow heterogeneous spread of electrical activation within the region addressed by the electrode or electrodes at which they are recorded. (Fractionated EGMs can occur intermittently in normal atrial tissue in AF due to spatial variation in activation time- history where some regions are refactory and activation cannot be initiated while others are partially repolarised and able to support slow, heterogenous electrical propagation. However, it also occurs more regularly in regions where structural remodeling, for instance due to atrial muscle cell death and subsequent infiltration scar tissue (replacement fibrosis) causes conduction slowing and conduction block.) Under these circumstances, it is difficult using the state of the art to identify activation times accurately at an individual extracellular electrode. This is because EGM deflections caused by activation a few millimeters away from the electrode may not be immediately distinguishable from those caused by depolarisation of
Attorney Docket No.246-021PCT smaller tissue volumes immediately adjacent to it. It is utilitarian to identify the location of sources of fractionated activity more precisely to improve the classification and mapping of this behavior. [00311] Embodiments thus include devices for, systems for and/or methods of determining an origin of energy or data, etc., or, at least determining or identifying energy or data or portions of the signal or signals that does not originate in a certain area, such as the tissue that is the area of interest or otherwise applicable to a certain channel, or otherwise that the energy or data signal originates from outside a specific area of interest or otherwise is more applicable to be utilized in activation time development for another channel. Embodiments thus include any one or more of the methods detailed herein along with such action. Corollary to this is that these devices, systems and/or methods can be utilized for into respectively, identify which maxima from a plurality of maxima can be utilized or otherwise should be utilized in the activation timing identification. [00312] Wavelet transformation with a ten derivative Gaussian mother wavelet is useful here. Fractionated unipolar atrial EGMs are decomposed into a set of time-varying components at different wavelet scales which carry information about the distribution of frequencies in EGM deflections associated with activation. In embodiments, activation sources adjacent to the electrode typically carry relatively more "power" in low-order (short-duration) wavelet scales whereas more distant sources should carry relatively more power in higher-order (longer duration) wavelet scales. Hence the different scales as taught herein to differentiate between the durations / identify the durations and determine the origin thereof so that the ones that are applicable to tissue associated with the channel can be utilized and the ones that are not can be removed (or at least weighted so that they impact the evaluation less than that which would otherwise be the case / reduce so that they are effectively negligible with respect to implementing the teachings herein). Comparison of wavelet decompositions for adjacent electrodes over time is also utilitarian to infer the location of sources from fractionated EGMs because this enables separation of passive non-near field components (which can occur on adjacent electrodes at the same time) from propagating electrical activity (where there are internally consistent time delays between activation detected at adjacent clusters of electrodes). Embodiments can include identifying the location of non-near field components (or identifying that such is not a near field component) in a fractionated EMG at less than grid spacing dimensions by, for example, "triangulation" (based on the attenuation of simultaneous wavelet components across adjacent clusters of electrodes) or by for example, increasing the
Attorney Docket No.246-021PCT resolution of our inverse mapping methods. Embodiments can use these techniques to do so, by way of example only and not by way of limitation. These rules can be utilized to identify wavelet components in a complex fractionated EGM that are associated with activation spread. Where more than one such component is identified at an electrode within the accepted absolute refractory period (this can happen if there is more than one local activation pathway), a weighted average of the activation time selected can be used. The process can be automated and could be made more adaptive using artificial intelligence (AI). By way of example only and not by way of limitation, any of the machine learning implementations detailed herein or otherwise the machine learning teachings where the products of machine learning detailed herein can be utilized in at least some exemplary embodiments to implement the weighted average or otherwise to identify the weights and to apply such. [00313] Accordingly, teachings herein include utilizing one or more of the techniques thus detailed to identify the particular maxima in the fractionated activity to utilize as the activation timing. The techniques herein provide for the development (extraction) of the maxima, and then the identification of the one maxima out of the various extracted maxima (if there are more than one). And note that it could be that there is no maxima that is used – that is, the maxima are not for the tissue associated with the channel. Any device for, system for and/or method that can enable utilitarian selection of a maxima to implement the teachings herein can be used providing that the art enables such and such can support the teachings herein, unless otherwise noted. [00314] In the above, the timing examples provide additional grounds for selecting or excluding a scale. Because in at least some exemplary embodiments, the purpose of choosing a scale is to choose the scale that can provide a utilitarian timing, if the scale does not provide a utilitarian timing according to the given algorithm that is utilized to identify the timing, that scale could be eliminated or otherwise not used in the evaluation of the positive power to determine timing. By way of example, if there is no clean mode of maxima with respect to timing (e.g. four or six maxima as opposed to three or five maxima, the latter being a clean mode because it would be the second maxima or the third maxima respectively) where the timing is based on the mode of the maxima, that scale might be dropped in lieu of another scale that does provide a clean maxima. In this underscored the fact that the scale that is selected is the scale that can provide the utilitarian value to select the timing. And this is where, for example, the trained neural network, or more accurately, the product of the trained neural network, can come into play in at least some exemplary embodiments. Still, a predetermined hard coded algorithm /
Attorney Docket No.246-021PCT predetermined chip and/or predetermined firmware can be utilized as well. Note also that big data could be used by way of example only and not by way of limitation, as will be described in greater detail below. [00315] FIG. 34 presents an exemplary algorithm for an exemplary method, method 1090, which includes method action 1092, which entails obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation. This can be executed according to any of the teachings herein. Accordingly, in an exemplary embodiment the data based on the first data is the obtained one or more unipolar EGMs. Conversely, in another exemplary embodiment, the data based on the first data is based on one or more unipolar EGMs with ventricular far-field artifact subtracted, concomitant with the teachings associated with method 1030, etc. But in an alternate embodiment of method 1090, as represented in FIG. 34A by method 1091, method action 1092 instead includes obtaining first data based on one or more unipolar EGMs of electrical properties in a living human (this is method action 1093 of FIG. 34A). Concomitant with the teachings herein, the electrical properties are electrical properties in an atrium of the living human. So, in an embodiment, the electrical properties are electrical properties in part of the living human. In some instances, the one or more unipolar EGMs are recorded in an atrium of a heart of the human. But in other embodiments, the one or more unipolar EGMs are recorded outside an atrium of a heart of the human. Any device or system that can enable the acquisition of, and any method of acquiring the data, can be used in some embodiments providing that the art enables such. Again, the method action 1093 can be executed in a human that has a normally functioning heart (method action 1092 can be replaced with method action 1093). Alternatively and/or in addition to this, the human is suspected of having atrial fibrillation but is not afflicted by atrial fibrillation. And while the teachings detailed herein can be utilized to diagnose or otherwise detect atrial fibrillation, the teachings detailed herein can also be utilized to determine that the person is in fact not afflicted with atrial fibrillation. [00316] To be clear, in some embodiments, the person is afflicted with atrial fibrillation, in an embodiment, the electrical activity is electrical activity in a heart not affiliated with atrial fibrillation, wherein the human has a pacemaker for his/her heart. In an embodiment, the electrical properties are properties of disorganized electrical activity, while in other embodiments, the electrical properties are properties of organized electrical activity. This may or may not be known beforehand. Method 1090 and method 1091 includes method action 1094, which entails identifying regional atrial activation times based on wavelet processing of
Attorney Docket No.246-021PCT data based on the first data. This can be done according to, for example, the teachings just detailed. The actor of method action 1094 can be the actor executing the processing, etc., or can be the actor that receives the processed data and from that data, identifies the regional atrial activation times therefrom. Note also that method action 1094 can be executed by evaluating data that already indicates where the activation times are present. [00317] Still, embodiments include a single actor implementing the processing, and identifying the activation. Thus, FIG.35 shows an exemplary algorithm for an exemplary method, method 1110, which includes method action 1112, which entails executing method action 1092 or method action 1093 (the text of method action 1112 in FIG.35 is an amalgamation of the two in the interests of textual economy) or a variation of one or both method actions. Method 1110 further includes method action 1114, which includes wavelet processing data based on the first data to obtain second data, and method action 1116, which includes identifying regional atrial activation times based on data based on the second data. This can be directly from the second data, or a result of manipulation of the second data (e.g., such as developing positive power). Again, any of the above teachings can be utilized to implement this method. [00318] As noted above, the wavelet processing can include developing first derivative Gaussian wavelets. Embodiments can include identifying, from wavelets produced by the wavelet processing, period(s) when power due to depolarization is maintained at a level sufficiently the greater than background power. In an exemplary embodiment, the power is at least 100, 125, 150, 175, 200, 250, 300, 350, 400, 450, 500, 600 or 700% or more or any value or range of values therebetween in 1% increments greater than the background power. [00319] In an embodiment, there is a method 2010, as seen in the algorithm presented in FIG. 36, which includes method action 2012, which includes executing method 1110. Method 2010 includes method action 2014, which includes processing the second data to obtain positive power data. Alternatively, such as where method 1090 is executed, the action of identifying regional atrial activation times is based on positive power data that is obtained from the results of the wavelet processing of the data based on the first data. In an exemplary embodiment with respect to method 2010, the positive power data includes data in a plurality of scales. Further, in an exemplary embodiment, method 2010 includes identifying data in at least one scale from the plurality of scales and using the data in the at least one scale for detection of positive power maxima, wherein the action of identifying regional activation times is based on the detected positive power maxima. And again, such as where method 1090 is executed, the action of identifying regional activation times is based on detected positive power maxima based on the
Attorney Docket No.246-021PCT results of the wavelet processing of the data based on the first data. In an embodiment, there is a method 2110, as seen in the algorithm presented in FIG.37, which includes method action 2112, which includes executing method 1110. Method 2110 includes method action 2114, which includes manipulating and/or processing the second data to develop the data based on the second data. By way of example, method action 2114 can include the action of selecting one or more wavelet scales of the second data that better matches local atrial electrical activity than other scales of the second data, wherein the selected one or more wavelet scales minimize confounding effects of noise more than the other scales of the second data. This can be done in accordance with the teachings above by way of example only and not by way of limitation, such as based on the number of maxima above a threshold value where all of the maxima are between zero power values, etc. [00320] In an embodiment, the wavelet processing of data results in a CWT at a plurality of different scales. In an embodiment, the data based on second data is rectified wavelet decompositions of the CWT. Thus, in an embodiment, the methods detailed herein can further include rectifying wavelet decomposition of the CWT. This can be done to extract positive deflections due to depolarization. Accordingly, in an embodiment, the data based on second data is based on positive deflections due to depolarization. Moreover, as noted above, only some scales of the CWT are used (sometimes only 1). Thus, in an embodiment, the method further includes rectifying wavelet decomposition of the CWT to extract positive deflections due to depolarization, the method further includes selecting one or more scales that are a subset of the total number of different scales (e.g., scale 5 from the 10 scales). Further with respect to methods that include rectifying the decompositions, in an embodiment, the method includes obtaining instantaneous measures of corresponding power at at least one scale of the plurality of different scales, and the data based on the second data includes maxima of the instantaneous measures. In an embodiment with respect to the rectification embodiments, concomitant with the teachings above associated with FIG. 33B and FIG. 33C, the method includes obtaining instantaneous measures of corresponding power at a plurality of scales of the plurality of different scales, as well as identifying one or more scales of the plurality of scales that provide a utilitarian balance between temporal resolution maxima identification during time periods of fractionated activity and the data based on the second data includes maxima (one or more) of the instantaneous measures from the identified one or more scales. Also, embodiments can include executing one or more of the method actions detailed above, such as those associated with method 1110, etc., and also classifying at least one of normal or fractionated activation
Attorney Docket No.246-021PCT complexes based on the data based on the second data to identify the regional atrial activation times. Corollary to this is that embodiments can include executing one or more of the method actions detailed above, such as those associated with method 1090, etc., and also classifying at least one of normal or fractionated activation complexes based on the wavelet processing of data based on the first data to identify the regional atrial activation times. [00321] To round things out, the action of executing wavelet processing can include decomposing the data based on the first data using CWT. Also, in an embodiment, the action of executing wavelet processing includes calculating wavelets at three (3) to fifteen (15) different time scales with first derivative Gaussian mother wavelets. And in the interest of textual economy, it is noted that any one or more of the teachings detailed above with respect to the embodiments associated with removing the artifact can apply to the embodiments of determining atrial activation times, etc., and vice versa, providing that the art enables such, unless otherwise noted. Accordingly, by way of example, the number of scales developed with respect to the embodiments related to removing the artifact can apply to the embodiments of determining atrial activation times, the times to which CWT is implemented detailed above can correspond in part or in whole to the times to which CWT is implemented with respect to the embodiments associated with the atrial activation, etc. And of course all this is in reverse as well, all in the interest of textual economy. [00322] And embodiments include combined methods. In an embodiment, any one or more of the method actions associated with the V artifact removal embodiments can be combined in a method that includes any one or more of the method actions associated with the atrial activation timing embodiment. Indeed, as would be understood from the above, the methods associated with V artifact removal are a precursor to the methods associated with atrial activation timing. [00323] Accordingly, in an exemplary embodiment, such as with respect to method 1110, in an exemplary embodiment, the first data obtained in method action 1112 can correspond to the results of the methods to remove the V artifact. In an exemplary embodiment, the first data can be the extracted atrial activation complexes thereof. Thus, with reference to FIG.38, there is a method, method 2310, which includes method action 2312, which includes executing method 1030 or method 1031 or a method or a method action associated therewith (such as method 1040 or method 1050 or method 1060 or method 1070 or the other methods actions detailed above to remove V artifact). Method 2310 also includes method action 2314, which includes executing method 1090 and/or method 1110 or a method or method action associated therewith, wherein the first data of method 1090 and/or 1110 is the second data of method 1030
Attorney Docket No.246-021PCT or the manipulated third data that results in the extracted atrial activation complexes. As will be understood, method 2310 is a broad method that encompasses combining any one or more of the teachings associated with V artifact removal with one or more of the teachings of atrial activation timing determination, all in the interests of textual economy. [00324] In an embodiment, the accuracy of atrial activation times identified in unipolar signals can be validated by comparing with corresponding Laplacian difference signals. Unipolar recordings can be acquired from, for example, 5 adjacent electrodes in an HD grid™ catheter. The Laplacian difference signal VL(t) can be estimated as: ^^^^ ^^^ ^ ^, ^^ ^^ ^^^ ( ^^^^) = ^^^^ ^^^^, ^^^^ − 0.25 ∗� ^^^^ ^^^^, ^^^^+1 ( ^^^^) + ^^^^ ^^^^, ^^^^−1 ( ^^^^) + ^^^^ ^^^^+1, ^^^^ ( ^^^^) + ^^^^ ^^^^−1, ^^^^ ( ^^^^)� (6) where ^^^^ ^^^^, ^^^^+1 ( ^^^^) is the voltage recorded at the centre electrode and the four remaining voltages are recorded on the electrodes surrounding it. Activation times for Laplacian signals can be identified as maxima in the time series using a threshold of 1.5x standard deviation. [00325] Embodiments also include reconstruction of near-field atrial electrograms. Here, atrial EGMs can be reconstructed using a wavelet-based filter that is synchronized with local activation and matches the known temporal variation of electrical activation, based on an occurrence where the highest instantaneous frequencies occur during depolarization whereas subsequent activity during repolarization is characterized by lower frequencies. FIG.40 shows an exemplary flowchart for an exemplary method. To start with, there is FIG. 39, which presents representative atrial EGMs (with V artifact subtracted) acquired from a patient with persistent AF, where the broken red lines / broken vertical lines are activation times determined in accordance with the teachings above." That is, FIG. 39 shows, a wavelet-based reconstruction of near-field atria EGMs, where the vertical lines indicate previously detected activation times. [00326] After identification of atrial activation times in unipolar recordings (V artifact subtracted – FIG. 39), a CWT can be computed at various scales (e.g., 10 scales, but other numbers can be used – we refer to the above statements about the applicability of the teachings of one embodiment being applicable to the other and not being repeated in the interests of textual economy – that applies for this reconstruction embodiment as well, and vis-a-versa) using a 2nd derivative Gaussian mother wavelet. That is, wavelet decomposition can be performed, here, in 10 scales, using 2nd derivative Gaussian wavelets, but again, more or fewer scales can be used (any of the numbers detailed herein can be used, and the number and/or scales can be different from each other (e.g., the scales for V subtraction can be different in
Attorney Docket No.246-021PCT type and/or number than those for activation time and/or for the reconstruction of the atrial electrogram, and visa-versa (the scales for the reconstruction can be different from the V subtraction and/or the timing, and the timing can be different from the V subtraction and the reconstruction). [00327] The following will focus on a processing workflow for an atrial signal segment (300 msec – the grey area of FIG.39, expanded in FIG.41A for clarity) encompassing a complete activation cycle. FIG. 41B shows the wavelet decomposition for each scale, with the lowest scale at the top and the highest (coarsest) scale at the bottom, some of which are labeled (11701 is scale 1 and 11710 is scale 10, 11709 is scale 9 and 11704 is scale 4, and the scales that are not labeled would follow the pattern in 1 integer increments). [00328] In an embodiment, the temporal period of the wavelet decompositions computed across time windows from -150 to +150 ms by way of example with respect to the maxima of the EGMs and/or the activation times. In an embodiment, the base raw signal extends (in some embodiments contiguously) for less than, greater than, and/or equal to 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, or 900 ms, or more or any value or range of values therebetween in 1 ms increments for respective windows, and the time need not be the same for each window. It is to be understood that some of these values may start to overlap with the next window, depending on the spread of the maxima and/or activation timing, and thus the window could end at the beginning of the next window. In an embodiment, the CWT can be computed across time windows from plus and/or minus less than and/or greater than and/or equal to 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 275, 300, 325, 350, 375, 400, 425, 450, 475, 500, 525, 550, 575, 600, 625, 650, 675, 700, 725, 750, 775, 800 ms, or more, or any value or range of values therebetween in 1 ms increments from the atrial activation time (and the minus need not be the same absolute value as the positive, and the numbers need not be the same in each window, although in other embodiments, the numbers will be the same in each window), and this can be different for different windows, and again, this can be truncated upon running up onto the next window). In an embodiment, the CWT can be computed continuously and/or for time windows that form a contiguous set of time windows for the given data from the channel, where in some embodiments, each window has a maxima and/or activation times or two and/or three or more of such (whatever basis for window length that can have utilitarian value can be used – indeed, window length can be an artificial construct – a single window can exist where certain times
Attorney Docket No.246-021PCT therein are the basis from which to extract the data that has utilitarian value). In an embodiment, the temporal limitation is to prevent interference from coming through or otherwise unwanted data from coming through / being included in the calculations. In an embodiment, there are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350 or 400 or more, or any value or range of values therebetween in 1 increment time windows for each scale (this can be the case for all the windows herein for the other embodiments in the interest of textual economy). [00329] In an embodiment, each window (wavelet window) can include 1 or 2 or 3 or more (and that includes only 1 or 2 or 3 – any disclosure of values herein corresponds to an alternate disclosure of having only those values) activation times. [00330] In the figures, wavelet decomposition and power segments are presented horizontally for same scales, with scale becoming more coarse from top to bottom. The data is also normalized. Maxima in both decompositions and power segments identify contributions of different wavelet scales. [00331] Next, in this embodiment, the wavelet decomposition for each scale is tapered with window function, here a Tukey window aligned with activation time and centered on the activation time for finer wavelet scales. This is seen in FIG. 41C. Note that in some other embodiments, other types of window functions can be used, such as those disclosed above, providing that such can enable the teachings herein and provide utilitarian results. Tukey is simply an exemplary window function for this embodiment. And as noted above, windows based on any of those herein can be utilized. [00332] Again, in an embodiment, the product of a trained neural network, trained on a sufficient number of examples, can be used to select the window function and/or the parameters of the window function (length of time, centering / off-centering, which decompositions will be off centered and which will be centered, for example). [00333] Window length and the period of cosine tapering for one or more or all scales can be set to preserve components associated with activation. For remaining scales, the onset and duration of the initial cosine taper, which are often consistent with finer scales, the window length can be prolonged so that final cosine taper and can be synchronized with the next atrial activation, as seen in FIG.41C and FIG.42 (where FIG.42 is a larger view of FIG.41C, with the scales labeled – as seen scales 11710 and 11709 are synchronized with the next atrial
Attorney Docket No.246-021PCT activity / the window is prolonged – additional scales (scale 8 and/or scale 7 by way of example) can also be synchronized with the next atrial activity / the windows therefore can be prolonged). To be clear, in an embodiment, for each scale, the decomposition can be tapered with Tukey windows aligned as indicated. The finest scales can be centered on the activation time, while window lengths for the coarsest scales can be extended to coincide with the start of the next activation cycle. [00334] In an embodiment, the symmetric portion of the window for the relatively high frequencies and/or lower scales is 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 260, 27, 280, 290, 300, 310, 320, 330, 340 or 350 ms, or more, or any value or range of values therebetween in 1 ms increments (e.g., in the embodiment of FIG. 41C, 140 ms). Concomitant with the CWT windows noted above, some of this might result in the window overlapping with the next window and if the window starts abutting the next window, depending on the activation timing, that would truncate the window. Outside of this, the data is zeroed. Conversely, the non-symmetric portion of the window for the relatively low frequencies and/or higher scales is 60, 65, 70, 75, 80, 90, 100, 110, 120, 130, 140, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650 ms, or more, or any value or range of values therebetween in 1 ms increments (but again, if the window starts abutting the next window, depending on the activation timing, that would truncate the window). That said, in embodiments, the non-symmetrical portion (relative to the activation times – the window is symmetrical – this can be more accurately explained as window with a tail that is extended past what would be the end based on symmetry regarding activation time) of the window can be extended to the beginning of the symmetrical portion (symmetrical relative to the activation time) of the next window. [00335] Note that in at least the embodiment depicted here, it is the symmetric portion of the windowing that drives the tail of the non-symmetric / adjusted window. In this regard, here, the tail is moved to the right until the beginning of the next window for the next activation time. In an embodiment, the tail can be stopped before that time, such as, for example, less than, greater than and/or equal to 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95 or 100 ms or more or any value or range of values therebetween in 1 ms increments before the beginning of the next window, and thereagain, if the end starts running onto the symmetric window, it could end there.
Attorney Docket No.246-021PCT [00336] In an embodiment, as can be the case with respect to determining which scale to use in the embodiment to identify atrial activation times, the person of ordinary skill in the art can perform an inspection of the wavelet decomposition and determine which scales should have the prolonged window length. That said, in embodiments, the highest scales can be prolonged as a matter of course. In an embodiment, the highest 2, 3, 4, 5, 10, 15, 20, 25, 30, 35, or 40%, or more, or any value or range of values therebetween in 1% increments scales can be automatically prolonged by default, while the remaining scales are subjected to the Tukey windows centered on the activation time. That said, curve fitting techniques and/or statistical data analysis techniques can be utilized to identify the scales where onset and/or duration of the initial cosine taper should be prolonged. In an embodiment, a judgment call must be made as to which frequencies should be subjected to the strict Tukey windowing in which frequencies are more appropriate to be expanded outside the more limited window. In an embodiment, this can be predetermined. In an embodiment, a product of a trained neural network can be utilized to identify which scales are to be manipulated in which manner, based on observation or otherwise analysis of data where a technician made the judgment as to which scales should be modified in which manner. In an embodiment, this can depend on the mother wavelet time scale and/or the rate of fibrillation. In some embodiments, the judgement call can be whether to use the longer duration window for scale 10 only or to extend to scale 9, scale 9 and 8 (or potentially just 8), scale 9, 8 and 7 (or one or two of them and not 9 and/or not 8), scale 9, 8, 7 and 6, and so on. In an embodiment, a trial and error approach can be used and the results can be evaluated, which evaluation can be based on a predetermined set of requirements or can be based on knowledge of one of skill based on visual inspection. In an embodiment, the judgement(s) will be related to what underlying frequency is intended to be captured / desired to be captured. Shorter timescales can use lower scales, which capture faster activation. In an embodiment, the shorter timescales can capture up to 80, 90, 100, 110, 120, 130, 140, 150 Hz or more or any value or range of values therebetween in 1 Hz increments. In the embodiment presented in the figures, it is desired to capture up to 60 or 70 Hz. 5 to 10 Hz would be considered a low frequency component, which components can be present for repolarization. Accordingly, the scales that are extended can be the scales that provide fair indication of repolarization. Thus, in an embodiment, the scales are adjusted to the frequency one seeks to address. One of skill will make a tradeoff between the increased scale and the costs associated therewith, such as, for example, the fact that longer scales can give the ability to get rid of the slowly varying background noise (which is useful in some embodiments). This can be something that is not obtained as cleanly with a standard continuous filter.
Attorney Docket No.246-021PCT [00337] In an embodiment, the frequencies / scales that are subjected to “strict windowing” (symmetric tapering about the activation time) are those that have a maximum amplitude that is less than and/or equal to 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97 or 98% or more or any value or range of values therebetween in 1% increments of the average amplitude (mean, median and/or mode, depending on the embodiment) within the window and/or the average (mean, median and/or mode) of the positive and/or negative amplitude (for the positive, the negative is ignored (rectification) and vis-a-versa) - the maximum amplitude would be the maximum positive if positive average and negative if average negative for example). In an embodiment, the frequencies / scales that are subjected to the “adjusted windowing” (non-symmetric tapering about the activation time) are those that have a maximum amplitude that is equal to and/or greater than 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98 or 99% or more or any value or range of values therebetween in 1% increments of the average amplitude (mean, median and/or mode, depending on the embodiment) within the window and/or the average (mean, median and/or mode) of the positive and/or negative amplitude. [00338] In an embodiment, the frequencies / scales that are subjected to “strict windowing” are those that have a variation from zero or from a predetermined range of no more than a certain amount. In an embodiment, the frequencies / scales that are subjected to “adjusted windowing” are those that have a variation from zero or from a predetermined range of more than a certain amount. In an embodiment, the frequencies / scales that are subjected to “strict windowing” are those that have more than a certain number of maxima and/or minima, such as for example, more than or equal to for example 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 or 20 or more or any value or range of values therebetween in 1 increment maxima and/or minima (collectively or singularly depending on the embodiment) within the timescale of the given window and the opposite would be the case for the adjusted windowing (those that have less than or equal to 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 or more or any value or range of values therebetween in 1 increment maxima and/or minima (collectively or singularly depending on the embodiment). [00339] In an embodiment, at least 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98 or 99% or more or any value or range of values therebetween in 1% increments of the values over the time are within 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85 or 90% or any value or range of values therebetween in
Attorney Docket No.246-021PCT 1% increments of the average (mean median and/or mode) or of a predetermined value / range of values for the scale to be subjected to strict windowing, and at least 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60 or 65% or more or any value or range of values therebetween in 1% increments of the values over the time are outside of 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85 or 90% or any value or range of values therebetween in 1% increments of the average (mean median and/or mode) or of a predetermined value / range of values for the scale to be subjected to adjusted windowing. [00340] The timing noted above can be the timing of the taper windows. Conversely, the timing could be a timing before and/or after the activation timing in the embodiments above. In an embodiment, such can be used to establish the windows for the other embodiments, all by way of example. [00341] FIG.41D shows the results of the wavelet decomposition modified with the results of the Tukey windows (where scales 1-8 show the result of the symmetrical window and where scales 9 and 10 show the results of the non-symmetrical window / the elongate window). FIG. 41E shows the reconstructed electrogram (upon the application of the inverse transform). The result of this processing is shown in FIG. 43 where high frequency electrical activity in the period after local activation is attenuated but low frequency variation is preserved. FIG. 43 shows reconstructed electrograms from the processed wavelet components. [00342] Window lengths and cosine tapering periods for each scale can be adjusted but a single set of default values can be used (and is used for all results presented here, all by way of example). [00343] In view of the above, there is a method, method 1190, as seen in FIG.44, which includes method action 1192, which includes obtaining first data based on one or more unipolar EGMs recorded in an atrium of a living human afflicted with atrial fibrillation. The method further includes method action 1194, which includes reconstructing one or more unipolar EGMs based on data based on wavelet processing of data based on the first data. The wavelet processing can be any of the processing detailed above / herein. In an embodiment, the actor of method 1190 executes the wavelet processing, while in another embodiment, the actor obtains the data based on wavelet processing of data based on the first data, and identifies the regional activation times accordingly. In an embodiment, the action of reconstructing is executed by filtering across time scales in CWTs calculated with second derivative Gaussian mother wavelets. Corollary to this is that the data based on wavelet processing is data based on results of filtering
Attorney Docket No.246-021PCT across time scales in CWTs calculated with second derivative Gaussian mother wavelets. In an embodiment, the action of reconstructing is executed by filtering across time scales in CWTs calculated with only second derivative Gaussian mother wavelets. Corollary to this is that the data based on wavelet processing is data based on results of filtering across time scales in CWTs calculated only with second derivative Gaussian mother wavelets. [00344] In an embodiment, there is an alternate version of method action 1192, which includes obtaining first data based on electrical activity in a living human. In an embodiment, the first data is based on one or more unipolar EGMs, which correspond to the electrical activity. In embodiment, the first data is based on one or more unipolar EGMs, which represent the electrical activity, and the one or more unipolar EGMs are recorded in a heart of the living human, and in embodiment, the electrical activity is electrical activity in an atrium of the living human. Again, in an embodiment, the electrical activity etc. can be in another part of the living human. In an embodiment, the electrical activity is electrical activity in an organ of the living human. The electrical activity is electrical activity in a heart of the living human. In an embodiment, electrical activity and/or electrical phenomenon detailed herein, etc., is electrical activity in a muscle of the human. In an embodiment, the electrical activity and/or electrical phenomenon detailed here, etc., is electrical activity in an arm muscle and/or a leg muscle and/or a finger muscle and/or an eye muscle and/or a muscle of the face, etc. Consistent with the teachings detailed herein, human may or may not be a human afflicted with atrial fibrillation, and the human may or may not be a healthy human or may or may not have a heart that is healthy. The heart could have a heart with a hole therein and/or the living human has at least one heart stent. Again, the human could have one or more ailments but not atrial fibrillation. Again, the electrical activity can be organized or disorganized. And as noted above, the human may not be afflicted with atrial fibrillation. Thus, there is method 1191 as represented in FIG.44A, which has method action 1193 instead of method action 1192, which includes obtaining first data based on one or more unipolar EGMs of activity in a living human. In an embodiment, method action 1193 is substituted for method action 1192. [00345] In an embodiment, the action of reconstructing is executed by using a wavelet-based filter that is synchronized with local activation and that matches known temporal variation of electrical activation, and thus the data based on wavelet processing is synchronized with local activation and that matches known temporal variations of electrical activation. Embodiments can include the action of filtering for high instantaneous frequency components proximate atrial activation times and filtering for lower frequency components away from the atrial
Attorney Docket No.246-021PCT activation times. By way of example, the Tukey windows detailed above can be used to implement this filtering, where the lower frequency components are in the scale(s) that have the extended windows. Thus, in an embodiment, the data based on wavelet processing is data where the data based on wavelet processing has been filtered accordingly. [00346] Note that the method 1190 is a method that includes expanded actions, such as any of the actions associated with V artifact removal and/or the activation time identification. In this regard, there is method 2190, as represented in FIG.45, which includes method action 2192, which includes executing method action 1192. Further, method 2190 includes method action 2194, which includes identifying atrial activation times based on data based on the first data. In an expanded embodiment, there are some actions associated with method action 2194, such as any one or more of the method actions detailed above with respect to the embodiment of identifying atrial activation times. The method also includes action 2196, which includes executing method action 1194, where the action of reconstructing includes filtering out components based on the identified activation times. Figure 46 presents an exemplary flowchart for an exemplary algorithm for an exemplary method, method 2199, which includes the actions of method 2190 detailed above, with the caveat that follows, with the additional action of method action 2193, which entails removing V artifact from data based on the first data. And note that this data based on the first data could be different data based on the first data in that for example, the data based on the first data utilized in method 1194A could have the Tukey windows applied thereto. With respect to the caveat, method action 2194A parallels method action 2194, except that this action requires that the action based on the first data is data resulting from the removal of V artifact from the first data. And any of the actions detailed above associated with removal of the V artifact can be undertaken in method action 2193. [00347] In an embodiment, the wavelet processing of methods 1190, 2190 and/or 2199 includes obtaining wavelet decompositions for a plurality of scales and the method further includes tapering the obtained wavelet decompositions based on atrial activation times. In this exemplary embodiment, the atrial activation times can be based on data based on the decompositions for the plurality of scales. Note that these decompositions could also be utilized to execute method action 1194 by way of example. Alternatively, the decompositions could be reexecuted or otherwise can be different in some other embodiments. In this regard, it is noted that in some embodiments, the scales that are utilized across the various embodiments can be the same while in other embodiments, the scales that are utilized across the various embodiments could be different. Any scale of use that can enable the teachings
Attorney Docket No.246-021PCT detailed herein can be utilized in at least some exemplary embodiments, providing that the art enables such, unless otherwise noted. Note further that in some embodiments, more scales can be utilized for, for example, the V artifact removal process than that which is utilized for the atrial activation time identification and/or the reconstruction of the atrial EGM. In an embodiment, more utilized for the former than the latter to and in another exemplary embodiment, more utilized for the latter than the former to, in an exemplary embodiment, more utilized for the second than the first, and more utilized for the third than the second, and in an embodiment, more utilized for the second than are utilized for the first and third etc. [00348] Corollary to the above is that in an exemplary embodiment, the data based on wavelet processing of data based on the first data in method action 1194 can be the results of the tapering of the obtained wavelet decompositions based on atrial activation times. [00349] But to bring things back a step or two, in an exemplary embodiment, the tapering includes tapering for at least some finer wavelet scales, which tapering for the at least some finer wavelet scales is centered on the atrial activation times. Consistent with the teachings above with respect to the coarser scales, in an embodiment, the tapering includes tapering for at least one coarser wavelet scale(s), which tapering for at least one coarser wavelet scale(s) is off centered from the atrial activation times. [00350] While embodiments above have been described in terms of the Tukey window, any window function and/or tapering function that can enable the teachings detailed herein and otherwise provide utilitarian result can be utilized in at least some exemplary embodiments, providing that the art enables such, unless otherwise noted. Is also the case with the Hann windows detailed above. Any arrangement for implementation of a mathematical function that is zero valued outside some chosen interval, whether symmetric or not symmetric around the middle of the interval, which may or may not approach a maximum in the middle or may or may not approach a maximum offset from the middle, and may or may not be tapered away from the middle or wherever the offset is determined to be, can be utilized in some embodiments providing that the art enables such. [00351] In an embodiment, the wavelet processing includes obtaining wavelet decompositions for a plurality of scales. In an embodiment, the methods detailed herein such as methods 2199, 2190, and/or 1190, further include the action of window function processing the obtained wavelet decompositions based on atrial activation times. In this exemplary embodiment, window length and a period of tapering for at least some scales are set to preserve components
Attorney Docket No.246-021PCT associated with atrial activation. In an embodiment, onset and duration of initial taper for at least some other scales different from the at least some scales (e.g., scale 10 in FIG. 41B) is consistent with that of the at least some scales and window length for the at least some other scales is prolonged relative to the window length of the at least some scales. In an embodiment, window length and a period of tapering for at least some scales are set to preserve components associated with atrial activation and window length for at least some other scales is prolonged relative to the window length of the at least some scales so that final taper is synchronized with next atrial activations. [00352] In view of the above, it can be seen that in some embodiments, the wavelet processing attenuates higher frequency electrical activity in a period after local atrial activation and low frequency variation is preserved. [00353] In an embodiment, correspondence between ground-truth (GT) and processed (P) EGMs was quantified by evaluating the correlation coefficient (CC) and normalized root- mean-squared error (nRMSE). ^^^^
where ^^^^ ^^^^ is the number of activation times identified correctly (±5 msec) and ^^^^ ^^^^ the total number detected. Activation time differences between ground-truth (GT) and processed (P) recordings can be estimated as ∆T=|ATGT - ATP| (9) Signal processing functions can be written in the MATLAB programming language (The Mathworks, Natick, Massachusetts). Distributions can be plotted as box plots and the significance of differences between them was estimated using the non-parametric Mann- Whitney U test. Data can be expressed as mean±SD when normally distributed but can
Attorney Docket No.246-021PCT otherwise represented as median [interquartile range]. Below we explain an exemplary scenario based on this noted data analysis regime. [00354] The frequency content of the activation complex in atrial EGMs for patients in longstanding (persistent of permanent AF) is from 10-100Hz whereas the frequency content of the repolarization complex ranges from ~2-10 Hz, all by way of example. Wavelet scale 1 equates roughly with 80 Hz and scale 9 equates roughly with 8 Hz in this embodiment and scale 10 equates to roughly 5 or 6 Hz in this embodiment. In an embodiment, the scale 1 is any value or range of values between 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 105, 110, 115, 120, 125, or 130 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 2 is any value or range of values between 35, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 105, 110 or 115 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 3 is any value or range of values between 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100 or 105 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 4 is any value or range of values between 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, or 85 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 5 is any value or range of values between 20, 30, 35, 40, 45, 50, 55, 60, 65 or 70 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 6 is any value or range of values between 15, 20, 30, 35, 40, 45 or 50 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 7 is any value or range of values between 10, 15, 20, 30, 35, 40 or 45 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 8 is any value or range of values between 5, 10, 15, 20, 30 or 35 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 9 is any value or range of values between 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 15, 17, 18, 19, 20, 21, 22, 23, 24 or 25 Hz or any value or range of values therebetween in 1 Hz increments. In an embodiment, the scale 10 is any value or range of values between 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 15, 17, 18, 19 or 20 Hz or any value or range of values therebetween in 1 Hz increments. All of these are by example only. In an embodiment, any one or more scales can have these values increased or decreased by 5, 10, 15, 20, 25, 30, 35, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 175, 200, 250, 300, 350 or 400% or more or any value or range of values therebetween in 1% increments, and such need not be the same across all scales. [00355] In an exemplary embodiment, the teachings above can be validated vis-à-vis ventricular far-field artifact subtraction. In this regard, Fig. 47 presents a representative
Attorney Docket No.246-021PCT example of wavelet-based V subtraction from unipolar EGMs recorded from adjacent HD grid electrodes in a patient with PeAF. Overlaid records before and after processing in Figs 47A & 47B indicate that this subtraction algorithm removes V artifact while preserving near synchronous atrial EGMs. In Fig 47C, bipolar EGMs constructed from the processed unipolar records are near identical to corresponding bipolar signals output from the mapping system showing that V subtraction has had minimal impact on atrial components in the two unipolar channels. Further, it is noted that FIG. 47 provides validation of wavelet-based V artifact subtraction. FIG. 47A provides unipolar EGMs recorded at adjacent electrodes of Advisor™ HD grid catheter before (grey) and after (blue) V artifact subtraction. FIG.47B provides bipolar EGMs for electrodes in (A) recorded from mapping system (red) and constructed from unipolar EGMs in (A) after V artifact subtraction. FIG. 47C provides for, following addition of a V signal component to "gold standard" unipolar atrial EGMs, artifact was removed with wavelet- based method and by subtracting mean and median QT templates synchronized with V activation. The correspondence of all three approaches with the "gold standard" atrial signal is indicated by the box-plots of correlation coefficient and normalized RMS error above. P value <1.0 x10-5 indicated by ***. [00356] Systematic comparison of our wavelet-based artifact cancellation with subtraction of fixed mean or median QT templates demonstrates that this technique has utilitarian value beyond either alternative. The effects of V subtraction on synchronous atrial EGMs can be evaluated as outlined herein across 50 successive QT intervals. Figure 47C shows that with wavelet-based correction CC (0.94 [0.01]) was significantly better (p <1.0x10-5) than both mean or median QT template subtraction (0.64 [0.11] and (0.65 [0.13], respectively) and that these were not different from each other (p=0.873). Corresponding results for nRMSE were 0.08 [0.01] compared with 0.22[0.06] and 0.22 [0.07] with the wavelet-based correction again a highly significant (p <1.0x10-5) improvement on mean or median QT template subtraction, which were not different from each other (p=0.764). Embodiments include methods, devices, and/or systems disclosed herein where the results are at least 50%, 55%, 60, 65, 70, 75, 80, 85, 90, 95, or 100%, or more as good as these results. [00357] With respect to the identification of atrial activation, the wavelet-based method of identifying atrial activation is based on detection of elevated signal power across characteristic combinations of time and frequency. Representative results as noted above with respect to FIGs. 33A and 33B indicate that thresholds for the duration and magnitudes of power in individual wavelet scales can be set to segment different classes of local electrical activity. In
Attorney Docket No.246-021PCT cycle a, power exceeds threshold for 18ms with a single peak during this period in wavelet decompositions for all scales. In cycle b, however, lower levels of power are sustained for much longer (78ms) with multiple lower amplitude peaks in the wavelet decomposition during this period, both supporting classification as a fractionated EGM. Three fractionated EGMs identified on this basis are shown in the figures. Again, in an embodiment, there are methods of identifying the fractionated activity based on identifying the presence of two or three or four or five or six or more lower amplitude peaks relative to the single high peak in the non- fractionated activity over a sufficiently longer time, such as 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, 6, 6.5, 7, 7.5, 8, 9, or 10, or more, or any value or range of values therebetween in 0.1 increments times the time of the single peak that exceeds the threshold, such as the average thereof with respect to the nonfractionated activities. In an embodiment, the prerequisite is that the occurrence of the peaks must occur by at least and/or equal to the aforementioned times over the given period that is specified, and in some embodiments, there must be a certain amount of time between the peaks, such as at least 5, 10, 15, 20, 25 or 30% or more or any value or range of values therebetween in 1% increments of the total time to which the analysis is limited (e.g., five times the time that the single peak was above the threshold). [00358] In an embodiment, the methods used here for detection and classification of atrial activation are not affected by the magnitude of EGM components. As shown in FIG. 33B, thresholds used to detect the beginning and end of activation can be set for individual scales based on the previous power in that scale. Likewise, activation times can be estimated from maxima during activation in decompositions at different wavelet scales. The independence of identification on magnitude across wavelet scales is underscored by the fact that time series records for wavelet decomposition and associated power for the 5 scales presented in 33B and 33C are normalized. In some embodiments, there is a high level of consistency between activation characteristics detected at different scales and in this case we have selected scale 5 as best matching the EGM components that we are seeking to identify. Finally, a second possible activation is evident at scales 1 and 3 in cycle a, 37ms after initial detection of activation. This can be judged as being nonlocal because similar synchronous deflections were recorded at adjacent electrodes and markedly attenuated in associated Laplacian EGMs. [00359] In Fig. 48, an exemplary comparison of the performance of the teachings herein with Laplacian difference EGMs in the presence of wide-band non common-mode noise. Briefly, this shows a comparison of wavelet-based and Laplacian atrial activation detection in presence of uncorrelated wide-band noise. FIG. 48A is a Schematic of Advisor™ HD grid catheter
Attorney Docket No.246-021PCT showing 5 electrodes used to construct Laplacian difference signal B3. FIG. 48B shows a unipolar signal recorded at B3 with V artifact removed. Green dots represent activation times detected using wavelet-based method. FIG.48C shows Laplacian signal at B3 without added noise. FIG.48D shows Unipolar signal with white noise added to reduce SNR to -10dB. Red dots indicate activation times detected using wavelet-based approach. FIG. 48E shows Laplacian signal at B3 with white noise with the same distribution as at B3 also added at A3,B2,C3 and B4. FIG. 48F shows comparison of detection accuracy and timing error for activation times detected in unipolar signal in B) using wavelet-based method (red) and from corresponding Laplacian signal with progressive addition of wide-band white noise. Noise and SNR are referred to the unipolar signal B3 but the same noise power is also applied to the 4 other electrodes in the Laplacian signal. [00360] More specifically, unipolar recordings acquired at 5 adjacent sites with an HD grid catheter (Fig.48A) are processed as outlined herein and a Laplacian difference signal is formed from these data. While the Laplacian difference signal attenuates components that are common across electrodes, it also sharpens atrial activation complexes and amplifies uncorrelated noise (compare Figs. 48B and 48C). Addition of bandlimited Gaussian white noise sufficient to reduce SNR in unipolar B3 to -10dB lowered wavelet-based detection accuracy in Fig.48D to 88%, with the same number of activation complexes detected but incorrectly identified in 3 cases as preceding the actual time by 46 to 52 msec. However, the same noise has a substantially greater impact on Laplacian EGMs; SNR can be reduced to -14dB and detection accuracy dropped to 10.4% with a timing error of 43.2±25.0 msec. The robustness of the wavelet-based detection in the presence of increasing levels of uncorrelated noise is demonstrated in Figure 48F where accuracy and timing error are plotted as functions of decreasing SNR for unipolar electrode E5. Detection accuracy for the former is >80% with minimal timing error until SNR is < -10dB. In contrast, both are rapidly degraded with Laplacian difference signals when SNR drops below +10dB. Embodiments include methods, devices, and/or systems disclosed herein where the results are at least 50%, 55%, 60, 65, 70, 75, 80, 85, 90, 95, or 100%, or more, or any value or range of values therebetween in 1% increments of those presented. [00361] With respect to near-field atrial signal reconstruction, representative results obtained with wavelet-based reconstruction of unipolar atrial EGMs are presented in Fig. 49. Application of our time-dependent filter synchronized with atrial activation time preserved signal components associated with local atrial activation and this is verified by the consistency
Attorney Docket No.246-021PCT of timing and magnitude between reconstructed unipolar and corresponding Laplacian EGMs (compare FIG.49B and 49C). As expected though, there are clear differences in morphology; Laplacian EGMs are shorter in duration with peak amplitudes overlapping the maximum negative slope of unipolar EGMs (Fig. 49D). Attenuation of fine wavelet scales following activation resulted in stable electrical activity during this period, enabling estimation of an activation recovery interval of -220.8±13.5 msec in Fig.49B at a base cycle length of 363.2±7.0 msec. Wavelet-based EGM reconstruction proved extremely robust in the presence of noise. Addition of white Gaussian noise to the unipolar EGMs in Fig.49A sufficient to reduce SNR by > 20 dB had surprisingly little effect on the EGMs recovered. CC and nRMSE were preserved with progressive degradation occurring only when SNR was reduced below -10 dB (see Fig.49E). That is, performance of wavelet-based reconstruction of atrial unipolar EGMs can be shown by way of example by FIG.49. FIG.49A shows unipolar EGMs acquired using Advisor™ HD grid catheter with V artifact subtracted and FIG.49B shows corresponding near- field EGMs signals reconstructed as outlined above. Reconstruction is synchronized with activation times detected from FIG.49A, dashed red lines indicate activation times identified. FIG. 49C shows Laplacian EGM corresponding to unipolar EGM in FIG.49A. Voltage and time scales in upper panels apply to FIGs.49 (A), (B) and (C). FIG.49D shows representative reconstructed unipolar EGM cycle (blue) with corresponding Laplace EGM (red) overlaid. Timing within record indicated by grey shading in (B) and (C). FIG. 49E shows correspondence between reconstructed EGM in (B) with progressive addition of band-limited white Gaussian noise sufficient to reduce SNR to < -20dB. Embodiments include methods, devices, and/or systems disclosed herein where the results are at least 50%, 55%, 60, 65, 70, 75, 80, 85, 90, 95 or 100% or more as good as these results. [00362] Spatial difference EGMs that can provide a streamlined way of removing common mode noise recorded with multi-electrode arrays (including ventricular far-field artifacts) and enable robust identification of local activation times. However, important information is lost and difference signals cannot be used for inverse intracardiac potential mapping. The teachings herein present novel wavelet-based methods for extracting near-field atrial EGMs recorded during AF that are fast (2, 3, 4, 5, 6, 7, 8, 9 or 10 times faster than alternative methods mentioned above), enable more reliable estimation of atrial activation times in the presence of noise than difference-based methods such as the Laplacian filter (where reliability can be increased by 10, 20, 30, 40, 50, 75, 100, 125, 150, 175, 200, 250, 300, 350 or 400% or more or any value or range of values therebetween in 1% increments over the other methods, for at least
Attorney Docket No.246-021PCT 50, 55, 60, 65, 70, 75, 80, 85 or 90% of patients out of 100 patients tested), and provide robust direction-independent information on local activation and recovery that cannot be obtained with spatial difference methods. In addition to the standard signal processing workflow (bandpass filtering and application of a notch filter to remove power line noise), the teachings herein can include an approach involves 3 key steps: 1) estimation and subtraction of artifact due to ventricular activation using a method that accounts for beat-to-beat variation of this far- field signal component 2) identification and classification of local atrial activation from distributions of instantaneous power following removal of V artifact, and 3) reconstruction of near-field atrial unipolar EGMs using a "matched" filter synchronised with atrial activation times. The teachings herein can be based on the fact that both V artifact and atrial near-field EGMs have characteristic time-varying frequency “fingerprints” that can be identified and separated using wavelet-based methods. The teachings herein contrast from the variety of methods have been used to subtract V artifact from unipolar EGMs in AF ( Slocum et al., 1985; Shkurovich et al., 1998; Castells et al., 2005; Salinet et al., 2010; Ahmad et al., 2011; Salinet et al, 2013). The teachings herein can include ensemble-averaging of unipolar EGMs synchronized with ECG timing accurately characterizes the mean V artifact in AF, while accounting for beat-to-beat variation in this far-field signal component. As a result, removal of mean or median electrical activity within fixed QT templates aligned with V activation reduces artifact, and in embodiments, there is little to no subtraction residues. The teachings herein can be implemented without principal component analysis to account for variation in QRST morphology (Castells et al., 2005) and/or in a computationally efficient manner. [00363] The teachings herein can have near real-time cancellation achieved by continuous estimation of a QT subtraction template weighted heavily by the R peak of the current V activation in some embodiments. The teachings can have wavelet based methods instead to scale the QRS complex in the average QT template so that it matches current V activation amplitude. The analysis presented in Figure 47 indicates that atrial activation complexes are recovered from overlying V artifact with excellent fidelity although some residue remains with nRMSE across the QT window reduced to ~8% from ~22% for fixed QT template subtraction. [00364] Precise identification and classification of local activation can be utilitarian for effective analysis of factors responsible for atrial rhythm disturbances, and the teachings herein can provide such. The teachings herein provide methods that do not magnify error introduced by non common-mode noise. In AF, classification accuracy is also compromised by the complexity of EGMs which are often fractionated and this has been addressed using template
Attorney Docket No.246-021PCT matching schemes based on data sets of “characteristic” unipolar atrial activation patterns. The wavelet decomposition used here in some embodiments is a generalization of a “matched filter” approach. Embodiments use the morphology of 1st-differential Gaussian wavelets closely resembling that of the unipolar activation complex, and corresponding wavelet decompositions effectively reliably identify such events in the presence of wide-band noise because the probability of random occurrence of such characteristics across scales is very low. Furthermore, wavelet based methods provide a sensitive means of identifying activation that is not biased by the magnitude of signal components in unipolar EGMs as is the case with conventional gradient measures. Embodiments include recording from 2D (or 3D) arrays with appropriate resolution. [00365] Duration of activation and number of peaks detected provide a straightforward basis for segmenting and further classifying normal and fractionated activation. [00366] As seen above, a wavelet-based filter can also be used to reconstruct atrial electrograms. In this case, scale-weighted windows synchronized with atrial activation time can be used to attenuate fine scale wavelet components not associated with the current activation but to pass coarse scale components throughout most of the cycle. The filter according to an exemplary embodiment passes only high frequency signal components only during activation, and captures the subsequent low frequency recovery of potential and then restores the baseline immediately prior to the next activation. As with wavelet-based identification of activation time, these reconstruction methods can be extremely robust in the presence of noise and for very similar reasons. This can be a matched filter, in this case based on a priori knowledge that electrical activity is characterized by high frequency components during activation but lower frequencies following this. The recovery in potential following activation may not be linked with the delayed attenuation of coarse-scale wavelet components. In an embodiment, this is not interpreted as local atrial repolarization. [00367] The three wavelet-based filters outlined above can have a relatively large number of parameters which specify wavelet time scales, scale-based attenuation characteristics etc. and adjustment of some of these is utilitarian to optimize performance. For example, the wavelet- scale at which activation time is identified impacts the accuracy with which it can be estimated and the most appropriate setting can vary between studies. That said, the methods described are relatively consistent across scales and the default values tabulated are used for all examples presented here (with respect to the figures – other examples are presented). Despite the apparent complexity of the signal processing methods used, in at least some embodiments, they
Attorney Docket No.246-021PCT are computationally efficient and can be completed in near real-time, in part because the Fast Fourier Transform is used to compute CWT decompositions. In this study, most emphasis is placed on estimation of atrial activation times and recovery of near-field electrical activity from individual unipolar recordings. However, the results demonstrate that these methods are particularly powerful for analysing signals acquired with high resolution multi-electrode catheter arrays where signals recorded on adjacent electrodes can be used to validate and extend the findings. In this setting, these wavelet-based signal processing tools provide a platform for more sophisticated 2D (or 3D) analysis, for instance using phase analysis to investigate underlying local mechanisms responsible for rhythm disturbance. [00368] The above thus presents a rich set of novel, computationally efficient, wavelet-based signal processing tools that can be used to extract detailed information about local electrical activity from unipolar atrial EGMs in AF. The teachings herein are extremely robust in the presence of noise because it can employ matched filtering informed by a priori knowledge of the temporal variation of frequency content in cardiac extracellular potentials generated by electrical activation. At least some of these embodiments are such that the methods are readily generalizable to clinical mapping applications with high-resolution multi-electrode catheter arrays. [00369] Embodiments include spatial mapping technologies, or more accurately, the implementation of the spatial mapping technologies detailed herein. The spatial mapping technologies may or may not be associated with atrial fibrillation. The spatial mapping technologies may be implemented for a heart that is not afflicted with atrial fibrillation. The teachings detailed herein do have utilitarian value with respect to mapping and evaluating a heart where there is irregular activity. But the teachings detailed herein are also applicable to mapping a heart where there is regular activity (whether known beforehand or not, or at least suspected beforehand or not), which regular activity falls within any of the human factors engineering values detailed herein unless otherwise noted. [00370] In an embodiment, there are methods that include using the teachings herein to identify re-entrant pathways in atrial tachycardia and/or ventricular tachycardia. Exemplary embodiment, these are regular rhythms that can be some instances be traced utilizing existing mapping techniques. In this regard, it is noted that in at least some exemplary scenarios, those existing mapping techniques cannot be utilized to trace the irregular rhythms of atrial fibrillation. Conversely, the teachings detailed herein can be utilized to trace the irregular rhythms of atrial fibrillation, which is thus an example of a utilitarian value of the teachings
Attorney Docket No.246-021PCT detailed herein. So, embodiments of the teachings detailed herein can be distinguished from such prior mapping techniques because such prior mapping techniques cannot be used trace such rhythms, at least not in a manner that has efficacy, or at least medical efficacy, and certainly not in the manner providing the accuracy and/or specificity that results from the teachings detailed herein. [00371] Still, referring back to the reentrant pathways in atrial tachycardia and ventricle tachycardia, the techniques detailed herein can be utilized to identify such. Accordingly, the teachings detailed herein are not limited to irregular rhythms, but include the regular rhythms of such. Any disclosure herein of tracing and/or mapping irregular rhythms or otherwise applying the teachings detailed herein disclosed in relationship to working with a heart that is afflicted with atrial tribulation corresponds to an alternate disclosure of implementing any one or more the teachings detailed herein to a heart that has regular rhythms, and otherwise implementing any one or more of the teachings detailed herein identifying reentrant pathways in atrial tachycardia and or ventricle tachycardia providing that the art enable such, unless otherwise noted, all in the interest of textual economy. [00372] Embodiments can also include implementing the teachings detailed herein with respect to an analysis that does not involve all of the specifics of the sophisticated (relative) spatial mapping technologies detailed herein. In an exemplary embodiment, there is its analysis that involves by way of example only and not by way of limitation, assessing a single lead electrogram and extracting atrial components in the presence of ventricular components. In an exemplary embodiment, the ventricular components are likely to appear to be dominant. The teachings detailed herein can be applied to such in some embodiments by way of example. In this regard, the teachings detailed herein can be utilized to overcome or otherwise account for the fact that ventricular components are often dominant in the resulting data collection, thus the techniques detailed herein can be utilized to assess the atrial components even in the presence of the dominant ventricular components. Embodiments include devices and systems that enable such and/or methods of doing so. To be clear, embodiments can include extracting atrial data, which can be valuable atrial data, from signals that appear to be dominated by noise. Typically, the noises from ventricle activity. But to be clear, the noise could come from other activity or other sources. Ventricle activity is disclosed herein as but one example of activity or otherwise a phenomenon that can cause noise where the teachings detailed herein can be utilized to extract the valuable data from the signal that is dominated by data that is not wanted
Attorney Docket No.246-021PCT (which data could also be valuable, it is just not wanted, or at least the underlying hidden data is wanted more than the dominant data). [00373] Embodiments include extracting atrial data, which atrial data can be utilitarian, from signals that appear to be dominated by noise, again which noise can be but is not limited to ventricle activity. This atrial data may or may not be associated with atrial fibrillation, as noted above. Again, the heart from which the signals are obtained for otherwise the body tissue of the human from which the signals are obtained may be one that is not afflicted by atrial fibrillation or otherwise any ailment. [00374] Also, there is utilitarian value with respect to avoiding, in part or in totality, reliance on coupling improved atrial signals into a modeling environment for assessing spatial atrial activity. Embodiments include devices and systems that when utilized, avoid such, in part or in total, and methods that also avoid such, again in part or in total. Embodiments can and do include utilizing such (such can provide utilitarian value with respect to improving the accuracy of the spatial analysis). Embodiments may also not use such. In this regard, there can be utilitarian value or otherwise just a desirable goal of simply having atrial signals which are clean or otherwise more clean than that which is the case with respect to implementing other techniques, and or are not swamped or otherwise camouflaged by the ventricular activity, or by way of example only and not by way of limitation. [00375] Note that in some embodiments, the potentials measured using the basket catheter for example, the potentials can be between the respective electrodes of the one or more electrodes and a common reference electrode or a plurality of respective common reference electrodes - in some embodiments, a plurality of electrodes but not all the electrodes could use one reference electrode, and a plurality of electrodes but not all the electrodes could use another reference electrode, etc.). Note that in principle, the unipolar potential is the potential measured at the electrode. In practice, one can at least usually only measure potential difference and thus in an embodiment the reference electrode is set to zero volts, and here, this can be done by forming a Wilson central terminal or otherwise averaging voltages around closed "triangle" on the body surface as far from the heart as possible. Conservation should make this as close to zero as possible. Embodiment can use catheters with a far-field reference which might be argued to allow V artifact to be subtracted. While such certainly minimizes V artifact, there are differences in the V artifact measured at different electrodes and this results in subtraction residues.
Attorney Docket No.246-021PCT [00376] Embodiments can include techniques that are at least effectively agnostic, or otherwise not affected, by the direction in which activation wavefronts propagate with respect to the recording electrodes. Embodiments can also be implemented where useful data is obtained when activation spreads perpendicular to the electrode array (that is, the data obtained by the arrays can be effectively used in such a scenario, including used effectively). [00377] Embodiments can include obtaining one or more of the utilitarian data herein without interpolation across the interval from the onset to the end of ventricular depolarization. Embodiments include obtaining data where at least 80, 85, 90, 95, or 100%, or any value or range of values therebetween in 1% increments of the near-field atrial activity (as measured based on energy level) that overlaps ventricular activation is not lost / is preserved and is otherwise not distorted. Embodiments can avoid the use of subtraction of time-averaged estimates of far-field activity during ventricular activation and repolarization to obtain the data detailed herein. Adaptive algorithms can also be avoided, at least with respect to such that are designed to account for beat-to-beat variation in ventricular (V) artifact. [00378] Embodiments include utilizing the results of one or more of the method actions detailed above with respect to V artifact removal, atrial activation and/or atrial EGM reconstruction. [00379] In some embodiments, the actions of executing any one or more or all of: (1) the wavelet processing and the implementation of the window functioning and the Fourier transformation and the subtraction of the results from the initial EGM to develop the EGM without the V artifact, starting with the obtained first data based on one or unipolar EGMs recorded in an atrium (2) the wavelet processing and the positive power development (and in some embodiments, the power maxima identification and/or the selection of scale and/or the activation timing) starting with the obtained first data based on one or unipolar EGMs recorded in an atrium; and/or (3) the wavelet processing and the application of the window functioning and the Fourier transformation and the reconstructing the atrial EGM are executed within a period of no more than 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes or within 50, 40, 35, 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4 or 3 seconds of the beginning of the calculations or any value or range of values therebetween in one second increments. Note also that in some embodiments, the actions of developing a time-varying electrical potential map (based on the wavelet filtering, with in some embodiments and without another embodiments actually performing the calculations therefore (as opposed to starting with data that is already wavelet processed, such as for example, starting with a reconstructed atrial EGM), developing the time- varying phase map, and identifying the repeating phase signatures are executed within a period
Attorney Docket No.246-021PCT of no more than 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes. Indeed, in some embodiments, the time averaging phase data reaches convergence within 40, 35, 30, 25, 20, 15, 10, 9, 8, 7, 6, 5, 4, or 3 seconds of the beginning of the calculations of the maximum phase gradients. (Convergence can mean that the statistically non-aberrant numbers (e.g., the zero values) average out to a level number. [00380] In the interest of textual economy, any temporal disclosure herein of executing any one or more or all of the method actions detailed herein corresponds to a disclosure where any one or more or all of the method actions associated with the wavelet filtering embodiments to develop the V artifact data and/or the EGM with the V artifact removed and/or the atrial activation timing based on wavelet filtering and/or the reconstructed EGM based on wavelet filtering are executed within that temporal period. Embodiments also include scenarios where the temporal disclosures herein do not include one or more or all of those actions. For example, the time varying electrical potential map could start with the reconstructed atrial electrograms based on wavelet filtering. Still, the temporal features detailed herein can apply to embodiments where, for example, the action of developing a time-varying electrical potential map is based on and includes actually taking invasive readings taken while a human in which the beating heart resides is in an operating room and where the method includes actually taking those invasive readings from the start of potential recording for example (potential recording of data that is actually used to implement the teachings herein, as opposed to potential recordings for the purposes of calibration and/or for verification that the system is functioning properly to enable the recordation of potentials that are usable). [00381] As noted herein, embodiments, include non-transitory computer readable medium having recorded thereon, a computer program for executing at least a portion of a method, the computer program including any one or more all the actions detailed herein providing that the art enables such. In an exemplary embodiment, the computer readable medium includes code for executing one or more or all of the method actions associated with methods 1030, 1040, 1050, 1060, 1070, 1090, 1110, 2010, 2110, 2310, 1190, 2190 and/or 2199, and the variations thereof and the various permutations detailed above. More specifically, in at least some exemplary embodiments, the computer readable medium includes code for executing one or more of the method actions detailed herein associated with the wavelet processing and the implementation of the window functioning and the Fourier transformation and the subtraction of the results from the initial EGM to develop the EGM without the V artifact, starting with the obtained first data based on one or unipolar EGMs recorded in an atrium. In at least some
Attorney Docket No.246-021PCT exemplary embodiments, the computer readable medium includes code for executing one or more of the method actions detailed herein associated with the wavelet processing and the positive power development (and in some embodiments, the power maxima identification and/or the selection of scale and/or the activation timing) starting with the obtained first data based on one or unipolar EGMs recorded in an atrium. In at least some exemplary embodiments, the computer readable medium includes code for executing one or more of the method actions detailed herein associated with the wavelet processing and the application of the window functioning and the Fourier transformation and the reconstructing the atrial EGM. Any computer readable medium disclosed herein corresponds to a disclosure in the interests of textual economy of a medium that executes any one or all of the method actions associated with the wavelet processing and the wavelet filtering based embodiments disclosed herein unless otherwise noted, providing that the art enables such. [00382] Embodiments include a model based on results from or that is a product of machine learning to execute and/or otherwise develop any one or more of the results of any one or more of the method actions detailed herein associated with the wavelet filtering embodiments. Embodiments also include utilizing that model to develop or otherwise obtain the results of those method actions / execute any one or more of those method actions. Embodiments disclosed herein have briefly addressed this feature with respect to certain aspects of the teachings herein. It is noted that this does not indicate a limitation on the utilization of the machine learning techniques. In an embodiment, those results are developed or otherwise obtained and/or the method actions are executed utilizing by way of example, a neural network, such as a deep neural network, that is trained. Corollary to this is that at least some exemplary embodiments include training the machine learning system, such as training the neural network. Embodiments include executing one or more of the actions herein a certain number of times, such as at least and/or equal to 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 70, 80, 90, 100, 125 or 150 or any value or range of values therebetween in one increment manually, and providing the results thereof to a neural network to train the neural network. In an embodiment, the resulting product of the machine learning can be based on the just detailed actions. In an embodiment, the selection of the scale(s) to utilize with respect to the positive power can be an example where the initial selection of scales are developed based on the judgment of the skilled technician and/or the skilled healthcare professional, and those selections are provided to the neural network which upon training, provides for a product that automatically selects the scale.
Attorney Docket No.246-021PCT [00383] To be clear, in an exemplary embodiment, any action of analyzing or evaluating or calculating or identifying, etc. can correspond to an action of utilizing the results of machine learning to obtain the results thereof. By way of example only and not by way of limitation, in an exemplary embodiment, the atrial activation timing can be identified by the results of machine learning. Corollary to this is that any disclosure herein with respect to identifying or choosing, etc., something vs. something else corresponds to a disclosure of doing so utilizing the results of machine learning, and also a disclosure where such choices and/or identification are provided to a machine learning system to develop the results. An embodiment, the machine learning develops the algorithms utilized by internally identifying features, such as important features, from the data provided. [00384] In an exemplary embodiment, there is an exemplary method that includes analyzing any one or more of the data presented herein using a computer chip or a logic circuit or electronics or software to develop the resulting data. In an embodiment, the computer chip or logic circuit or electronics or software is based on a statistically significant population of persons who were afflicted with afib. [00385] Corollary to the above is that in an exemplary embodiment, the code detailed herein for the computer readable media is code that results from machine learning. In an exemplary embodiment, the teachings detailed herein associated with machine learning can be implemented to produce the code, or at least a portion of the code that is for executing one or more of the actions detailed herein. [00386] In view of the above, in at least some exemplary embodiments, there is a system that includes an input subsystem configured to receive input regarding information about potentials or based on potentials, and output subsystem that provides output, such as any one or more of the results of any one or more of the method actions detailed herein. In an embodiment, there is an artificial intelligence subsystem interposed between the input subsystem and the output subsystem, wherein the system is configured to determine with the use of the artificial intelligence subsystem, any one or of the results herein. In an embodiment, the artificial intelligence subsystem is a neural network that is at least partially taught. In an exemplary embodiment, it can be a fully taught neural network. Note also that embodiments include not only training, but retraining a neural network or otherwise continuously training the neural network. In this regard, the results of the neural networks decisions can be evaluated and ratified or corrected in this ratification and/or corrective data can be provided to the neural network for retraining purposes. In this regard, the neural network is a partially taught neural
Attorney Docket No.246-021PCT network that is trainable with feedback provided through, for example, the input subsystem or another subsystem of the system. Any one or more or all of the variables detailed herein can be identified by using a product of machine learning / NN / DNN. Embodiments include providing a sufficient number of examples and/or solutions to a NN / DNN or otherwise a machine learning algorithm or device and training such accordingly, which examples / solutions can include any one or more or all of the variables herein for a sufficient number of individual patients or otherwise individual examples to train the system and otherwise obtain a product that results therefrom, which can be used to identify one or more of the solutions and/or results presented herein. [00387] Corollary to all of this is that in embodiments where, for example, there is the data, a database can be utilized instead of or in addition to the artificial intelligence arrangements detailed above. Here, in an exemplary embodiment, straightforward algorithms can be utilized to arrive at certain results based on a statistically significant amount of prior data. In an exemplary embodiment, there are methods that include the action of consulting a database that may include at least 100, 150, 200, 250, 300, 400, 500, 600, 700, 800, 900, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 5000, 7000, 10,000 or more or any value or range of values therebetween in one increment set of data respectively associated with individuals suffering from atrial fibrillation, where this database includes any one or more of the data disclosed herein. By way of example only and not by way of limitation, for a given positive power plot over different scales, the database can be consulted to identify a statistically significant similar plot that was utilized, and thus provide the identification of the scale to be utilized. Here, a non-AI algorithm and/or an algorithm that is not based on the results of machine learning can be utilized. This is because the amount of data collected in the database is sufficient to utilize statistical techniques outside of the realm of artificial intelligence to select a given scale. In the interest of textual economy, this concept can be applied to any of the other selection / determination actions detailed herein unless otherwise noted, providing that the art enables such. This is the case with any of the machine learning teachings as well, unless otherwise noted, providing that the art enables such. Of course, embodiments include developing the actual statistical analysis algorithm or other algorithm to make the determination or otherwise to select a given scale. [00388] Embodiments include dedicated electronic circuitry configured to analyze any one or more of the data provided to the system detailed herein or otherwise any one or the data
Attorney Docket No.246-021PCT obtained that is disclosed herein that is needed or otherwise as a basis to move forward on any one or more of the method actions detailed herein. [00389] In an embodiment, there is a device and/or a system, comprising an input terminal including one of a monitor, keyboard and mouse combination or a server and/or a USB port or an electronic signal reception port, the input terminal being configured to receive input relating to any one or more of the base data detailed herein (e.g., the raw EKG signals, the processed EKG signals, etc.). The device and/or system includes dedicated electronics circuitry configured to analyze data based on input into the input terminal, wherein the electronics circuitry includes at least one of a model that is used by the device/system to automatically execute one or more of the method actions herein based on the input into the input suite. An embodiment includes a computer system, comprising an input subsystem configured to receive input regarding data based on EKG information, and an output subsystem and an artificial intelligence subsystem interposed between the input subsystem and the output subsystem, wherein the system is configured to execute one or more of the method actions detailed herein based on input into the input subsystem and output the results using the output subsystem. [00390] In an embodiment, there is a method, comprising obtaining heart phase data for a plurality of activation cycles of a living human afflicted with atrial fibrillation, and analyzing the heart phase data to identify specific heart tissue locations where there are repeated and consistent temporal discrepancies of electrical activation relative to other tissue locations, wherein the obtained heart phase data can be based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of near field atrial EGMs, and this embodiment can include obtaining such based on any one or more of the techniques described herein. [00391] In an embodiment, the action of analyzing the heart phase data includes implementing a statistical analysis on the heart phase data and/or the action of analyzing the heart phase data includes implementing time averaging analysis on the heart phase data and/or the action of analyzing the heart phase data includes: for a plurality of spatial locations on an interior surface of the heart, which spatial locations include the identified specific heart tissue locations, identifying respective maximum phase gradients for respective locations of the plurality of spatial locations over a length of time;
Attorney Docket No.246-021PCT time averaging the respective maximum phase gradients for the respective locations; and identifying corresponding locations where the time averaged results are statistically aberrant and/or are not statistically aberrant. [00392] In an embodiment, the identified corresponding locations of the time average results that are statistically aberrant are the identified specific heart tissue locations and/or the identified corresponding locations of the time average results that are not statistically aberrant are not the identified specific heart tissue locations and/or the respective maximum phase gradients are the respective maximum phase gradients between the respective locations and a plurality of proximate locations on the surface of the heart and/or the proximate locations are effectively North-South-East-West adjacent locations and/or wherein the proximate locations are the locations immediately surrounding the respective location. [00393] In an embodiment, any one or more of the method actions herein and/or methods include abating a surface of the heart based at least on the identified specific heart tissue locations and/or identifying the surface of the heart to be ablated based on at least the identified specific heart tissue locations. Note that any disclosure herein of ablating includes a corresponding disclosure of identifying what to ablate. [00394] In an embodiment, the plurality of spatial locations includes at least 24 locations; and there are at least 36 temporally spaced respective maximum phase gradients for the respective locations. [00395] In an embodiment, the method further comprises: obtaining respective plurality of temporally spaced electrical potentials for respective electrodes of at least 24 electrodes of a catheter located in a heart chamber at a first location in the heart chamber; converting the obtained respective plurality of temporally spaced electrical potentials to the heart phase data based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein, wherein the converting results in the obtention of heart phase data, wherein the actions of identifying respective maximum phase gradients, time averaging, identified corresponding locations where the time average results
Attorney Docket No.246-021PCT are statistically aberrant and/or not statistically aberrant, are based on the obtained respective plurality of temporally spaced electrical potentials for the catheter located at the first location. [00396] In an embodiment the method further includes: [00397] obtaining respective second plurality of temporally spaced electrical potentials for respective electrodes of at least 64 electrodes of the catheter located in the heart chamber at a second location in the heart chamber different from the first chamber; [00398] converting the obtained respective second plurality of temporally spaced electrical potentials to second heart phase data based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein, for a plurality of second spatial locations on the interior surface of the heart, which second spatial locations include the identified specific heart tissue locations, identifying respective second maximum phase gradients for respective second locations of the plurality of second spatial locations over second length of time; second time averaging the second respective maximum phase gradients for the second respective locations; and identifying corresponding second locations where the time averaged second results are statistically aberrant and/or not statistical aberrant, wherein the identified corresponding second locations of the second time average results that are statistically aberrant are included in the identified specific heart tissue locations and/or the identified corresponding second locations of the time second average results that are not statistically aberrant are not included in the identified specific heart tissue locations. [00399] In an embodiment, the action of obtaining heart phase data and analyzing is executed in real time vis-à-vis a catheter located in a heart chamber. [00400] In an embodiment the action of analyzing the heart phase data includes: for a plurality of spatial locations on an interior surface of the heart, which spatial locations include the identified specific heart tissue locations, identifying respective maximum phase gradients for respective locations of the plurality of spatial locations over a length of time; time averaging the respective maximum phase gradients for the respective locations; and
Attorney Docket No.246-021PCT identifying corresponding locations where the time averaged results are non- zero and/or statistically zero, wherein the identified corresponding locations of the time average results that are non-zero are the identified specific heart tissue locations and/or the identified corresponding locations of the time average results that are statistically zero are not the identified specific heart tissue locations. [00401] In an embodiment, there is a method, comprising: developing a time-varying electrical potential map of a surface of a cavity of a beating heart; developing a time-varying phase map of the surface of the cavity based on the developed time-varying electrical potential map; and identifying repeating phase signatures for respective locations on the surface of the atrial cavity from the time-varying phase map that repeat in a statistically aberrant manner relative to other phase signatures at other respective locations, wherein in an embodiment, the developed phase map and/or the developed potential map is based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein. [00402] In an embodiment, the electrical potential map has at least 500 electrical potential spatial locations and at least respective 5,000 temporal potential values for the respective potential spatial locations, the phase map has at least 400 phase spatial locations and at least respective 4,000 temporal phase values for respective phase locations. [00403] In an embodiment, the respective electrical potential locations of the at least 500 electrical potential locations have respective phase locations of the at least 500 phase locations. In an embodiment, the electrical potential map has at least 500 electrical potential locations and at least respective 20,000 temporal locations for respective potential locations of the 500 potential locations, the phase map has at least 500 phase locations and at least respective 20,000 temporal locations for respective phase locations of the 500 phase locations. In an embodiment, the actions of developing a time-varying electrical potential map, developing the time-varying phase map, and identifying the repeating phase signatures are executed within a period of no more than 20 minutes. In an embodiment, the action of developing a time-varying electrical potential map of a surface of a cavity of a beating heart is based at least in part on
Attorney Docket No.246-021PCT time-varying readings from electrodes located in the cavity; and the action of identifying the repeating phase signatures is executed within 20 minutes of the electrodes being removed from the chamber. [00404] In an embodiment, the action of developing a time-varying electrical potential map of a surface of a cavity of a beating heart is based at least in part on invasive readings taken while a human in which the beating heart resides is in an operating room; and the action of identifying the repeating phase signatures is executed before the human leaves the operating room. [00405] In an embodiment, the method further includes executing a medical procedure targeted at tissue of the heart corresponding to at least some of the respective locations identified as having the repeating phase signatures that repeat in the statistically aberrant manner before the human leaves the operating room. [00406] In an embodiment, the method further includes: after executing the medical procedure, developing second time-varying electrical potential map of the surface of the cavity of the beating heart; developing second time-varying phase map of the surface of the cavity based on the second developed time-varying electrical potential map; and evaluating whether and/or how many repeating phase signatures for respective locations on the surface of the atrial cavity from the second time-varying phase map that repeat in a statistically aberrant manner relative to other phase signatures at other respective locations, and based on the evaluation, evaluation whether the medical procedure was successful, wherein one or both of the obtained maps is / can be based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein. [00407] In an embodiment, the medical procedure is ablation of the targeted tissue. [00408] In an embodiment, there is a method, comprising: developing data including at least X spatial locations and at least Y respective phase gradients for the respective spatial locations of the X spatial locations; statistically analyzing the developed data; and
Attorney Docket No.246-021PCT identifying locations of the respective locations that are indicative of tissue influencing atrial fibrillation based on the statistical analysis, wherein X is at least 64 and Y is at least 50, and wherein the phase gradients is based on data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein. [00409] In an embodiment, the statistical analysis is time averaging and/or the action of identifying locations includes identifying locations where the statistical analysis of the developed data indicates non-random activation of respective heart tissue cells at the identified locations and/or the action of identifying locations includes identifying locations where averaging of the maximum phase gradients yields a statistically meaningful non-zero value. In an embodiment, the action of identifying locations includes identifying other locations where averaging of the maximum phase gradients of at least a majority of the Y phase gradients yields a statistically zero value. [00410] In an embodiment, the action of identifying locations includes identifying locations where averaging of the maximum phase gradients of at least a majority of the Y phase gradients yields a statistically meaningful non-zero value and the action of identifying the locations includes further statistically analyzing the values of the non-zero values. In an embodiment, X is at least 300 and Y is at least 75. In an embodiment, X is at least 1,000 and Y is between 60 and 1,000, inclusive. [00411] In an embodiment, the statistical analysis of the developed data identifies statistically consistent patterns of electrical activity that repeat in a statistically meaningful manner over time. [00412] In an embodiment, there is a non-transitory computer readable medium having recorded thereon, a computer program for executing at least a portion of a method, the computer program including: code for statistically analyzing first data based on phase gradients for at least 150 locations on a surface of a chamber of a human heart; and code for identifying a plurality of locations from the at least 150 locations, based on the statistical analysis of the first data, that should be targeted for treatment; and code for obtaining / developing data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the
Attorney Docket No.246-021PCT reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein, wherein such is used to develop the phrase gradients in some embodiments. [00413] The medium can further include code for transforming respective electrograms for respective locations of the at least 150 locations to a phase record including the phase gradients and/or code for creating the electrograms from data based on electrical potentials obtained from electrodes within the human heart, the number of electrodes within the human heart being less than 150. In an embodiment, the code for creating the electrograms uses inverse solution methods and/or the code for statistically analyzing the first data time averages respective maximum phase gradients for the at least 150 locations and/or the code for statistically analyzing the first data time averages respective maximum phase gradients for respective locations of the at least 150 locations and/or the code for identifying the plurality of locations from the at least 150 location identifies respective locations where time averages of the respective maximum phase gradients are statistically significantly non-zero. In an embodiment, the at least 150 locations includes at least 2,000 locations, respective locations of the at least 2,000 locations have at least 1000 respective maximum phase gradients and/or the medium creates the electrograms and identifies the plurality of locations from the at least 2,000 locations for the at least 1000 respective maximum phase gradients within 20 minutes when run on a Dell ™ laptop with an Intel Core i9 Microprocessor with at least a 2.8 GHz clock frequency, at least 16 by 1024 KB L2 cache, at least 22.00 MB L3 cash, a TDP of at least 160 W, a DMI 3.0 I/O bus and a 4 x DDR4-2666 memory. In an embodiment, the code for creating the electrograms uses at least 500 measurements from each electrode per second. [00414] In an embodiment, there is code for developing data based on any one or more of the techniques described herein for V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM, and this embodiment can include obtaining such based on any one or more of the techniques described herein and this is done within 30, 29, 28, 27, 26, 25, 24, 23, 22, 21, 20, 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2 or 1 minutes when run on a Dell ™ laptop with an Intel Core i9 Microprocessor with at least a 2.8 GHz clock frequency, at least 16 by 1024 KB L2 cache, at least 22.00 MB L3 cash, a TDP of at least 160 W, a DMI 3.0 I/O bus and a 4 x DDR4-2666 memory. This code can execute any one or more or all of the method actions associated with the V artifact removal, atrial activation timing identification and/or the reconstruction of the near field atrial EGM disclosed herein.
Attorney Docket No.246-021PCT [00415] Embodiments can include a system and/or simply an embodiment that includes a non- transitory computer readable medium having recorded thereon, a computer program for executing at least a portion of a method, the computer program including code for executing any one or more of the method actions and/or functionalities detailed herein. Thus, any disclosure herein of a method action or functionality corresponds to a disclosure of a non- transitory computer readable medium having programed thereon code to execute one or more of those actions and also a product to execute one or more of those actions. [00416] Embodiments include any functionality disclosed herein and/or method action disclosed herein being executed by a computer chip, a processor, software, logic circuitry and/or electronics, and all are not mutually exclusive. Any circuit that can enable the teachings herein can be used providing that the art enables such. Thus, in the interests of textual economy, and disclosure herein of a functionality of an article of manufacture corresponds to any one or more of the aforementioned structures being configured to execute such and otherwise for such, and the same is for any method action disclosed herein, where any such action corresponds to a disclosure of any one or more of the aforementioned structures being configured to execute such and otherwise for such. [00417] In some embodiments, a neural network, such as a DNN, is used to directly interface to the input into the systems / devices detailed above, and process this input via its neural net, and determine the information detailed above. The network can be, in some embodiments, either a standard pre-trained network where weights have been previously determined (e.g., optimized) and loaded onto the network, or alternatively, the network can be initially a standard network, but is then trained to improve specific recipient results based on outcome oriented reinforcement learning techniques. [00418] Any disclosure herein of a processor corresponds to a disclosure in an embodiment of a non-processor device or a combined processor-non-processor device where the non-processor is a result of machine learning. Embodiments can include a link from the cloud to a clinic to pass information back and forth, enabling the remote processing noted above and/or enabling the obtaining of additional data for retraining purposes. Information can be uploaded to the cloud to the clinic, where the information can be analyzed. Another exemplary system includes a smart device, such as a smart phone or tablet, etc., that is running a purpose built application to implement some of the teachings detailed herein. This can be used by the clinician, and can contain at least the front end portions of the systems and devices detailed herein, or otherwise provide the interface portal to the back end. Any disclosure herein of a processor corresponds
Attorney Docket No.246-021PCT to a disclosure of a non-processing device, or includes non-processing devices, such as a chip or the like that is a result of a machine learning algorithm or machine learning system, etc. [00419] In an exemplary embodiment, the smart device can be configured to present the windows for the interface that will be used by the user. [00420] Any method action and/or functionality disclosed herein where the art enables such corresponds to a disclosure of a code from a machine learning algorithm and/or a code of a machine learning algorithm and/or a product of machine learning for execution of such. Still as noted above, in an exemplary embodiment, the code need not necessarily be from a machine learning algorithm, and in some embodiments, the code is not from a machine learning algorithm or the like. That is, in some embodiments, the code results from traditional programming. Still, in this regard, the code can correspond to a trained neural network. In an embodiment, the trained neural network can be utilized to provide (or extract therefrom) an algorithm that can be utilized separately from the trainable neural network. In one embodiment, there is a path of training that constitutes a machine learning algorithm starting off untrained, and then the machine learning algorithm is trained and “graduates,” or matures into a usable code – code of trained machine learning algorithm. With respect to another path, the code from a trained machine learning algorithm is the “offspring” of the trained machine learning algorithm (or some variant thereof, or predecessor thereof), which could be considered a mutant offspring or a clone thereof. That is, with respect to this second path, in at least some exemplary embodiments, the features of the machine learning algorithm that enabled the machine learning algorithm to learn may not be utilized in the practice some of the method actions, and thus are not present the ultimate system. Instead, only the resulting product of the learning is used. [00421] And to be clear, in an exemplary embodiment, there are products of machine learning algorithms (e.g., the code from the trained machine learning algorithm) that are included in any one or more of the systems / subsystems detailed herein, that can be utilized to analyze any of the data obtained or otherwise available disclosed above that can be utilized or otherwise is utilized to evaluate the data obtained herein. This can be embodied in software code and/or in computer chip(s) that are included in the system(s). [00422] An exemplary system includes an exemplary device / devices that can enable the teachings detailed herein, which in at least some embodiments can utilize automation. That is, an exemplary embodiment includes executing one or more or all of the methods and/or
Attorney Docket No.246-021PCT functionalities detailed herein and variations thereof, at least in part, in an automated or semiautomated manner using any of the teachings herein. Conversely, embodiments include devices and/or systems and/or methods where automation is specifically prohibited, either by lack of enablement of an automated feature or the complete absence of such capability in the first instance. [00423] Automated actions can be executed by an algorithm where circuitry receives the input (embodied in an analogue or a digital signal), where the input suite converts the “physical” input into electronic signals using analog to digital converters for example, or in the case of the input suite corresponding to an Internet server, receives the digital signal from a remote location, and the digital data is stored in a memory and/or received by the electronics. The electronics, which is a result of the machine learning by way of example, takes the digital signal and deconstructs the digital signal to evaluate properties, and then, using its “knowledge” from its training, provides an output based on the knowledge. [00424] Database herein can be a database such as Microsoft ™ Access, where the computer automatically matches the data instead of the human matching the data. The results of machine learning and/or a product thereof can be used to perform the automatic matching. [00425] In an exemplary embodiment, the some or all of the teachings herein are implanted on a computer chip and/or a computer circuit. There are comparators based on big data noted above. In an exemplary embodiment, comparison can be represented by an algorithm where circuitry receives the input (embodied in an analogue or a digital signal), where the input suite converts the “physical” input into electronic signals using analog to digital converters for example, or in the case of the input suite corresponding to an Internet server, receives the digital signal from a remote location, and the digital data is stored in a memory and/or received by the electronics. The electronics takes the digital data and “looks” for certain strings of zeros and ones that correspond to a match with signatures / identifiers linked to prestored data regarding performance capabilities. The data linked to the signatures / identifiers is the cohort identified. [00426] It is further noted that any disclosure of a device and/or system detailed herein also corresponds to a disclosure of otherwise providing that device and/or system and/or utilizing that device and/or system. [00427] The systems detailed herein can be configured to transform input into numerical form, and the artificial intelligence subsystem can be configured to, using the numerical form, produce an estimated outcomes measure, produce one or more of the results detailed herein.
Attorney Docket No.246-021PCT More specifically, in an embodiment, the system or device is configured to automatically transform input into numerical form. This can be executed using a computer chip or a logic circuit or electronics or software or a processor, that is programmed to take the input and transform the input. In some embodiments, the input subsystem is configured to execute this functionality. Thus, the input subsystem can be more than just a mouse and computer screen and keyboard, etc. Embodiments include an input subsystem that includes a processor and/or software and/or firmware and/or hardware and/or a computer chip or a logic circuit otherwise electronics that is specifically designed and configured to execute one or more of the functionalities of the input subsystem detailed herein. In an embodiment, the artificial intelligence subsystem is configured to, using the numerical form, automatically produce results based on this input, as transformed. [00428] As can be seen, embodiments of the output subsystem can be more than just a “dumb” computer screen or the like. Embodiments include an output subsystem that includes a processor and/or software and/or firmware and/or hardware and/or a computer chip (herein a computer chip also corresponds to a plurality of such, interconnected with a motherboard, etc.) or otherwise electronics that is specifically designed and configured to execute one or more of the functionalities of the output subsystem detailed herein. Any disclosure herein of software corresponds to an alternate disclosure of a computer chip or a logic circuit or electronics. [00429] It is also noted that any disclosure herein of any process of manufacturing or providing a device corresponds to a disclosure of a device and/or system that results therefrom. Is also noted that any disclosure herein of any device and/or system corresponds to a disclosure of a method of producing or otherwise providing or otherwise making such. [00430] An exemplary system includes an exemplary device / devices that can enable the teachings detailed herein, which in at least some embodiments can utilize automation, as will now be described in the context of an automated system. That is, an exemplary embodiment includes executing one or more or all of the methods detailed herein and variations thereof, at least in part, in an automated or semiautomated manner using any of the teachings herein. [00431] Any embodiment or any feature disclosed herein can be combined with any one or more or other embodiments and/or other features disclosed herein, unless explicitly indicated and/or unless the art does not enable such. Any embodiment or any feature disclosed herein can be explicitly excluded from use with any one or more other embodiments and/or other features
Attorney Docket No.246-021PCT disclosed herein, unless explicitly indicated that such is combined and/or unless the art does not enable such exclusion. [00432] Any function or method action detailed herein corresponds to a disclosure of doing so an automated or semi-automated manner. [00433] It is further noted that any disclosure of a device and/or system detailed herein also corresponds to a disclosure of otherwise providing that device and/or system and/or utilizing that device and/or system. [00434] It is also noted that any disclosure herein of any process of manufacturing other providing a device corresponds to a disclosure of a device and/or system that results there from. Is also noted that any disclosure herein of any device and/or system corresponds to a disclosure of a method of producing or otherwise providing or otherwise making such. [00435] It is further noted that any disclosure of a device and/or system detailed herein also corresponds to a disclosure of otherwise providing that device and/or system and/or utilizing that device and/or system. [00436] It is also noted that any disclosure herein of any process of manufacturing other providing a device corresponds to a disclosure of a device and/or system that results there from. Is also noted that any disclosure herein of any device and/or system corresponds to a disclosure of a method of producing or otherwise providing or otherwise making such. [00437] Any function or method action detailed herein corresponds to a disclosure of doing so an automated or semi-automated manner. [00438] While various embodiments of the present invention have been described above, it should be understood that they have been presented by way of example only, and not limitation. It will be apparent to persons skilled in the relevant art that various changes in form and detail can be made therein without departing from the spirit and scope of the invention.