WO2025207628A1

WO2025207628A1 - Arrhythmia origin identification system

Info

Publication number: WO2025207628A1
Application number: PCT/US2025/021336
Authority: WO
Inventors: Christopher J. T. VILLONGCO; David E. KRUMMEN
Original assignee: Vektor Group Inc; University of California Berkeley; University of California San Diego UCSD
Current assignee: Vektor Group Inc; University of California Berkeley; University of California San Diego UCSD
Priority date: 2024-03-25
Filing date: 2025-03-25
Publication date: 2025-10-02
Anticipated expiration: 2026-09-25

Abstract

A system for identifying sites of origin of an arrhythmia is provided. The system inputs an arrhythmia ECG. The system identifies outflow tract (OT) sites that may be the origin of the arrhythmia by applying a mapping system to the arrhythmia ECG. Each OT site may be associated with an OT site score. The system calculates an adjusted OT site score for each OT site that is based on its OT site score adjusted by an OT region accuracy of an OT region identification system. The system also calculates an adjusted OT site probability that is the adjusted OT site score divided by the sum of the adjusted OT site scores for the OT sites. The system then generates and displays a graphic of a heart with an indication of the adjusted OT site probability for an OT site.

Description

ARRHYTHMIA ORIGIN IDENTIFICATION SYSTEM

CROSS-REFERENCE TO RELATED APPLICATION(S)

[0001] This application claims priority to U.S. Provisional Patent Application No. 63/569,649, titled “ARRHYTHMIA ORIGIN IDENTIFICATION SYSTEM” and filed on March 25, 2024, which is incorporated herein by reference in its entirety.

BACKGROUND

[0002] Many heart disorders can cause symptoms, morbidity (e.g., syncope or stroke), and mortality. Common heart disorders include atrial fibrillation (“AF”), ventricular fibrillation (“VF”), atrial tachycardia (“AT”), ventricular tachycardia (“VT”), atrial flutter, and premature ventricular contractions (“PVC”). The sources of heart disorders include stable electrical rotors, recurring electrical focal sources, and so on. These sources are important drivers of sustained or clinically significant episodes of heart disorders. These heart disorders can be treated with therapeutic ablation, radiofrequency waves, pulsed fields, cryogenic, ultrasound, external radiation sources, and so on by targeting the source of the heart disorder. To target the source of a heart disorder, the source location of the heart disorder should be identified.

[0003] Unfortunately, many methods for reliably identifying the source locations of a heart disorder can be complex, cumbersome, and expensive. For example, one method uses an electrode basket catheter that needs to be inserted into the heart (e.g., left ventricle) to collect from within the heart measurements of the electrical activity of the heart, such as during an induced VF. The measurements can then be analyzed to help identify a possible source location. Such an electrode basket catheter is expensive. Moreover, the use of an electrode basket catheter can lead to serious complications. Another method uses a body surface vest with electrodes to collect from the patient’s body surface measurements, which can be analyzed to help identify a possible source location. A body surface vest is expensive, is difficult to manufacture, and may interfere with the placement of defibrillator pads needed after inducing a fibrillation to collect measurement during fibrillation. In addition, the vest analysis requires a CT scan and is unable to sense the interventricular and interatrial septa. [0004] The accurate identification of the origin of a PVC has been particularly challenging. A PVC may originate near the left ventricular outflow tract (LVOT) or the right ventricular outflow tract (RVOT). Techniques for identifying the origin typically identify the outflow tract (OT) in which the PCV originates without identifying a specific location within the OT. The identification has often been based on analysis of a 12-lead electrocardiogram (ECG) to identify patterns and criteria indicative of whether the arrhythmia is an LVOT arrhythmia (i.e., originating from the LVOT) or an RVOT arrhythmia (i.e., originating from the RVOT). An LVOT arrhythmia may be indicated when the ECG shows a right bundle branch block (RBBB) pattern or shows a less pronounced inferior axis due to its closer proximity to the left side of the heart. In contrast, an RVOT arrhythmia may be indicated when an ECG shows a left bundle branch block (LBBB) pattern with an inferior axis.

[0005] If an LVOT arrhythmia or an RVOT arrhythmia is incorrectly indicated, the treatment based on that indication may be suboptimal. For example, if an LVOT arrhythmia is incorrectly indicated as an RVOT arrhythmia, a patient may be treated less aggressively than if the OT was correctly indicated. Such a less aggressive treatment may initially start with lifestyle changes or medication. In contrast, if the LVOT arrhythmia was correctly indicated, the patient may be treated more aggressively such as with an ablation procedure because of LV’s critical role in systemic circulation. In addition to identifying the correct OT, it is important to identify the precise source location within the OT to help ensure that an ablation treatment is as effective as possible.

[0006] In general, both atrial arrhythmias and ventricular arrhythmias can originate at various locations throughout the heart. For example, VTs may originate in various regions of the left ventricle such as near the LVOT, fascicular regions, the mitral annulus, or the septal area. As another example, ATs may originate at various regions near the pulmonary veins, the tricuspid or mitral valve annuli, or the coronary sinus. The accurate identification of the origin of any arrhythmia and subsequent treatment based on that identification can help improve patient outcomes.

BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 is a flow diagram that illustrates an origin identification method for treating a patient in some embodiments. [0008] Figure 2 is a block diagram that illustrates components of an origin identification system in some embodiments.

[0009] Figure 3 is a flow diagram that illustrates the processing of a generate adjusted site probabilities component of the object identification system in some embodiments.

[0010] Figure 4 is a flow diagram that illustrates the processing of an adjusted site probability graphic component of the origine identification system in some embodiments.

DETAILED DESCRIPTION

[0011 ] Methods and systems are provided for treating a patient with an arrhythmia based on the identification of the origin of the arrhythmia. In some embodiments, an origin identification (OID) system combines techniques for identifying a site that is an origin (i.e., source location) to improve the accuracy of the identification. The OID system combines a region identification technique or a region accuracy assessment technique with a site identification technique. For example, a region identification system may employ a region identification technique to identify a region such as the LVOT or the RVOT as the likely region of the origin of a PVC. The region identification system may assign an OT region score (e.g., or probability) to each OT region as being the likely OT region of origin. The region identification system may assign an OT region score based on analysis of characteristics of a patient ECG. A site identification system may identify a specific site within the LVOT and a specific site within the RVOT as possible sites of the PVC. The site identification system may assign an OT site score to an LVOT site and to an RVOT site as being the likely OT site of the origin. The site identification system may assign an OT site score based on similarity of a patient ECG to an ECG (simulated or clinical) associated with an OT site. The OID system may identify likely OT sites that may be the origin by combining the OT region scores and the OT site scores.

[0012] As another example, a region accuracy assessment technique (e.g., a heuristic) may have an accuracy for the LVOT region, which may be the top match accuracy. In such a case, the LVOT region score and the RVOT region score are derived from that accuracy. The OID system may identify likely OT sites by combining the OT site scores and the OT region scores that are derived from the region accuracy assessment.

[0013] The OID system provides a physician (or other health care provider) with indications of likely OT sites based on a combination of OT region scores and OT site scores. Such indications of the likely OT sites may provide more accurate information on the actual origin of the arrhythmia that may result in an improved treatment (e.g., ablation) outcome. In the following, the OID system is described primarily in the context of sites in the LVOT and the RVOT. However, the OID system may be employed to identify sites in other regions such as the left ventricle (LV) and right ventricle (RV), the anterior and posterior papillary muscle, sub-regions of an OT region (e.g., septal area and aortic annulus), the endocardium and epicardium, and so on.

[0014] In some embodiments, the OID system generates adjusted OT site scores by adjusting the OT site scores based on a region accuracy assessment (e.g., OT region scores) of one or more region identification techniques. An OT region score may be the percentage of the time that a region identification system correctly identifies an OT region as containing the site of the origin (i.e., top match accuracy), which reflects the region accuracy assessment. For example, if the LVOT region score is 75%, the OID system may set the RVOT region score to 25% (i.e., 100% minus 75%). An adjusted LVOT site score is the LVOT site score multiplied by the LVOT region score, and an adjusted RVOT site score is the RVOT site score multiplied by the RVOT region score. Alternatively, the OID system may multiply each OT site score by the percentage of the time that a region identification system correctly identifies an OT site in that OT region. For example, if the region accuracy assessment indicates the identification of the LVOT region is accurate 75% of the time and that the RVOT is accurate 50% of the time, the OID system multiplies an LVOT site score by 75% and an RVOT site score by 50%.

[0015] The OID system may generate an adjusted OT site probability for an OT site by first generating an adjusted OT site score for that OT site (i.e., the OT site score for that OT site multiplied by the OT region score for the OT region that contains that OT site). The OID system then generates the adjusted OT site probability for that OT site by dividing the adjusted OT site score for that OT site by the sum of the adjusted OT site scores for all the identified OT sites. [0016] In some embodiments, the OID system identifies possible OT sites that are the source of the arrhythmia based on an arrhythmia cardiogram collected from a patient. The OID system may apply a mapping system to the cardiogram to identify the OT sites that may be the origin of the arrhythmia. The cardiogram may be represented as, for example, an electrocardiogram (ECG) or a vectorcardiogram (VCG) that may be derived from an ECG. Some techniques for implementing mapping systems are described in U.S. Pat. No. 10,860,754 titled “Calibration of Simulated Cardiograms” and issued on December 8, 2020, which is hereby incorporated by reference. The mapping system may employ mappings of a mapping ECG (clinical or simulated) to one or more OT sites that may be the origin that would likely result in that mapping ECG. The mappings of the mapping ECGs may also be referred to as a library of library ECGs. To identify an OT site, the mapping system identifies one or more mapping ECGs that are similar to a patient ECG. Similarity may be based on a similarity metric such as cosine similarity, Euclidean distance, Pearson Correlation Coefficient, L1 Norm, and so on. A mapping ECG and the patient ECG may be considered to be similar when the similarity metric satisfies a similarity criterion such as greater than 0.9 for cosine similarity. An OT site score for an OT site may be based on the similarity metric. The mapping system may select some of the highest OT site scores and generate OT site probabilities for those OT sites. The OT site probability for an OT site may be the OT site score for that OT site divided by the sum of those highest OT site scores. The OID system may employ a mapping system that is based on mapping machine learning (ML) model to identify an OT site. The mapping ML model inputs a patient ECG and outputs one or more OT sites along with an OT site score (e.g., that may be a probability) for each OT site. A mapping ML model is described in the 754 patent.

[0017] The OID system may employ an ECG analysis technique to identify the OT regions or region accuracy assessment derived from the ECG analysis technique to adjust OT site scores. An ECG analysis technique analyzes a patient ECG to identify characteristics that are likely indicative of the region in which the arrhythmia originates. For example, a characteristic may be based on an S-wave of V2 and an R-wave of the V3 of the patient ECG or ratio of an R-wave amplitude to an S-wave amplitude on V2. Various ECG analysis techniques are described in Tzeis, S., Asvestas, D., Ho, S.Y. and Vardas, P., 2019. Electrocardiographic landmarks of idiopathic ventricular arrhythmia origins. Heart, 105(14), pp.1 109-1 1 16, which is hereby incorporated by reference. An ECG analysis technique may generate an OT region score that indicates the likelihood of an OT region being the OT region of the origin. For example, an OT region score may be based on the ratio of the amplitudes of the S-wave of V2 and the R-wave of V3. Such an OT region score that is less than 1 .5 may indicate that the region is the LVOT. The OT region probability that the OT region is the LVOT or RVOT may be based on how far the OT region score is below or above 1 .5. For example, the LVOT region probability for an OT region score of 1 .5 may be 0.5 indicating a 50/50 chance of the OT region being the LVOT and for an OT region score of 1 .0 may be 0.72 (i.e., 1 ,0-(x- 0.4)/2.2 where 0.4 is the minimum OT region score and x is the OT region score). If the OT region score is below 1.5, the RVOT probability may be the 1.0 minus the LVOT probability which in this example would be 0.28. If the OT region score is above 1.5, the RVOT probability may be based on a maximum OT region score of 12.0. (See, Yoshida, N., Yamada, T., McElderry, H.T., Inden, Y., Shimano, M., Murohara, T., Kumar, V., Doppalapudi, H., Plumb, V.J. and Kay, G.N., 2014. A novel electrocardiographic criterion for differentiating a left from right ventricular outflow tract tachycardia origin: the V2S/V3R index. Journal of cardiovascular electrophysiology, 25(7), pp.747-753 and Doste, R., Lozano, M., Jimenez-Perez, G., Mont, L., Berruezo, A., Penela, D., Camara, O. and Sebastian, R., 2022. Training machine learning models with synthetic data improves the prediction of ventricular origin in outflow tract ventricular arrhythmias. Frontiers in Physiology, 13, p.909372., which are hereby incorporated by reference.) In some embodiments, the OID system may allow a user to input the OT region scores or specify the ECG analysis technique from which the OT region scores are derived (or combinations of them) that are to be used to generate the adjusted OT site scores. The OID system may generate multiple adjusted OT site probabilities for each site that are based on different mapping systems, ECG analysis techniques, and/or user inputs. The OID system may display the multiple adjusted OT site probabilities to help inform treatment of a patient.

[0018] Because a mapping system may not have an accuracy of 100%, the OID system may generate adjusted OT site probabilities by adjusting the OT site probabilities of the mapping system based on the OT region scores of an ECG analysis technique. An adjusted OT site score may be calculated by multiplying the OT site score by the OT region score and dividing the result by the sum of the OT site scores multiplied by their OT region scores. AOTSP(x) = OTSS(x) * OTRS(x) / Sum[i = 1 ...n] (OTSS(i) * OTRS(i)) where AOTSP is adjusted OT site probability, OTSS(x) is OT site score, OTRS is OT region score, and n is the number of OT sites. For example, if an LVOT site score is 0.97, the LVOT region score is 0.75 and the sum of the OT site scores is 3.73, the adjusted LVOT site probability is 0.38. The OID system may display the adjusted OT site probability for each OT site to assist in informing the treatment selected for the patient such as identifying the target of an ablation procedure.

[0019] In some embodiment, the OID system may display a graphic of a ventricle with the OT sites of that ventricle indicated and may display an indication of the adjusted OT site probabilities and the OT region score for that ventricle. For example, the OID system may display a graphic of the LV with the LVOT sites identified on the graphic and may display an indication of the adjusted LVOT site probabilities (or adjusted LVOT site scores) and the LVOT region score. As another example, the OID system may display a graphic of the RV with the RVOT sites identified along with an indication of the adjusted RVOT site probabilities. The indications of the adjusted OT site probabilities may be numeric and/or may be non-textual such as indications of varying sizes or varying colors near the OT site. When based on the size of the indication, the size may be inversely proportional to the adjusted OT site score or the adjusted OT site probability resulting in a smaller indication for an OT site with a higher adjusted OT site score or a higher adjusted OT site probability. The OID system may display graphics for the LV and the RV at the same time or may display only one graphic at a time. When a user selects the indication of an OT site, the OID system may display the adjusted OT site probability and/or an adjusted OT site score associated with that OT site.

[0020] A technique for generating the adjusted OT site probabilities (AOTSPs) is illustrated by the following example. The OID system receives a patient ECG representing a PVC. The OID system applies the mapping system to the patient ECG to identify OT sites and associated OT site scores. The OT site scores may be based on similarity between a patient ECG and an ECG associated with an OT site. The mapping system may select four sites having the highest OT site scores for further processing. The OT site scores for the four sites for this example are as follows:

The OID system then sums the OT site scores resulting in a sum of 3.73 and divides each OT site score by that sum to generate OT site probabilities. The calculation of the OT site probability for this example are as follows:

[0021] The OID system may adjust the OT site scores based on the accuracy of the mapping system itself in accurately identifying LVOT regions and RVOT regions as containing the OT site of origin. The accuracy may be determined based on applying the mapping system to patient ECGs with OT sites that are known to be an origin (e.g., based on the OT site of a successful ablation). For example, if the top match accuracy is 75% for LVOT sites, the OID system may calculate adjusted OT site scores by multiplying each OT site score by 0.75 for LVOT sites and by 0.25 for RVOT sites. The calculation of the adjusted OT site scores for this example are described as follows:

The OID system may then generate an adjusted OT site probability for each OT site. The OID system sums the adjusted OT site scores resulting in the sum of 1.885 and divides each adjusted OT site score by that sum to generate adjusted OT site probabilities for each OT site. The calculation of the adjusted OT site probabilities for this example is described as follows:

The 01 D system may then output the adjusted OT site probabilities to help inform treatment of the patient. The adjusted OT site probabilities may alternatively be generated by first calculating the products of the OT site probabilities and their corresponding OT region scores. The 01 D system then calculates the adjusted OT site probability for an OT site by dividing the product for that OT site by the sum of the products.

[0022] The OID system may combine the accuracy of the mapping system with the accuracy of an ECG analysis technique to generate adjusted OT site probabilities that factor in both accuracies. The ECG analysis technique may more accurately distinguish OT sites in the RVOT than LVOT and have a top match accuracy of 83% for the RVOT. Continuing with the example, the OT region score (OTRS1 ) for the mapping system are 0.75 and 0.25, and the OT region scores (OTRS2) for the ECG analysis technique are 0.17 and 0.83. The OID system calculates an adjusted OT site score for each OT site by multiplying the OT site score for that OT site by the OT region scores (OTSS * OTRS1 * OTRS2) that contain that OT site. The OID system then generates an adjusted OT site probability for each OT site that is the adjusted OT site score for that OT site divided by the sum of the adjusted OT site scores. The calculations of the adjusted OT site scores using the accuracy of both the mapping system and the ECG analysis system for this example are described as follows:

Because the accuracy of the ECG analysis technique in identifying RVOT sites is better than that of the mapping system as described above, the combination of the accuracies results in OT site 3 having the highest adjusted OT site probability rather than OT site 1 when only the accuracy of the mapping is employed. The OID system may output the final adjusted OT site probabilities in probability order for this example as follows:

[0023] In some embodiments, the OID system may generate site metrics (e.g., adjusted OT site scores and adjusted OT site probabilities) for a target site that is identified by a care provider. For example, during an ablation procedure, a care provider may be interested in determining whether to ablate a target site that is not a site or not one of the top sites identified by the mapping system. To generate a site score for such a target site, the OID system may employ a site score ML model that inputs a patient ECG and the site and outputs a site score. The training dataset for the site score ML model may be generated using the mapping system. For example, the OID system may retrieve arrhythmia ECGs from electronic health records (EHRs) and input each arrhythmia ECG to the mapping system which outputs a site and site score. The training dataset includes training examples that each has a feature vector and a label. The feature vectors include features derived from an arrhythmia ECG (e.g., a QRS-integral and/or a voltage-time series derived from the arrhythmia ECG) and a site identified based on the arrhythmia ECG, and the labels indicate the site scores. The site score ML model is then trained using the training dataset to learn weights for the ML model. The OID system may generate an adjusted site probability for that target site based on that target site and sites identified by the mapping system. The OID system may also be provided a list of target sites, apply the site score ML model to the patient ECG and each target site to generate site scores, and generate adjusted site probabilities (or other site metrics) based on the site scores. [0024] Figure 1 is a flow diagram that illustrates an origin identification method for treating a patient in some embodiments. The OID method 100 identifies potential sites of the origin of an arrhythmia based on a patient ECG, adjusts site scores based on a region accuracy assessment of identifying sites within regions, displays a graphic of a heart with indications of the adjusted site scores and/or adjusted site probabilities derived from the site scores, and performs on ablation on a patient targeting a potential site. In block 101 , the method accesses a patient ECG collected from the patient. In block 102, the method applies a mapping system to the patient ECG to determine sites that are potential origins of an arrhythmia of the patient and determines site scores for the sites. The sites may be in various regions of the patient’s heart such as LVOT and RVOT. In block 103, the method applies a region identification technique to identify a region score for one or more regions of the patient’s heart such as LVOT and RVOT. Techniques for identifying regions other than LVOT or RVOT are described in Kistler, P.M., Chieng, D., Tonchev, I.R., Sugumar, FL, Voskoboinik, A., Schwartz, L.A., McLellan, A.J., Prabhu, S., Ling, L.H., Al-Kaisey, A. and Parameswaran, R., 2021. P- wave morphology in focal atrial tachycardia: an updated algorithm to predict site of origin. Clinical Electrophysiology, 7(12), pp.1547-1556, which is hereby incorporated by reference. A region score indicates the likelihood that a region contains the origin of the patient’s arrhythmia. The region score may be determined based on analysis of the patient ECG. In block 104, the method generates a heart graphic. In block 105, the method adds indications of the sites and adjusted site scores to the heart graphic. In block 106, the method displays the heart graphic to help inform a care provider on treatment of the arrhythmia of the patient. In block 107, the method includes performing an ablation on the patient targeting a site that is identified based on the sites and adjusted site scores. The ablation may be performed by an electrophysiologist guiding an ablation catheter (manually or via a catheter guidance device) to the target site and activating the ablation energy (e.g., radiofrequency and/or pulsed field). The method of blocks 101 -106 are performed by the OID system executing on one or more computing systems.

[0025] Figure 2 is a block diagram that illustrates components of an origin identification system in some embodiments. The OID system 200 includes a generate adjusted site probability component 201 , a retrieve patient data component 202, a retrieve accuracy data component 203, a mapping system 204, and a display adjusted site probability graphic component 205. The OID system interfaces with an EHR data store 211 , an ECG analysis accuracy data store 212, and a mapping data store 213. The generate adjusted site probability component controls the overall processing of invoking the retrieve patient data component to retrieve a patient arrhythmia ECG, the retrieve accuracy data component to retrieve a region accuracy assessment, the mapping system to identify potential sites of the origin of the patient’s arrhythmia (e.g., PVC) and site scores and the display adjusted site probability graphic component to display a graphic of the heart along with an indication of the sites and the adjusted site probabilities. The EHR data store stores electronic health records of patients that include arrhythmia ECGs. The ECG analysis accuracy data store stores information related to the accuracy of various ECG analysis techniques and the mapping system. The mapping data store is a library of mappings of ECGs to sites or, when the mapping system is an ML model, parameters of the ML model such as weights and biases associated with neurons of a neural network.

[0026] The computing systems (e.g., network nodes or collections of network nodes) on which the OID system and the other described systems may be implemented may include a central processing unit, input devices, output devices (e.g., display devices and speakers), storage devices (e.g., memory and disk drives), network interfaces, graphics processing units, communications links (e.g., Ethernet, Wi-Fi, cellular, and Bluetooth), global positioning system devices, and so on. The input devices may include keyboards, pointing devices, touch screens, gesture recognition devices (e.g., for air gestures), head and eye tracking devices, microphones for voice recognition, and so on. The computing systems may include high-performance computing systems, distributed systems, cloud-based computing systems, client computing systems that interact with cloud-based computing system, desktop computers, laptops, tablets, e-readers, personal digital assistants, smartphones, gaming devices, servers, and so on. The computing systems may access computer- readable media that include computer-readable storage mediums and data transmission mediums. The computer-readable storage mediums are tangible storage means that do not include a transitory, propagating signal. Examples of computer- readable storage mediums include memory such as primary memory, cache memory, and secondary memory (e.g., DVD), and other storage. The computer-readable storage media may have recorded on them or may be encoded with computer-executable instructions or logic that implements the OID system and the other described systems. The data transmission media are used for transmitting data via transitory, propagating signals or carrier waves (e.g., electromagnetism) via a wired or wireless connection. The computing systems may include a secure crypto processor as part of a central processing unit (e.g., Intel Secure Guard Extension (SGX)) for generating and securely storing keys, for encrypting and decrypting data using the keys, and for securely executing all or some of the computer-executable instructions of the OID system. Some of the data (e.g., EHRs) sent by and received by the OID system may be encrypted, for example, to preserve patient privacy (e.g., to comply with government regulations such the European General Data Protection Regulation (GDPR) or the Health Insurance Portability and Accountability Act (HIPAA) of the United States). The OID system may employ asymmetric encryption (e.g., using private and public keys of the Rivest-Shamir- Adleman (RSA) standard) or symmetric encryption (e.g., using a symmetric key of the Advanced Encryption Standard (AES)).

[0027] The one or more computing systems may include client-side computing systems and cloud-based computing systems (e.g., public or private) that each executes computer-executable instructions of the OID system. A client-side computing system may send data to and receive data from one or more servers of the cloud-based computing systems of one or more cloud data centers. For example, a client-side computing system (e.g., an ablation planning system) may send a request to a cloudbased computing system (e.g., executing the OID system) to perform tasks such as applying a mapping system to determine sites and site scores or train a site score ML model. A cloud-based computing system may respond to the request by sending to the client-side computing system data derived from performing the task such as a source location of an arrhythmia. The servers may perform computationally expensive tasks in advance of processing by a client-side computing system such as training an ML model or in response to data received from a client-side computing system. A clientside computing system may provide a user experience (e.g., user interface) to a user of the OID system. The user experience may originate from a client computing device or a server computing device. For example, a client computing device may generate a patient-specific graphic of a heart and display the graphic. Alternatively, a cloud-based computing system may generate the graphic (e.g., in a Hyper-Text Markup Language (HTML) format or an extensible Markup Language (XML) format) and provide it to the client-side computing system for display. A client-side computing system may also send data to and receive data from various medical devices such as an ECG monitor, an ablation therapy device, an ablation planning device, and so on. The data received from the medical devices may include an ECG, actual ablation characteristics (e.g., ablation location and ablation pattern), and so on. The data sent to a medical device may include, for example, data in a Digital Imaging and Communications in Medicine (DICOM) format. A client-side computing device may also send data to and receive data from medical computing systems of medical facilities. The data may include patient medical history data, descriptions of medical devices (e.g., type, manufacturer, and model number) that store results of procedures, and so on. The term cloud-based computing system may encompass computing systems of a public cloud data center provided by a cloud provider (e.g., Azure provided by Microsoft Corporation) or computing systems of a private server farm (e.g., operated by the provider of the OID system).

[0028] The OID system and the other described systems may be described in the general context of computer-executable instructions, such as program modules and components, executed by one or more computers, processors, or other devices. Generally, program modules or components include routines, programs, objects, data structures, and so on that perform tasks or implement data types of the OID system and the other described systems. Typically, the functionality of the program modules may be combined or distributed as desired in various examples. Aspects of the OID system and the other described systems may be implemented in hardware using, for example, an application-specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

[0029] Figure 3 is a flow diagram that illustrates the processing of a generate adjusted site probabilities component of the object identification system in some embodiments. The object identification system 300 receives a patient ECG and generates adjusted site probabilities for potential origins of an arrhythmia. In block 301 , the component applies the mapping system to the patient ECG to generate sites and site scores of potential origins of the arrhythmia. In block 302, the component selects sites with the highest site scores. In block 303, the component selects the next site of the sites with the highest site scores for further processing. In decision block 304, if all such sites have already been selected, then the component continues at block 306, else the component continues at block 305. In block 305, the component generates an adjusted site score based on a region accuracy of the mapping system or an ECG analysis system. The component then loops to block 303 to select the next site. In block 306, the component sums the adjusted site scores of the sites. In block 307, the component selects the next site. In decision block 308, if all the sites have already been selected, then the component completes, else the component continues at block 309. In block 309, the component sets the adjusted site probability for that site to the adjusted site score for that site divided by the sum of the adjusted site scores and then loops to block 307 to select the next site.

[0030] Figure 4 is a flow diagram that illustrates the processing of an adjusted site probability graphic component of the origin identification system in some embodiments. The display adjusted site probability graphic component 400 receives site data related to potential sites of the arrhythmia. The site data may include site locations, site scores, site probabilities, adjusted site scores, region scores, and/or adjusted site probabilities for potential sites. In block 401 , the component accesses a 3D mesh of the heart that includes vertices corresponding to locations within the heart wall of the patient. In block 402, the component generates a 3D graphic based on the 3D mesh. In block 403, the component selects the next site of the site data. In decision block 404, if all the sites have already been selected, then the component continues at block 406, else the component continues at block 405. In block 405, the component adds to the 3D graphic an indication of the adjusted site probability of the site data that is associated with the selected site and loops to block 403 to select the next site. The adjusted site probabilities and other site data may be displayed as text in a table. In block 406, the component displays the 3D graphic augmented with the indications derived from the site data. In decision block 407, if the processing of the component is to terminate, then the component completes, else the component continues at block 408. In block 408, the component receives a request to update the 3D graphic. The request may be indicated in various ways such as a cursor hovering over an indication of a site, selecting a site from a list of displayed site data, and so on. In block 409, the component updates the displayed 3D graphic to reflect the request and loops to block 406 to display the updated 3D graphic.

[0031] In some embodiments, the OID system associates colors (e.g., shades of a color) with the vertices near the vertex corresponding to the location of a site. The colors indicate that the locations of those vertices may be an origin but with a likelihood that decreases as the distance from the site increases. For example, the colors may be selected based on a normal distribution as represented by the following equation: where x is the coordinates of a vertex, is the coordinates of the vertex corresponding to the site, and o is an integer representing the rank ordering of the sites (e.g., 1 , 2, 3, etc.). If multiple sites are to be illustrated, then vertices are associated with a color representing the color indicating the highest likelihood associated that vertex.

[0032] In some embodiments, the 3D mesh used to generate the graphic of a heart may be patient specific in the sense that it is generated based on the 3D images (e.g., computed tomography images) of the patient’s heart. The 3D mesh may alternatively be a generic 3D mesh that is based on the cardiac geometry of a generic heart. The 3D mesh may be a simulated 3D mesh that was employed to generate mappings of ECGs to sites used by the mapping system. The simulated 3D mesh may be a 3D mesh of a simulation from which a simulated ECG was generated that matches the patient’s ECG. Since the sites identified by the mapping system are based on a simulated 3D mesh (or a clinical 3D mesh when the mapping system is based on clinical data), the OID system may map those sites to corresponding sites of the patient-specific 3D mesh. After the sites are identified by the mapping system, the OID system may adjust those sites based on differences between the simulated 3D mesh and the patientspecific 3D mesh. The OID system may calculate site scores for the adjusted sites by applying the site score ML model to the patient ECG and the adjusted sites. Techniques for generating adjusted sites are described in PCT App. No. 2023/168017 entitled “Overall Ablation Workflow System” and published on September 7, 2023, which is hereby incorporated by reference.

[0033] In some embodiments, a mapping system ML model may be trained using training data using a computational model of a heart. The computational model models electromagnetic output of the heart over time based on a source configuration of the heart. The electromagnetic output may represent electrical potential, a current, a magnetic field, and so on. The source configuration may include information on geometry and muscle fibers of the heart, torso anatomy, scar locations, rotor locations, electrical properties, and so on, and the EM output is a collection of the electric potentials at various heart locations over time. To generate the EM output, a simulation may be performed for simulation steps of a step size (e.g., 1 ms) to generate an EM mesh for that step. The EM mesh may be a finite-element mesh that stores the value of the electric potential at each heart location for that step. For example, the left ventricle may be defined as having approximately 70,000 heart locations with the EM mesh storing an electromagnetic value for each heart location. If so, a three-second simulation with a step size of 1 ms would generate 3,000 EM meshes that each include 70,000 values. The collection of the EM meshes is the EM output for the simulation. A computational model is described in C. T. Villongco, D. E. Krummen, P. Stark, J. H. Omens, & A. D. McCulloch, “Patient-specific modeling of ventricular activation pattern using surface ECG-derived vectorcardiogram in bundle branch block,” Progress in Biophysics and Molecular Biology, Volume 1 15, Issues 2-3, August 2014, Pages 305- 313, which is hereby incorporated by reference.

[0034] In some embodiments, the training data may be generated by running many simulations, each based on a different source configuration, which is a set of different values for the configuration parameters of the computational model. For example, the configuration parameters for the heart may be cardiac geometry, rotor location, focal source location, ventricular orientation in the chest, ventricular myofiber orientation, cardiomyocyte intracellular potential electrogenesis and propagation, and so on. Each configuration parameter may have a set or range of possible values. For example, a focal source location parameter may have 20 possible locations in an LVOT. Since a simulation may be run for each combination of possible values, the number of simulations may be in the millions.

[0035] In some embodiments, EM outputs of the simulations may be used to generate derived EM data such as a simulated ECG or a simulated VCG from the EM output of a simulation and the torso anatomy of the source configuration. The library mappings may be the derived EM data and the source location (e.g., site of the origin). The mapping ML model may be trained using the library mappings.

[0036] In some embodiments, the library mapping may include source configuration parameters used in the simulation from which the simulated ECG was derived. In such a case, the mapping system may perform a calibration as described in the 754 patent for the identification of a simulated ECG that is similar to the patient ECG. Also, the region scores may factor in characteristics of the patient’s heart such as those of the configuration parameters as described above. For example, the accuracy of a mapping system or an ECG analysis system may vary between patients with different characteristics. The mapping system may be more accurate at identifying sites in patient who are younger than 50. In such a case, the OID system uses different region scores for those younger than 50 and those older than 50.

[0037] An ML model employed by the OID system may be any of a variety or combination of supervised, semi-supervised, self-supervised, unsupervised, or reinforcement learning ML models including a neural network such as fully connected, convolutional, recurrent, or autoencoder neural network, a restricted Boltzmann machine, a support vector machine, a Bayesian classifier, K-means clustering, K Nearest Neighbors (KNN), and so on. When the ML model is a deep neural network, the model is trained using training data that includes features derived from data and labels corresponding to the data. For example, the data may be images of ECGs with a feature being the image itself, and the labels may be sites. The training results in a set of weights for the activation functions of the layers of the deep neural network. The trained deep neural network can then be applied to new data to generate a label for that new data. An ML model may generate values of discrete domain (e.g., classification), probabilities, and/or values of a continuous domain (e.g., regression value, classification probability).

[0038] A neural network model has three major components: architecture, loss function, and search algorithm. The architecture defines the functional form relating the inputs to the outputs (in terms of network topology, unit connectivity, and activation functions). The search in weight space for a set of weights that minimizes the loss function is the training process. A neural network model may use a radial basis function (RBF) network and a standard or stochastic gradient descent as the search technique with backpropagation.

[0039] The following paragraphs describe various aspects of the OID system. An implementation of the OID system may employ any combination or sub-combination of the aspects and may employ additional aspects. The processing of the aspects may be performed by one or more computing systems with one or more processors that execute computer-executable instructions that implement the aspects and that are stored on one or more computer-readable storage mediums.

[0040] In some aspects, the techniques described herein relate to a method for treating a patient who has an arrhythmia, the method including: under control of one or more computing systems, identifying one or more outflow tract (OT) sites from which the arrhythmia may originate by, receiving a patient cardiogram collected from a patient; identifying OT sites and OT site scores by applying a mapping system to the patient cardiogram, each OT site associated with an OT site score indicative of a likelihood that the arrhythmia originates from that OT site, the mapping system based on mappings of simulated cardiograms to OT sites; for each of a plurality of OT sites, calculating an adjusted OT site score for that OT site that is based on the OT site score for that OT site adjusted by an OT region accuracy of an OT region identification system; and calculating an adjusted OT site probability that is the adjusted OT site score for that OT site divided by the sum of the adjusted OT site scores for the plurality of OT sites; generating a graphic of a heart with an indication of an adjusted OT site probability for each of one or more OT sites; and displaying the generated graphic of the heart; and performing an ablation procedure on the patient targeting a target OT site identified factoring in the adjusted OT site probability for one or more OT sites. In some aspects, the techniques described herein relate to a method wherein the mapping system is a machine learning (ML) model that inputs a cardiogram and outputs OT sites and OT site scores, the ML model being trained with training data that includes simulated cardiograms labeled with OT sites. In some aspects, the techniques described herein relate to a method wherein the ML model is a neural network that inputs features derived from a cardiogram and outputs an OT site and an OT site score. In some aspects, the techniques described herein relate to a method wherein the ML model is a convolutional neural network that inputs an image of a cardiogram and outputs an OT site and an OT site score. In some aspects, the techniques described herein relate to a method wherein the cardiogram is represented as a voltage-time series. In some aspects, the techniques described herein relate to a method wherein the mapping system compares the patient cardiogram to a library of library cardiograms mapped to OT sites and OT site scores to identify OT sites mapped to library cardiograms that are similar to the patient cardiogram. In some aspects, the techniques described herein relate to a method wherein the OT region accuracy is derived from an assessment of accuracy of the mapping system in identifying OT regions that include OT sites. In some aspects, the techniques described herein relate to a method wherein the assessment of accuracy of the mapping system is based on the OT sites identified by the mapping system being targets of successful ablation procedures. In some aspects, the techniques described herein relate to a method wherein the OT region accuracy is derived from an assessment of accuracy of an algorithm that is not based on the mapping system in identifying OT regions that include OT sites. In some aspects, the techniques described herein relate to a method wherein the OT region accuracy is derived from an assessment of accuracy of the mapping system in identifying OT regions that include OT sites and from an assessment of accuracy of an algorithm that is not based on the mapping system in identifying OT regions that include OT sites. In some aspects, the techniques described herein relate to a method wherein the OT scores are based on a similarity metric derived from similarity between the patient cardiogram and simulated cardiograms. In some aspects, the techniques described herein relate to a method wherein the OT scores are based on a probability of an OT site being origin of the arrhythmia.

[0041 ] In some aspects, the techniques described herein relate to a method for treating a patient having an arrhythmia, the method including: under control of one or more computing systems, receiving a patient cardiogram collected from the patient; identifying outflow tract (OT) sites within the heart of the patient by applying a mapping system to the patient cardiogram, each OT site associated with an OT site score indicating a likelihood that that OT site is an origin of the arrhythmia; for each of a plurality of OT sites, generating an adjusted OT site score based on the OT site score for that OT site and an accuracy associated with distinguishing between OT regions; and outputting indications based on the adjusted OT site scores to inform an ablation treatment for the arrhythmia; and performing an ablation procedure to treat the arrhythmia, the ablation procedure targeting an OT site selected based on the adjusted OT site scores. In some aspects, the techniques described herein relate to a method wherein the outputting includes outputting an indication of an OT site to an ablation navigation device that controls movement of an ablation catheter to that OT site. In some aspects, the techniques described herein relate to a method wherein the mapping system is a machine learning (ML) model that inputs a cardiogram and outputs OT sites and OT site scores that are probabilities, the ML model being trained with training data that includes simulated cardiograms labeled with OT sites. In some aspects, the techniques described herein relate to a method wherein the mapping system compares the patient cardiogram to library cardiograms mapped to OT sites to identify OT sites mapped to library cardiograms that are similar to the patient cardiogram. In some aspects, the techniques described herein relate to a method wherein the OT site score is based on similarity of the patient cardiogram to a library cardiogram. In some aspects, the techniques described herein relate to a method wherein the accuracy is derived from an assessment of accuracy of the mapping system in identifying OT regions that include OT sites that are determined to be actual origins of an arrhythmia. In some aspects, the techniques described herein relate to a method wherein the assessment of accuracy of the mapping system is based on the OT sites identified by the mapping system that have been target sites of successful ablation procedures. In some aspects, the techniques described herein relate to a method wherein the accuracy is derived from an assessment of accuracy of an algorithm that is not based on the mapping system in identifying OT regions that include OT sites that are origins of arrhythmias. In some aspects, the techniques described herein relate to a method wherein the accuracy is derived from an assessment of accuracy of the mapping system in identifying OT regions that include OT sites and from an assessment of accuracy of an algorithm that is not based on the mapping system in identifying OT regions that include OT sites.

[0042] In some aspects, the techniques described herein relate to one or more computing systems for treating a patient having an arrhythmia, the one or more computing systems including: one or more computer-readable storage mediums that store computer-executable instructions for controlling the one or more computing systems to: receive from a cardiogram collection device a patient cardiogram collected from the patient; identify a plurality of sites within the heart of the patient by applying a mapping system to the patient cardiogram to identify the sites, each site associated with a site score indicative of that site being an origin of the arrhythmia; generate an adjusted site score for each site based on a region score relating to identification of origins of arrhythmias with regions of a heart; and output indications of the adjusted sites scores; and one or more processors for controlling the one or more computing systems to execute one or more of the computer-executable instructions. In some aspects, the techniques described herein relate to one or more computing systems wherein the adjusted site score is based on accuracy in identifying regions that include origins of arrhythmias. In some aspects, the techniques described herein relate to one or more computing systems wherein the accuracy is based on accuracy of the mapping system. In some aspects, the techniques described herein relate to one or more computing systems wherein the accuracy is based on accuracy of a technique other than the mapping system. In some aspects, the techniques described herein relate to one or more computing systems wherein the regions include a left ventricular outflow tract and a right ventricular outflow tract. In some aspects, the techniques described herein relate to one or more computing systems wherein the regions include the endocardium and the epicardium. In some aspects, the techniques described herein relate to one or more computing systems wherein the regions include the anterior papillary muscle and the posterior papillary muscle. In some aspects, the techniques described herein relate to one or more computing systems wherein the regions include sub regions of an outflow tract.

[0043] In some aspects, the techniques described herein relate to a method performed by one or more computing systems for assessing potential target sites for an ablation procedure to treat a patient having an arrhythmia, the method including: receiving a patient cardiogram collected from the patient; identifying outflow tract (OT) sites within the heart of the patient by applying a mapping system to the patient cardiogram, each OT site associated with an OT site score indicating a likelihood that that OT site is an origin of the arrhythmia; for each of a plurality of OT sites, generating an adjusted OT site score based on the OT site score for that OT site and an accuracy associated with distinguishing between OT regions as containing an origin of an arrhythmia; and outputting indications based on the adjusted OT site scores to inform an ablation treatment for the arrhythmia.

[0044] All documents incorporated by reference are incorporated in their entirety for the full extent of their disclosures. In the event of inconsistencies between the language in this document and any incorporated-by-reference document, the language in the incorporated-by-reference document should be considered supplementary to that of this document and the language in this document controls.

[0045] Although the subject matter has been described in language specific to structural features and/or acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims. In addition, the claims cover a practical application relating to improved computer techniques that provide for more effective medical treatment and cover techniques that cannot be practically performed by a human without the aid of a computer.

Claims

CLAIMS I/We claim:

1 . A method for treating a patient who has an arrhythmia, the method comprising: under control of one or more computing systems, identifying one or more outflow tract (OT) sites from which the arrhythmia may originate by, receiving a patient cardiogram collected from a patient; identifying OT sites and OT site scores by applying a mapping system to the patient cardiogram, each OT site associated with an OT site score indicative of a likelihood that the arrhythmia originates from that OT site, the mapping system based on mappings of simulated cardiograms to OT sites; for each of a plurality of OT sites, calculating an adjusted OT site score for that OT site that is based on the OT site score for that OT site adjusted by an OT region accuracy of an OT region identification system; and calculating an adjusted OT site probability that is the adjusted OT site score for that OT site divided by the sum of the adjusted OT site scores for the plurality of OT sites; generating a graphic of a heart with an indication of an adjusted OT site probability for each of one or more OT sites; and displaying the generated graphic of the heart; and performing an ablation procedure on the patient targeting a target OT site identified factoring in the adjusted OT site probability for one or more OT sites.

2. The method of claim 1 wherein the mapping system is a machine learning (ML) model that inputs a cardiogram and outputs OT sites and OT site scores, the ML model being trained with training data that includes simulated cardiograms labeled with OT sites.

3. The method of claim 2 wherein the ML model is a neural network that inputs features derived from a cardiogram and outputs an OT site and an OT site score.

4. The method of claim 2 wherein the ML model is a convolutional neural network that inputs an image of a cardiogram and outputs an OT site and an OT site score.

5. The method of claim 2 wherein the cardiogram is represented as a voltage-time series.

6. The method of claim 1 wherein the mapping system compares the patient cardiogram to library cardiograms mapped to OT sites and OT site scores to identify OT sites mapped to library cardiograms that are similar to the patient cardiogram.

7. The method of claim 1 wherein the OT region accuracy is derived from an assessment of accuracy of the mapping system in identifying OT regions that include OT sites.

8. The method of claim 7 wherein the assessment of accuracy of the mapping system is based on the OT sites identified by the mapping system being targets of successful ablation procedures.

9. The method of claim 1 wherein the OT region accuracy is derived from an assessment of accuracy of an algorithm that is not based on the mapping system in identifying OT regions that include OT sites.

10. The method of claim 1 wherein the OT region accuracy is derived from an assessment of accuracy of the mapping system in identifying OT regions that include OT sites and from an assessment of accuracy of an algorithm that is not based on the mapping system in identifying OT regions that include OT sites.

1 1 . The method of claim 1 wherein the OT scores are based on a similarity metric derived from similarity between the patient cardiogram and simulated cardiograms.

12. The method of claim 1 wherein the OT scores are based on a probability of an OT site being origin of the arrhythmia.

13. A method for treating a patient having an arrhythmia, the method comprising: under control of one or more computing systems, receiving a patient cardiogram collected from the patient; identifying outflow tract (OT) sites within the heart of the patient by applying a mapping system to the patient cardiogram, each OT site associated with an OT site score indicating a likelihood that that OT site is an origin of the arrhythmia; for each of a plurality of OT sites, generating an adjusted OT site score based on the OT site score for that OT site and an accuracy associated with distinguishing between OT regions; and outputting indications based on the adjusted OT site scores to inform an ablation treatment for the arrhythmia; and performing an ablation procedure to treat the arrhythmia, the ablation procedure targeting an OT site selected based on the adjusted OT site scores.

14. The method of claim 13 wherein the outputting includes outputting an indication of an OT site to an ablation navigation device that controls movement of an ablation catheter to that OT site.

15. The method of claim 13 wherein the mapping system is a machine learning (ML) model that inputs a cardiogram and outputs OT sites and OT site scores that are probabilities, the ML model being trained with training data that includes simulated cardiograms labeled with OT sites.

16. The method of claim 13 wherein the mapping system compares the patient cardiogram to a library of library cardiograms mapped to OT sites to identify OT sites mapped to library cardiograms that are similar to the patient cardiogram.

17. The method of claim 16 wherein the OT site score is based on similarity of the patient cardiogram to a library cardiogram.

18. The method of claim 13 wherein the accuracy is derived from an assessment of accuracy of the mapping system in identifying OT regions that include OT sites that are determined to be actual origins of an arrhythmia.

19. The method of claim 18 wherein the assessment of accuracy of the mapping system is based on the OT sites identified by the mapping system that have been target sites of successful ablation procedures.

20. The method of claim 13 wherein the accuracy is derived from an assessment of accuracy of an algorithm that is not based on the mapping system in identifying OT regions that include OT sites that are origins of arrhythmias.

21. The method of claim 14 wherein the accuracy is derived from an assessment of accuracy of the mapping system in identifying OT regions that include OT sites and from an assessment of accuracy of an algorithm that is not based on the mapping system in identifying OT regions that include OT sites.

22. One or more computing systems for treating a patient having an arrhythmia, the one or more computing systems comprising: one or more computer-readable storage mediums that store computerexecutable instructions for controlling the one or more computing systems to: receive from a cardiogram collection device a patient cardiogram collected from the patient; identify a plurality of sites within the heart of the patient by applying a mapping system to the patient cardiogram to identify the sites, each site associated with a site score indicative of that site being an origin of the arrhythmia; generate an adjusted site score for each site based on a region score relating to identification of origins of arrhythmias with regions of a heart; and output indications of the adjusted sites scores; and one or more processors for controlling the one or more computing systems to execute one or more of the computer-executable instructions.

23. The one or more computing systems of claim 22 wherein the adjusted site score is based on accuracy in identifying regions that include origins of arrhythmias.

24. The one or more computing systems of claim 23 wherein the accuracy is based on accuracy of the mapping system.

25. The one or more computing systems of claim 23 wherein the accuracy is based on accuracy of a technique other than the mapping system.

26. The one or more computing systems of claim 22 wherein the regions include a left ventricular outflow tract and a right ventricular outflow tract.

27. The one or more computing systems of claim 22 wherein the regions include the endocardium and the epicardium.

28. The one or more computing systems of claim 22 wherein the regions include the anterior papillary muscle and the posterior papillary muscle.

29. The one or more computing systems of claim 22 wherein the regions include sub regions of an outflow tract.

30. A method performed by one or more computing systems for assessing potential target sites for an ablation procedure to treat a patient having an arrhythmia, the method comprising: receiving a patient cardiogram collected from the patient; identifying outflow tract (OT) sites within the heart of the patient by applying a mapping system to the patient cardiogram, each OT site associated with an OT site score indicating a likelihood that that OT site is an origin of the arrhythmia; for each of a plurality of OT sites, generating an adjusted OT site score based on the OT site score for that OT site and an accuracy associated with distinguishing between OT regions as containing an origin of an arrhythmia; and outputting indications based on the adjusted OT site scores to inform an ablation treatment for the arrhythmia.