WO2024141350A1

WO2024141350A1 - Producing combined error values

Info

Publication number: WO2024141350A1
Application number: PCT/EP2023/086775
Authority: WO
Inventors: Marianna MEO; Ankur Rajiv SHAH; Youssef ROUCHDY; Jacob Laughner
Original assignee: Koninklijke Philips NV
Current assignee: Koninklijke Philips NV
Priority date: 2022-12-26
Filing date: 2023-12-20
Publication date: 2024-07-04
Anticipated expiration: 2025-06-26

Abstract

A mechanism for producing a combined error value for each of a plurality of electrophysiological values, which each represent a value of an electrophysiological property at a different location about an anatomical cavity. For each electrophysiological value, a plurality of error values from different sources is received. A subset of the error values are combined to produce the combined error value, the subset being identified in a received indication.

Description

PRODUCING COMBINED ERROR VALUES

FIELD OF THE INVENTION

The present invention relates to the field of electrophysiological monitoring.

BACKGROUND OF THE INVENTION

It is becoming increasingly common to make use of electrophysiology datasets to monitor or determine the condition of an anatomical structure of a subject/patient. Electrophysiology datasets identify an electrophysiological value for each of a plurality of different locations about the anatomical structure. It is possible to combine these values with a model of the anatomical structure, e.g., to produce electrophysiology maps that identify a shape of the anatomical structure and indicate the electrophysiological value at the different locations.

One example of an electrophysiological value is a local activation time (LAT). Indeed, local activation time (LAT) mapping is routinely performed during electrophysiological (EP) studies to assess cardiac wavefront propagation and allow clinicians to diagnose the subject’s rhythm and define a treatment strategy. LAT maps are usually generated by the interpolation of a finite number of LAT measurements from intracardiac electrograms (EGMs) acquired through diagnostic catheters at specific locations of the heart.

Other examples of electrophysiological values will be readily apparent to the skilled person, such as impedance measurements (useful for identifying the success of ablation) or even voltage measurements (e.g., in the form of an ECG or similar measurement system).

However, accurate interpretation of electrophysiological datasets and/or maps strongly relies on the accuracy of the measurements. These can be compromised by several factors.

For instance, LAT estimation may be more challenging when there is complex morphology, longer duration and/or multiple deflections, as the automatic assignment of a time marker may be more prone to errors.

As another example, uncertainty may be introduced when creating an electrophysiological map, e.g., due to errors introduced by performing projection of a electrophysiological value onto a model of the anatomical structure. If interpolation is performed to produce electrophysiological values, then the interpolation process brings additional uncertainty, as the fitting error is strongly influenced by the choice of the interpolant, the density of the locations for which electrophysiological values are obtained and their mutual distance.

There is therefore a desire to produce information that indicates or reflects any errors in the electrophysiology dataset, which would provide a clinician with useful information for making a clinical decision, as they are able to take possible errors into account when assessing the condition of the anatomical structure and/or subject.

SUMMARY OF THE INVENTION

The invention is defined by the claims.

According to examples in accordance with an aspect of the invention, there is provided a processing system for providing error information about an electrophysiology dataset comprising a plurality of electrophysiological values for a respective plurality of locations about an anatomical structure, each electrophysiological value being derived from one or more electrical measurements of the anatomical structure.

The processing system comprises an input interface configured to receive for each electrophysiological value of the electrophysiological dataset, a different set of error values, each error value indicating a predicted error of the electrophysiological value from a different, respective error source; and an indication of one or more desired error sources, being a subset of all error sources associated with the plurality of different error values of each electrophysiological values.

The processing system also comprises an output interface configured to communicatively couple to an output user interface for displaying information to an individual.

The processing system also comprises a data processor configured to: for each electrophysiological value, process the set of error values for the electrophysiological value to produce a combined error value representing a combined error of the electrophysiological value from each of the one or more desired error sources indicated by the indication, to thereby produce a plurality of combined error values; and when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation of the plurality of combined error values.

The proposed system facilitates production of a combined error value for a subset (e.g., not all) of possible/known error sources for errors in an electrophysiological dataset. The proposed approach allows an operator or user monitoring the electrophysiological dataset to readily identify the contribution of different error sources to the overall or combined error at different locations about an anatomical structure. This aids in the making of a clinical decision as to whether to further investigate the anatomical structure (e.g., to reduce error) or whether the causes of error are acceptable for making a clinician diagnosis or assessment of the anatomical structure (e.g., using the electrophysiological dataset).

Embodiments thereby facilitate improved and safer assessment an anatomical structure represented by an electrophysiological dataset.

In the context of the present disclosure, an electrophysiological value is a value of an electrophysiological parameter, such as a voltage, an impedance or a response time to an electrical stimulus (e.g., a local activation time). Other examples include a sinus node recovery time, sinoatrial, intraatrial and interatrial conduction times, and effective refractory periods (ERP) of right and left atria and atrioventricular node. These examples are non-exhaustive and other suitable examples of electrophysiological values will be apparent to the skilled person.

The anatomical structure may, for instance, be a heart of a subject. This is particularly advantageous as the electrophysiological parameters of the heart are a key indicator of health or condition of the heart, and are important for assessing the condition of the subject and determining possible treatments (e.g., whether or not to perform ablation).

It is noted that the precise mechanism(s) used to produce the error values is/are immaterial to achieving the underlying inventive concept of the present invention. Rather, the underlying concept relies upon an approach for controlling the combination of produced error values to provide a combined error value to which only desired error sources contribute.

For the avoidance of doubt, it is noted that each error value may in fact represent a predicted error or predicted uncertainty that the corresponding electrophysiological value represents the true electrophysiological value. Where appropriate, the term “error” may be replaced by the term “uncertainty” throughout this disclosure.

Preferably, the indication is a user input signal provided by an individual at an input user interface communicatively coupled to the input interface. This approach provides the individual or user with a level of control over which error sources contribute to the combined error values. This technique allows the individual to selectively control and identify the contribution of each error individually, to aid in the assessment of whether further investigative work is required.

In other examples, the indication is a predefined indication. This may, for instance, allow the combined error values to follow or align with recommended error sources for assessing the error values - which may be dependent upon the specific electrophysiological value.

In some examples, the input interface is further configured to receive, for each electrophysiological value, a position within a Euclidean space, the position representing the location about the anatomical structure associated with the electrophysiological value.

The data processor may be configured to define, using the combined error value and the position within the Euclidean space of each electrophysiological value, an error map that represents the combined error value at each of the plurality of locations of the anatomical structure; and, when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation of the error map.

This approach provides an individual with a readily interpretable visualization of the anatomical structure and the relationship of the combined error value with relevant locations about the anatomical structure. This approach thereby credibly assists the individual in performing a technical task of assessing the condition of the anatomical structure, by providing a direct understanding or link between a combined error value and a position/location on an anatomical model of the anatomical structure.

The data processor may be configured to process the position within the Euclidean space of each electrophysiological value to generate an anatomical map of the anatomical structure; and process the combined error value of each electrophysiological value and the anatomical map to produce the error map.

This provides an error map that is directly derived from the locations at which data for deriving the error values was obtained. This provides a direct link between locations and error values, to improve the individual’s understanding of the properties of the anatomical structure.

The plurality of electrophysiological values may comprise at least one electrophysiological value derived by interpolating between two or more sets of one or more electrical measurements of the anatomical structure.

The data processor may be configured to: determine, for each error source, an average error value; and when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation for each error source, wherein, for each error source, the appearance of the visual representation for said error source is responsive to the determined average value of that error source.

This provides a user or individual with a direct and readily interpretable indication of which error sources contribute most to the overall error in the electrophysiological dataset. This aids the individual in performing a technical task of assessing the condition of the anatomical structure, as they are able to identify the impact of causes of error in the electrophysiological dataset, where different causes may result in different interpretations of whether the error is significant.

In some examples, the input interface is configured to receive the plurality of electrophysiological values; and the data processor is configured to when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation of the plurality of electrophysiological values.

The visual representation of the plurality of electrophysiological values and the plurality of combined error values may be provided at a same time, to allow immediate and direct understanding of the error in the displayed electrophysiological values.

The data processor may be configured to: define, using each electrophysiological value and the position within the Euclidean space associated with each electrophysiological value, an electrophysiological map that represents the shape of the anatomical structure and the electrophysiological value across the anatomical structure.

The step of providing the visual representation of the plurality of electrophysiological values may comprise providing a visual representation of the electrophysiological map.

The data processor may be configured to: determine, for each desired error source, whether or not the error values associated with the desired error source follow a Gaussian distribution.

Responsive to determining that the error values associated with each desired error source follow a Gaussian distribution, the data processor may produce a combined error value for each electrophysiological value by processing the error values for the electrophysiological value from the one or more desired error sources using a Gauss error function.

Responsive to determining that the error values associated with each desired error source do not follow a Gaussian distribution, the data processor may produce a combined error value for each electrophysiological value by processing the error values for the electrophysiological value from the one or more desired error sources using a non-parametric technique.

There is also provided computer-implemented method for providing error information about an electrophysiology dataset comprising a plurality of electrophysiological values for a respective plurality of locations about an anatomical structure, each electrophysiological value being derived from one or more electrical measurements of the anatomical structure.

The computer-implemented comprises: receiving for each electrophysiological value of the electrophysiological dataset, a different set of error values, each error value indicating a predicted error of the electrophysiological value from a different, respective error source; receiving an indication of one or more desired error sources, being a subset of all error sources associated with the plurality of different error values of each electrophysiological values; for each electrophysiological value, processing the set of error values for the electrophysiological value to produce a combined error value representing a combined error of the electrophysiological value from each of the one or more desired error sources indicated by the indication, to thereby produce a plurality of combined error values; and controlling an output user interface to provide a visual representation of the plurality of combined error values.

The indication is preferably a user input signal provided by an individual at an input user interface communicatively coupled to the input interface.

There is also proposed a computer program product comprising computer program code means which, when executed on a computing device having a processing system, cause the processing system to perform all of the steps of any herein described method.

These and other aspects of the invention will be apparent from and elucidated with reference to the embodiment(s) described hereinafter.

BRIEF DESCRIPTION OF THE DRAWINGS

For a better understanding of the invention, and to show more clearly how it may be carried into effect, reference will now be made, by way of example only, to the accompanying drawings, in which:

Figure 1 illustrates a system for producing electrophysiological values;

Figure 2 illustrates an error in an EGM;

Figure 3 shows a signal with a LAT annotation uncertainty range;

Figure 4 illustrates a processing system according to an embodiment; and Figure 5 is a flowchart illustrating a method according to an embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

The invention will be described with reference to the Figures.

It should be understood that the detailed description and specific examples, while indicating exemplary embodiments of the apparatus, systems and methods, are intended for purposes of illustration only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatus, systems and methods of the present invention will become better understood from the following description, appended claims, and accompanying drawings. It should be understood that the Figures are merely schematic and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the Figures to indicate the same or similar parts.

The invention provides a mechanism for producing a combined error value for each of a plurality of electrophysiological values, which each represent a value of an electrophysiological property at a different location about an anatomical cavity. For each electrophysiological value, a plurality of error values from different sources is received. A subset of the error values are combined to produce the combined error value, the subset being identified in a received indication.

Embodiments are based on the realization that different error sources introduce different errors into an electrophysiological value. However, the impact of different error sources on the interpretability of the electrophysiological dataset (e.g., to make a clinical decision) is dependent upon the error source. By facilitating control over which error sources contribute to a combined error value, there is improved control and understanding of the accuracy/certainty of the electrophysiological dataset and whether inaccuracies or uncertainties can be ignored (e.g., as they relate to non-important error sources).

Proposed approaches can be employed in any suitable environment in which electrophysiological (EP) values of an anatomical structure are determined, but find particular use in the processing of EP values producing by sampling data using an interventional device.

More generally, the present disclosure proposes a technique for processing error values, for each of a plurality of electrophysiological values, to produce a combined error value for each electrophysiological value. The combined error value represents a combined error from each of a plurality of desired error sources. For the sake of contextual understanding, a brief description shall be hereafter provided of an example approach of generating electrophysiological values and producing error values for each electrophysiological value.

Figure 1 illustrates a system 100 for generating an electrophysiological dataset containing a plurality of electrophysiological values, as well as error values for each electrophysiological value.

Each electrophysiological value is associated with a different location about an anatomical structure, meaning that is it possible to produce an electrophysiological map or model representing the shape of the anatomical structure and identifying the electrophysiological values at different locations of the map/model.

The system 100 comprises an interventional device 110, such as a catheter, a device tracking system 120, a recording system 130, a data processing system 140 and a database/memory 150.

The interventional device 110 is sized/configured to be movable within an anatomical cavity 191 (bounded by an anatomical structure 192) of a subject 190. For instance, the interventional device may comprise a catheter positionable within the heart of the subject 190.

The interventional device 110 comprises at least one sampling electrode 111 configured to be electrically responsive to changes in an electrophysiological property of material/tissue in the vicinity. Thus, as the interventional device moves within the anatomical cavity 191, the electrical response of (i. e. , data sampled at) the at least one sampling electrode will change for different electrophysiological values at different positions of the anatomical structure 192.

The device tracking system 120 is configured to track the location or position of the interventional device within the anatomical cavity. The location or position is defined in a Euclidean space.

Location/position tracking can be performed using any known device tracking technique, particularly those that track a location of a field-sensitive sensor or field-sensitive electrode carried by the interventional device by monitoring the response of the sensor to crossing electric and/or magnetic fields. The sampling electrode 111 may also act as the fieldsensitive sensor, or it may be a separate element carried by the interventional device. Known approaches are disclosed by: International Patent Application having Publication No. WO 9,404,938 Al (which makes use of magnetic fields); European Patent Application having Publication No. EP 2,568,880 Al (which uses fields generated by injecting an electric current); or European Patent Application having Publication No. EP 4,026,494 Al (which tracks using electric fields). Other approaches use non-invasive imaging to track the location of the interventional device, but are less preferred due to exposure to potentially harmful rays and equipment complexity.

The device tracking system may thereby comprise one or more field generators (not shown) configured to induce one or electric and/or magnetic fields in the subject and a sensor monitoring system (not shown) for monitoring the response of the field-sensitive sensor to track or determine a location of the interventional device and/or sampling electrode 111.

The recording system 130 is configured to record or sample, for each of a plurality of positions of the interventional device within the anatomical cavity 191, an electrical response of (i.e. data at) the sampling electrode of the interventional device. More particularly, each of the plurality of positions may represent a position at which the sampling electrode 111 makes contact with the anatomical structure 192 in the anatomical cavity 191.

The recorded electrical response can, by itself, act as an electrophysiological value (e.g., a voltage measurement or an impedance measurement) or may be processable to derive an electrophysiological value (e.g., a local activation time). Sampling (and optionally further processing) data at the sampling electrode 111, for multiple locations/positions of the sampling electrode or interventional device, thereby produces an electrophysiological dataset comprising a plurality of electrophysiological values, one for each of a plurality of positions. The electrophysiological dataset may be stored in the memory 150.

As previously mentioned, the recording system 130 may perform some further processing of the recorded electrical response of the sampling electrode of the interventional device in order to produce an electrophysiological value. For instance, the electrical response may be averaged over time (e.g., to determine an average voltage or average impedance, which can reduce measurement error) or may undergo a more complex function (e.g., to determine a local activation time or other time-dependent value as the electrophysiological value).

Some examples of electrophysiological values may be generated following artificial stimulation of the material/tissue in the vicinity of the sampling electrode 111. For instance, if the electrophysiological value is a local activation time, then the anatomical structure may be artificially stimulated or paced using a pacing electrode carried by the interventional device. In some examples, the system thereby comprises a pacing system 135 for controlling the artificial pacing. However, other approaches make use of non-artificial stimulation (e.g., a heartbeat or stimulation by the cardiac pacemaker) to act as the stimulus for determining the electrophysiological value. Yet other approaches do not rely on pacing to determine or produce the electrophysiological value.

The recording system 130 may also record a tracked location of the interventional device for each position, as determined by the device tracking system 120. Thus, each electrophysiological value produced for the electrophysiological dataset by the recording system may have an associated tracked location.

The data processing system 140 is configured to determine a plurality of error values for each electrophysiological value. Thus, the data processing system produces a different set of error values for each electrophysiological value. Each set of error values comprises a predicted error (value) for each of a plurality of error sources. The same error sources are represented in different sets of error values, i.e., for different electrophysiological values.

The determined sets of error values may, for instance, be stored in the memory or database 150.

The data processing system 140 is optionally also configured to synthesize or produce additional electrophysiological values for the electrophysiological dataset, e.g., update the electrophysiological dataset stored in the database or memory 150. This can be performed, for instance, by interpolating between neighboring or nearby (in terms of location) electrophysiological values and/or extrapolation. An example of a suitable technique is disclosed by European Patent No. EP 3,744,240 Bl.

Thus, the plurality of electrophysiological values may comprise at least one electrophysiological value derived by interpolating between two or more sets of one or more electrical measurements of the anatomical structure.

In some examples, the data processing system is configured to produce an electrophysiological map (EP map) using an anatomical map or model of the anatomical structure and the electrophysiological dataset. An anatomical map or model is formatted in a data structure that defines the location of surfaces of the anatomical structure, e.g., in the form of a mesh-based or a volume-based data structure.

The map or model of the anatomical structure may, for instance, be produced or constructed by monitoring the location of the interventional device within the anatomical structure and producing a map based on the monitored locations, e.g., as disclosed in the European Patent Applications having publication numbers EP 3,558,151 Al and EP 4,036,930 Al or using the KODEX-EPD system, such as explained by Tovia Brodie, Oholi, et al. "Anatomical accuracy of the KODEX-EPD novel 3D mapping system of the left atrium during pulmonary vein isolation: A correlation with computer tomography imaging." Journal of Cardiovascular Electrophysiology 33.4 (2022): 618-625.

Alternatively, the model of the anatomical structure may be produced in a preoperative process and/or be a template.

The data processing system may be configured to process the anatomical model and the electrophysiological dataset (which includes a location for each electrophysiological value) to produce an electrophysiological map or electrophysiological model of the anatomical structure. This defines values for the electrophysiological property at different positions along the surface of the anatomical model.

In some examples, electrophysiological values for locations not on the surface of the anatomical model may be projected onto the surface of the anatomical model to define the values for the electrophysiological model. This can be performed by defining a projection cone from the location at which the electrophysiological sample was taken (with respect to the anatomical model) and the surface defined by the anatomical model. This approach facilitates the defining of electrophysiological values at surface locations for which no electrophysiological value has been directly sampled.

A suitable approach for producing an electrophysiological map or model is set out in European Patent Applications having Publication Numbers EP 0,661,948 Al and EP 1,166,714 Al.

There are a wide variety of potential error sources for electrophysiological data, and a similarly large variety of approaches for determining or predicting an error of an electrophysiological value attributable to said error source.

It can be assumed that the estimation Emeas of the true electrophysiological value Etrue is hampered by an error e accounting for measurement noise according to a linear model: meas f(Etrue> ) (1)

The value Emeas represents the electrophysiological value sampled and/or computed by the recording system 130. The error e is assumed to account for multiple sources of estimation uncertainty. Thus, the error e can be conceptually modelled as:

£ = £1 + £2 - + £_n (2) where e_n represents the error provided by the n_th error source. The function f can represent (but it is not limited to) a linear model, such as that put forward by J. Farrance and R. Frenkel, "Uncertainty of Measurement: A Review of the Rules for Calculating Uncertainty Components through Functional Relationships," Clin Biochem Rev, vol. 33, no. 2, pp. 49-75, 2012.. The function f may be predetermined or, in some examples, could be chosen by the user before starting the procedure.

Depending on the mathematical formulation of the function f, several expressions can be used to compute the standard error associated with e.

In the simplest-case scenario, it is possible to assume a linear relationship between estimated and true electrophysiological values, with a normally distributed error e describing the joint probability of all sources of uncertainty (inline with Bayes’ theorem). This hypothesis will enable the use of easy-to-parametrize Gaussian functions, with a more straightforward computation of the e index.

Alternatively, if a Gaussianity assumption cannot be met (for instance, if the number of electrophysiological values is extremely low), suitable transformations can be applied to the data to remove any form of skewness before further processing, or non-Gaussian inverse cumulative distribution functions (ICDFs) can be fitted to derive the uncertainty metrics.

It will be appreciated that an error value could alternatively be represented by a measure of uncertainty o, as a measure of uncertainty indicates a predicted or possible error for the electrophysiological value.

Similarly, a measure of uncertainty could be represented by a measure of confidence. There is an inverse relationship between a measure of uncertainty and a measure of confidence.

Various examples of possible error sources, and approaches for quantifying the error introduced by such error sources, are hereafter described. However, it will be appreciated that the precise nature of an error source is dependent upon the type of electrophysiological parameter being measured and/or the mechanism by which the electrophysiological parameter is measured. The hereafter described examples should therefore be seen as a small selection of exemplary approaches for determining an error value for a small selection of possible error source for a small selection of possible electrophysiological parameters.

A first example of an error source (“first error source”) is poor signal quality when sampling the electrical response of a sampling electrode. This may result from poor contact between the sampling electrode and the anatomical structure and/or certain pathologies of the anatomical structure (e.g., scar tissue, focal ectopy and so on). Where the anatomical structure is a heart, one approach for determining an error value for the first error source is to make use of EGM characteristics, e.g., duration of a heartbeat, voltage, fragmentation, wherein that long-duration, low-voltage and highly fragmented signals can be quantified by a lower confidence score. An error value or uncertainty value derived from EGM characteristics can be labelled an EGM error or EGM uncertainty OEGM.

A representative EGM example 200 is illustrated in Figure 2. The onset and the offset of the EGM complex may be automatically detected as indicated by label 210 and used to measure its total duration. The voltage 220 can be automatically detected. Low-voltage values may be converted into milliseconds by using EGM energy and/or power metrics. The EGM firing rate (i.e., number of deflections 230 per time unity) and/or the time distance between multiple consecutive peaks may be used to assess fractionation. These indices can be computed either from raw EGMs or from signals obtained in one or more specific frequency bands after filtering (e.g., by wavelet analysis or empirical mode decomposition), based on the characteristics of the pathology under investigation and/or user’s decision.

All these elements can be summarized into the EGM error/uncertainty OEGM. One specific approach for calculating the EGM uncertainty is disclosed by S. Coveney et al., “Probabilistic Interpolation of Uncertain Local Activation Times on Human Atrial Manifolds,” IEEE Trans. Biomed. Eng., vol. 67, no. 1, pp. 99-109, 2020, doi:

10.1109/TBME.2019.2908486.

A second example of an error source is movement of the subject and/or the anatomical structure whilst obtaining the electrophysiological value. Example movements may result from respiration, cardiac wall motion, ectopic beats and so on. This affects the stability of the electrical response at the sampling electrode, thereby affecting the accuracy of any electrophysiological value obtained therefrom. This form of error value can be a stability error value or stability uncertainty o_stab.

Higher confidence levels can be attributed to electrophysiological values that are sampled or produced when the electrical response is stable across multiple consecutive locations in a certain area, thus resulting in a lower error/uncertainty o_stab. By contrast, highly different values in the same spot may hint at potential outlier(s) and denote higher stability error/uncertainty o_stab. A variety of different solutions can be envisaged to quantify this factor, e.g., by using an existing local activation time (LAT) stability criterion embedded in the KODEX-EPD system developed by Philips ®, or computing an autoregressive model estimation error accounting for LAT data temporal structure (e.g., using BIC or AIC criteria as set forth in V. A. Profillidis and G. N. N. Botzoris, Trend Projection and Time Series Methods. 2019).

A third source of error is contact level between the sampling electode 111 and the anatomical structure 191. It is possible to take advantage of the dielectric response of the anatomical structure to determine a measure of the contact level (and thereby a predicted error or uncertainty). In particular, the higher the measure of contact force or amount of contact, the lower the predicted error. As an example, a higher impedance value will denote better contact with the anatomical structure.

Approaches for determining a measure of contact force is known the art, for instance, as described in the European Patent Applications having Publication Number No. EP 3,932,351 Al or EP 3,294,174 Al. Measures of a contact force can be mapped to an uncertainty or error value, e.g., based on a calibration between contract force and error values.

A fourth source of errror is low (local) sampling density. The lower the sampling density, the less confident or accurate results in the sampled region will be, as there is increased inaccuracies if performing outlier detection, noise reduction and/or interpolation/ extrapolation techniques.

Information on local point density can therefore be accounted for, so as to assign a higher density error Odense to electrophysiological values derived from samples taken in more more sparsely sampled areas. More particularly, for any given electrophysiological value, a density error can be assigend or calculated based on the number of neighboring locations having assocaited electrophysiological values within a predefined area surrounding the location of said electrophysiological value (e.g. a circular region centered on the LAT point). Alternatively, network-based kernel density approaches may also be used to assess the location density distribution across multiple cluster regions and predict the desired local density for any given location to achieve optimal point coverage. The density error for an electrophysiological value may be responsive to a difference between the local density error at the location of the electrophysiological value and this desired local density error.

A fifth source of error, found in some embodiments, is an uncertainty of projection of the electrophysiological values onto a surface of an anatomical model to produce an electrophysiological map (if performed). This uncertainty can, for instance, be a result of increased location-to-surface distance (i.e., the distance between the location at which the electrophysiological value is sampled and the distance to the surface of the anatomical structure). More specifically, it can be assumed that locations that are farther from the surface of the anatomical model are more prone to position registration errors, and their coordinates may be annotated by the system on a potentially wider surface area of projection (for a fixed solid angle of projection). A greater level of confidence can be expected for projection onto flatter surfaces and over shorter location-to-surface distances, e.g. the middle of the posterior wall. This is because the resulting projection cone will be narrower and cover a smaller area. By contrast, higher ambiguity is expected at surfaces with higher curvature and for increased location-to-surface distances, e.g. the pulmonary veins, as the location of the sampling electrode in such regions may potentially be equidistant to multiple candidate locations for projection. It is possible to produce a projection uncertainty or error value o_Prqj based on a number of factor, including physiological information about the surface to which the electrophysiological value is projected and/or location-to-surface distances. Regarding physiological information about the surface, information about electrical propagation can be integrated and modeled by computing a conduction velocity vector field, so that sensitive hot spots e.g. slow-conducting pathways, fibrotic and scarred areas will result in lower projection confidence, i.e. higher projection uncertainty o_proj. Point projection uncertainty can be more accurately modeled by integrating shifts due to cardiac motion and respiration, which may be computed either via current KODEX built-in modules or by modeling their impact on catheter tip motion in a probabilistic framework such as that put forward by M. A. Constantinescu, S. L. Lee, S. Ernst, A. Hemakom, D. Mandic, and G. Z. Yang, “Probabilistic guidance for catheter tip motion in cardiac ablation procedures,” Med. Image Anal., vol. 47, pp. 1-14, 2018, doi: 10.1016/j.media.2018.03.008.

A sixth source of error, found in some embodiments, is an error incurred by interpolation. In particular, if additional electrophysiological values are produced via interpolation, then this naturally introduces some error in the interpolated values. A. Shah, M. Meo, Y. Rouchdy, and J. Laughner, “Spatial analysis of localized uncertain local activation times for creation of activation maps,” 2022 descibes how the computation of the spatial distribution of electrophysiological value (here: LATs) comes with an interpolation error between the interpolated values and true ones. Specifically, in the probabilistic interpolation framework (e.g., kriging), this prediction error accounts for spatial data structure and its changes across the mesh manifold, with data covariance expected to decrease as a function of the distance between electrophysiological values.

It is therefore possible to use an interpolation error Winter as an index of error or uncertainty, with lower errors being assigned to electrophysiological values where lower prediction error is measured, and interpolation will be less likely introduce artefacts. Various approaches for determining an error value or uncertainty value for different error sources are disclosed or suggested by S. Coveney et al., “Probabilistic Interpolation of Uncertain Local Activation Times on Human Atrial Manifolds,” as referenced above, and can be used in proposed embodiments.

For the avoidance of doubt, it is noted that the above provided examples aim to improve the contextualization of the present disclosure and that the underlying concept of the present approach is not reliant upon the precise mechanism(s) used to produce the error values for each electrophysiological value.

The present disclosure provides a technique for controlling how the error values for each electrophysiological value are combined, and provides a processing system and method for doing so.

An explanation will now be provided of how an uncertainty can be quantified. In particular, an example will be given of a combined error using two sources of uncertainty: EGM morphology and position of LAT points.

For this purpose, it is assumed that the estimation of the true LAT value LAT_true is hampered only by two sources of uncertainty: EGM and position of the LAT point. The difference between the two LAT values is modelled by an error (E) and can be expressed as: l|LAT_meas — LAT_true || < E - EGM d- ^pos

In this model, E_EGM and E_pos are assumed to be independent and follow respectively the normal distributions N(0, <J_EGM) and N(0,

Therefore, the error s has a Gaussian distribution with mean 0 and variance:

The LAT error s can be estimated at a specific probability p using confidence intervals. It can be expressed as:

is the inverse cumulative distribution function. The terms F^^-1) and E can be calculated easily as the value cr is computed from the source of uncertainty.

As an example of a confidence map, the following quantity is defined:

The max and min values are computed over all of the LAT points on the LAT map. Points with high confidence will have a small value for the error term £. The variation of the confidence term Ci is determined by the error £. The term (max + min) can be seen as a constant, as it doesn’t change the direction of the variation.

A small error £ implies high confidence (high Ci score). A small error £ means the measured LAT value is very close to the true LAT value at a probability p. p is for example set at 0.95 (i.e. 95% confidence interval).

_ For the EGM uncertainty, it is assumed that the measured LAT value LAT_meas = t₅₀ is computed based on a center of mass algorithm, for example as described in the S. Coveney et al. paper “Probabilistic Interpolation of Uncertain Local Activation Times on Human Atrial Manifolds,” referenced above. A center of mass annotation marker is defined as the time when the cumulative area under the rectified envelope of the signal reaches 50% from the onset of the complex.

The standard deviation <J_EGM is defined as:

The timing value t₂₅ is defined as the time where the cumulative area under the rectified envelope reaches 25% (starting from the onset of the activation) and t₇₅ is defined as the time where the cumulative area under the rectified envelope reaches 75%.

Figure 3 shows a signal with a LAT annotation (tso) uncertainty range 250. Figure 3 shows the original signal and the processed envelope of the signal.

The standard deviation for the position error £_pos may be defined as:

0 7

Wherein:

CV = 350 mm/s d is the distance between the catheter point and its projection on the cardiac mesh.

This position error standard deviation is for example explained in more detail in S. Coveney et al., “Probabilistic Interpolation of Uncertain Local Activation Times on Human Atrial Manifolds,” as referenced above.

Figure 4 illustrates a processing system 300 according to an embodiment.

The processing system 300 is configured to provide error information about an electrophysiology dataset comprising a plurality of electrophysiological values for a respective plurality of locations about an anatomical structure, each electrophysiological value being derived from one or more electrical measurements of the anatomical structure.

The processing system 300 comprises an input interface 310, an output interface 320 and a data processor 330.

The input interface 310 is configured to receive for each electrophysiological value of the electrophysiological dataset, a different set of error values, each error value indicating a predicted error of the electrophysiological value from a different, respective error source.

It will be apparent that each electrophysiological value is preferably originally produced by an interventional device within the anatomical cavity of the subject. Nonetheless, all of the steps of the proposed method are performed externally to the subject, i.e., by the processing of data.

The input interface 310 may receive the sets of error values from the data processing system 140 or the memory/database 150 of the system 100 described in Figure 1.

The input interface is also configured to receive an indication of one or more desired error sources, being a subset of all error sources associated with the plurality of different error values of each electrophysiological values.

The indication may, for instance, be received from an input user interface 390. More particularly, the indication may be carried by a user input signal provided by an individual at an input user interface communicatively coupled to the input interface 310.

In other examples, the indication is received from the memory/database of the system 100 (or another memory/database). Thus, the indication may be a predefined indication.

The output interface 320 is configured to communicatively couple to an output user interface 390 for displaying information to an individual. If present, and as illustrated, the input and output user interfaces may be integrated into a same device, e.g., a same computer, phone, laptop or tablet. Alternatively, they may form part of different or separate systems. The data processor is configured to for each electrophysiological value, process the set of error values for the electrophysiological value to produce a combined error value representing a combined error of the electrophysiological value from each of the one or more desired error sources indicated by the indication, to thereby produce a plurality of combined error values.

In this way, a combined error value is produced for each electrophysiological value, where the combined error value represents the error attributable to a desired set of one or more error sources.

The data processor 330 is also configured to, when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation of the plurality of combined error values.

In this way, the user is provided with a visual representation of the combined error values attributable to a particular or desired selection of error sources. This provides a level of control over the presentation of the error values, allowing the user to identify particularly relevant or important sources of error to aid in their interpretation of the electrophysiological values.

Figure 5 is a flowchart illustrating a process 400 or computer-implemented method performed by the processing system 200 to produce the combined error values.

The method 400 comprises a step 410 of receiving for each electrophysiological value of the electrophysiological dataset, a different set of error values, each error value indicating a predicted error of the electrophysiological value from a different, respective error source.

The method 400 also comprise a step 420 of receiving an indication of one or more desired error sources, being a subset of all error sources associated with the plurality of different error values of each electrophysiological values.

The method also comprises a step 430 of, for each electrophysiological value, processing the set of error values for the electrophysiological value to produce a combined error value representing a combined error of the electrophysiological value from each of the one or more desired error sources indicated by the indication, to thereby produce a plurality of combined error values.

The method 400 also comprises a step 440 of controlling an output user interface to provide a visual representation of the plurality of combined error values.

Steps 430 and 440 are performed by a data processor of the processing system. Step 440 may be performed via an output interface of the processing system. The present disclosure envisages a number of approaches for combining error values, for one or more desired error sources, to produce the combined error value, i.e., for use in step 430.

In a simple example, e.g., where each error value is on a same scale or has been normalized to he in a same scale, the combined error value can be produced by summing or averaging the error values for the one or more desired error sources.

In some examples, step 430 comprises normalizing or transforming error values before calculating the combined error value. In particular, additional statistical analysis and/or proceeding can be performed on the error values when calculating the combined error value.

As an example, it is possible to transform all error values associated with each source of error so that they follow a zero-mean normal distribution N(0,o;). In this way, it is possible to describe the dispersion around the true electrophysiological values in terms of a probability density function (PDF) standard deviation oi, where Oi G [oi, 02, ..., On |. 01, 02 and o_n represent the error values for each of n desired error sources. Once the error values for each desired error source have been modelled separately, all their dispersion contributions can be condensed into the overall variance of the process o, so that the probability density function of the error e will be fully characterized.

The combined error value e_COm at a specific probability p can be estimated using confidence intervals.

In some examples, the choice between a Gaussian and a non-Gaussian PDF modelling approach can be automated and iteratively updated, e.g., using PDF goodness-of-fit metrics, or by running suitable statistical tests to assess data distribution skewness (e.g. the Shapiro-Wilk test).

When non-normal PDF functions are selected, the single parameters Oi will denote dispersion in a non-parametric sense (e.g., interquartile range). The proposed probabilistic formulation only takes into account LAT measurement information and informs the operator to which extent the LAT estimate differs from the model value, thus considerably reducing inter-observer variability.

One suitable non-parametric technique that could be used is disclosed by A. Dong, J. S. Chan and G. w. Peters, "Risk Margin Quantile Function Via Parametric and NonParametric Bayesian Quantile Regression," ASTIN Bulletin, vol. 45, no. 3, pp. 503-550, 2015.

Alternatively, if gaussianity assumption cannot be met (for instance, if the number of measurements is extremely low), suitable transformations can be applied to the data to remove any form of skewness before further processing, or non-Gaussian inverse cumulative distribution functions (ICDFs) can be fitted to derive the combined error values.

In view of this, step 430 may comprise determining, for each desired error source, whether or not the error values associated with the desired error source follow a Gaussian distribution. This is illustrated as an optional sub-step 431.

Responsive to determining that the error values associated with each desired error source follow a Gaussian distribution, step 430 may comprise producing a combined error value for each electrophysiological value by processing the error values for the electrophysiological value from the one or more desired error sources using a Gauss error function. This is illustrated as optional sub-step 432.

Step 430 may comprise, responsive to determining that the error values associated with each desired error source do not follow a Gaussian distribution, produce a combined error value for each electrophysiological value by processing the error values for the electrophysiological value from the one or more desired error sources using a non-parametric technique. This is illustrated as optional sub-step 433.

In some examples, the method 400 further comprises a step 450 of receiving, for each electrophysiological value, a position within a Euclidean space, the position representing the location about the anatomical structure associated with the electrophysiological value.

The method 400 may further comprise a step 460 of defining, using the combined error value and the position within the Euclidean space of each electrophysiological value, an error map that represents the combined error value at each of the plurality of locations of the anatomical structure.

In this way, a model or map of the anatomical structure is produced. The model represents the appearance of the surface of the anatomical structure, and indicates, for each of a plurality of locations about the surface, the combined error value at that location.

Approaches for producing a model or map using the error values and the locations (associated with the error values) will be readily apparent to the skilled person.

For instance, the locations can be processed using a point-to-model mechanism to produce a model of the anatomical structure, with shading or coloring of the model being responsive to the combined error value at each location. Interpolation of the combined error values can be performed in order to appropriately shade or color the entirety of the model, where different shades/colors represent different combined error values. Shading/coloring can be performed based on a continuous distribution, a categorical distribution (based on two or more thresholds) or a binarized distribution (e.g., based on a single threshold such as a user- defined threshold).

In this way, step 460 may comprise processing the position within the Euclidean space of each electrophysiological value to generate an anatomical map of the anatomical structure; and processing the combined error value of each electrophysiological value and the anatomical map to produce the error map.

As another example, each combined error value may be projected onto a surface of an existing anatomical model using a previously described approach. The projected combined error values can then be used to color or shade the anatomical model appropriately (i.e., such that different colors or shades represent different combined error values). As before, interpolation of the projected combined error values may be performed to complete shading of the anatomical model. Shading/coloring can performed either a continuous distribution, a categorical distribution (based on two or more thresholds) or a binarized distribution (e.g., based on a single threshold such as a user-defined threshold).

An alternative to shading/coloring, in any above described embodiment, is to associate each (e.g., projected) combined error value with a symbol (e.g., a cross or a circle) positioned at the (e.g., projected) location of the combined error value, in which the size of the symbol changes responsive to the magnitude of the combined error value (e.g., larger symbols indicate larger combined error values).

Step 440 may be accordingly modified to provide a visual representation of the determined or derived error map. Approaches for providing a visual representation of a map are well established in the art. In particular, step 440 may comprise controlling the output user interface to provide a visual representation of the error map.

The proposed approach allows a user or individual to control which error sources contribute to the combined error value, e.g., by changing the indication received in step 420. This allows the user to quickly identify and understand the causes underlying any errors in the electrophysiological values at specific sites. The proposed technique thereby provides the user with significant freedom on the choice of the error sources to be visualized. This is important, as local anatomical reinspection (i.e., re-exploring the sampled space) may be prompted by several motivations (technical uncertainty, pathological mechanisms etc.), and not all of them may be necessary in all scenarios.

The proposed approach thereby allows a user to make an informed decision as to whether further invasive analysis of the individual (e.g., with the interventional device) is required. The method 400 may further comprise a step 470 of determining, for each error source, an average error value. Step 470 can be performed using a simple mean function on all error values produced for that error source.

Step 440 may be corresponding modified to comprise controlling the output user interface to provide a visual representation for each error source, wherein, for each error source, the appearance of the visual representation for said error source is responsive to the determined average value of that error source. This provides a user with useful information for understanding the relevance or contribution of each error source to the overall error experienced by the electrophysiological dataset.

In these embodiments, step 470 may comprise controlling the visual representation for each error source to order or rank the error sources by the average error value of each error source. This allows operator to prioritize the sources with higher impact on the estimation error, e.g., so as to selectively include only the most relevant in the combined error value display.

Preferably, step 440 is performed at a same time as a step of controlling the output user interface to provide a visual representation of the electrophysiological dataset, e.g., in the form of an electrophysiological map or model.

In this way, the combined errors for each electrophysiological value and the electrophysiological values themselves are displayed at the output user interface. This allows a clinician to assess the cardiac rhythm and depict arrhythmia mechanisms while exploring the cardiac chamber. Multiple graphical solutions are possible. For instance, the error e could be interpolated over the cardiac mesh and displayed as a shaded heat map over the LAT distribution.

The present disclosure considers that determining an error value of an electrophysiology value is considered analogous to determining a quality of a transmitted audio/image/video signal, in that it represents a quality or accuracy of information representing a true physiological feature which is important for making a technical decision.

The skilled person would be readily capable of developing a processing system comprising an input interface, a data processor and an output interface for carrying out any herein described method. In particular, each step of the flow chart may represent a different action performed by a processing system, and may be performed by a relevant module of the processing system.

Embodiments make use of a data processor. The data processor can be implemented in numerous ways, with software and/or hardware, to perform the various functions required. A processor is one example of a data processor which employs one or more microprocessors that may be programmed using software (e.g., microcode) to perform the required functions. A data processor may however be implemented with or without employing a processor, and also may be implemented as a combination of dedicated hardware to perform some functions and a processor (e.g., one or more programmed microprocessors and associated circuitry) to perform other functions.

Examples of data processor components that may be employed in various embodiments of the present disclosure include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs).

In various implementations, a processor or data processor may be associated with one or more storage media such as volatile and non-volatile computer memory such as RAM, PROM, EPROM, and EEPROM. The storage media may be encoded with one or more programs that, when executed on one or more processors and/or data processors, perform the required functions. Various storage media may be fixed within a processor or data processor or may be transportable, such that the one or more programs stored thereon can be loaded into a processor or data processor.

It will be understood that disclosed methods are preferably computer- implemented methods. As such, there is also proposed the concept of a computer program comprising code means for implementing any described method when said program is run on a processing system, such as a computer. Thus, different portions, lines or blocks of code of a computer program according to an embodiment may be executed by a processing system or computer to perform any herein described method.

There is also proposed a non-transitory storage medium that stores or carries a computer program or computer code that, when executed by a processing system (e.g., having a data processor), causes the processing system to carry out any herein described method.

In some alternative implementations, the functions noted in the block diagram(s) or flow chart(s) may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved.

Variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. If a computer program is discussed above, it may be stored/distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems. If the term "adapted to" is used in the claims or description, it is noted the term "adapted to" is intended to be equivalent to the term "configured to". If the term "arrangement" is used in the claims or description, it is noted the term "arrangement" is intended to be equivalent to the term "system", and vice versa. Any reference signs in the claims should not be construed as limiting the scope.

Claims

CLAIMS:

1. A processing system (300) for providing error information about an electrophysiology dataset comprising a plurality of electrophysiological values for a respective plurality of locations about an anatomical structure, each electrophysiological value being derived from one or more electrical measurements of the anatomical structure, the processing system comprising: an input interface (310) configured to receive: for each electrophysiological value of the electrophysiological dataset, a different set of error values, each error value indicating a predicted error of the electrophysiological value from a different, respective error source; and an indication of one or more desired error sources, being a subset of all error sources associated with the plurality of different error values of each electrophysiological values; an output interface (320) configured to communicatively couple to an output user interface for displaying information to an individual; and a data processor (330) configured to: for each electrophysiological value, process the set of error values for the electrophysiological value to produce a combined error value representing a combined error of the electrophysiological value from each of the one or more desired error sources indicated by the indication, to thereby produce a plurality of combined error values; and when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation of the plurality of combined error values.

2. The processing system of claim 1, wherein the indication is a user input signal provided by an individual at an input user interface communicatively coupled to the input interface.

3. The processing system of claim 1, wherein the indication is a predefined indication.

4. The processing system of any of claims 1 to 3, wherein: the input interface is further configured to receive, for each electrophysiological value, a position within a Euclidean space, the position representing the location about the anatomical structure associated with the electrophysiological value; and the data processor is configured to: define, using the combined error value and the position within the Euclidean space of each electrophysiological value, an error map that represents the combined error value at each of the plurality of locations of the anatomical structure; and when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation of the error map.

5. The processing system of claim 4, wherein the data processor is configured to: process the position within the Euclidean space of each electrophysiological value to generate an anatomical map of the anatomical structure; and process the combined error value of each electrophysiological value and the anatomical map to produce the error map.

6. The processing system of any of claims 1 to 5, wherein the plurality of electrophysiological values comprises at least one electrophysiological value derived by interpolating between two or more sets of one or more electrical measurements of the anatomical structure.

7. The processing system of any of claims 1 to 6, wherein the data processor is configured to: determine, for each error source, an average error value; and when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation for each error source, wherein, for each error source, the appearance of the visual representation for said error source is responsive to the determined average value of that error source.

8. The processing system of any of claims 1 to 7, wherein: the input interface is configured to receive the plurality of electrophysiological values; and the data processor is configured to when the output interface is communicatively coupled to the output user interface, control the output user interface to provide a visual representation of the plurality of electrophysiological values.

9. The processing system of claim 8, when dependent on any of claims 4 or 5, wherein the data processor is configured to: define, using each electrophysiological value and the position within the Euclidean space associated with each electrophysiological value, an electrophysiological map that represents the shape of the anatomical structure and the electrophysiological value across the anatomical structure.

10. The processing system of any of claims 1 to 9, wherein the data processor is configured to: determine, for each desired error source, whether or not the error values associated with the desired error source follow a Gaussian distribution; and responsive to determining that the error values associated with each desired error source follow a Gaussian distribution, produce a combined error value for each electrophysiological value by processing the error values for the electrophysiological value from the one or more desired error sources using a Gauss error function.

11. The processing system of claim 10, wherein the data processor is configured to responsive to determining that the error values associated with each desired error source do not follow a Gaussian distribution, produce a combined error value for each electrophysiological value by processing the error values for the electrophysiological value from the one or more desired error sources using a non-parametric technique.

12. A computer-implemented method for providing error information about an electrophysiology dataset comprising a plurality of electrophysiological values for a respective plurality of locations about an anatomical structure, each electrophysiological value being derived from one or more electrical measurements of the anatomical structure, the computer- implemented comprising:

(410) receiving for each electrophysiological value of the electrophysiological dataset, a different set of error values, each error value indicating a predicted error of the electrophysiological value from a different, respective error source; (420) receiving an indication of one or more desired error sources, being a subset of all error sources associated with the plurality of different error values of each electrophysiological values; for each electrophysiological value, (460) processing the set of error values for the electrophysiological value to produce a combined error value representing a combined error of the electrophysiological value from each of the one or more desired error sources indicated by the indication, to thereby produce a plurality of combined error values; and

(440) controlling an output user interface to provide a visual representation of the plurality of combined error values.

13. The computer-implemented method of claim 12, wherein the indication is a user input signal provided to the input interface.

14. The computer-implemented method of claim 12 or 13, wherein the plurality of electrophysiological values comprises at least one electrophysiological value derived by interpolating between two or more sets of one or more electrical measurements of the anatomical structure.

15. A computer program product comprising computer program code means which, when executed on a computing device having a processing system, cause the processing system to perform all of the steps of the method according to any of claims 12 to 14.