Error estimation of local activation time (LAT) measured by multi-electrode catheters

Abstract

The present disclosure is entitled "Error Estimation of Local Activation Times (LAT) Measured by a Multi-Electrode Catheter." The present disclosure relates to a method comprising receiving a set of acquisitions via a plurality of electrodes in a heart, wherein each acquisition comprises a set of electrophysiological (EP) signals measured by the electrodes. For at least some of the acquisitions, respective directions of arrival (DOAs) and respective distances relative to the electrodes from which the set of EP signals originate are estimated. Based on the estimated DOAs and distances, a timing error in at least one of the EP signals is estimated. The timing of the EP signals is adjusted to fit the estimated DOAs and distances and correct for the error. An EP map of at least a portion of the heart is generated using the set of EP signals, including the adjusted EP signals.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application is related to U.S. patent application entitled "Automatic Identification of a Location of Focal Source in Atrial Fibrillation (AF)" filed on the same date, attorney docket number 1002-1729, the disclosure of which is incorporated herein by reference. Technical Field

[0003] The present invention relates generally to electrophysiological mapping, and particularly to cardiac electrophysiological mapping. Background Art

[0004] Invasive cardiac techniques for mapping the electrophysiological (EP) properties of cardiac tissue have been previously proposed in the patent literature. For example, U.S. Patent Application Publication No. 2017 / 0042449 describes a system and method for localized EP characterization of the cardiac substrate using a multi-electrode catheter. The system selects at least one electrode cluster from a plurality of electrodes to derive at least one orientation-independent signal from the at least one electrode cluster based on information content corresponding to a weighted portion of an electrogram signal. The system displays or outputs electrophysiological information to a user or process that is independent of catheter orientation.

[0005] As another example, U.S. Patent Application Publication No. 2015 / 0366476 describes a system and method for mapping the electrical activity of the heart. The system may include a catheter shaft having a plurality of electrodes. A processor of the system is capable of collecting a set of signals from at least one of the plurality of electrodes. The set of signals may be collected over a period of time. The processor is also capable of calculating at least one propagation vector from the set of signals, generating a data set from the at least one propagation vector, generating a statistical distribution of the data set, and generating a visual representation of the statistical distribution, such as a circular histogram of angles. The direction (e.g., propagation angle) and speed of cell wavefront propagation can be determined by comparing the activation time sensed by an adjacent electrode with the target electrode for which the propagation vector is being determined.

[0006] U.S. Patent Application Publication No. 2017 / 0202470 describes a system and method for identifying focal origins. The method may include detecting electrocardiogram (ECG) signals over time via a sensor, each ECG signal detected via one of the sensors having a location in the heart and indicating electrical activity of the heart, each signal including at least an R wave and an S wave; creating an RS map, the RS map including an R to S ratio for each of the ECG signals, the R to S ratio including a ratio of an absolute magnitude of the R wave to an absolute magnitude of the S wave; for each of the ECG signals, identifying a local activation time (LAT); and correlating the R to S ratio of the ECG signals on the RS map with the identified LAT, and identifying a focal origin using the correlation.

[0007] U.S. patent application publication 2017 / 0281031 describes an electroanatomical mapping method that is performed by the following steps: inserting a multi-electrode probe into the heart of a living subject; recording concurrent electrograms from electrodes at corresponding locations in the heart; defining corresponding activation time intervals in the electrograms; generating an electrical propagation wave map from the activation time intervals; maximizing the coherence of the waves by adjusting the local activation times within the activation time intervals of the electrograms; and reporting the adjusted local activation times.

[0008] U.S. Patent Application Publication No. 2004 / 0243012 describes a method and system for identifying and locating an intracorporeal isthmus in a subject's heart during sinus rhythm. The method may include: (a) receiving electrogram signals from the heart during sinus rhythm via electrodes, (b) generating a map based on the electrogram signals, (c) determining a location of the intracorporeal isthmus in the heart based on the map, and (d) displaying the location of the intracorporeal isthmus. Summary of the Invention

[0009] Embodiments of the present invention provide a method comprising receiving a collection of acquisitions via a plurality of electrodes in a heart, wherein each acquisition comprises a set of electrophysiological (EP) signals measured by the electrodes. For each acquisition in the acquisition, a corresponding direction of arrival (DOA) and a corresponding distance relative to the electrodes from which the set of EP signals originate are estimated. The acquisitions are aggregated to form a statistical distribution of the acquisitions as a function of the estimated DOA and distance. The statistical distribution of the acquisitions is checked for consistency according to predefined consistency criteria using a statistical test. If the statistical distribution of the acquisitions is found to be consistent, an estimated location in the heart of a focal origin of arrhythmogenic activity of the received EP signals is derived from the statistical distribution. The estimated location of the focal origin is superimposed on an anatomical map of at least a portion of the heart.

[0010] In some exemplary embodiments, for a given acquisition, estimating the DOA and range includes extracting a corresponding set of relative times of arrival from a set of EP signals in the given acquisition, and estimating the DOA and range using the extracted relative times of arrival.

[0011] In some exemplary embodiments, aggregating the acquisitions includes pre-filtering the acquisitions according to a corresponding set of relative arrival times extracted from each acquisition, and including only the pre-filtered acquisitions in the statistical distribution of the acquisitions.

[0012] In an exemplary embodiment, pre-filtering acquisitions based on the extracted set of relative arrival times comprises the steps of: (a) using the estimated DOA and distance, calculating, for each acquisition, a modeled set of relative arrival times that would be obtained by EP waves originating from focal origins at the estimated DOA and distance, and (b) for each acquisition, determining the similarity between the extracted set of relative arrival times and the modeled set of relative arrival times by applying a predefined geometric test.

[0013] In some exemplary embodiments, estimating the similarity comprises calculating a cosine similarity geometric test between the two groups. In other exemplary embodiments, estimating the similarity comprises calculating an estimated error for each relative time of arrival and comparing the estimated error to a given threshold.

[0014] In an exemplary embodiment, the method further includes adjusting a time value of an annotation on an EP signal using the modeled set of relative arrival times, the EP signal having a voltage-time slope less than a pre-specified slope.

[0015] In another exemplary embodiment, pre-filtering the acquisitions includes discarding one or more acquisitions determined to have distinct sets of arrival times.

[0016] In some exemplary embodiments, deriving the estimated position includes fitting a curve to the statistical distribution and finding the maximum of the curve as a function of the estimated DOA and distance.

[0017] In some exemplary embodiments, estimating the DOA and the distance comprises minimizing a cost function. In other exemplary embodiments, minimizing the cost function comprises minimizing a weighted cost function. In yet other exemplary embodiments, minimizing the cost function comprises iteratively minimizing the cost function by removing the EP signal value with the largest estimation error in each iteration.

[0018] In one exemplary embodiment, deriving the estimated locations includes applying k-means analysis to the statistical distribution, projecting the estimated locations on the anatomical structure, and selecting locations having a projected distance less than a given value.

[0019] According to an exemplary embodiment of the present invention, a system comprising an interface and a processor is also provided. The interface is configured to receive a set of acquisitions acquired by a plurality of electrodes in a heart, wherein each acquisition comprises a set of electrophysiological (EP) signals. The processor is configured to (a) estimate, for each acquisition in the acquisitions, a corresponding direction of arrival (DOA) and a corresponding distance relative to the electrodes from which the set of EP signals originate, (b) aggregate the acquisitions to form a statistical distribution of the acquisitions as a function of the estimated DOA and distance, (c) check whether the statistical distribution of the acquisitions is consistent according to a predefined consistency criterion using a statistical test, (d) if the statistical distribution of the acquisitions is found to be consistent, derive from the statistical distribution an estimated location in the heart of a focal origin of arrhythmogenic activity of the received EP signals, and (e) superimpose the estimated location of the focal origin on an anatomical map of at least a portion of the heart.

[0020] Another exemplary embodiment of the present invention provides a method comprising receiving a set of acquisitions via a plurality of electrodes in a heart, wherein each acquisition comprises a set of electrophysiological (EP) signals measured by the electrodes. For at least some of the acquisitions, estimating a respective direction of arrival (DOA) and a respective distance relative to the electrodes from which the set of EP signals originate. Based on the estimated DOA and distance, estimating a timing error in at least one of the EP signals. Adjusting the timing of the EP signals to fit the estimated DOA and distance and correcting for the error. Generating an EP map of at least a portion of the heart using the set of EP signals, including the adjusted EP signals.

[0021] In some exemplary embodiments, generating the EP map includes generating a local activation time (LAT) map.

[0022] In some exemplary embodiments, estimating the DOA and the distance includes deriving the DOA and the distance that minimizes a cost function.

[0023] In an exemplary embodiment, adjusting the timing of the EP signal includes: (a) selecting an initial annotation in the EP signal, (b) determining a corrected annotation corresponding to the initial annotation based on the estimated DOA and distance, and (c) adjusting the timing of the EP signal upon confirming that the corrected annotation meets a predefined condition.

[0024] In another exemplary embodiment, confirming that the corrected annotation satisfies the predefined condition comprises confirming that the corrected annotation falls in a notch between adjacent peaks in the EP signal. In yet another exemplary embodiment, confirming that the corrected annotation satisfies the predefined condition comprises confirming that the corrected annotation and the initial annotation are located on the same monotonically decreasing segment of the EP signal.

[0025] In some exemplary embodiments, confirming that the corrected annotation satisfies the predefined condition includes confirming that the corrected annotation and the initial annotation are located on the same monotonic segment of the EP signal, and confirming that the slope of the segment is below a predefined threshold slope.

[0026] According to an exemplary embodiment of the present invention, a system comprising an interface and a processor is also provided. The interface is configured to receive a set of acquisitions acquired by a plurality of electrodes in a heart, wherein each acquisition comprises a set of electrophysiological (EP) signals. The processor is configured to: (a) estimate, for at least some of the acquisitions, respective directions of arrival (DOAs) and respective distances relative to the electrodes from which the set of EP signals originate, (b) estimate a timing error of at least one of the EP signals based on the estimated DOAs and distances, (c) adjust the timing of the EP signals to fit the estimated DOAs and distances and correct for the error, and (d) generate an EP map of at least a portion of the heart using the set of EP signals (including the adjusted EP signals). BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention will be more fully understood through the following detailed description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, in which:

[0028] Figure 1 is a schematic illustration of an electrophysiological (EP) mapping system according to an exemplary embodiment of the present invention;

[0029] Figure 2 1 is a flow chart schematically illustrating a method for automatically identifying the location of a focal origin of a cardiac arrhythmia according to an exemplary embodiment of the present invention;

[0030] Figure 3A and Figure 3B To illustrate the exemplary embodiment of the present invention, Figure 1 Two graphs of the EP signal collected by the system;

[0031] Figure 4A and Figure 4B To illustrate the respective uses according to an exemplary embodiment of the present invention Figure 3A and Figure 3B A graph of the relative arrival time of the EP signal extracted and modeled;

[0032] Figure 5 To schematically illustrate a method for following an exemplary embodiment of the present invention Figure 4A and Figure 4B Flowchart of a method for deriving direction of arrival (DOA) and range by steps;

[0033] Figure 61 is a flow chart schematically illustrating a method for deriving direction of arrival (DOA) from a focal origin according to another exemplary embodiment of the present invention;

[0034] 7A to 7C (a) shows the exemplary embodiment of the present invention respectively. Figure 1 Graph of a graph of unipolar EP signals acquired by a system of (a) a catheter, (b) the position of the catheter, and (c) an isochronal map showing the corresponding estimated error of the extracted EP values;

[0035] Figure 8A and 8B According to an exemplary embodiment of the present invention, Figure 1 a graph of a unipolar EP signal acquired by a system of FIG and an isochronal map showing the corresponding estimated error of the extracted EP value;

[0036] Figure 9A and Figure 9B are diagrams showing a graph of a unipolar EP signal containing an estimation error above a given threshold, and an initial estimated position of a focal origin in XY space, respectively, according to an exemplary embodiment of the present invention;

[0037] Figure 10 To illustrate the nine iterations of the iterative DOA model according to an exemplary embodiment of the present invention, Figure 9B Plot of the estimated location of focal origin;

[0038] Figure 11A and Figure 11B is a histogram of direction of arrival (DOA) and distance to focal origin according to an exemplary embodiment of the present invention; and

[0039] Figure 12 FIG. 1 is a diagram illustrating DOA clusters analyzed by a k-means clustering model according to an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0040] Overview

[0041] In the case of focal arrhythmias, abnormal electrophysiological (EP) wave impulses propagate abnormally from ectopic lesions in the heart. Focal arrhythmias may occur due to local abnormal cardiac tissue that triggers abnormal EP waves or because of local abnormal cardiac tissue that forms small reentrant pathways that cause existing EP waves to propagate erroneously. In some patients, local arrhythmogenic tissue can be ablated to eliminate the focal arrhythmia. Therefore, identifying the location of focal arrhythmogenic tissue can be clinically valuable.

[0042] The exemplary embodiments of the present invention described below provide EP mapping systems and methods for automatically identifying the location of focal arrhythmogenic activity in the heart. Additionally or alternatively, some exemplary embodiments provide methods for estimating annotation errors in measured EP values ​​and correcting the annotation errors based on their causes. In some exemplary embodiments, a processor generates an EP map (e.g., a LAT map) of at least a portion of the heart based on a set of annotations including the corrected annotations.

[0043] EP mapping systems use multi-electrode catheters such as A catheter (manufactured by Biosense-Webster, Irvine, California) is used to obtain multiple acquisitions from the region of the heart covered by the electrodes. Each acquisition includes a set of EP signals measured by the electrodes, where the size of the set of signals is set according to the number of electrodes. However, other multi-electrode catheters can be used mutatis mutandis with the disclosed techniques.

[0044] In some exemplary embodiments, the processor then estimates, for each acquisition, the direction of arrival (DOA) and the distance R relative to the electrode from which the acquired set of EP signals originates. DOA The processor aggregates the acquisitions to form a statistical distribution of the acquisitions as a function of the estimated DOA and distance (eg, a histogram, or a cluster map in XY space), and checks whether the statistical distribution of the acquisitions is consistent using statistical tests.

[0045] The constancy is checked according to predefined consistency criteria. Examples of relevant consistency tests include, but are not limited to, the use of a constancy estimator or confidence interval. As another example, a cluster map in XY space may be consistent if one or more clusters in the map each contain at least a given percentage (e.g., 10%) of data points, as described below.

[0046] If the acquired statistical distribution is found to be consistent, the processor generates an estimated location of the focal origin of the arrhythmogenic activity of the received EP signal in the heart from the statistical distribution. Finally, the processor superimposes the estimated location of the focal origin on an anatomical map of at least a portion of the heart.

[0047] In some exemplary embodiments, to estimate DOA and range, the processor annotates each EP signal with the arrival time of the EP wavefront, also referred to hereinafter as the initial annotation. The processor extracts a corresponding set of relative arrival times from the initially annotated set of signals (i.e., from a given acquisition). Using a geometric model, the processor analyzes the extracted set of relative times to indicate the nature of the EP wave under consideration, as described below. The model assumes that each acquisition is uniquely associated with a single, traveling, broad EP wavefront with a constant velocity over the area where the EP signal was acquired.

[0048] In some exemplary embodiments, when aggregating acquisitions, the processor pre-filters the acquisitions and includes only the pre-filtered acquisitions in the statistical distribution of the acquisitions. The processor pre-filters the acquisitions by applying the following steps: (a) using the estimated DOA and distance, for each acquisition, calculates a set of modeled relative arrival times that would be obtained by EP waves originating at a focal origin that has the estimated DOA and distance relative to the catheter, and (b) for each acquisition, applies a test to determine the similarity of the extracted set of relative arrival times and the modeled set of relative arrival times, and discards any acquisitions that generate dissimilar sets of timings. Examples of relevant tests include geometric similarity tests, and comparison of the estimation error (also referred to as timing error hereinafter) to a given threshold.

[0049] In some exemplary embodiments, the geometric similarity test includes applying a cosine similarity geometric test between the extracted set of relative arrival times and the modeled set of relative arrival times. The similarity can range between zero (indicating complete dissimilarity) and one (indicating complete similarity). In an alternative exemplary embodiment, a least squares method is used as the geometric test.

[0050] Abnormal EP waves may not necessarily be of focal origin, as indicated by similarity checks. In one exemplary embodiment, regardless of the nature of the EP wave, i.e., focal or abnormal, the derived modeled relative times can be used to adjust the time values ​​of the initial annotations that were not well defined, i.e., where the voltage-time slope of the wavefront was less than a pre-specified slope. The adjusted annotations are also referred to hereinafter as corrected annotations.

[0051] Typically, the processor is programmed with software containing specific algorithms that enable the processor to perform each of the processor-related steps and functions listed above.

[0052] The disclosed techniques for automatically identifying focal origins of arrhythmogenic activity in the heart can improve clinical outcomes of catheter-based treatment of related arrhythmias.

[0053] System Description

[0054] Figure 1FIG1 is a schematic illustration of an electrophysiological (EP) mapping system 10 according to an exemplary embodiment of the present invention. System 10 includes a catheter 14 that is inserted by a physician 32 through a patient's vascular system into a chamber or vascular structure of a heart 12. Physician 32 brings the distal tip 18 of the catheter into contact with the heart wall, for example, at an EP mapping target site. Catheter 14 typically includes a handle 20 having suitable controls that enable physician 32 to steer, position, and orient the distal end of catheter 14 as required for EP mapping.

[0055] Catheter 14 is a multi-electrode catheter, such as the aforementioned one shown in FIG. The Pentaray catheter 14 includes five flexible arms 15, each of which carries four electrodes 16. Thus, the system 10 obtains a total of twenty EP signals in each instance of EP signal acquisition, such as Figure 2 As further described in.

[0056] The catheter 14 is coupled to a console 24 that enables a physician 32 to observe and adjust the functions of the catheter 14. To assist the physician 32, the distal portion of the catheter 14 may include various sensors, such as a contact force sensor (not shown) and a magnetic sensor 33 that provide position, direction, and orientation signals to a processor 22 located in the console 24. The processor 22 may perform several processing functions as described below. Specifically, electrical signals may be transmitted back and forth between the heart 12 and the console 24 via a cable 31 from electrodes 16 located at or near the distal end 18 of the catheter 14. Pacing signals and other control signals may be transmitted from the console 24 to the heart 12 via the cable 31 and electrodes 16.

[0057] The console 24 includes a monitor 29 driven by the processor 22. Signal processing circuitry in the electrical interface 34 generally receives, amplifies, filters, and digitizes signals from the catheter 14, including signals generated by the aforementioned sensors and the plurality of sensing electrodes 16. The digitized signals are received and used by the console 24 and the positioning system to calculate the position and orientation of the catheter 14 and to analyze the EP signals from the electrodes 16 as described in further detail below.

[0058] During the disclosed procedure, the relative positions of the electrodes 16 are tracked. Tracking can be performed using, for example, a microscopy device manufactured by Biosense-Webster. Such a system measures the impedance between the electrode 16 and a plurality of external electrodes 30 coupled to the patient's body. For example, three external electrodes 30 may be coupled to the patient's chest and another three external electrodes may be coupled to the patient's back. (For ease of illustration, Figure 1Only one chest electrode is shown. ) Wire connections 35 connect the console 24 to the body surface electrodes 30 and other components of the positioning subsystem for measuring the position and orientation coordinates of the catheter 14. The method of tracking the position of the electrode 16 based on electrical signals (called active current location (ACL)) is implemented in various medical applications, such as in the above-mentioned Details of the ACL subsystem and method are provided in U.S. Patent No. 8,456,182, which is assigned to the assignee of the present patent application and the disclosure of which is incorporated herein by reference.

[0059] In some exemplary embodiments, in addition to or in lieu of the ACL tracking subsystem, system 10 includes a magnetic position tracking subsystem that determines the position and orientation of a magnetic sensor 33 at the distal end of catheter 14 by generating magnetic fields in a predefined working space using field generating coils 28 and sensing these fields at the catheter. Because electrodes 16 have known positions on arm 15 and a known relationship to each other, once catheter 14 is magnetically tracked in the heart, the position of each of electrodes 16 in the heart becomes known. Suitable magnetic position tracking subsystems are described in U.S. Patents 7,756,576 and 7,536,218, which are assigned to the assignee of the present patent application and whose disclosures are incorporated herein by reference.

[0060] Based on the EP signals from electrodes 16 having tracked positions, electrical activation maps may be prepared according to the methods disclosed in US Patents 6,226,542, 6,301,496, and 6,892,091, which are assigned to the assignee of the present patent application and whose disclosures are incorporated herein by reference.

[0061] The processor 22 operates the system 10 using software stored in the memory 25. For example, the software may be downloaded to the processor 22 in electronic form over a network, or alternatively or in addition, the software may be provided and / or stored on a non-transitory tangible medium such as magnetic memory, optical memory, or electronic memory. Specifically, the processor 22 executes the software disclosed herein, including Figure 2 , which enables the processor 22 to perform the disclosed steps, as further described below.

[0062] Figure 1 The exemplary illustrations shown are chosen solely for the sake of conceptual clarity. Other types of sensing geometries may also be employed, such as a basket catheter or Sensing geometry of the catheter (manufactured by Biosense-Webster).

[0063] Automatic identification of focal origins of atrial fibrillation (AF)

[0064] Figure 2 1 is a flow chart schematically illustrating a method for automatically identifying the location of a focal origin of an arrhythmia according to an exemplary embodiment of the present invention. According to the exemplary embodiment presented, the algorithm executes a process starting with a physician 32 inserting a catheter 14 having a plurality of sensing electrodes 16 into a patient's heart 12 at a catheterization step 100.

[0065] Next, at EP signal acquisition step 102, system 10 receives a collection of multiple EP signal acquisitions from electrodes 16, which physician 32 places in contact with cardiac tissue. In a typical diagnostic interval of thirty seconds used in some of the exemplary embodiments disclosed herein, the system collects 100 to 200 acquisitions, comprising ECG segments, each having a typical duration of 100-200 milliseconds. In some exemplary embodiments, automatic segmentation of the 30-second window is performed by processor 22 to generate 100-200 millisecond segments, each corresponding to a single activation propagating through the atria.

[0066] Next, at a DOA derivation step 104, processor 22 derives estimates of the DOA and distance of the hypothesized focal origin from each acquisition. Then, at a pre-filtering step 106, processor 22 pre-filters each acquisition to discard acquisitions that are unsuitable for inclusion in subsequent statistical analysis. In some exemplary embodiments, processor 22 pre-filters the acquisitions by comparing the estimated relative time error between the extracted EP values ​​and the modeled EP values. This stage may include attempting further processing, such as iterative calculations, to improve the DOA estimate and thereby reduce the estimation error, as described below.

[0067] In other exemplary embodiments, the processor pre-filters the acquisition by running a geometric test of consistency of the modeled EP values ​​and the extracted EP values ​​(eg, running a cosine similarity test).

[0068] Either way, processor 22 summarizes the confirmed DOA and distance values ​​at aggregation step 108. Next, at statistical analysis step 110, processor 22 runs statistical tests to find one or more candidates for DOA and distance (ie, candidate focal locations), if any.

[0069] At a projection verification step 112 , the processor confirms which of the candidate positions are valid positions by projecting the positions on the anatomical structure.

[0070] Furthermore, at a direct verification step 114 , the processor confirms the verification candidate location at which at least a minimum number of focal indicative ECG signals described below are acquired.

[0071] Finally, at a focal origin presentation step 116 , processor 22 superimposes one or more identified locations of focal origins of arrhythmogenic activation onto the anatomical map of at least a portion of heart 12 .

[0072] Figure 2 The exemplary flow chart shown in is chosen purely for the sake of conceptual clarity. More details and specific exemplary embodiments of the steps briefly described above will be given below, including in Figure 5 and Figure 6 Flowchart of .

[0073] Derivation and confirmation of direction of arrival (DOA) and range by the first method

[0074] Figure 3A and Figure 3B To illustrate the exemplary embodiment of the present invention, Figure 1 For example, the two graphs of the EP signal curve 42 collected by the system are shown in FIG. Figure 1 The EP mapping system of FIG. 1 uses catheter 14 to acquire the EP signals shown. These two acquisitions are part of a collection of acquisitions numbering between tens and hundreds. Such a collection may include acquisitions acquired at different intracardiac placements of the catheter and / or repeated acquisitions acquired during the same placement. Using a tracking system, each of the multiple electrodes has a location in the heart.

[0075] The graph shows the initial annotation time 44 at which the EP wave "hit" each of the twenty electrodes of the catheter 14. Annotation is performed by methods known in the art, such as described in U.S. Patent 8,700,136, which is assigned to the assignee of the present patent application and the disclosure of which is incorporated herein by reference.

[0076] exist Figure 3A In the example, the EP wave first hits electrodes "13" and "14", then hits electrodes "15" and "7", and so on. Figure 3B In the example, the EP wave first hits electrodes "5" and "6", then hits electrodes "7" and "8", and so on.

[0077] like Figure 3A and Figure 3B As shown, some of the initially annotated voltage-time slopes of the EP signal are not well defined, i.e., are shallow (e.g., Figure 3B curve Figure 10 In one exemplary embodiment, the disclosed technology improves the accuracy of such annotations by deriving corrected annotations, as follows Figure 7A 、 Figure 8A and Figure 9AIn one exemplary embodiment, the time values ​​of the initially undefined annotations are adjusted using the modeled time based on the known geometry of the catheter 14 derived below, i.e., where Figure 3A and Figure 3B The voltage-time slope shown is smaller than the pre-specified slope.

[0078] Figure 3A and Figure 3B Provided by way of example. If another catheter with multiple electrodes is used, such as a basket catheter or a Lasso catheter, the size of the acquisition (e.g., the number of plots in a set) and the annotation time will reflect the geometry of the given catheter and be similarly used by the disclosed techniques.

[0079] Figure 4A and Figure 4B To illustrate the respective uses according to an exemplary embodiment of the present invention Figure 3A and Figure 3B The processor 22 will pre-filter the acquisition according to the corresponding set of extracted relative arrival times by performing the following steps.

[0080] Figure 4A and Figure 4B The extraction time 66 in the process is calculated by the processor 22. Figure 3A and Figure 3B The time difference between the initial annotation time 44 is derived. Figure 4A and Figure 4B The corresponding modeled relative time 68 in is then derived by the processor 22 using the corrected annotated time (not shown, as described below).

[0081] Figure 4A and Figure 4B The color scale 48 in the upper portion of encodes the relative arrival times by color coding each of the depicted twenty electrodes 16 of the catheter 14 . Figure 4A and Figure 4B The upper portion of the diagram also shows the tracked positions of electrodes 16 (on the schematically defined arm 15 of the Pentaray catheter 14) in two cases where the electrodes are collecting EP signals. The position of each electrode 16 is tracked in 3D space using, for example, the aforementioned ACL tracking technique. The XYZ axes (Z not shown) belong to a fixed reference axis system, such as a position tracking system 20 using the ACL method.

[0082] Based on the extracted relative arrival times 66, the processor 22 geometrically estimates (e.g., as shown by arrows 40a and 40b) a hypothetical focal origin 50 from which the EP wave appears to originate, as further marked by the distance shown by the line 60 connecting the common location where arrows 40a and 40b originate from the distal tip 18 of the catheter 14.

[0083] Figure 4A and Figure 4B The lower portion of FIG. 5 shows, on the same graph, a set of extracted relative arrival times 66 and a corresponding set of modeled relative arrival times 68. The relative arrival times 66 are derived by processor 22 from the initial annotated time at which the actual EP wave "hit" the electrode. The modeled times 68 are calculated by processor 22 using the estimated DOA 55 and distance 60, assuming the simulated EP wave originates from a focal origin 50 having an estimated DOA and distance relative to electrode 16.

[0084] In one exemplary embodiment, the processor determines that an acquisition is indicative of focal origin only if the cosine similarity timing match derived from the acquisition exceeds a value of 0.9. Geometric tests (e.g., metrics) other than cosine similarity may be used that examine the degree of similarity between the extracted and modeled groups, such as the Hamming distance.

[0085] The processor estimates the similarity of each pair of such groups (denoted herein as vector S) for each acquisition using the following cosine similarity equation EX and S MD ) Degree of similarity:

[0086] Equation 1

[0087] The cosine similarity, in which the normalized inner product of two ordered groups is calculated, can give any value between -1 and 1. In practice, cosine similarity is particularly used in positive space, where the result is confined to [0, 1). For example, a value of 1 corresponds to complete similarity, while a value of zero or any negative value indicates complete dissimilarity. In an exemplary embodiment, the processor determines that the groups are similar if the calculated cosine similarity gives a value above a pre-specified minimum value (such as above 0.9).

[0088] Processor 22 runs a similarity check on all groups derived from the set of acquisitions and discards acquisitions that have a cosine similarity below a pre-specified minimum (eg, <0.9).

[0089] Next, the processor 22 calculates the vector (r) defined as follows using equations 2 and 3 only for the acquisitions that successfully pass the cosine similarity test. 50 -r 18 ) and the distance 60 from which the signal group originated:

[0090] Equation 2 DOA = phase (r 50 -r 18 )

[0091] Equation 3: Distance = || r 50 -r 18 ||,

[0092] where r 50 and r 18 are the vector coordinates of the assumed focal origin 50 and the distal tip 18 of the catheter 14, respectively. In some exemplary embodiments, when converting the system coordinates from 3D to 2D, r 18 will be the zero vector since the center of the catheter is placed at the origin of XY space.

[0093] Figure 5 To schematically illustrate a method for following an exemplary embodiment of the present invention Figure 4A and Figure 4B Flowchart of a method for deriving direction of arrival (DOA) and range by the steps of FIG. The process begins with processor 22 extracting relative time of arrival 66 from the initially annotated EP signal at relative time extraction step 200 .

[0094] Next, at a DOA and distance estimation step 202 , processor 22 derives each acquired estimate of the DOA 55 and distance 60 of the EP wave based on the assumption that the EP signal is generated by a single EP wave that (a) has a wide wavefront and (b) propagates at a constant velocity, and based on the known geometry of catheter 14 .

[0095] Next, at a relative time modeling step 204, based on the values ​​of the estimated DOA 55 and the distance 60, the processor 22 calculates the relative times that will generate the focal wave with the tentative DOA and distance in step 106. Then, at a similarity checking step 206, the processor 22 checks the degree of similarity between the extracted set of relative times and the modeled set of relative times, for example, by using a cosine similarity test.

[0096] If the groups are found to be dissimilar, then the processor 22 stores or discards the bad DOA and distance values ​​as non-indicative at an acquisition discard step 208. At an aggregation step 210, the processor 22 aggregates all acquisitions with extracted and modeled relative time groups that were found to be similar (i.e., by pre-filtering) into a single distribution as a function of DOA and distance (e.g., into the following Figure 11A and Figure 11B 70 and 72). As shown below Figure 11A and Figure 11BIf deemed present in the histogram, the aggregated DOAs and distances were statistically analyzed to find the location of the focal origin, as shown.

[0097] Figure 5 The exemplary flow chart shown in is selected solely for conceptual clarity. Additional steps may typically be performed, such as the physician 32 initially anatomically mapping the relevant portion of the heart 12 (e.g., using the Fast Anatomical Mapping (FAM) procedure) to obtain an anatomical map. The criteria may vary depending on the type of statistical tool used. In an exemplary embodiment, the discarded modeling timing groups may still be used to adjust the corresponding initial annotation times that were not clearly defined, as described below under "LAT Improvement."

[0098] Derivation and confirmation of direction of arrival (DOA) and range by the second method

[0099] Figure 6 1 is a flow chart schematically illustrating a method for deriving a direction of arrival (DOA) from a focal origin according to another exemplary embodiment of the present invention.

[0100] Figure 6 The illustrated process begins with processor 22 extracting relative arrival times 66 from the initially annotated EP signals at a relative time extraction step 300. Next, at a projection check step 304, the algorithm checks whether to apply a 3D or 2D weighted DOA model estimate by checking whether the projection of the catheter (i.e., projection step 302) into 2D space is valid, as described below.

[0101] Next, depending on whether the projection step 304 is found to be an invalid step or a valid step, the processor 22 runs a 3D weighted DOA discovery model at a 3D modeling step 306 or a 2D weighted DOA discovery model at a 2D modeling step 308 , respectively.

[0102] At an estimate error step 310 , using either the 3D or 2D model, the processor 22 then checks whether the estimated error between the modeled relative times and the extracted relative times is below a given threshold.

[0103] If the estimation error is within a given threshold, the processor 22 applies a LAT improvement calculation to make the DOA estimate more accurate at a LAT improvement step 312. LAT improvement is described further below.

[0104] On the other hand, if the estimation error is above a given threshold, then at a DOA iteration estimation step 314 , the processor 22 runs the DOA iteration model.

[0105] At a subsequent estimate error step 316, processor 22 then checks whether the estimated error recalculated using the iterative model is below a given threshold. If not, processor 22 stores or discards the poor DOA and range values ​​as non-indicative at an acquisition discard step 318. However, if the iterative model is successful, processor 22 applies the LAT improvement step 312 to the results.

[0106] In either approach, the successfully pre-filtered LAT-improved estimates of DOA and distance are aggregated by processor 22 at an aggregation step 320. As shown below, if the statistical model indicates the presence of a focal origin, the aggregated DOA values ​​are statistically analyzed to find the location or locations of the focal origin.

[0107] In an exemplary embodiment, Figure 6 The disclosed method for deriving DOA and direction described in

[15] utilizes a cost function in 3D space, such as 7A to 7C As stated.

[0108] 7A to 7C (a) shows the exemplary embodiment of the present invention respectively. Figure 1 , (a) a graph of a unipolar EP signal acquired by the system, (b) the position of the catheter 14, and (c) an isochronal map showing the corresponding estimated error 550 of the extracted EP value. Figure 7B The position of the distal tip 18 of the catheter 14 in XY space is shown, as well as the actual position 338 of the catheter on the anatomy of the left atrium 340 .

[0109] The estimated error 550 (ie, the timing error 550) is shown in FIG. Figure 7A 、 Figure 8A and Figure 9A , as the time difference between the initial annotation and the corrected annotation.

[0110] Figure 7A A set of monopolar signals 330 is shown with measured and initially annotated local activation times 332 (ti - circles) and corresponding estimated local activation times 334 That is, the corrected annotation is derived using the cost function model described below. As described further below, the estimated error 550 between the measured value of activation time and the modeled EP value is calculated for each electrode as the time difference.

[0111] A cost-function-based DOA model was applied to each acquisition consisting of a set of at least 10 local atrial activations. i , local atrial activation time of electrode i, i = 1, ..., m10 ≤ m ≤ N, where N is the number of active electrodes of the catheter, e.g. catheters, N = 20. If a single EP wave is assumed to originate from any point in 3D space and travels towards the catheter with constant conduction velocity (CV), a cost function J(θ) can be defined for the “total cost” of the model:

[0112] Equation 4

[0113]

[0114] In Equation 4, is defined as the DOA point located at (x0, y0, z0) and at t i Arrived at (x i ,y i , z i ) is the distance between the electrodes i at the position t0. Time t0 is defined as the arrival time of the bias wave at all electrodes, and v is 1 / CV of the wave. The term in J(θ) is a regularization term, and it effectively prefers solutions closer to the distal tip 18 of the catheter, thereby increasing the probability of finding a solution within the anatomy of the atrium. The goal of our model is to minimize the cost J(θ) by finding the "best" θ = (x0, y0, z0, t0, v) that minimizes the cost J(θ), which can be done using a gradient descent estimation procedure with the constraint that v is greater than zero. Gradient descent is based on the observation that if the multivariable function J(θ) at the kth iteration k ) at point θ k is confined and differentiable in the neighborhood of , then J(θ k ) from θ k Along J(θ k ) decreases fastest when moving in the negative gradient direction, making and denotes the differential operation and γ is the learning rate factor. γ should be small to ensure transitions, but not too small to overcome slow transitions or convergence to local minima of J(θ). For a formal description of the gradient descent algorithm, we derive the differential equations for J(θ) with respect to each of the parameters (x0, y0, z0, t0, v):

[0115] Equation 5

[0116]

[0117] Figure 7B The upper panel depicts the resulting estimated focal activity. Figure 7B In the middle, the color code (with Figure 4A and Figure 4B The relative arrival times are encoded by color coding each of the twenty depicted electrodes 16 of the catheter 14 ) using the same color code 48 in . Figure 7BAlso shown are the tracked positions of electrodes 16 (on the schematically defined arm 15 of the Pentaray catheter 14) in two separate instances of the electrodes acquiring EP signals. The position of each electrode 16 is tracked in 3D space using, for example, the aforementioned ACL tracking technique. The XYZ axes (Z not shown) belong to a fixed reference axis system, such as a position tracking system 20 employing the ACL method.

[0118] Finally, the cost function model derived DOA 55 and the distance 60 from which a set of signals 330 originate are also shown.

[0119] Figure 7C (Isochronous Map) shows the propagation of the analyzed EP wave from the focal origin location 336 within the circular line 344, which represents the Figure 7C The color bar on the right side shows the arrival time in milliseconds. The circle 350 indicates the location of the electrode, and the numbers inside the circle indicate the arrival time in milliseconds. Figure 7A The cost function is derived to obtain the individual (ie, per-electrode) estimated errors 550 .

[0120] LAT Improvement

[0121] In some exemplary embodiments of the present invention, the processor adjusts the timing of the initial annotation of the EP signal by selecting an initial annotation in the EP signal, determining a corrected annotation corresponding to the initial annotation based on the estimated DOA and distance, and adjusting the timing of the EP signal upon confirming that the corrected annotation meets a predefined condition, as described below.

[0122] Processor 22 derives (e.g., calculates) estimated error 550 by: (a) using a cost function θ=(x0, y0, z0, t0, v) that minimizes the set of positions and conduction velocities, and the measured positions of the electrodes to calculate the impact time and (b) calculate the difference between each electrode

[0123] In an exemplary embodiment, if one of the predefined conditions 1 to 3 is met, Replace LAT value t i To improve LAT estimation:

[0124] 1. In the fragmentation signal (not shown) or dual potential LAT (such as Figure 9A The value 333 of the EP signal of the middle electrode 15 is found within

[0125] 2. is not an anchor point, which means that the weight of the LAT is less than 0.3, as described in the weighting model below (Equation 6). LATs with low weights are LATs with “shallow” voltage deflections (i.e., the voltage-time slope of the EP signal is less than the pre-specified slope), so their initial annotation time (such as Figure 9A The initial comment in 525) is not very "reliable".

[0126] 3.t i and Both are located in the unipolar negative deflection (such as Figure 9A Between the starting point and end point of the negative deflection 555 of the EP signal of electrode 16).

[0127] The description continues with another topic, which describes a simplified implementation of the cost function.

[0128] In some exemplary embodiments, a cost function in 2D space can be applied. In the 2D model, the catheter is projected onto the surface; this is performed by obtaining the two eigenvectors with the highest eigenvalues. If the energy retained by the two eigenvectors is greater than 95%, the model assumes that the projection from 3D space to the surface is valid, and the set of equations is simpler, θ = (x0, y0, t0, v), without the z dimension.

[0129] In some exemplary embodiments, an alternative DOA estimation step comprising estimating DOA using a weighted cost function is used by the algorithm and is described in Figure 8A and Figure 8B middle.

[0130] Figure 8A and 8B According to an exemplary embodiment of the present invention, Figure 1 4 and isochronous maps showing the corresponding estimation errors of the extracted EP values. The main concept behind the weighted cost function DOA model described below is that "sharp" activations are more "reliable" than shallow activations, where the sharpness level is based on t i The dv / dt of the unipolar signal at each t i Based on its dv / dt, it is mapped to a weight w between 0 and 1. i .exist Figure 8A In , the number next to each circle indicates the weight of the slope.

[0131] It should also be noted that in Figure 8AIn the embodiment of the present invention, some EP signals include the earliest S wave pattern, such as in the signal 444 sensed by electrodes E19 and E20 (together 448). Such a negative slope pattern (without the signal amplitude rising first as the EP wave approaches the electrode) indicates an abnormal focal EP wave propagating away from the electrode. This condition indicates that the catheter 14 is "right on target", where some of the electrodes are near the focal origin of the arrhythmia (e.g., the distance 60 is less than the length of the arm 15).

[0132] like Figure 8B As shown in , the estimated location 446 of the focal origin derived using the weighted cost function is at least partially surrounded by the measured location of electrode 16.

[0133] The required alternation in the set of equations for the 2D cost function model (i.e., Equation 5 without z-dependence) is as follows:

[0134] Equation 6

[0135]

[0136] In some exemplary embodiments, if the estimated error of the relative time is above a given threshold, an iterative DOA estimation process is applied and Figure 9A and Figure 9B 、 Figure 10 as well as Figure 11A and Figure 11B Described in.

[0137] Figure 9A and Figure 9B and FIG. 5 , respectively, are diagrams showing a graph of a unipolar EP signal with an estimated error 550 above a given threshold, and a corresponding initial estimated position 560 of the focal origin in XY space, according to an exemplary embodiment of the present invention. In addition, some EP values ​​exist within the dual potential LAT (such as Figure 9A In addition, some EP values ​​(measured and estimated) exist between the start and end of a unipolar negative deflection (such as at Figure 9A The negative deflection 555 of the EP signal of electrode 16 in the EP signal is within).

[0138] like Figure 9B As shown, position 560 is very close to the location of the distal tip of the catheter, however, given the above observations, this location may be incorrect.

[0139] In one exemplary embodiment, if the average estimation error is above a given threshold (eg, 7 milliseconds), processor 22 runs an iterative calculation to estimate the DOA. Figure 9AThe average estimation error in is 12.4 milliseconds. In each iteration, the local activation time with the highest estimation error is removed from the DOA estimate. This process is repeated as long as there are more than ten valid local activation time values.

[0140] Figure 10 To illustrate the nine iterations of the iterative DOA model according to an exemplary embodiment of the present invention, Figure 9B Graph of the estimated location of the focal origin 560. In iteration zero and iteration one, the focal origin location 560 almost coincides with the location of the distal tip 18 of the catheter, however, from iteration two to iteration nine, the location of the focal activity shifts and is ultimately placed near (570) electrode "1". Figure 10 In the figure, full circles represent active electrodes used for DOA estimation, while donut-shaped circles represent inactive electrodes. The label "Iteration x" next to the donut-shaped electrodes indicates that the particular electrode was eliminated from the DOA estimation at iteration x. The percentage of inactive segments is a good measure of the "complexity" of AF for that subject.

[0141] Between iteration 0 and iteration 9, the maximum estimation error dropped from approximately 25 milliseconds to less than 5 milliseconds. The conduction velocity CV, also used as an estimate of the cost in Equation 4, dropped from more than 100 mm / mSeC to a minimum of 0.5 mm / mSeC.

[0142] The iterative model disclosed in the present invention is used to handle acquisitions with noisy waveforms or situations with more than one wave propagating toward the catheter.

[0143] The first statistical test method and the second statistical test method

[0144] The duration of a typical acquisition is 100-200 milliseconds. A typical recording has a 30-second unipolar signal and therefore contains approximately 120-200 acquisitions. All valid DOA estimates from approximately 120-200 acquisitions are stored and subsequently aggregated until all acquisitions have been processed, and then statistical methods are applied to the database of valid DOA estimates.

[0145] First statistical method

[0146] As noted above and described in this section, processor 22 places the aggregated DOA and distance values ​​in a histogram and performs statistical analysis on the histogram. In one exemplary embodiment, the processor is configured to derive the estimated position from the histogram by fitting a curve to the histogram and finding the maximum of the curve as a function of the estimated DOA and distance.

[0147] Figure 11A and Figure 11B70A and 70B, respectively, are histograms of direction of arrival (DOA) and histograms 72A and 72B of distance to focal origin, according to an exemplary embodiment of the present invention. As shown, the DOA distribution consists of the number of acquisitions for each DOA value, and the distance distribution consists of the number of acquisitions for each distance value. Typically, the processor 22 compiles (i.e., summarizes) the histograms 70A and 70B and the histograms 72A and 72B from a number ranging from tens to hundreds of acquisitions that have passed the pre-filtering stage, such as Figure 2 shown and analyzed in Figure 3.

[0148] Using statistical tests, the processor 22 first checks if Figure 11A and Figure 11B The DOA distribution shown in the example of the DOA histogram produces consistent DOA values. Examples of consistency testing tools include, but are not limited to, the use of constancy estimators and confidence intervals.

[0149] If the DOAs are found to be inconsistent, for example, by a distribution indicating two or more DOA values ​​with different affinities, the processor 22 ends the disclosed focal origin identification process. In an exemplary embodiment, the processor 22 presents a notification to the user that the process did not identify a focal origin of arrhythmogenic activity.

[0150] If processor 22 derives a consistent DOA value from the acquired distribution as a function of DOA, processor 22 best estimates the DOA and the distance to the focal origin in question based on the distribution. Processor 22 then uses the best estimated DOA and distance to identify the location of the focal origin of arrhythmogenic activity in heart 12 that generated the received EP signal.

[0151] like Figure 11A As shown, through Figure 11A The most common DOA values ​​for the clinical cases analyzed by histogram 70A fall near a DOA of 0.85π. The corresponding most common distance, indicated by histogram 72A, is approximately 300 mm. Therefore, processor 22 can identify the location of the focal arrhythmia at an angle of 0.85π relative to the X-axis for this patient at a distance of approximately 300 mm from the location of distal tip 18.

[0152] Figure 11B Shown by Figure 11B The most common DOA values ​​for the clinical cases analyzed by histogram 70B fall near a DOA of 0.5π. The corresponding most common distance, indicated by histogram 72B, is approximately 300 mm. Therefore, processor 22 can identify the location of the focal arrhythmia at an angle of 0.5π relative to the X-axis for this patient at a distance of approximately 300 mm from the location of distal tip 18.

[0153] The second statistical method

[0154] Figure 12 FIG2 is a diagram illustrating DOA clusters analyzed by a k-means clustering model according to an exemplary embodiment of the present invention. The processor 22 applies the k-means clustering model to a set of DOA values ​​that are pre-filtered and aggregated in the XY space.

[0155] exist Figure 12 In the figure, the circles represent DOA estimates from multiple acquisitions of LAT values ​​included in a single recording session. In the recorded results shown, there are two DOA clusters that "explain" the data. In the recording, the first cluster 80 (red circle at (-3.7 mm, -0.2 mm)) contains 80.5% of the DOA, and the second cluster 88 contains 19.5% of the DOA.

[0156] A focal origin is identified if the estimated position of one of the major clusters (more than 10% of the DOA segments) can be projected to the anatomical structure, i.e. the distance from the estimated position, such as the estimated position 84, to the anatomical structure is less than a given value, such as 6 mm (configurable). Figure 12 As shown, the distance from k-means clustering estimated position 84 to anatomical structure position 90 (given by the length of arrow 85) is approximately 3 mm, well below the 6 mm upper limit. Therefore, position 90 is validated as a focal origin by the disclosed techniques.

[0157] Focal origin can also be verified if we find at least 10 indications (configurable) of the earliest S-wave pattern in the electrode located within a 6 mm radius from the focus. It is important to note that DOA based on focus detection can be demonstrated in anatomical locations without placing the catheter in the area of ​​focal activity, so the verification process is optional.

[0158] While the exemplary embodiments described herein primarily relate to cardiac applications, the methods and systems described herein can also be used in other applications, such as neurology. The disclosed methods can also be applied to any dataset that contains spatiotemporal "clues" of focal activity, and a processor is required to find this focal activity, such as for focal estimation in epilepsy patients using EEG / MEG.

[0159] It should therefore be understood that the exemplary embodiments described above are cited by way of example, and the present invention is not limited to what is specifically shown and described above. On the contrary, the scope of the present invention includes combinations and subcombinations of the various features described above, as well as variations and modifications thereof, which will occur to those skilled in the art upon reading the above description and which are not disclosed in the prior art. The documents incorporated by reference into this patent application are considered an integral part of this application, except that if any term defined in these incorporated documents conflicts with a definition explicitly or implicitly given in this specification, only the definition in this specification shall be considered.

Claims

1. A method for error estimation of local activation time, the method comprising: receiving a collection of acquisitions via a plurality of electrodes in the heart, wherein each acquisition comprises a set of electrophysiological (EP) signals measured by the electrodes; for at least some of the acquisitions, estimating respective directions of arrival (DOAs) and respective distances relative to the electrodes from which the set of EP signals originate; estimating a timing error in at least one of the EP signals based on the estimated DOA and the range; adjusting the timing of the EP signal to fit the estimated DOA and range and correct for the error; and generating an EP map of at least a portion of the heart using the set of EP signals, including the adjusted EP signal, Wherein adjusting the timing of the EP signal comprises: selecting an initial annotation in the EP signal; determining a corrected annotation corresponding to the initial annotation based on the estimated DOA and distance; and The timing of the EP signal is adjusted upon confirming that the corrected annotation satisfies a predefined condition. 2 . The method of claim 1 , wherein generating the EP map comprises generating a local activation time (LAT) map. 3 . The method of claim 1 , wherein estimating the DOA and the distance comprises deriving the DOA and the distance that minimize a cost function. 4 . The method of claim 1 , wherein confirming that the corrected annotation satisfies the predefined condition comprises confirming that the corrected annotation falls into a notch between adjacent peaks in the EP signal. 5 . The method of claim 1 , wherein confirming that the corrected annotation satisfies the predefined condition comprises confirming that the corrected annotation and the initial annotation are located on a same monotonically decreasing segment of the EP signal.

6. The method of claim 1 , wherein confirming that the corrected annotation satisfies the predefined condition comprises confirming that the corrected annotation and the initial annotation are located on a same monotonic segment of the EP signal, and confirming that a slope of the segment is below a predefined threshold slope.

7. A system for error estimation of local activation time, the system comprising: an interface configured to receive a set of acquisitions acquired by a plurality of electrodes in the heart, wherein each acquisition comprises a set of electrophysiological (EP) signals; as well as A processor configured to: for at least some of the acquisitions, estimating respective directions of arrival (DOAs) and respective distances relative to the electrodes from which the set of EP signals originate; estimating a timing error in at least one of the EP signals based on the estimated DOA and the range; adjusting the timing of the EP signal to fit the estimated DOA and range and correct for the error; and generating an EP map of at least a portion of the heart using the set of EP signals, including the adjusted EP signal, wherein the processor is configured to adjust the timing of the EP signal by: selecting an initial annotation in the EP signal; determining a corrected annotation corresponding to the initial annotation based on the estimated DOA and distance; and The timing of the EP signal is adjusted upon confirming that the corrected annotation satisfies a predefined condition.

8. The system of claim 7, wherein the EP map comprises a local activation time (LAT) map.

9. The system of claim 7, wherein the processor is configured to estimate the DOA and the distance by deriving the DOA and the distance that minimize a cost function.

10. The system of claim 7, wherein the processor is configured to confirm that the corrected annotation satisfies the predefined condition by confirming that the corrected annotation falls into a notch between adjacent peaks in the EP signal.

11. The system of claim 7, wherein the processor is configured to confirm that the corrected annotation satisfies the predefined condition by confirming that the corrected annotation and the initial annotation are located on a same monotonically decreasing segment of the EP signal.

12. The system of claim 7, wherein the processor is configured to confirm that the corrected annotation satisfies the predefined condition by confirming that the corrected annotation and the initial annotation are located on the same monotonic segment of the EP signal, and confirming that the slope of the segment is below a predefined threshold slope.

Images