Retrospectively optimizing the mapping of ECG signals by detecting inconsistencies

Abstract

The invention is entitled "Retrospectively optimizing mapping of ECG signals by detecting inconsistencies". The invention discloses a system comprising a signal acquisition circuit and a processor. The signal acquisition circuit is configured to receive a plurality of intracardiac signals acquired by a plurality of electrodes of an intracardiac probe in a heart of a patient. The processor is configured to perform a sequence of annotation visualization operations at subsequent times by performing in each operation the following steps: extracting a plurality of annotation values from the intracardiac signals; selecting a group of the intracardiac signals; identifying in the group one or more annotation values that are statistically deviating beyond a predefined deviation measure; and visualizing the annotation values to a user while omitting and avoiding visualization of the statistically deviating annotation values. The processor is further configured to estimate an omission rate of annotation values over one or more of the annotation visualization operations, and to take a corrective action in response to detecting that the omission rate exceeds a predefined threshold.

Description

[0001] Cross Reference to Related Applications

[0002] This patent application is related to U.S. Patent Application entitled "Using Statistical Characteristics of Multiple Grouped ECG Signals to Detect Inconsistent Signals" and Attorney Docket No. BIO6196USNP1, filed on the same date, the disclosure of which is incorporated herein by reference. TECHNICAL FIELD

[0003] The present invention relates generally to in-vivo medical procedures and instruments, and in particular to in-vivo cardiac electrocardiogram (ECG) sensing. BACKGROUND

[0004] When measuring and annotating internal electrocardiogram (iECG) signals generated by a large number of electrodes, it can be desirable to process these signals (e.g., by a computer) in order to reduce embedded noise.

[0005] Various methods exist for such iECG signal processing. For example, U.S. Patent Application 2009 / 0089048 describes an automated method of determining local activation times (LATs) for four or more multichannel cardiac electrogram signals, including a ventricular channel, a mapping channel, and a plurality of reference channels. SUMMARY

[0006] One embodiment of the invention described herein provides a system comprising a signal acquisition circuit and a processor. The signal acquisition circuit is configured to receive a plurality of intracardiac signals acquired by a plurality of electrodes of an intracardiac probe in a heart of a patient. The processor is configured to perform a sequence of annotation visualization operations at subsequent times by performing, in each operation, the following steps: extracting a plurality of annotation values from the intracardiac signals; selecting a group of the intracardiac signals; identifying one or more of the annotation values in the group that are statistically deviant beyond a predefined measure of deviation; and visualizing the annotation values to a user while omitting and avoiding visualization of the statistically deviant annotation values. The processor is further configured to estimate an omission rate of the annotation values in one or more of the annotation visualization operations and to take a corrective action in response to detecting that the omission rate exceeds a predefined threshold.

[0007] In some embodiments, the processor is configured to take the corrective action by reapplying one or more of the omitted annotation values and re-identifying the statistically deviant annotation values. In some embodiments, the processor is configured to take the corrective action by re-extracting one or more of the annotation values from the intracardiac signals.

[0008] In one embodiment, the processor is configured to define the measure of deviation in terms of a standard score of the annotation values. In another embodiment, the processor is configured to define the measure of deviation in terms of one or more percentiles of the annotation values.

[0009] In one exemplary embodiment, in a given annotation visualization operation, the processor is configured to compute the deviation of the annotation values on intracardiac signals acquired by a selected subset of spatially related electrodes in the heart that are no more than a predefined distance apart from each other. In another embodiment, in computing the deviation of the annotation values of a given annotation visualization operation, the processor is configured to average the intracardiac signals over a plurality of time-dependent cardiac cycles occurring within a predefined time period.

[0010] In yet another embodiment, in a given annotation visualization operation, the processor is configured to correct one or more of the annotation values in a given intracardiac signal acquired by a given electrode in the set to compensate for a displacement of the given electrode relative to other electrodes in the set.

[0011] In some embodiments, the annotation values include local activation times (LATs). In some embodiments, the processor is configured to visualize the annotation values by superimposing the annotation values (except for the statistically deviated annotation values) on a model of the heart.

[0012] According to one embodiment of the present application, there is additionally provided a method comprising receiving a plurality of intracardiac signals acquired by a plurality of electrodes of an intracardiac probe in a heart of a patient. A sequence of annotation visualization operations is performed at subsequent times by performing in each operation the steps of: (i) extracting a plurality of annotation values from the intracardiac signals, (ii) selecting a set of intracardiac signals, (iii) identifying in the set one or more annotation values that are statistically deviated beyond a predefined measure of deviation, and (iv) visualizing the annotation values to a user while omitting and avoiding visualization of the statistically deviated annotation values. An omission rate of the annotation values is estimated over one or more of the operations in the annotation visualization operations. A corrective action is taken in response to detecting that the omission rate exceeds a predefined threshold.

[0013] The present application will be more fully understood from the following detailed description of the embodiments thereof, taken together with the drawings in which: BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 Schematic illustration of an electroanatomical system for multi-channel measurement of intracardiac ECG signals according to embodiments of the present application;

[0015] Figure 2Fig. 1 is a diagram schematically illustrating a plurality of electrodes acquiring signals in a plurality of cardiac cycles according to an embodiment of the present application;

[0016] Figure 3A Fig. 4 is a flowchart schematically illustrating a first method for enhancing the reliability of annotated values according to an embodiment of the present application;

[0017] Figure 3B Fig. 5 is a flowchart schematically illustrating a second method for enhancing the reliability of annotated values according to an embodiment of the present application;

[0018] Figure 4 Fig. 6 is a flowchart schematically illustrating an improved method for enhancing the reliability of annotated values according to an embodiment of the present application.

[0019] Figure 5A Fig. 7 is a diagram schematically illustrating a series of LAT values from a set of spatially related electrodes in an electroanatomical system according to an embodiment of the present application being annotated;

[0020] Figure 5B Fig. 8 is a diagram schematically illustrating statistical property calculations on LAT values received by a processor (in Figure 5A ) when no corrective action is taken according to an embodiment of the present application;

[0021] Figure 5C Fig. 9 is a diagram schematically illustrating avoiding error propagation by reapplying omitted values according to an embodiment of the present application;

[0022] Figure 6A Fig. 10 is a diagram schematically illustrating the probability of a series of LAT values acquired by a set of spatially related electrodes in an electroanatomical system according to an embodiment of the present application;

[0023] Figure 6B Fig. 11 is a diagram schematically illustrating LAT values of a set 1 signal being annotated according to an embodiment of the present application;

[0024] Figure 6C Fig. 12 is a diagram schematically illustrating LAT values of a set 2 signal being annotated according to an embodiment of the present application;

[0025] Figure 6D Fig. 13 is a diagram schematically illustrating a set 1 signal being reannotated according to an embodiment of the present application;

[0026] Figure 6E Fig. 14 is a diagram schematically illustrating set 2 annotation values being recalculated after reannotation according to an embodiment of the present application; and

[0027] Figure 7A flowchart schematically showing a method for avoiding error propagation by reannotation according to an embodiment of the application. DETAILED DESCRIPTION

[0028] SUMMARY

[0029] Intracardiac probe-based (e.g., catheter-based) cardiac diagnostic and treatment systems can measure a plurality of intracardiac signals, such as electrocardiograms (ECGs), during invasive procedures. Such systems can use electrodes (hereinafter also referred to as "distal electrodes") fitted at the distal end of the probe to acquire the plurality of intracardiac signals. The measured signals can be used to provide visual cardiac information to a physician, such as 3-D mapping of intracardiac pathologic electrograms sources of a patient, and support corrective medical procedures, such as ablation.

[0030] The measured signals are typically weak, with low signal-to-noise ratio (SNR). In addition, the current connection of some electrodes to tissue can be poor or non-existent. On the other hand, many electrodes are used, and thus there can be some redundancy in the data received by the system from the electrodes.

[0031] Embodiments of the application disclosed herein provide intracardiac probe-based electroanatomical measurement and analysis systems and methods of statistical properties of signals collected using distal electrodes to improve the quality and reliability of the collected data.

[0032] In the description below, we will refer to the annotated values of local activation times (LATs). However, the disclosed techniques are not limited to LATs; in various embodiments of the application, annotated values of various other suitable signal parameters can be used.

[0033] In some embodiments according to the application, the processor extracts the annotated values of the signals (e.g., LATs), and then calculates statistical properties of the LAT values of a set of signals acquired by a corresponding set of electrodes (which can include all or some of the electrodes). In one embodiment, the statistical properties include the mean of the LAT values of the set of signals (e.g., ); in other embodiments, the properties also include the standard deviation of the set (e.g., ). The processor then uses statistical methods to determine, for each of the set of signals, whether the annotated value of the signal is a valid value or a value that should be ignored.

[0034] In another embodiment, the statistical property comprises the quartiles of the set of LAT values. The processor calculates the first quartile Ql and the third quartile Q3, and then ignores all values below Ql or above Q3 (the first quartile (Ql) is defined as the median between the minimum and the median of the data set; the third quartile (Q3) is the median between the median and the maximum of the data set). Alternatively, the processor can define the measure of deviation of the LAT values according to any other suitable percentile (or multiple percentiles) of the LAT values. Further alternatively, any other suitable procedure of discarding abnormal LAT values can be used.

[0035] The technology disclosed above assumes that the set of electrodes exhibit similar annotation values in the absence of noisy and irregular current connections. Typically, the annotation values taken by electrodes that are far from each other can vary significantly. Moreover, the signal from each electrode can be periodically annotated with each heartbeat ("cardiac cycle"), and the annotation values derived from cardiac cycles that are far from each other in time can vary. In one embodiment, the set of signals is interrelated. In some embodiments, the tracking system measures the geometric positions of the electrodes, and the set comprises annotation values derived only from neighboring electrodes ("spatially related", i.e. electrodes that are not more than a predefined distance apart from each other). In other embodiments, the set comprises annotation values only from neighboring cardiac cycles ("temporally related", i.e. cardiac cycles that all occur within no more than a predefined time period); and in one embodiment, the set comprises values that are both spatially and temporally related (simply "related values").

[0036] In some embodiments, the processor, after calculating the statistical property of the set of related LAT values, omits LAT values of the set that are statistically deviant (e.g. significantly different from the average value of the set of values) (the remaining set of LAT values will be referred to as the set of valid LAT values). Thus, LAT values corresponding to electrodes with poor connections or to electrodes subject to extreme noise can be eliminated from the set of valid LAT values.

[0037] In an embodiment, in order to determine whether a LAT value is statistically deviant from the average LAT of a set of signals, the processor measures the deviation of the annotated LAT value from the average value of the set of LAT values. In one embodiment, the measure of deviation is the standard score of the value (defined as the difference between the value and the average value divided by the standard deviation), which is compared to a preset limit. For example, a value that is more than 3.5 standard deviations above the average value (standard score = 3.5) or more than 1.5 standard deviations below the average value (standard score = -1.5) can be considered statistically deviant, and thus omitted. In another embodiment, the processor omits values below the first quartile or above the third quartile.

[0038] In some embodiments of the application, the processor can mitigate changes in LAT values of spatially related electrodes due to different time delays in cardiac signal propagation within the heart. According to embodiments, the processor can correct LAT annotations taken by a given electrode by compensating for the displacement of the given electrode relative to other electrodes, so as to cancel the differences in propagation delays.

[0039] ECG signals can sometimes be ambiguously interpreted, and two LAT values (correct and incorrect) can be annotated with different probabilities. For example, a signal can have two peaks close to each other. Thus, in some cases, the computed average of the subset can be closer to the incorrect LAT value than to the correct LAT value.

[0040] In some embodiments, the processor repeats the selection of the group over time (e.g., once every 32 cardiac cycles), with partial overlap between subsequent groups. The processor forms a new group by removing some (but not all) of the old values (e.g., values extracted in the oldest 16 cardiac cycles) and adding new values (e.g., values extracted in the newest 16 cardiac cycles) (“old” and “new” refer in this context to the sequential number of cardiac cycles in which the signal was taken). If the average value computed by the processor for the first group is incorrect (e.g., due to ambiguity of the signal), the processor can omit the correct overlapping LAT values in the next group, thus breaking the average computation, so that the error in the first average computation can propagate to later groups, but the newly annotated LAT values can be correct.

[0041] Embodiments according to the application disclosed herein avoid such error propagation. In one embodiment, the processor monitors the number of omitted LAT values, and in response to the rate at which values are omitted, the processor can decide to consider the omitted values in the computation of new average values, so that the error does not propagate beyond the erroneous group. In other embodiments, if the processor detects a large number of omitted values, the processor can re-annotate some of the values with alternative interpretations of the corresponding signal, and then re-count the omitted values; if the number of omitted values will decrease, the processor will select the alternative annotations.

[0042] In general, a processor according to embodiments of the present application can improve the quality and reliability of a set of annotated values of a spatially and / or temporally dependent intracardiac signal by computing statistical properties of the set of annotated values, comparing the annotated values to the set of mean values, and omitting values from the set of valid values that are far from the mean values. In some embodiments, prior to the statistical property computation, the processor can modify the set of annotated values to correct for propagation delays of the signal. To avoid error propagation, the processor can monitor the number of omitted values, and in response to the rate at which omissions are occurring, the processor can reevaluate the set and restore previously omitted values; in other embodiments, in response to the rate of omitted values, the processor can re-annotate the values, looking for alternative LAT interpretations.

[0043] System Description

[0044] Figure 1 A schematic illustration of an electroanatomical system 21 for multichannel measurement of intracardiac ECG signals according to embodiments of the present application. In some embodiments, the system 21 is used for electroanatomical mapping of the heart.

[0045] Figure 1 A physician 22 is depicted using an electroanatomical catheter 23 to perform electroanatomical mapping of a heart 24 of a patient 25. The catheter 23 includes one or more arms 26 at a distal end thereof that can be mechanically flexible, with one or more distal electrodes 27 coupled to each of the one or more arms. It will be appreciated that although a catheter with five arms is depicted, other types of catheters can be used in alternative embodiments according to the present application. The electrodes are coupled to a processor 34 through an interface 32. Figure 1 A physician 22 is depicted using an electroanatomical catheter 23 to perform electroanatomical mapping of a heart 24 of a patient 25. The catheter 23 includes one or more arms 26 at a distal end thereof that can be mechanically flexible, with one or more distal electrodes 27 coupled to each of the one or more arms. It will be appreciated that although a catheter with five arms is depicted, other types of catheters can be used in alternative embodiments according to the present application. The electrodes are coupled to a processor 34 through an interface 32.

[0046] During an electroanatomical mapping procedure, a tracking system is used to track the intracardiac locations of the distal electrodes 27 so that each of the acquired electrophysiological signals can be associated with a known intracardiac location. One example of a tracking system is the active current location (ACL) system described in U.S. Patent 8,456,182. In the ACL system, the processor estimates the respective locations of the distal electrodes 27 based on impedance measurements between each of the distal electrodes and a plurality of surface electrodes 28 coupled to the skin of the patient 25 (for ease of illustration, only one surface electrode is shown). The processor can then associate any electrophysiological signals received from the distal electrodes 27 with the location at which the signal was acquired. Figure 1 A physician 22 is depicted using an electroanatomical catheter 23 to perform electroanatomical mapping of a heart 24 of a patient 25. The catheter 23 includes one or more arms 26 at a distal end thereof that can be mechanically flexible, with one or more distal electrodes 27 coupled to each of the one or more arms. It will be appreciated that although a catheter with five arms is depicted, other types of catheters can be used in alternative embodiments according to the present application. The electrodes are coupled to a processor 34 through an interface 32.

[0047] In some embodiments, the plurality of distal electrodes 27 acquire intracardiac ECG signals from the tissue of a heart chamber of the heart 24. The processor includes a signal acquisition circuit 36 coupled to receive the intracardiac signals from the interface 32, a memory 38 that stores data and / or instructions, and a processing unit 42 (e.g., a CPU or other processor).

[0048] The signal acquisition circuit 36 digitizes the intracardiac signals to produce a plurality of digital signals. The acquisition circuit then transmits the digitized signals to a processing unit 42 included in the processor 34.

[0049] Among other tasks, the processing unit 42 is configured to extract annotated parameters from the signals, compute statistical properties such as mean values for annotated parameters of possibly similar groups of adjacent signals (in this context, adjacent signals refer to signals from electrodes that are close to each other ("spatial correlation"), and / or annotated values extracted from cardiac cycles that are close in time ("temporal correlation")).

[0050] The processing unit is further configured to discard (i.e., omit) possibly invalid annotated values (such as annotations from electrodes with poor current connections or subject to strong temporal noise) from the group after computing the statistical properties. The remaining annotated values will be referred to hereinafter as "valid annotated values".

[0051] The processor 34 visualizes the valid annotated values to the user, i.e., the annotated values excluding the statistically deviant annotated values that have been omitted. In some embodiments, the processor 34 visualizes the valid annotated values, for example, by superimposing the valid annotated values on an electroanatomical map 50 of the heart and displaying the map on the screen 52 to the physician 22. Alternatively, the processor 34 can visualize the valid annotated values (after omitting the invalid annotated values) in any other suitable manner.

[0052] Figure 1 The exemplary illustration shown is chosen purely for conceptual clarity. In alternative embodiments of the present application, for example, position measurements can also be performed by applying voltage gradients between pairs of surface electrodes 28 and measuring the resulting electric potentials with the distal electrode 27 (i.e., using the CARTO® 4 technology produced by Biosense-Webster, Irvine, California). Thus, embodiments of the present application are applicable to any position sensing method. 4technology) produced by Biosense-Webster, Irvine, California. Thus, embodiments of the present application are applicable to any position sensing method.

[0053] Other types of catheters can be equivalently employed, such as catheters (produced by Biosense-Webster, Inc.), or basket catheters. The contact sensor can be fitted at the distal end of the electroanatomical catheter 23. Intracardiac electrophysiological signals can be acquired in a similar manner on the distal electrode 27 with other types of electrodes (such as electrodes for ablation).

[0054] Figure 1Only components relevant to embodiments of the present application are shown. Other system elements, such as external ECG recording electrodes and their connections, are omitted. Various ECG recording system elements, as well as elements for filtering, digitizing, protection, etc. of the electrical circuit, are omitted.

[0055] In alternative embodiments, a readout application specific integrated circuit (ASIC) is used to measure the intracardiac ECG signals. Various elements for routing the signal acquisition circuit 36 can be implemented in hardware, e.g., using one or more discrete components, such as a field programmable gate array (FPGA) or an ASIC. In some embodiments, some elements of the signal acquisition circuit 36 and / or the processing unit 42 can be implemented in software, or by using a combination of software elements and hardware elements.

[0056] The processing unit 42 generally comprises a general purpose processor having software programmed to perform the functions described herein. For example, the software can be downloaded to the network in electronic form, or alternatively or additionally, the software can be provided and / or stored on non-transitory tangible media, such as magnetic, optical or electronic memory.

[0057] Related Note Values

[0058] The correlation annotated values are derived from spatially correlated electrodes (e.g., electrodes that are geometrically close to each other, i.e., not more than a predefined distance apart from each other) and / or temporally correlated signals (e.g., values extracted from cardiac cycles that are close to each other, i.e., occurring not more than a predefined duration apart). More precisely, the correlation annotated values are annotated values for which the combined distance between the geometric distance between the electrodes and the temporal distance between the cardiac cycles is below some predefined threshold.

[0059] Figure 2 is a diagram 200 schematically showing the acquisition of signals over multiple cardiac cycles by multiple electrodes. The horizontal axis 202 shows the cardiac cycles (one cardiac cycle per vertical line), and the vertical axis 204 shows the distance of the electrodes from a reference point (only one spatial dimension is shown; it is understood, however, that two or three dimensions can be used in practice, but are not shown for clarity). According to Figure 2 In the exemplary embodiment shown, there is an electrode in all horizontal lines, and a LAT annotated value is registered for all intersections of horizontal and vertical lines (each intersection will be referred to hereinafter as a LAT point).

[0060] Curve 206 is an iso-LAT line showing the location of the indicated LAT values, and the electrodes can measure values interpolated from adjacent iso-LAT curves at the corresponding cardiac cycle. For example, the expected registration value for LAT point 208, which is vertically midway between iso-LAT line 714 and iso-LAT line 716, is 715, while the expected registration value for LAT point 210 is 708.5.

[0061] It can be seen that the LAT values of adjacent vertical lines and adjacent horizontal lines are similar. Circle 212 represents a set of related LAT values 214 that are close to each other in terms of geometric (vertical) distance and temporal (horizontal) distance.

[0062] Figure 2 The illustrated exemplary graphs are simplified and shown entirely for conceptual clarity. In alternative embodiments, for example, the distances between the electrodes are not uniform, and the set of related signals can not be circular.

[0063] Figure 3A is a flowchart 300 schematically illustrating a first method for enhancing the reliability of annotated values according to embodiments of the present application. The flow is performed by the processor 34 ( Figure 1 ). The flow starts with a step 302 of recording signals, in which the processor records ECG signals monitored by the electrodes 27 and acquired by the acquisition circuit 36 ( Figure 1 ). Next, at a step 304 of extracting annotated values, the processor calculates an annotated value for each electrode and for each cardiac cycle.

[0064] The processor then proceeds to a step 306 of obtaining electrode positions, in which the positions of the electrodes are acquired (e.g., using ACL technology) and the spatial positions of each electrode are registered, and then proceeds to a step 308 of selecting a set.

[0065] At step 308, the processor selects a set of related annotated values. As described above, the set includes annotated values that are likely similar according to spatial and / or temporal correlation of the signals.

[0066] Next, at a step 310 of calculating mean and SD, the processor calculates the mean and standard deviation of all the annotated values of the set. In the context of the present application, any suitable type of mean can be used, such as an arithmetic mean, a geometric mean, a median, a root mean square (RMS) value, a centroid, or any other mean.

[0067] The processor then repeats and sequentially performs steps 312, 314, and 316 or 318 for each annotation value in the set. In step 312 of calculating a standard score, the processor calculates a standard score for the annotation value (e.g., by dividing the difference between the annotation value and the mean by the standard deviation). In step 314 of comparing the standard score, the processor compares the standard score calculated in step 312 to a preset limit. In step 316 of discarding values, if the standard score exceeds the preset limit, the processor discards the statistically deviating annotation value; and in step 318 of adding values, if the standard score is within the preset limit, the processor adds the annotation value to a set of valid annotation values.

[0068] The processor repeats the sequence of steps 312, 314, and 316 or 318 for all annotation values of the set. The flowchart can then be repeated for other relevant electrode sets (starting from step 308).

[0069] When the flow ends, the set of valid annotation values replaces the original set with better reliability, as outliers (e.g., from electrodes with poor current connections) are omitted.

[0070] Figure 3B is a flowchart 350 schematically illustrating a second method for enhancing the reliability of annotation values according to an embodiment of the present application. Figure 3B The illustrated method differs from Figure 3A The illustrated method differs only in the statistical properties and selection of omitted values. Thus, apart from steps 310, 314 of Figure 3A , steps 360 and 364, which differ from and will be described below, Figure 3A Steps 302-318 of the illustrated method are identical to Figure 3B Steps 352-368 of the illustrated method, respectively.

[0071] In step 362 of calculating quartiles, the processing unit 42 Figure 1 ) calculates the first quartile (Q1) and the third quartile (Q3) of the set of LAT values (Q1 is defined as the median between the minimum and the median of the set of LAT values; Q3 is the median between the median and the maximum of the set of LAT values).

[0072] In step 364 of comparing the annotation value, the processing unit compares the annotated LAT value to Q1 and Q3. If the value is smaller than Q1 or higher than Q3, the processing unit will perform step 366 of discarding the annotation value, wherein if the value is between Q1 and Q3, step 368 of adding the value will be performed.

[0073] Figure 3A , Figure 3BThe exemplary flowchart shown is chosen purely for conceptual clarity. In alternative embodiments, for example, the annotation values can be extracted as the signals are acquired (rather than after the signals are recorded). In one embodiment, the selection of the set of signals can be done by a physician; in other embodiments, the processor will select the set according to a region and / or time range indicated by the physician.

[0074] In some embodiments, step 318( Figure 3B ) is not required - in step 316 (366), the processor will discard the extreme values from the set, and when the flow is completed, only the good values will be retained. In other embodiments, all of the annotation values are initially marked as invalid, and step 316 (366) is not required.

[0075] In some embodiments, other statistical properties are used than those described above; for example, in one embodiment, octiles can be used instead of quartiles, and the processing unit can omit values below the first octile or above the last octile. Further alternatively, any other suitable percentile can be used.

[0076] In alternative embodiments, any other suitable statistical method can be used to detect and omit extreme values.

[0077] Propagation Delay Compensation

[0078] In some embodiments, the techniques described above can be improved by correcting the extracted LAT values for expected variations in the values due to the different spatial locations of the electrodes, before the statistical property calculations. For example, it can be assumed that the wave travelling through the heart travels at a given speed (e.g. 1 m / s). By using the known locations of the electrodes that acquired the signals, the theoretical difference in the LAT can be applied when calculating the mean value.

[0079] Figure 4 is a flowchart 400 schematically showing an improved method for enhancing the reliability of the annotation values, according to an embodiment of the application. The flow is performed by the processor 34 Figure 1 ). The flow starts with a step 402 of recording the signals, followed by a step 404 of calculating the annotation values, a step 406 of obtaining the electrode locations and a step 408 of selecting the set, which can be identical to steps 302, 304, 306 and 308 (Fig. 3), respectively.

[0080] Next, the processor performs a step 410 of correcting the LAT values, in which, for each LAT value of the set, the processor calculates and applies an estimated correction according to the spatial location of the electrode and the assumed wave travel speed. After step 410, the flow returns to Fig. 3 at step 310 of calculating the mean and SD.

[0081] Thus, the estimate of bias caused by signal propagation delay can be removed from the group, further enhancing the reliability of the annotated signal.

[0082] Figure 4 The exemplary flowchart shown in FIG. 4 is chosen for conceptual clarity. In alternative embodiments, for example, the correction for expected signal delay can be incorporated into the steps of calculating the mean and SD. In other embodiments, the correction is done prior to selecting the group (and thus, the group can include a larger number of relevant LAT values).

[0083] Error Propagation and Its Prevention

[0084] In some embodiments according to the present application, the processor 34 continuously selects a group of relevant LAT values in response to receiving a signal from a subsequent heartbeat. In some cases, the processor can select groups of values corresponding to signals that overlap in time. For example, if a first group includes LAT values extracted from signals corresponding to cardiac cycles n through m, and a second group includes values corresponding to cycles x through y, these groups partially overlap if x > m and y > n.

[0085] In the foregoing description, reference was made to Figures 5A-5C , Figures 6A-6E and Figure 7 to LAT values annotated according to different cardiac cycles; in fact, each of the LAT values can include a plurality of spatially relevant LAT values (including LAT values that have been corrected for signal propagation delay). For simplicity, however, we will show a single LAT value for each cardiac cycle.

[0086] Figures 5A-5C is a timing diagram depicting events along a horizontal time axis (not shown). Figure 5A is a diagram 500 schematically showing the annotation of a series of LAT values in an electroanatomical system according to an embodiment of the present application. In the exemplary embodiment shown in Figure 5A each group includes ten LAT values corresponding to ten consecutive cardiac cycles; each pair of consecutive groups shares five common LAT values.

[0087] For clarity, in the example shown in Figure 5A the annotated LAT values can be numerically close to "X" or close to "Y" (to be referred to simply as X values and Y values). It should be understood that this simplification in no way limits the scope of the present disclosure; embodiments of the present application can annotate any suitable collection of LAT values.

[0088] In Figure 5AIn the example shown, the first five LAT values are "X", but when the subsequent intracardiac signals are received, the majority of the LAT values are "Y", which means that the initial annotations and decisions made for the first group can be in error. As will be described below (with reference to Figure 5B ), in some embodiments according to the present application, no corrective action is taken, and the error decisions made for the first group can propagate to additional groups. As will be further described (with reference to Figure 5C ), in other embodiments, corrective action is taken, and error propagation is prevented.

[0089] Figure 5B is a graph 510 that schematically illustrates the statistical property calculations performed on the LAT values received by the processor (in Figure 5A ) when no corrective action is taken, according to embodiments of the present application. Graph 510 includes group 1 averaging scheme 512, group 2 averaging scheme 514, and group 3 averaging scheme 516 (as will be shown, group 4 averaging is the same as group 2 and group 3 and is not shown).

[0090] Group 1 averaging scheme 512 represents the averaging of the first ten LAT values. There are seven X values and three Y values, and therefore, the average will be close to X. Thus, the Y values of the last five samples will be omitted from the subsequent groups.

[0091] Group 2 averaging scheme 514, which represents the averaging of LAT values 6-15, includes four X values and three Y values (two of the original Y values were omitted). The average will again be close to X. Thus, the Y values of the last five samples will be omitted from the subsequent groups. As can be seen, the error from the first group has propagated to the second group.

[0092] Group 3 averaging scheme 516, which represents the averaging of LAT values 11-20, also includes four X values and three Y values, as two of the original Y values were omitted, and again, the average is close to X and the Y values will be omitted. Thus, the error from the first group propagates to this group and will continue to additional groups. (According to Figure 5A , Figure 5B the example shown, the propagation will stop only if the majority (e.g., 80%) of the subsequent LAT values are Y.)

[0093] Figure 5C is a graph 520 that schematically illustrates the avoidance of error propagation by reapplying the omitted values, according to embodiments of the present application. Graph 520 includes group 1 averaging scheme 522, group 2 preliminary averaging scheme 524, group 2 reassessed averaging scheme 526, and group 3 averaging scheme 528.

[0094] Group 1 averaging scheme 522 represents the averaging of the first ten LAT values, and is the same as group 1 averaging scheme 512 ( Figure 5B) is the same. Again, the average is close to X, and the Y value will be omitted from the next set.

[0095] The Group 2 preliminary averaging scheme 524 is the same as the Group 2 averaging scheme 514 Figure 5B ) and again, the average is close to X, and the Y value will be omitted from the next set. However, according to the exemplary embodiment shown in Figure 5C the frequency at which LAT values are omitted is monitored. Since now there are six omitted values from a set of ten samples, the processor decides to reapply the omitted values in the statistical property calculation. The decision to reapply the omitted values (e.g., Group 2 in Figure 5C ) is indicated by reference numeral 524) is a "corrective action" to be taken.

[0096] The Group 2 recalculation averaging scheme 526 includes the omitted Y values in the statistical calculation. There will now be six Y values and four X values in Group 2, the average will be close to Y, and the X values will be omitted from the next set. The Group 3 averaging scheme 528 will now include two X values and six Y values; the average will be close to Y and the X values will be omitted. This pattern will continue through Group 4.

[0097] Thus, according to the exemplary embodiment shown in Figure 5C the processor can stop the propagation of errors that can occur due to omitted values by monitoring the frequency at which values are omitted, and can take a corrective action that can include recalculating the average of values without omission in response to the frequency at which values are omitted.

[0098] Other embodiments according to the present application can prevent the propagation of errors by reannotating at least some of the LAT values in response to the frequency at which values are omitted. This can be effective when the intracardiac signal is ambiguous and can be interpreted in more than one way.

[0099] Figures 6A-6E is a timing diagram that depicts events along a horizontal time axis (not shown).

[0100] Figure 6A is a graph that schematically shows the probability of a LAT value of a series of signals acquired by a set of spatially dependent electrodes in an electroanatomical system, according to an embodiment of the present application. Figure 6A Each rectangle in represents the probability that the LAT value of the corresponding signal is either X (the upper limit value) or Y (the lower limit value). The first five signals represent a 60% probability of an X LAT value and a 40% probability of a Y value. In most cases, the later signals (associated with later cardiac cycles) have a higher probability of Y than X. The signals are divided into partially overlapping groups (1-3) with ten samples in each group and a five-sample overlap between subsequent groups.

[0101] Figure 6B is a diagram schematically illustrating the annotation of LAT values for a group 1 signal according to an embodiment of the application. The processor annotates the X values for six samples and the Y values for four samples. The average is close to X, and the processor omits four Y values (three of the omitted signals indicated by the dashed rectangles overlap with group 2).

[0102] Figure 6C is a diagram schematically illustrating the annotation of LAT values for a group 2 signal according to an embodiment of the application. From the five signals shared with group 1, the processor has omitted three signals with higher y probability. Four of the remaining seven values are X values, and three are Y values. The average is again close to X, and the processor can omit Y values from the next group. However, the processor now determines that the number of omitted values is large (e.g., exceeds a preset limit), and attempts to re-annotate the LAT values to reduce the number of omitted values.

[0103] Figure 6D is a diagram schematically illustrating the re-annotation of the signals of group 1 according to an embodiment of the application. The processor now annotates the second most likely values (Y, Y, Y, X, Y for signals 1-5, respectively) for the first five samples. The LAT values that are now annotated for the first group will include three X values and seven Y values, and the average will be close to Y. The seventh and ninth values will now be omitted from the next group.

[0104] Figure 6E is a diagram schematically illustrating the re-annotation of the group 2 values after re-annotation of group 1 according to an embodiment of the application. Since two of the five overlapping signals from group 1 were not omitted, group 2 includes six Y values and two X values. The average is now close to Y, and error propagation is prevented.

[0105] Referring to 5A-5D Figure 5C and Figures 6A-6E The error propagation prevention method described is an example method referenced for conceptual clarity. Alternative methods can be employed to prevent error propagation according to embodiments of the application. For example, the length and overlap of the groups can be any suitable number, including group sizes and / or overlaps that vary over time. In some embodiments, the threshold can be dynamic. In one embodiment, the processor can not annotate the second most likely LAT value if the probability of the second most likely LAT value is less than a preset limit, or alternatively less than a limit that is a function of the frequency of LAT values being omitted.

[0106] Finally, in some embodiments, a combination of recovering omitted values and re-evaluating annotations can be employed.

[0107] Figure 7is a flowchart 700 schematically illustrating a method for avoiding error propagation by re-annotation according to an embodiment of the application. The flowchart is executed by the processor 34 Figure 1

[0108] The flow begins with a receive overlapping annotations step 702 in which the processor retrieves from memory annotations of the intracardiac ECG signal common to the previous and current groups (as described above, the processor can omit values that deviate substantially from the mean of the previous group).

[0109] Next, in a receive new ECG signal step 704, the processor receives from the electrodes (via the acquisition circuit 36 Figure 1 ) an ECG signal relating to a subsequent cardiac cycle and annotates the corresponding LAT value. The processor then proceeds to a compute statistical values step 706 and computes statistical properties (e.g., mean and standard deviation) of a group of values, including the overlapping values from the previous group and the newly annotated value.

[0110] The processor then proceeds to a scan group step 708 in which the processor compares each of the LAT values of the group to the group mean (computed in step 706) and omits values that deviate from the mean by more than a preset threshold (or alternatively, omits values if the z-score is not within a preset limit). The processor also counts the number of omitted LAT values.

[0111] Next, the processor proceeds to a compare counter step 710 in which the processor compares the number of omitted values to a preset limit. If the number of omitted values does not exceed the limit, the processor proceeds to the next group (e.g., returns to step 702).

[0112] If the number of omitted values has exceeded the preset limit in step 710, the processor assumes that error propagation has occurred due to a false annotation of a previous LAT value. The processor then proceeds to a receive overlapping signal step 712 and retrieves from memory the overlapping signal. Next, the processor proceeds to an extract alternative values step 714 and re-annotates the LAT values from the signal. Since the processor has already annotated the LAT values (in step 704 of the previous group), the processor now looks for alternative annotations (e.g., the second most likely LAT value for each signal).

[0113] After step 714, the processor re-proceeds to step 706 and re-computes the statistical values with the alternative annotation values. If the number of omitted values is now less than the previous count of step 708, the new values will take effect.

[0114] In summary, according to Figure 7 ​The exemplary embodiment shown annotates the LAT values of the relevant intracardiac signals of the consecutive groups. If an error occurs in the first group, the processor can detect that values in the subsequent groups are frequently omitted. The processor then attempts an alternative annotation of the signals and selects the LAT value corresponding to this alternative annotation if this results in a smaller number of omitted annotations.

[0115] Figure 7 The exemplary flowchart shown in FIG. 7 is chosen for conceptual clarity. In alternative embodiments, for example, the processor always extracts a first set of the most likely annotated values and a second set of alternative annotated values, and thus step 714 is replaced by an alternative values step. In some embodiments, more than two possible LAT values are extracted, and the flowchart is modified accordingly. In one embodiment, step 720, in which the number of omitted values is compared to a threshold, can be replaced by other measures that compare frequencies rather than counts. In embodiments, instead of extracting alternative annotations, the processor recomputes the group statistics with the omitted values. In still other embodiments, the processor attempts both statistical value recomputation and alternative annotated values.

[0116] It is to be understood that the embodiments described above are cited by way of example, and that the present application is not limited to what has been particularly shown and described hereinabove. Rather, the scope of the present application includes both combinations and sub-combinations of the various features described above, as well as variations and modifications thereof, which would occur to persons of ordinary skill in the art upon reading the foregoing description and which are not disclosed in the prior art. Documents incorporated by reference in the present patent application are to be considered an integral part of the application except that to the extent any terms are defined in such incorporated documents in a manner that conflicts with the definitions made explicit herein, the definitions made explicit in this application shall prevail. In closing, the claims are what is regarded as the disclosure that defines the application.

Claims

1. A system comprising: a signal acquisition circuit configured to receive a plurality of intracardiac signals acquired by a plurality of electrodes of an intracardiac probe in a heart of a patient; and a processor configured to: perform a sequence of annotation visualization operations at subsequent times by performing the following steps in each operation: (i) extract a plurality of annotation values from the intracardiac signals; (ii) select a group of the intracardiac signals; (iii) identify one or more annotation values in the group that are statistically deviant beyond a predefined measure of deviance; and (iv) visualize the annotation values to a user while omitting and refraining from visualizing the statistically deviant annotation values; estimate an omission rate of annotation values over one or more of the annotation visualization operations; and in response to detecting that the omission rate exceeds a predefined threshold, take a corrective action comprising one of: reapply one or more of the omitted annotation values and re-identify the statistically deviant annotation values; or re-extract one or more of the annotation values from the intracardiac signals.

2. The system of claim 1, wherein the processor is configured to define the measure of deviance in terms of a z-score of the annotation values.

3. The system of claim 1, wherein the processor is configured to define the measure of deviance in terms of one or more percentiles of the annotation values. In a given annotation visualization operation, the processor is configured to compute a deviance of the annotation values over intracardiac signals acquired by a selected subset of spatially correlated electrodes in the heart that are no more than a predefined distance apart from each other. In computing the deviance of the annotation values of a given annotation visualization operation, the processor is configured to average the intracardiac signals over a plurality of time- dependent cardiac cycles occurring within a predefined time period.

4. The system of claim 1, wherein, In a given annotation visualization operation, the processor is configured to correct one or more of the annotation values in a given intracardiac signal acquired by a given electrode in the group to compensate for a displacement of the given electrode relative to other electrodes in the group.

5. The system of claim 1, wherein, 7. The system of claim 1, wherein the annotation values comprise local activation times (LATs).

6. The system of claim 1, wherein, 8. The system of claim 1, wherein the processor is configured to visualize the annotation values by superimposing the annotation values, except for the statistically deviant annotation values, on a model of the heart.

9. A method comprising: receiving a plurality of intracardiac signals acquired by a plurality of electrodes of an intracardiac probe in a heart of a patient; performing a sequence of annotation visualization operations at subsequent times by performing the following steps in each operation: (i) extracting a plurality of annotation values from the intracardiac signals; (ii) selecting a group of the intracardiac signals; (iii) identifying one or more annotation values in the group that are statistically deviant beyond a predefined measure of deviance; and (iv) visualizing the annotation values to a user while omitting and refraining from visualizing the statistically deviant annotation values; ​ ​ (iv) visualizing the annotation values to the user while omitting and avoiding visualization of the statistically deviating annotation values; estimating an omission rate of annotation values on one or more of the annotation visualization operations; and taking a corrective action in response to detecting that the omission rate exceeds a predefined threshold, the corrective action comprising one of: reapplying one or more of the omitted annotation values and re-identifying the statistically deviating annotation values; or re-extracting one or more of the annotation values from the intracardiac signals.

10. The method of claim 9, wherein the measure of deviation is defined in terms of a z-score of the annotation values.

11. The method of claim 9, wherein the measure of deviation is defined in terms of one or more percentiles of the annotation values.

12. The method of claim 9, and comprising, in a given annotation visualization operation, computing a deviation of the annotation values on intracardiac signals acquired by a selected subset of spatially correlated electrodes in the heart that are no more than a predefined distance apart from each other.

13. The method of claim 9, and comprising, in computing the deviation of the annotation values for a given annotation visualization operation, averaging the intracardiac signals over a plurality of time-dependent cardiac cycles occurring within a predefined time period.

14. The method of claim 9, and comprising, in a given annotation visualization operation, correcting one or more of the annotation values in a given intracardiac signal acquired by a given electrode in the set to compensate for a displacement of the given electrode relative to other electrodes in the set.

15. The method of claim 9, wherein the annotation values comprise local activation times (LATs).

16. The method of claim 9, wherein visualizing the annotation values comprises superimposing the annotation values, except for the statistically deviating annotation values, on a model of the heart.

Images