Spatial correlation for identifying pattern-matched ventricular locations
Through processor analysis and machine learning models, the problem of difficult to determine the direction of catheter movement in cardiac electrical propagation evaluation is solved, the recognition and ablation efficiency of arrhythmia positions is improved, and visual auxiliary tools are provided.
Patent Information
- Application Number
- CN202410582245.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-05-11
- Filing Date
- 2024-05-11
- Publication Date
- 2024-11-12
AI Technical Summary
When evaluating electrical propagation of the heart, it is difficult for physicians to find ventricular regions that provide good correlation between sampling data and storage mode, especially when a few channels fail to be associated with storage mode, resulting in difficulty determining the direction of catheter movement to improve all channel dependency.
Analyzing large amounts of electrophysiological data through a processor, applying statistical analysis methods such as spindle analysis and regression methods, or using machine learning models of loss functions, instructing physicians to move catheters towards ventricular areas with better correlations, and construct correlation mappings Ablation with auxiliary guidance.
It improves the probability that physicians find correlation with storage mode when catheters move in the heart cavity, enhances the recognition and ablation efficiency of arrhythmia locations, and provides visual aid tools to help physicians determine ablation targets.
Smart Images

Figure CN118924299A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates generally to electrophysiological (EP) signals, and in particular to the assessment of electrical propagation in the heart. Background Art
[0002] Electrophysiological signal estimation for determining the location of ventricular arrhythmias has been proposed previously in the patent literature. For example, U.S. Patent 7,907,994 describes how a ventricular tachycardia (VT) signal is induced in a living subject. Pacing mapping signals are then obtained from multiple points within the ventricle and automatically compared numerically to the induced signal. Identifying a high cross-correlation between the induced signal and one or more of the pacing mapping signals identifies an arrhythmogenic focus or pathway, which can then be ablated so that the arrhythmia becomes non-inducible.
[0003] The present disclosure will be more fully understood through the following detailed description of examples of the present disclosure in conjunction with the accompanying drawings, in which: BRIEF DESCRIPTION OF THE DRAWINGS
[0004] Figure 1 is a schematic illustration of a catheter-based electrophysiological (EP) mapping and ablation system according to one example of the present disclosure;
[0005] Figure 2 A method of guiding a physician to a candidate left ventricular location of an arrhythmia according to an example of the present disclosure is schematically illustrated;
[0006] Figure 3 The following schematically illustrates an example of an operation according to the present disclosure. Figure 2 The guidance method is to establish a correlation map of the left ventricle; and
[0007] Figure 4 is a schematic description of an example according to the present disclosure Figure 2 Flowchart of the boot method. DETAILED DESCRIPTION
[0008] Overview
[0009] To characterize arrhythmias (eg, reentrant arrhythmias) in a cardiac chamber such as the left ventricle (LV), physicians may use a multi-electrode mapping catheter to perform electrophysiological (EP) mapping of suspected tissue pathways and circuits within the chamber.
[0010] During an EP procedure, if a transient proarrhythmic event occurs, such as an ectopic beat, premature beat, or premature ventricular contraction (PVC), the physician can record the event and then attempt to stimulate repetition of the event through pacing, thereby enabling evaluation of the origin or cause of the proarrhythmic event.
[0011] To stimulate repetition of this event through pacing, the physician captures pacing mapping signals from multiple points within the ventricle and compares the recorded signals. Identification of a high degree of cross-correlation between the recorded signal and one or more of these pacing mapping signals indicates successful stimulation of the same stored transient event. During pacing to identify the mechanism of a complex ventricular arrhythmia, the physician samples different areas of the ventricle and performs pattern matching (e.g., electrogram waveform matching between the acquired waveform and a stored VT characteristic waveform pattern) to identify correlations with a defined pattern. In one example, pattern matching is performed on an electrode array to identify the location of the VT mechanism.
[0012] However, sometimes, it is difficult for the physician to find a ventricular region that provides a good correlation between the sampled data and the stored pattern. For example, the waveform acquired from one or more electrodes may not match the pattern. Sometimes, the waveforms from most channels correlate well with the stored pattern, while only one of a few channels fails to correlate with the stored pattern. When this happens, it may be particularly difficult for the physician to determine in what direction and / or orientation to move the catheter to improve the correlation of all channels.
[0013] The examples of the present disclosure described below provide techniques for guiding the physician to the ventricular region that provides good correlation between the sampled data and the stored patterns. The disclosed techniques rely on the benefit of a processor analyzing a large amount of EP data that is accumulated as the physician moves the catheter inside the heart chamber and paces at different locations therein.
[0014] In one example, the processor applies a statistical analysis associated with the spatial location of each of the captured EP heart signals to indicate a direction to the ventricular region toward which the physician may move the catheter. Movement in this direction is expected to provide a signal with better correlation to the stored pattern. Examples of statistical analysis methods include principal axis analysis and regression methods. Another statistical analysis used is to examine the correlation of each channel at each point and then find the maximum value.
[0015] In another example, the processor applies a machine learning model (ML) including a loss function that indicates a direction to a region of the ventricle toward which a physician may move the catheter.
[0016] In one example, the loss function includes a measure of possible directions for: (i) LV anatomical targets known to be potentially arrhythmic, (ii) analysis regions of ventricular regions having corresponding average correlations, and (iii) analysis regions exhibiting signals with the lowest correlations. At any given location on the ventricle considered by the physician, a processor applying the ML model finds a direction that minimizes the loss function.
[0017] The analysis zone may be the area covered by the multi-electrode catheter at any given location during the electrogram acquisition step. Alternatively, the zone may be defined by an area of a given size and shape suitable for analysis (e.g., a circle of a predefined radius including a sufficient number of correlation data points).
[0018] In an embodiment of the present disclosure, a physician paces cardiac tissue at different locations, and for each paced location, a processor compares the EP signal invoked in response to pacing with the stored signals. The processor searches for the tissue location with the best (e.g., highest average) correlation. Optionally, the correlation of the signals from each channel is required to be above a predefined threshold to meet the criterion. Optionally, the best correlation is the location where the greatest number of channels meet the criterion of having a correlation with its corresponding stored signal above a predefined threshold.
[0019] The output of the algorithm is an arrow pointing the physician in the direction in which the next pacing should be performed. Once the physician follows the direction to the region, a new EP signal is acquired therein, and the processor calculates the correlation at the new location. The processor can check whether the new location correlates better with the stored pattern, for example by achieving an average correlation value above a given threshold and / or achieving a minimum correlation value above a given threshold. If the region is considered to be arrhythmogenic, the physician can ablate the tissue therein to eliminate the arrhythmia (e.g., VT).
[0020] In some examples, after the ML model guides the physician, the processor constructs a visual teaching material in the form of a map of correlation coefficients to visualize the path constructed by the ML model and to visualize the accumulated progress of correlation information. The map acts as a visual aid to guide the physician whether and where to perform ablation.
[0021] To construct the map, the processor stores the correlation coefficients at each of the plurality of different locations and for each of the channels, and generates a spatial map of the calculated correlation coefficients. The correlation coefficient may be the average of all correlation coefficients from different electrodes (weighted or unweighted) within the analysis region, or may be the correlation coefficient from one or more selected electrodes within the analysis region.
[0022] In other examples, instead of an ML model, the processor applies a statistical model (eg, principal component analysis (PCA)) to find the direction.
[0023] System Description
[0024] Figure 1 is a schematic illustration of a catheter-based electrophysiological (EP) mapping and ablation system 10 according to one example of the present disclosure.
[0025] The system 10 includes a plurality of catheters that are inserted by a physician 24 through the patient's vascular system percutaneously into a chamber or vascular structure of the heart 12 (see illustration 45). Typically, a delivery sheath catheter is inserted into a cardiac chamber (such as the left atrium or right atrium) near a desired location in the heart 12. Multiple catheters may then be inserted into the delivery sheath catheter to reach the desired location. The multiple catheters may include a catheter dedicated to pacing, a catheter for sensing intracardiac electrogram signals, a catheter dedicated to ablation, and / or a catheter dedicated to both EP mapping and ablation. The example catheter 14 shown herein is configured to sense a bipolar electrogram. The physician 24 brings the distal end 28 of the catheter 14 (hereinafter also referred to as the distal end assembly 28) into contact with the heart wall for sensing a target site in the heart 12. For ablation, the physician 24 similarly brings the distal end of the ablation catheter to the target site.
[0026] As shown in inset 65, the catheter 14 is an exemplary catheter including a basket-shaped distal end 28, the one and preferably multiple electrodes 26 are optionally distributed on multiple splines 22 at the distal end 28 and are configured to sense IEGM signals. The catheter 14 may additionally include a position sensor 29 embedded in or near the distal end 28 on the shaft 46 of the catheter 14 for tracking the position and orientation of its distal end 28. Optionally and preferably, the position sensor 29 is a magnetic-based position sensor that includes three magnetic coils for sensing three-dimensional (3D) position and orientation. As shown, the distal end 28 also includes an expansion / collapse rod 42 of the expandable component 28, which is mechanically connected to the basket component 28 at the distal edge 41 of the component 28.
[0027] The magnetic-based position sensor 29 can operate with the positioning pad 25, which includes a plurality of magnetic coils 32 configured to generate a magnetic field in a predefined working space. The real-time position of the distal tip 28 of the catheter 14 can be tracked based on the magnetic field generated by the positioning pad 25 and sensed by the magnetic-based position sensor 29. Details of magnetic-based position sensing technology are described in U.S. Patents 5,5391,199, 5,443,489, 5,558,091, 6,172,499, 6,239,724, 6,332,089, 6,484,118, 6,618,612, 6,690,963, 6,788,967 and 6,892,091.
[0028] The system 10 includes one or more electrode patches 38 positioned in contact with the skin of the patient 23 to establish a position reference for impedance-based tracking of the positioning pad 25 and the electrodes 26. For impedance-based tracking, current is directed toward the electrodes 26 and sensed at the electrode skin patches 38 so that the position of each electrode can be triangulated via the electrode patches 38. Details of impedance-based position tracking techniques are described in U.S. Patents 7,536,218, 7,756,576, 7,848,787, 7,869,865, and 8,456,182.
[0029] Recorder 11 displays cardiac signals 21 (e.g., electrograms and corresponding cardiac tissue locations) acquired using surface ECG electrodes 18 and intracardiac electrograms acquired using electrodes 26 of catheter 14. Recorder 11 may include pacing capabilities for pacing cardiac rhythms and / or may be electrically connected to a separate pacemaker.
[0030] The system 10 may include an ablation energy generator 50 adapted to conduct ablation energy to one or more electrodes at the distal end of a catheter configured for ablation. The energy generated by the ablation energy generator 50 may include, but is not limited to, radio frequency (RF) energy or pulsed field ablation (PFA) energy (including monopolar or bipolar high voltage DC pulses to be used to achieve irreversible electroporation (IRE)), or a combination thereof.
[0031] The patient interface unit (PIU) 30 is an interface configured to establish electrical communication between the catheter, electrophysiology equipment, power supply, and workstation 55 to control the operation of the system 10 and receive EP signals from the catheter. The electrophysiology equipment of the system 10 may include, for example, multiple catheters, positioning pads 25, surface ECG electrodes 18, electrode patches 38, ablation energy generator 50, and recorder 11. Optionally and preferably, the PIU 30 additionally includes processing capabilities for achieving real-time calculation of catheter position and for performing ECG calculations.
[0032] The workstation 55 includes a memory 57, a processor 56 unit with a memory or storage device loaded with appropriate operating software, and user interface capabilities. The workstation 55 can provide multiple functions, optionally including: (i) three-dimensional (3D) modeling of the endocardial anatomy and rendering the model or anatomical map 20 for display on the display device 27; (ii) displaying on the display device 27 an activation sequence (or other data) compiled from the recorded cardiac signals 21 as representative visual markers or images superimposed on the rendered anatomical map 20; (iii) displaying the real-time position and orientation of multiple catheters within the heart chamber; and (iv) displaying on the display device 27 a site of interest (such as where ablation energy has been applied). A commercial product embodying elements of the system 10 can be CARTO TM 3 System was purchased from Biosense Webster, Inc., 31A Technology Drive, Irvine, CA 92618.
[0033] In some examples, processor 56 typically includes a general purpose computer that is programmed in software to perform the functions described herein. The software may be downloaded to the computer in electronic form over a network, for example, or it may alternatively or additionally be provided and / or stored on a non-transitory tangible medium such as magnetic, optical, or electronic memory.
[0034] This configuration of system 10 is shown by way of example in order to illustrate certain problems solved by the examples of the present disclosure and to demonstrate the application of these examples in enhancing the performance of such systems. However, the examples of the present disclosure are by no means limited to this particular class of exemplary systems, and the principles described herein may be similarly applied to other classes of medical systems. For example, other multipolar catheter types, such as OCTARAY TM catheter or basket catheter.
[0035] Identify the location of the ventricle exhibiting arrhythmia
[0036] Figure 2 A method of guiding a physician to a left ventricular location of an arrhythmia according to one example of the present disclosure is schematically illustrated. At any given location on the left ventricle considered by the physician, a processor applying a statistical analysis method or ML model provides the physician with a direction d towards a ventricular region having a better correlation between the sampled data and the stored pattern.
[0037] exist Figure 2, given a position on a guide path 210 on the anatomical surface 202 of the LV (e.g., a position that has been selected within an analysis region 212), directions d that are toward the region of interest (i.e., exhibiting increased correlation) are considered by deciding between candidate analysis regions 214. In one example, sufficient correlation means that each of the channels of a mapping catheter (e.g., catheter 14) has a correlation above a defined threshold. In this case, if the correlation of even one of these channels is too low, the physician needs to move the catheter to improve this. In another example, it is sufficient that the average correlation over the channels at a position is above a defined threshold.
[0038] As described above, a machine learning model (ML) may be used to make the determination. The ML model includes a loss function f that includes metrics about: (i) the direction v toward an anatomical target known to exhibit arrhythmias (also referred to as a known anatomical location); ai , (ii) pointing to the direction v of the region with the highest average correlation among the candidate analysis regions 214 bj , and (iii) a direction v pointing to the position in the analysis region that exhibits the lowest correlation within the candidate analysis region 214 cj , where the correlation at region 214 is below a given threshold so as to be disadvantageously considered in the loss function. An example of such a loss function is,
[0039] f(i;d)=w 1i (1-|d·v ai |)+∑ j [w2|d·v bj |+w3(1-|d·v cj |)]
[0040] where i is an index operating on suspicious anatomical regions within the LV (such as the LV apex 222, LV appendages, etc.), and j operates on the analysis region 214. The weight w k are the parameters of the ML model.
[0041] At any given current position on the path 210, the next direction d toward the region of interest is determined as the direction that minimizes the non-negative loss function f(d). If the physician guided to the analysis region 212 determines that the position inside the analysis region is the ablation target, the above-mentioned processor guidance ends. As described below, visual teaching can be provided in the form of a correlation map 200.
[0042] Guided Assisted Mapping Tools
[0043] Figure 3 The following schematically illustrates an example of an operation according to the present disclosure. Figure 2The process of guiding the physician using the ML model is to construct a correlation map 220 of the left ventricle. In the example shown, as the ML model guides the physician along the path 210 on the LV anatomical surface 202, the processor constructs a map 200 of correlation coefficients to visualize the path 210 constructed by the ML model and to visualize the progression in the correlation data in the form of regions 203-206 described by scales 218. The map acts as a visual aid to guide the physician toward a region of interest 208 (also referred to in the present disclosure as region of interest 208).
[0044] Methods to guide the physician to the ventricular location of the arrhythmia
[0045] Figure 4 is a schematic description of an example according to the present disclosure Figure 2 According to the example given, the algorithm performs a process that begins with the physician 24 acquiring signals from the left ventricle region using the multi-electrode catheter 14 in a signal acquisition step 402 .
[0046] In an analysis region generating step 404, the processor 56 generates a candidate analysis region 214 based on the current position of the catheter.
[0047] In the correlation analysis step 406, the processor 56 Figure 2 The ML model of is applied to each of the analysis regions 214 to calculate the next direction d on the path 210 towards the region deemed to be of interest by exhibiting the largest increase in correlation.
[0048] Next, in the next signal acquisition step 408, the physician moves the catheter to a new position along the calculated direction.
[0049] In a correlation calculation step 410, processor 56 calculates the correlation at the new position for each of the channels.
[0050] In a check for region of interest step 412, processor 56 checks whether the new location directed by the next direction d on path 210 is an arrhythmogenic region of interest. This may be accomplished, for example, by achieving an average correlation value above a given threshold and / or achieving a minimum correlation value above a given threshold.
[0051] If the answer is yes, the process ends at a marking step 414, where the processor 56 marks the current location as a region of interest. If the answer is no, the process returns to step 402.
[0052] Figure 4The example flow chart shown in is selected purely for conceptual clarity. For example, other analysis methods, such as PCA, may be used. Additional steps may be performed, such as receiving an indication of the degree of physical contact of the electrode with the diagnostic tissue from a contact force sensor. This step and other possible steps are intentionally omitted from the disclosure herein in order to provide a more simplified flow chart.
[0053] Example
[0054] Example 1
[0055] A method comprising: receiving cardiac signals (21) from a plurality of locations within a ventricle of a patient's heart (12). Comparing the received signals to a reference signal indicative of a cardiac arrhythmia. Based on the comparison, calculating a direction toward a location (208, 222) that exhibits an increased correlation between the received signals and the reference signal. Indicating the direction to a user.
[0056] Example 2
[0057] A method according to embodiment 1, wherein comparing includes applying a statistical analysis method to estimate the direction.
[0058] Example 3
[0059] A method according to embodiment 1, wherein comparing includes applying a machine learning (ML) model, and the machine learning model includes a loss function for the direction.
[0060] Example 4
[0061] According to the method described in any one of Examples 1 and 3, the method further includes including a directional loss term regarding a direction toward a known anatomical location (222) in the loss function.
[0062] Example 5
[0063] A method according to any one of embodiments 1, 3 and 4, wherein calculating the direction includes finding a direction towards a candidate analysis area (214) among directions, the direction minimizing the loss function.
[0064] Example 6
[0065] According to the method described in any one of Examples 1 to 5, the method further includes checking whether the position guided by the direction is an arrhythmogenic region of interest (208), and if so, marking the position, and if not, receiving cardiac signals from additional multiple positions and indicating the next direction to the user based on an analysis of the additional cardiac signals.
[0066] Example 7
[0067] The method according to claim 6, wherein checking whether the location is an area of interest comprises checking whether an average value of correlations within an analysis area (214) including the location is above a given threshold and / or whether a value of a minimum correlation in the analysis area (214) is above a given threshold.
[0068] Example 8
[0069] The method of claim 6, further comprising constructing a map (220) of the correlation to visualize the directions and the path (210) generated by the following directions and to visualize the progression in the correlation data.
[0070] Example 9
[0071] The method of claim 1, wherein receiving the cardiac signal comprises receiving one of a unipolar electrogram and a bipolar electrogram acquired using a catheter (14).
[0072] Example 10
[0073] A system (10) includes an interface (30) and a processor (56). The interface (30) is configured to receive cardiac signals from a plurality of locations within a ventricle of a patient's heart. The processor (560) is configured to (i) compare the received signals with a reference signal indicative of an arrhythmia, (ii) based on the comparison, calculate a direction toward a location (208, 222) that exhibits an increased correlation between the received signals and the reference signal, and (iii) indicate the direction to a user.
[0074] It should be understood that the above embodiments are cited by way of example, and the present disclosure is not limited to what is specifically shown and described above. On the contrary, the scope of the present disclosure includes combinations and sub-combinations of the aforementioned various features and their variations and modifications, which should be expected by those skilled in the art when reading the above description and which are not disclosed in the prior art.
Claims
1. A method, comprising: receiving cardiac signals from a plurality of locations within a ventricle of a heart of the patient; comparing the received signal to a reference signal indicative of arrhythmia; calculating, based on the comparison, a direction towards a location exhibiting an increased correlation between the received signal and the reference signal; as well as The direction is indicated to the user.
2. The method according to claim 1, wherein: The comparison includes applying statistical analysis methods to estimate the directions.
3. The method according to claim 1, wherein: The comparing includes applying a machine learning (ML) model that includes a loss function for the direction.
4. The method according to claim 3 further includes including a directional loss term regarding a direction toward a known anatomical location in the loss function.
5. The method according to claim 3, wherein: Calculating the direction includes finding a direction among directions towards a candidate analysis region that minimizes the loss function.
6. The method according to claim 1 further comprises checking whether the position guided by the direction is an area of interest as an arrhythmogenic region, and if so, marking the position, and if not, receiving cardiac signals from additional multiple positions and indicating the next direction to the user based on an analysis of the additional cardiac signals.
7. The method according to claim 6, wherein: Checking whether the location is a region of interest comprises checking whether an average value of correlations within an analysis area including the location is above a given threshold and / or whether a value of a minimum correlation in the analysis area is above a given threshold.
8. The method of claim 6, further comprising constructing a map of the correlation to visualize the directions and paths generated by the underlying directions and to visualize progression in correlation data.
9. The method according to claim 1, wherein: Receiving the cardiac signal includes receiving one of a unipolar electrogram and a bipolar electrogram acquired using a catheter.
10. A system, comprising: an interface configured to receive cardiac signals from a plurality of locations within a ventricle of a patient's heart; and A processor, the processor being configured to: comparing the received signal to a reference signal indicative of arrhythmia; calculating, based on the comparison, a direction towards a location exhibiting an increased correlation between the received signal and the reference signal; as well as The direction is indicated to the user.
11. The system according to claim 10, wherein: The processor is configured to compare by applying a statistical analysis method to estimate the directions.
12. The system according to claim 10, wherein: The processor is configured to perform the comparison by applying a machine learning (ML) model including a loss function for the direction.
13. The system according to claim 12, wherein: The processor is further configured to include a directional loss term related to a direction toward a known anatomical location in the loss function.
14. The system according to claim 12, wherein: The processor is configured to calculate the direction by finding a direction towards a candidate analysis region among directions, the direction minimizing the loss function.
15. The system according to claim 10, wherein: The processor is further configured to check whether the location guided by the direction is a region of interest as an arrhythmogenic region, and if so, mark the location, and if not, receive cardiac signals from additional multiple locations and indicate the next direction to the user based on analysis of the additional cardiac signals.
16. The system of claim 15, wherein: The processor is configured to check whether the location is a region of interest by checking whether an average value of correlations within an analysis area including the location is above a given threshold and / or whether a value of a minimum correlation in the analysis area is above a given threshold.
17. The system of claim 15, wherein: The processor is further configured to construct a map of the correlation to visualize the directions and paths generated by the following directions and to visualize progression in correlation data.
18. The system of claim 10, wherein: The interface is configured to receive the cardiac signal by receiving one of a unipolar electrogram and a bipolar electrogram acquired using a catheter.
