Local excitation time analysis system

Abstract

To process cardiac signals.SOLUTION: In one embodiment, a method for detecting local activation times of intracardiac electrogram (IEGM) signals includes receiving, from electro-physiological laboratory sub-systems, first IEGM signals and corresponding local activation time annotations of the first IEGM signals manually annotated by respective annotators, training an artificial neural network to detect local activation times of IEGM signals responsively to the first IEGM signals and the corresponding local activation time annotations, receiving a second IEGM signal, and applying the trained artificial neural network to the received second IEGM signal to provide an indication of a local activation time of the received second IEGM signal.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] (Cross - Reference to Related Applications) This application claims the benefit of U.S. Provisional Patent Application No. 63 / 077,780, filed on September 14, 2020, the disclosure of which is incorporated herein by reference.

[0002] (Field of the Invention) The present invention relates to medical systems, and more particularly, but not exclusively, to the processing of cardiac signals.

Background Art

[0003] Electrode catheters have been commonly used in the medical field for many years. Electrode catheters are used to stimulate and map the electrical activity within the heart and to ablate sites where abnormal electrical activity is observed. When in use, an electrode catheter is inserted into a major vein or artery, such as the femoral vein, and then guided into the cardiac chamber of interest within the heart. A typical ablation procedure involves inserting a catheter having one or more electrodes at its distal end into the cardiac chamber. A reference electrode can generally be provided by taping it to the patient's skin or by a second catheter placed within or near the heart. Radio - frequency (RF) current is applied between the catheter electrode of the ablation catheter and an indifferent electrode (which can be one of the catheter electrodes), and the current flows through the medium between the electrodes, i.e., between the blood and the tissue. The distribution of the current can depend on the amount of electrode surface in contact with the tissue compared to the blood, which has a higher conductivity than the tissue. Heating of the tissue occurs due to the electrical resistance of the tissue. When the tissue is heated sufficiently, cell destruction is caused in the cardiac tissue, and as a result, a lesion is formed within the non - conductive cardiac tissue. In some applications, irreversible electroporation can be performed to ablate the tissue.

[0004] Electrophysiological (EP) cardiac mapping is a diagnostic medical procedure that identifies the location of cardiac dysfunction within the heart. Time-varying electrocardiogram (ECG) signals are received by electrodes placed in contact with points along the surface of the patient's heart. The signals are processed, and various criteria for cardiac function are calculated from the processed (ECG) signals. The processed signals are then spatially mapped onto an image of the heart. The map is then output for analysis by medical professionals.

[0005] Analysis of cardiac signals may involve synchronization with the timing of ECG signals. For example, U.S. Patent Application Publication 2013 / 0123652 describes a method for analyzing signals that includes detecting time-varying intracardiac potential signals and finding a fit of the time-varying intracardiac potential signals to a predetermined oscillatory waveform. The method further includes estimating the annotation time of the signals in response to the fit.

[0006] Electrophysiological (EP) cardiac mapping, or electroanatomical cardiac mapping, is used to identify areas within cardiac tissue that are dysfunctional. An internal probe, typically a catheter with multiple mapping electrodes positioned along the body of the catheter near its distal end, is inserted into the cardiac cavity. Time-varying electrocardiogram (ECG) signals are recorded at multiple contact points between the mapping electrodes and cardiac tissue. The ECG electrodes are then moved to different contact points with the cardiac tissue, and the process is repeated. Criteria for cardiac function are then calculated from the local ECG signals, which are spatially mapped across the surface of the cardiac cavity. Mapping helps medical professionals identify areas of cardiac dysfunction.

[0007] Electrical sources within the heart, such as the sinoatrial (SA) and atrioventricular (AV) nodes, initiate electrical activity waves that propagate throughout the heart, triggering the atrium and ventricular muscle tissue to contract in their intrinsic sinus rhythm. As the activity wavefront reaches multiple mapping electrodes during each cardiac cycle, a unique ECG waveform is detected by multiple mapping electrodes. These waveforms are time-shifted due to the different arrival times of the same wavefront at different electrodes that contact the tissue at different spatial locations along the surface of the cardiac chambers.

[0008] The arrival times of ECG waveforms detected by multiple mapping electrodes can be used to map the propagation time and / or velocity of activity waves across the heart. Activity wave mapping is performed against a single time reference representing the cardiac cycle, known herein as the reference annotation time.

[0009] Reference annotation time can be calculated by processing ECG signals acquired from body surface (BS) electrodes or from intra-cardiac (IC) reference electrodes of an additional catheter placed in contact with the surface of the cardiac chamber. Typically, the physician specifies whether to calculate the reference annotation time from the BS or IC channel, depending on the suspected lesion. [Overview of the project] [Means for solving the problem]

[0010] According to one embodiment of the present disclosure, a method is provided for detecting local excitation times of an intracardiac electrogram (IEGM) signal, comprising: receiving a first IEGM signal and corresponding local excitation time annotations of the first IEGM signal, manually annotated by their respective annotators, from an electrophysiological laboratory subsystem; training an artificial neural network to detect local excitation times of an IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotations; receiving a second IEGM signal; and applying the trained artificial neural network to the received second IEGM signal to indicate the local excitation times of the received second IEGM signal.

[0011] Furthermore, according to one embodiment of the present disclosure, the method includes calculating weights for annotations made by each annotator according to the annotator's level of annotation experience for each annotator's local excitation time, and training includes training an artificial neural network to detect the local excitation time of an IEGM signal according to a first IEGM signal and the corresponding local excitation time annotation, weighted according to the respective calculated weights of each annotator who made the local excitation time annotation.

[0012] Furthermore, according to one embodiment of the present disclosure, the method includes searching a database of scientific literature publications for each of the annotators, which yields a number of search matches for each of the annotators indicating the level of annotation experience of each annotator's local excitement time, and calculating a weight for the annotations made by each of the annotators, in proportion to the number of search matches for each of the annotators.

[0013] In addition, according to one embodiment of the present disclosure, the search is limited to searching for scientific literature publications that include annotations on local excitation time.

[0014] Furthermore, according to one embodiment of the present disclosure, the number of search matches is the number of scientific publications that match each annotator.

[0015] Furthermore, according to one embodiment of the present disclosure, training includes inputting a first IEGM signal into an artificial neural network and iteratively adjusting the parameters of the artificial neural network in accordance with the output of the artificial neural network and annotations of the local excitation time of the first IEGM signal.

[0016] Furthermore, according to one embodiment of the present disclosure, the method includes minimizing a loss function which is a function of the output of the artificial neural network and annotations of the local excitation times of a first IEGM signal weighted according to each of the calculated weights, and the iterative adjustment is performed in accordance with minimizing the loss function.

[0017] Furthermore, according to one embodiment of the present disclosure, the loss function includes a binary cross-entropy loss function.

[0018] Furthermore, according to one embodiment of the present disclosure, the method includes generating an electroanatomical map in response to the local excitation time being indicated.

[0019] Another embodiment of the present disclosure also provides a system for detecting local excitation times of an intracardiac electrogram (IEGM) signal, comprising a remote server including processing circuitry configured to: receive a first IEGM signal and corresponding local excitation time annotations of the first IEGM signal, manually annotated by their respective annotators, from an electrophysiological laboratory subsystem; train an artificial neural network to detect local excitation times of an IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotations; receive a second IEGM signal; and apply the trained artificial neural network to the received second IEGM signal to indicate the local excitation times of the received second IEGM signal.

[0020] Furthermore, according to embodiments of the present disclosure, the system includes an electrophysiological laboratory subsystem, each electrophysiological laboratory subsystem comprising: a catheter inserted into at least one cardiac chamber of at least one living organism and configured to capture each first IEGM signal of a first IEGM signal from at least one cardiac chamber; a display; and a processing circuit, the processing circuit configured to render each first IEGM signal of a first IEGM signal on the display; receive corresponding local excitation time annotations of the displayed first IEGM signal, which have been manually annotated by each annotator; and provide each first IEGM signal of the first IEGM signal and the corresponding local excitation time annotations of the local excitation time annotations to a remote server.

[0021] Furthermore, according to one embodiment of the present disclosure, the processing circuit is configured to calculate weights for annotations made by each annotator according to the annotator's level of local excitation time annotation experience, and to train an artificial neural network to detect the local excitation time of the IEGM signal according to the first IEGM signal and the corresponding local excitation time annotations weighted according to the respective weights of the calculated weights of each annotator who annotated the local excitation time of the local excitation time annotation.

[0022] In addition, according to one embodiment of the present disclosure, the processing circuit is configured to search a database of scientific literature publications according to each annotator, yielding a number of search matches for each annotator indicating the level of annotation experience of each annotator's local excitation time, and to calculate weights for annotations made by each annotator according to the number of search matches for each annotator.

[0023] Furthermore, according to one embodiment of the present disclosure, the processing circuit is configured to limit the search of the database to scientific literature publications that describe annotations of local excitation times.

[0024] Furthermore, according to one embodiment of the present disclosure, each number of search matches is the respective number of scientific literature publications that match each respective annotator.

[0025] Still further, according to one embodiment of the present disclosure, the processing circuit is configured to input a first IEGM signal into an artificial neural network and iteratively adjust the parameters of the artificial neural network according to the output of the artificial neural network and the annotation of the local excitation time of the first IEGM signal.

[0026] In addition, according to one embodiment of the present disclosure, the processing circuit is configured to minimize a loss function that is a function of the output of the artificial neural network and the annotation of the local excitation time of the first IEGM signal weighted according to each of the calculated weights, and iteratively adjust the parameters of the artificial neural network according to minimizing the loss function.

[0027] Furthermore, according to one embodiment of the present disclosure, the loss function includes a binary cross-entropy loss function.

[0028] Furthermore, according to one embodiment of the present disclosure, the processing circuit is configured to generate an electroanatomical map in response to the local excitation time being indicated.

[0029] A further embodiment of the present disclosure also provides a software product comprising a non-temporary computer-readable medium in which program instructions are stored, wherein, when read by a central processing unit (CPU), the CPU causes the CPU to: receive a first IEGM signal and corresponding local excitation time annotations of the first IEGM signal, manually annotated by their respective annotators, from an electrophysiological laboratory subsystem; train an artificial neural network to detect local excitation times of an IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotations; receive a second IEGM signal; and apply the trained artificial neural network to the received second IEGM signal to indicate the local excitation times of the received second IEGM signal. [Brief explanation of the drawing]

[0030] The present invention will be understood from the following detailed description in conjunction with the attached drawings. [Figure 1] This is a diagram illustrating a system for performing a catheterization procedure in the heart, which is configured and operates according to an embodiment of the present invention. [Figure 2] Figure 1 is a flowchart showing the steps involved in operating the electrophysiological laboratory subsystem of the system. [Figure 3] Figure 1 shows a rendered, displayed, and annotated intracardiac electrocardiogram signal using the system shown in Figure 1. [Figure 4] Figure 1 is a block diagram of the system. [Figure 5] Figure 1 is a flowchart showing the steps involved in operating the remote server of the system. [Figure 6] This is a schematic diagram of the artificial neural network to be used in the system shown in Figure 1. [Figure 7] This is a flowchart including the sub-steps in the process of the method shown in Figure 5. [Figure 8]Figure 6 is a flowchart illustrating the steps involved in a method for processing intracardiac electrocardiogram signals using a trained artificial neural network. [Figure 9] Figure 1 is a schematic diagram of the displayed electroanatomical map, rendered by the system shown. [Modes for carrying out the invention]

[0031] Overview One of the major challenges in electrophysiology (EP) is finding an accurate annotation algorithm for intracardiac electrocardiogram (IEGM) signals to select the correct local activation time (LAT) from the IEGM signal and generate, for example, an LAT map. One method for automatically detecting LAT involves detecting the maximum negative gradient of the signal within a respective window of interest (WOI) and assigning the LAT to that point. The WOI is typically set by the system based on detecting a set of QRS complexes of signals captured by one or more surface electrodes. Detecting the maximum negative gradient does not provide a broad solution to the above problem, as there may be many such gradients within the WOI, and the algorithm may select the wrong gradient. Therefore, physicians have the option of manually changing the calculated LAT to a different point within the WOI.

[0032] Embodiments of the present invention solve the above problem by training an artificial neural network (ANN) using deep learning techniques to detect the LAT of a corresponding IEGM signal. The ANN is trained using IEGM signals and corresponding LAT annotations manually annotated by annotators (e.g., physicians or other medical professionals) provided from various EP (electropsychiatric) laboratories (labs). Manual annotation may involve correcting automatically calculated annotations or providing initial annotations to the IEGM signal.

[0033] In some embodiments, different LAT annotations are not given equal weights during ANN training. For example, the loss function used to train the ANN may include weights that can be associated with each IEGM signal and LAT annotation pair in order to weight the contribution of each IEGM signal and LAT annotation pair during training. In some embodiments, a binary cross-entropy (BCE) loss function may be used.

[0034] In some embodiments, the weights assigned to an IEGM signal and LAT annotation pair (for use, for example, in a loss function) may be calculated according to the LAT annotation experience level of the annotator who manually determined the LAT annotation. In this way, the ANN can be trained by assigning greater weights to more experienced annotators.

[0035] The level of LAT annotation experience of various annotators can be estimated by searching databases containing scientific publications to detect the number of search matches (e.g., the number of scientific publications) for various annotators within the LAT annotation range. For example, searching for "John Smith and LAT annotation" might yield 50 matches, while searching for "Tim Jones and LAT annotation" might yield 400 matches. In such cases, LAT annotations provided by Tim Jones would be given much greater weight in training than those provided by John Smith. Any suitable database, such as Google Scholar, Scopus, or Web of Science, can be searched.

[0036] System Description Hereinafter, we refer to Figure 1, a pictorial diagram of a medical system 10 configured according to an embodiment of the present invention for performing a catheter insertion procedure on a working heart 12. The medical system 10 may be configured to evaluate electrical activity and perform an ablation procedure on the living heart 12. The system 10 includes an EP laboratory subsystem 11 (only one is shown for simplicity) for capturing EP data. The medical system 10 also includes a remote server 26, e.g., a cloud computing device, to which the EP data is stored and / or transmitted via a network 27 for processing by the EP laboratory subsystem 11. The EP data may be compressed in the EP laboratory subsystem 11 and transmitted to the remote server 26 via the network 27 in a compressed form.

[0037] Next, one of the EP laboratory subsystems 11 is described in more detail below, merely as an example. The various EP laboratory subsystems 11 may comprise the same or different EP laboratory equipment that provides EP data to a remote server 26 for processing. The EP laboratory subsystem 11 in Figure 1 comprises a catheter 14 that is percutaneously inserted by an operator 16 into a ventricle or vascular structure of the heart 12 through the vascular system of a subject. The operator 16, usually a physician, brings the distal end 18 of the catheter into contact with the heart wall, for example, at the ablation target site. The electrical activity map may be prepared according to the methods disclosed in U.S. Patents No. 6,226,542, No. 6,301,496, and No. 6,892,091. One commercially available product embodying the elements of system 10 is available as the CARTO® 3 system (available from Biosense Webster, Inc., Irvine, CA). This system may be modified by those skilled in the art to embody the principles of the invention described herein.

[0038] Areas identified as abnormal by evaluation of the electrical activity map can be ablated by applying thermal energy, for example, by passing a high-frequency current through a wire inside the catheter to one or more electrodes at the distal end 18 that apply high-frequency energy to the myocardium. This energy is absorbed into the tissue and heats it until the tissue permanently loses its electrical excitability. If performed successfully, this procedure creates non-conductive damaged areas in the cardiac tissue, which block the abnormal electrical pathways that cause arrhythmias. The principle of the present invention can be applied to different cardiac chambers to diagnose and treat a number of different cardiac arrhythmias.

[0039] The catheter 14 typically includes a handle 20 having a control unit suitable for a handle, which enables the operator 16 to manipulate, position, and orient the distal end 18 of the catheter 14 as desired for ablation. To assist the operator 16, the distal portion of the catheter 14 houses a position sensor (not shown) that supplies signals to a processing circuit 22 located in a console 24. The processing circuit 22 can perform several processing functions, as described below.

[0040] The wire connection section 35 may connect the console 24 to the body surface electrode 30 and other components of the positioning subsystem for measuring the position and orientation coordinates of the catheter 14. A processing circuit 22 or another processor (not shown) may be an element of the positioning subsystem. As taught in U.S. Patent No. 7,536,218, the catheter electrode 31 and the body surface electrode 30 can be used to measure tissue impedance at the ablation site. A temperature sensor (not shown), typically a thermocouple or thermistor, may be placed on the ablation surface of the distal portion of the catheter 14, as described below.

[0041] The console 24 typically houses one or more ablation power generators 25. The catheter 14 can be adapted to conduct ablation energy to the heart using any known ablation technique, such as radiofrequency energy, ultrasonic energy, irreversible electroporation, and laser-generated light energy. Such methods are disclosed in U.S. Patents 6,814,733, 6,997,924, and 7,156,816.

[0042] In one embodiment, the positioning subsystem includes a magnetic position tracking arrangement that uses a magnetic field generating coil 28 to generate a magnetic field within a predetermined working volume and senses these magnetic fields in the catheter to determine the position and orientation of the catheter 14. The positioning subsystem is described in U.S. Patents No. 7,756,576 and No. 7,536,218.

[0043] As described above, the catheter 14 is coupled to the console 24, which allows the operator 16 to observe and adjust the function of the catheter 14. The console 24 includes a processing circuit 22, typically a computer with appropriate signal processing circuitry. The processing circuit 22 is coupled to drive a display 29 (e.g., a monitor). The signal processing circuitry typically receives, amplifies, filters, and digitizes signals from the catheter 14, which include signals generated by sensors such as electrical, temperature, and contact force sensors, as well as signals generated by a distally located location-sensing electrode 31 within the catheter 14. 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 electrical signals from the electrodes. In some embodiments, the digitized signals are sent to a remote server 26 (optionally compressed before transmission) to calculate position and orientation data and / or analyze electrical signals from electrodes 30, 31 and / or use the electrical signals and associated data to train an artificial neural network. These are described in more detail below.

[0044] To generate an electroanatomical map, the processing circuit 22 typically includes a mapping module, which includes an electroanatomical map generator, an image registration program, an image or data analysis program, and a graphical user interface configured to present graphical information on a display 29. In some embodiments, some or all of the functions of the mapping module are performed by a remote server 26.

[0045] Although not shown in the diagram for simplification, system 10 typically includes other elements. For example, system 10 may include an electrocardiogram (ECG) monitor, which is coupled to receive signals from one or more of the body surface electrodes 30 to supply an ECG synchronization signal to the console 24 or remote server 26. In some embodiments, some or all of the functions of the ECG monitor are performed by the remote server 26. As mentioned above, system 10 also typically has a reference position sensor, either on an externally attached reference patch attached to the outside of the patient's body or on an internally-placed catheter inserted into the heart 12 and maintained in a fixed position relative to the heart 12. Conventional pumps and lines may be provided for circulating fluid through the catheter 14 to cool the ablation site. System 10 and / or remote server 26 may receive image data from an external imaging modality such as an MRI unit or similar, and include an image processor that is incorporated into or can be activated (e.g., by processing circuit 22 and / or remote server 26) to generate and display images.

[0046] In practice, some or all of the functions of the processing circuit 22 can be combined into a single physical component, or alternatively implemented using multiple physical components. These physical components may include hardwired devices, programmable devices, or a combination of the two. In some embodiments, at least some of the functions of the processing circuit 22 may be executed by a programmable processor under the control of suitable software. This software may be downloaded to the device in electronic form, for example, via a network. Alternatively, this software may be stored in a tangible, non-temporary computer-readable medium such as optical memory, magnetic memory, or electronic memory.

[0047] Next, refer to Figures 2 and 3. Figure 2 is a flowchart 40 showing the steps in one operating method of the electrophysiological laboratory subsystem 11 of system 10 in Figure 1. Figure 3 is a diagram of the displayed and annotated IEGM signal 48 rendered using system 10 in Figure 1.

[0048] Each EP laboratory subsystem 11 includes a catheter 14 configured to be inserted into at least one cardiac chamber of at least one living organism and to capture the respective IEGM signal 48 from at least one cardiac chamber. Any preferred type of catheter 14 or more catheters 14 may be used in each EP laboratory subsystem 11. The processing circuit 22 is configured to receive the IEGM signal 48, render a representation of the IEGM signal 48 on a display 29 (block 42), and receive annotations 50 of the local excitation time (LAT) of the corresponding IEGM signal 48, which have been manually annotated by an annotator in the EP laboratory subsystem 11 (block 44). First, the IEGM signal 48 may be automatically annotated by an algorithm operating on the processing circuit 22. The annotator can then correct the automatic annotation with a manual annotation by marking the local excitation time annotation 50 on the displayed IEGM signal 48. In some embodiments, the commentator may view the displayed IEGM signal 48 on the display 29 (without the processing circuit 22 for calculating automated LAT annotations), and then mark the local excitation time annotation 50 on the displayed IEGM signal 48.

[0049] The processing circuit 22 of the EP laboratory subsystem 11 is configured to provide the IEGM signal 48 and the corresponding local excitation time annotation 50 to a remote server 26 (block 46). In some embodiments, the EP laboratory subsystem 11 is configured to provide the IEGM signal 48 and the corresponding local excitation time annotation 50 to the remote server 26 over a network 27 (Figure 1) via a preferred network interface.

[0050] Next, refer to Figures 4 and 5. Figure 4 is a block diagram of system 10 in Figure 1. Figure 5 is a flowchart 100 including the steps in the operation method of the remote server 26 of system 10 in Figure 1.

[0051] The remote server 26 includes a processing circuit 60, memory 62, a data bus 64, and a network interface 66. The processing circuit 60 is configured to launch software and perform various signal processing and computation tasks, and includes an artificial neural network 68, a training module 70, and a mapping module 72. The training module 70 is configured to train the artificial neural network 68, as will be described in more detail below with reference to Figures 5 to 7. The mapping module 72 is configured to generate an EP map in response to cardiac signals and other data captured from the biological system, as will be described in more detail with reference to Figures 8 and 9.

[0052] In practice, some or all of the functions of the processing circuit 60 can be combined in a single physical component, or alternatively implemented using multiple physical components. These physical components may include hardwired devices, programmable devices, or a combination of the two. In some embodiments, at least some of the functions of the processing circuit 60 may be executed by a programmable processor under the control of suitable software. This software may be downloaded to the device in electronic form, for example, via a network. Alternatively, this software may be stored in a tangible, non-temporary computer-readable medium such as optical memory, magnetic memory, or electronic memory.

[0053] Memory 62 is configured to store data used by the processing circuit 60. The data bus 64 is configured to transfer data between various elements of the remote server 26, for example, between the processing circuit 60 and the network interface 66.

[0054] The training module 70, operating on the processing circuit 60, is configured to receive IEGM signals 48 (captured within each EP laboratory subsystem 11) and corresponding local excitation time annotations 50 of the IEGM signals 48, manually annotated by each annotator, from the EP laboratory subsystem 11 (block 102). In other words, the training module 70 is configured to receive corresponding local excitation time annotations 50 and IEGM signals 48, manually annotated by a certain annotator, from one of the EP laboratory subsystems 11, and corresponding local excitation time annotations 50 and other IEGM signals 48, manually annotated by a different annotator, from another EP laboratory subsystem 11.

[0055] The training module 70, operating on the processing circuit 60, is configured to train an artificial neural network 68 (block 104) to detect the local excitation time of the IEGM signal, based on training data including the IEGM signal 48 and the corresponding local excitation time annotation 50. The process in block 104 includes the sub-processes of blocks 106 to 110, which are described in more detail below.

[0056] The training module 70, operating on the processing circuit 60, is configured to search a database of scientific literature publications (e.g., Google Scholar, Scopus, or Web of Science) as a search string (block 106) for each annotator (who provided the local excitation time annotations 50), and the search yields a number of search matches for each annotator indicating their level of local excitation time annotation experience. In other words, the database is searched using different search strings for each annotator (e.g., "J.Smith", "T.Jones", etc.) and yields a number of search matches for each annotator (e.g., 5 matches for J.Smith and 40 matches for T.Jones, etc.). In some cases, the search for a given annotator may yield no search matches. The search may be performed using any suitable software script, for example, using a web crawler such as pybliometrics2.5.0 to access Scopus. The number of search matches may also be the number of scientific publications that match each annotator (for example, 5 publications for J. Smith and 40 publications for T. Jones).

[0057] In some embodiments, a training module 70 operating on the processing circuit 60 is configured to limit the database search to scientific literature publications that include annotations for local excitation time and / or IEG and / or electrocardiogram and / or EP, etc. Limiting the search to one or more of the above is useful in preventing false results. For example, there may be a J. Smith who published a paper in the journal "Nuclear Physics," and therefore his experience is unrelated to the J. Smith who provided the local excitation time annotation 50.

[0058] The training module 70, which operates on the processing circuit 60, is configured to calculate weights for annotations made by each commentator according to each commentator's level of annotation experience in local excitation time (block 108). For example, a weight is calculated for annotations made by J. Smith, and another weight is calculated for annotations made by T. Jones, and so on.

[0059] In some embodiments, the training module 70 operating on the processing circuit 60 is configured to calculate weights for annotations made by each annotator based on the number of search matches for each annotator (e.g., the number of scientific publications). For example, an annotation made by J. Smith is weighted according to the 5 search matches (e.g., 5 publications) found in the database for J. Smith, and an annotation made by T. Jones is weighted according to the 40 search matches (e.g., 40 publications) found in the database for T. Jones. The weights can be calculated proportionally. For example, if there are N annotators and the j-th annotator is found to have P matches during a search... j If a number of publications is detected, the weight W of the i-th commentator i This is equivalent to the following:

[0060]

number

[0061] The training module 70, which operates on the processing circuit 60, is configured to train an artificial neural network 68 (block 110) to detect local excitation times of the IEGM signal in accordance with training data including the IEGM signal 48 (received from the EP laboratory subsystem 11) and the corresponding local excitation time annotations 50, which are weighted according to the respective calculated weights of each annotator who annotated each local excitation time annotation 50. For example, annotations provided by J. Smith are weighted according to the weight calculated for J. Smith, annotations provided by T. Jones are weighted according to the weight calculated for T. Jones, and so on.

[0062] Next, refer to Figures 6 and 7. Figure 6 is a schematic diagram of the artificial neural network 68 for use in the system 10 of Figure 1. Figure 7 is a flowchart including sub-steps in the block 110 process of the method of Figure 5.

[0063] A neural network is a network or circuit of neurons, or in the modern sense, an artificial neural network composed of artificial neurons or nodes. The connections of biological neurons are modeled as weights. Positive weights reflect excitatory connections, while negative values ​​indicate inhibitory connections. Inputs are modified by the weights and summed using linear combinations. An activation function can control the amplitude of the output. For example, the acceptable range of the output is usually 0 to 1, but it can also be -1 to 1.

[0064] These artificial networks can be used for predictive modeling, adaptive control, and applications, and can be trained via datasets. Self-learning derived from experience can occur within the network, allowing it to draw conclusions from complex and seemingly unrelated sets of information.

[0065] For completeness, a biological neural network consists of groups of chemically connected or functionally associated neurons. One neuron may be connected to many other neurons, and the total number of neurons and connections in the network can be wide-ranging. Connections called synapses are typically formed from axons to dendrites, but interdendritic synapses and other connections are also possible. Apart from electrical signaling, there are other forms of signaling resulting from the diffusion of neurotransmitters.

[0066] Artificial intelligence, cognitive modeling, and neural networks are information processing paradigms inspired by how biological nervous systems process data. Artificial intelligence and cognitive modeling attempt to simulate some of the characteristics of biological neural networks. In the field of artificial intelligence, artificial neural networks have been successfully applied to speech recognition, image analysis, and adaptive control, and are used to build software agents or autonomous robots (in computer and video games).

[0067] A neural network (NN), also known as an artificial neural network (ANN) or simulated neural network (SNN), is an interconnected group of natural or artificial neurons that uses mathematical or computational models based on an accessibility-theoretic approach to information processing. In most cases, an ANN is an adaptive system that modifies its structure based on external or internal information flowing through the network. More practically, neural networks are nonlinear statistical data modeling or decision-making tools. They can be used to model complex relationships between inputs and outputs and to find patterns in data.

[0068] In some embodiments, the artificial neural network 68 includes a fully connected neural network, such as a convolutional neural network. In other embodiments, the artificial neural network 68 may include any preferred ANN. The artificial neural network 68 may also include software executed by the processing circuit 60 (Figure 4) and / or hardware modules configured to perform the functions of the artificial neural network 68.

[0069] The artificial neural network 68 includes an input layer 80 from which input is received, and one or more hidden layers 82 that progressively process the input to an output layer 84 from which the output of the artificial neural network 68 is provided. The artificial neural network 68 may include layer weights between layers 80, 82, and 84 of the artificial neural network 68. The artificial neural network 68 manipulates the data received in the input layer 80 according to the various layer weight values ​​between layers 80, 82, and 84 of the artificial neural network 68.

[0070] The layer weights of the artificial neural network 68 are updated during the training of the artificial neural network 68 so that the artificial neural network 68 can perform the data manipulation tasks that it is trained to perform.

[0071] The number and width of layers within the artificial neural network 68 may be configurable. As the number and width of layers increase, the accuracy with which the artificial neural network 68 can manipulate data according to the task at hand increases. However, a larger number of layers and wider layers generally require more training data, i.e., more training time, and training may not converge. For example, the input layer 80 may contain 400 neurons (for example, to compress a batch of 400 samples), and the output layer may also contain 400 neurons.

[0072] Training the artificial neural network 68 is largely an iterative process. One method of training the artificial neural network 68 is described below. The training module 70, operating on the processing circuit 60 (Figure 4), is configured to iteratively adjust the parameters of the artificial neural network 68 (e.g., layer weights) (block 112) to reduce the difference between the output of the artificial neural network 68 and the annotation 50 of the local excitation time of the IEGM signal 48.

[0073] In some embodiments, a training module 70 operating on a processing circuit 60 (Figure 4) is configured to minimize a loss function, which is a function of the output of the artificial neural network 68 and annotation 50 of the local excitation time of the IEGM signal 48 weighted according to each of the calculated weights (calculated in step 108 of block 5 in Figure 5), and to iteratively adjust the parameters of the artificial neural network 68 (e.g., layer weights) in response to minimizing the loss function. In some embodiments, the loss function includes a binary cross-entropy (BCE) loss function. Pytorch.org provides an example of a suitable BCE loss function.

[0074] The sub-processes of the Block 112 process are described below.

[0075] The training module 70, operating on the processing circuit 60 of the processing circuit 22 (Figure 4), is configured to input the IEGM signal 48 to the input layer 80 of the artificial neural network 68 (block 114). The training module 70, operating on the processing circuit 60 (Figure 4), is configured to compare the output of the artificial neural network 68 (e.g., the output of the output layer 84) with a desired output, i.e., the corresponding local excitation time annotation 50 of the IEGM signal 48 (block 116), using a suitable loss function that takes into account, for example, the weights for each local excitation time annotation 50 (calculated for each annotator).

[0076] The output of the artificial neural network 68 contains various vectors corresponding to the IEGM signal 48 input to the artificial neural network 68. Each output vector contains a component whose components are floating-point values ​​(e.g., 0 to 1). Each desired output is represented as a one-hot vector where all components of the vector have zero values, except for one component whose value is 1, corresponding to the time value of the annotation 50 for each local excitation time.

[0077] For example, if there is a set of vectors A, B, and C output by the artificial neural network 68, and a corresponding set of vectors representing the corresponding local excitation time annotations A', B', and C', the training module 70 of the processing circuit 60 (Figure 4) uses a loss function to compare A with A', B with B', C with C', etc., based on the weights of the annotators who annotated A', B', and C'.

[0078] In the determination block 118, the training module 70 operating on the processing circuit 60 (Figure 4) is configured to determine whether the difference between the output of the artificial neural network 68 and the desired output is sufficiently small. If the difference between the output of the artificial neural network 68 and the desired output is sufficiently small (branch 120), the training module 70 operating on the processing circuit 60 (Figure 4) is configured to save the parameters of the artificial neural network 68 (e.g., layer weights) for future use (block 122).

[0079] If the difference is not sufficiently small (Branch 124), the training module 70 operating on the processing circuit 60 (Figure 4) is configured to modify the parameters of the artificial neural network 68 (e.g., layer weights) (Block 126) to reduce the difference between the output of the artificial neural network 68 and the desired output of the artificial neural network 68 according to the loss function. The difference to be minimized in the above embodiment is the overall difference between all the outputs of the artificial neural network 68 and all the desired outputs (e.g., local excitation time, note 50) according to the loss function. The training module 70 operating on the processing circuit 60 (Figure 4) is configured to modify the parameters using any suitable optimization algorithm, for example, a gradient descent algorithm such as Adam optimization. The process in Blocks 114 to 118 is then repeated.

[0080] Here, we refer to Figure 8, which is a flowchart 150 showing the steps in a method for processing intracardiac electrocardiogram signals using the trained artificial neural network 68 of Figure 6. See also Figure 4.

[0081] A mapping module 72 (or any other suitable module) operating on the processing circuit 60 is configured to receive an IEGM signal from one of the EP laboratory subsystems 11 (block 152). The mapping module 72 (or any other suitable module) operating on the processing circuit 60 is configured to apply a trained artificial neural network 68 to the received IEGM signal (block 154) to indicate the local excitation time of the received IEGM signal. The output of the artificial neural network 68 may include a vector having components (e.g., 400 components) where each component has a floating value (e.g., a decimal value) between 0 and 1. The floating values ​​represent the respective probabilities that each vector component is the LAT value for the input IEGM signal. Thus, the highest floating value is associated with the highest probability and therefore indicates the LAT value that should be used for the input IEGM signal. The mapping module 72, operating on the processing circuit 60, may be configured to receive the output of the artificial neural network 68, detect the highest floating-point value of the vector components output by the artificial neural network 68, and calculate the LAT value of the received IEGM signal according to the position of the vector component having the highest floating-point value.

[0082] Here, refer to Figure 9, a schematic diagram of the displayed electroanatomical map 160 rendered by the system 10 of Figure 1. See also Figure 8. The mapping module 72 (or any other suitable module) operating on the processing circuit 60 is optionally configured to generate the electroanatomical map 160 in response to the display of local excitation times and other similar EP data provided in the process of block 154 (block 156). In some embodiments, the processing circuit 60 is configured to provide the electroanatomical map 160 to the EP laboratory subsystem 11 that provided the IEGM signals used to generate the electroanatomical map 160. The mapping module 72 is optionally configured to provide the LAT detected in the process of block 154 to the EP laboratory subsystem 11 that provided the IEGM signals in which the LAT was detected (block 158).

[0083] When used herein, the terms “about” or “approximately” with respect to any number or range of numbers indicate a suitable dimensional tolerance that enables a part or set of components to function in accordance with its intended purpose as described herein. More specifically, “about” or “approximately” may refer to a range of values ​​within ±20% of the listed values; for example, “about 90%” may refer to a range of values ​​between 72% and 108%.

[0084] Various features of the present invention are described in the context of separate embodiments for clarity, but these may also be provided in combination in a single embodiment. Conversely, various features of the present invention described in the context of a single embodiment for brevity may be provided separately or in any preferred partial combination.

[0085] The embodiments described above are by reference only, and the present invention is not limited to those specifically illustrated and described in the above specification. Rather, the scope of the present invention includes both combinations and partial combinations thereof of the various features described in the above specification, as well as variations and modifications thereof not disclosed in the prior art, which would be conceivable to those skilled in the art by reading the above description.

[0086] [Implementation Method] (1) A method for detecting the local excitation time of intracardiac electrocardiogram (IEGM) signals, Receiving a first IEGM signal and corresponding local excitation time annotations of the first IEGM signal, manually annotated by each annotator, from the electrophysiological laboratory subsystem, Training an artificial neural network to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotation, Receiving the second IEGM signal, A method comprising applying a trained artificial neural network to the received second IEGM signal so as to indicate the local excitation time of the received second IEGM signal. (2) The method of Embodiment 1, further comprising calculating weights for annotations made by each of the annotators according to the annotator's level of annotator experience in local excitation time, and training the artificial neural network to detect local excitation time in the first IEGM signal and the corresponding local excitation time annotation weighted according to the respective weights of the calculated weights of each of the annotators who annotated the local excitation time annotation. (3) The method of Embodiment 2, further comprising searching a database of scientific literature publications according to each of the annotators of the annotators, which yields a number of search matches indicating the level of annotation experience of each of the annotators of the local excitation time, wherein the calculation comprises calculating the weights for the annotations made by each of the annotators of the annotators according to the number of search matches for each of the annotators of the annotators. (4) The method according to Embodiment 3, wherein the search is limited to searching for scientific literature publications that include annotations on local excitation time. (5) The method according to Embodiment 3, wherein each of the search matches is the number of scientific literature publications that match each of the annotators.

[0087] (6) The training described above is Inputting the first IEGM signal into the artificial neural network, The method according to Embodiment 3, comprising iteratively adjusting the parameters of the artificial neural network in accordance with the output of the artificial neural network and the annotation of the local excitation time of the first IEGM signal. (7) The method according to Embodiment 6, further comprising minimizing a loss function which is a function of the output of the artificial neural network and annotations of the local excitation times of the first IEGM signal weighted according to each of the calculated weights, wherein the iterative adjustment is performed in accordance with minimizing the loss function. (8) The method according to embodiment 7, wherein the loss function includes a binary cross-entropy loss function. (9) The method according to Embodiment 1, further comprising generating an electroanatomical map in response to the indicated local excitation time. (10) A system for detecting the local excitation time of intracardiac electrocardiogram (IEGM) signals, Receiving a first IEGM signal and corresponding local excitation time annotations of the first IEGM signal, manually annotated by each annotator, from the electrophysiological laboratory subsystem, Training an artificial neural network to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotation, Receiving the second IEGM signal, A system comprising a remote server including a processing circuit configured to apply the trained artificial neural network to the received second IEGM signal so as to indicate the local excitation time of the received second IEGM signal.

[0088] (11) The system further comprises the electrophysiological laboratory subsystem, each of which is an electrophysiological laboratory subsystem. A catheter inserted into at least one cardiac chamber of at least one living organism and configured to capture the first IEG signals of each of the first IEG signals from the at least one cardiac chamber, The display and A processing circuit, wherein the processing circuit is Rendering each of the first IEGM signals of the first IEGM signal onto the display, Receiving the corresponding local excitation time annotation from among the local excitation time annotations of the displayed first IEGM signal, which have been manually annotated by each of the aforementioned annotators, The system according to embodiment 10, comprising a processing circuit configured to provide the remote server with the respective first IEGM signals and the corresponding local excitation time annotations from the local excitation time annotations of the first IEGM signals. (12) The processing circuit is The weight of each of the aforementioned annotators' annotations is calculated according to the annotator's level of annotation experience with local excitement time, The system according to Embodiment 10, comprising: training the artificial neural network to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotations weighted according to the respective weights of the calculated weights of the respective annotations of the local excitation time annotations of the respective annotations of the respective local excitation time annotations of the respective annotations of the respective local excitation time annotations of the respective annotations of the respective local excitation time annotations of the respective annotations of the respective annotations of the respective local excitation time annotations. (13) The processing circuit is Searching a database of scientific literature publications according to each of the aforementioned annotators yields a number of search matches indicating the annotator's level of experience in local excitation time for each of the aforementioned annotators, The system according to embodiment 12, configured to calculate the weight of the annotation made by each of the annotations of the annotation, in accordance with the number of search matches for each of the annotations of the annotation. (14) The system according to embodiment 13, wherein the processing circuit is configured to limit the search of the database to scientific literature publications that include annotations on local excitation times. (15) The system according to Embodiment 13, wherein each of the search matches is the number of scientific literature publications that match each of the annotators.

[0089] (16) The processing circuit is Inputting the first IEGM signal into the artificial neural network, The system according to Embodiment 13, configured to iteratively adjust the parameters of the artificial neural network in accordance with the output of the artificial neural network and the annotation of the local excitation time of the first IEGM signal. (17) The processing circuit is Minimizing a loss function which is a function of the output of the artificial neural network and the annotation of the local excitation time of the first IEGM signal, weighted according to each of the calculated weights, The system according to embodiment 16, configured to iteratively adjust the parameters of the artificial neural network in accordance with minimizing the loss function. (18) The system according to embodiment 17, wherein the loss function includes a binary cross-entropy loss function. (19) The system according to Embodiment 10, wherein the processing circuit is configured to generate an electroanatomical map in response to the local excitation time being indicated. (20) A software product including a non-temporary computer-readable medium on which program instructions are stored, wherein when the instructions are read by a central processing unit (CPU), the CPU Receiving a first IEGM signal and corresponding local excitation time annotations of the first IEGM signal, manually annotated by each annotator, from the electrophysiological laboratory subsystem, Training an artificial neural network to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotation, Receiving the second IEGM signal, A software product that causes a trained artificial neural network to be applied to the received second IEGM signal so as to indicate the local excitation time of the received second IEGM signal.

Claims

1. A system for detecting local excitation time of intracardiac electrocardiogram (IEGM) signals, Receiving a first IEGM signal and corresponding local excitation time annotations of the first IEGM signal, manually annotated by each annotator, from the electrophysiological laboratory subsystem. Training an artificial neural network to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotation, Receiving the second IEGM signal, A remote server including a processing circuit is configured to apply the trained artificial neural network to the received second IEGM signal so as to indicate the local excitation time of the received second IEGM signal. The aforementioned processing circuit is The weight of each of the aforementioned annotators' annotations is calculated according to the annotator's level of annotation experience with local excitement time, The artificial neural network is trained to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotations, weighted according to the respective weights of the calculated weights of the annotations of each of the annotations that have annotated the local excitation time of each of the annotations, The aforementioned processing circuit is Searching a database of scientific literature publications according to each of the aforementioned annotators yields a number of search matches indicating the annotator's level of experience in local excitation time for each of the aforementioned annotators, A system configured to calculate the weight of the annotation made by each of the aforementioned annotations, according to the number of search matches for each of the aforementioned annotations.

2. The electrophysiological laboratory subsystem further comprises the electrophysiological laboratory subsystem, each of which is: A catheter inserted into at least one cardiac chamber of at least one living organism and configured to capture the first IEGM signals of each of the first IEGM signals from the at least one cardiac chamber, The display and A processing circuit, wherein the processing circuit is Rendering each of the first IEGM signals of the first IEGM signal onto the display, Receiving the corresponding local excitation time annotation from among the local excitation time annotations of the displayed first IEGM signal, which have been manually annotated by each of the aforementioned annotators, The system according to claim 1, comprising a processing circuit configured to provide the remote server with the respective first IEGM signals and the corresponding local excitation time annotations of the local excitation time annotations of the first IEGM signals.

3. The system according to claim 1, wherein the processing circuit is configured to limit the search of the database to scientific literature publications that include annotations on local excitation times.

4. The system according to claim 1, wherein each of the search matches is the number of scientific literature publications that match each of the annotators.

5. The aforementioned processing circuit is Inputting the first IEGM signal into the artificial neural network, The system according to claim 1, configured to iteratively adjust the parameters of the artificial neural network in accordance with the output of the artificial neural network and the annotation of the local excitation time of the first IEGM signal.

6. The aforementioned processing circuit is Minimizing a loss function which is a function of the output of the artificial neural network and the annotation of the local excitation time of the first IEGM signal, weighted according to each of the calculated weights, The system according to claim 5, configured to iteratively adjust the parameters of the artificial neural network in accordance with minimizing the loss function.

7. The system according to claim 6, wherein the loss function includes a binary cross-entropy loss function.

8. The system according to claim 1, wherein the processing circuit is configured to generate an electroanatomical map in response to the local excitation time being indicated.

9. A software product including a non-temporary computer-readable medium on which program instructions are stored, wherein, when the instructions are read by a central processing unit (CPU), the CPU... Receiving a first IEGM signal and corresponding local excitation time annotations of the first IEGM signal, manually annotated by each annotator, from the electrophysiological laboratory subsystem. Training an artificial neural network to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotation, Receiving the second IEGM signal, Applying the trained artificial neural network to the received second IEGM signal so as to indicate the local excitation time of the received second IEGM signal, The weight of each of the aforementioned annotators' annotations is calculated according to the annotator's level of annotation experience with local excitement time, The artificial neural network is trained to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotations weighted according to the respective weights of the calculated weights of the annotations of the annotations of the local excitation time of the first IEGM signal, Searching a database of scientific literature publications according to each of the aforementioned annotators yields a number of search matches indicating the annotator's level of experience in local excitation time for each of the aforementioned annotators, A software product that calculates the weights for the annotations made by each of the aforementioned annotators, according to the number of search matches for each of the aforementioned annotators.

10. A method for operating a system that detects the local excitation time of intracardiac electromagnetism (IEGM) signals, The processor of the system receives from the electrophysiological laboratory subsystem a first IEGM signal and annotations of the corresponding local excitation times of the first IEGM signal, which are manually annotated by each annotator. The processor trains an artificial neural network to detect the local excitation time of the IEGM signal in accordance with the first IEGM signal and the corresponding local excitation time annotation. The processor receives a second IEGM signal, The processor applies the trained artificial neural network to the received second IEGM signal so as to indicate the local excitation time of the received second IEGM signal. Includes, The processor further includes calculating weights for annotations made by each of the annotators according to the annotator's level of local excitation time annotation experience, and the training includes training the artificial neural network to detect the local excitation time of the IEGM signal according to the first IEGM signal and the corresponding local excitation time annotation weighted according to the respective weights of the calculated weights of each of the annotators who annotated the local excitation time annotation, A method of operating the system, wherein the processor further includes searching a database of scientific literature publications according to each of the annotators, yielding a number of search matches representing the level of annotation experience of each of the annotators' local excitation times, and the calculation includes the processor calculating the weights for the annotations made by each of the annotators according to the number of search matches for each of the annotators.

11. The method of operating the system according to claim 10, wherein the search is limited to searching for scientific literature publications that include annotations on local excitation time.

12. The method of operating the system according to claim 10, wherein each of the search matches is the number of scientific literature publications that match each of the annotators.

13. The aforementioned training is, The processor inputs the first IEGM signal to the artificial neural network, A method of operating the system according to claim 10, wherein the processor iteratively adjusts the parameters of the artificial neural network in accordance with the output of the artificial neural network and the annotation of the local excitation time of the first IEGM signal.

14. The method of operating the system according to claim 13, further comprising the processor minimizing a loss function which is a function of the output of the artificial neural network and annotations of the local excitation times of the first IEGM signal weighted according to each of the calculated weights, wherein the iterative adjustment is performed in accordance with minimizing the loss function.

15. The method of operating the system according to claim 14, wherein the loss function includes a binary cross-entropy loss function.

16. A method of operating the system according to claim 10, further comprising the processor generating an electroanatomical map in response to the local excitation time being indicated.

Images