Processing method of intracardiac electrophysiological examination data based on artificial intelligence technology, program product, electronic equipment and storage medium

Abstract

The invention provides an intracardiac electrophysiological examination data processing method based on an artificial intelligence technology, a program product, electronic equipment and a storage medium, and is applied to the technical field of medical signal processing. The method for processing the intracardiac electrophysiological examination data based on the artificial intelligence technology comprises the following steps: acquiring the intracardiac electrophysiological examination data corresponding to a to-be-detected object; the electrophysiological examination data in the cardiac cavity are dynamically tracked, a first detection result corresponding to the to-be-detected object is obtained, and the first detection result represents arrhythmia information corresponding to the to-be-detected object. The electrophysiological examination data in the cardiac cavity are dynamic data obtained by recording cardiac electrical activity for a long time, so that arrhythmia information can be determined by dynamically tracking the electrophysiological examination data in the cardiac cavity. Compared with manual detection in the prior art, the arrhythmia detection efficiency can be improved, subjective misjudgment can be avoided, and therefore the arrhythmia detection accuracy is improved.

Description

Technical Field

[0001] The present application relates to the field of medical signal processing technology, and in particular to a method, program product, electronic device and storage medium for processing intracardiac electrophysiological examination data based on artificial intelligence technology. Background Art

[0002] Intracardiac electrophysiological study (EPS) is an accurate and invasive method for evaluating cardiac electrical conduction function. It is an important tool for cardiovascular medicine in diagnosing complex fast and slow arrhythmias. It is extremely widely used and is the cornerstone of current interventional surgical treatment of arrhythmias. The basic principle of intracardiac electrophysiological study is to record intracardiac electrical signals using electrode catheters placed in different parts of the heart using a multi-channel recorder and stimulator (for example, using graded incremental or programmed pre-stimulation of specific parts of the heart (such as the atria, ventricles, or coronary sinus) to obtain characteristics such as the direction, sequence, and conduction velocity of electrical conduction in various parts of the heart). The above electrical signals are used as the "gold standard" for diagnosing arrhythmias, providing a diagnostic basis for further treatment of arrhythmias (such as performing radiofrequency ablation surgery), or verifying the success of radiofrequency ablation treatment for arrhythmias.

[0003] In the prior art, doctors generally perform manual detection of arrhythmias based on intracardiac electrophysiological examination data, resulting in low detection efficiency and accuracy. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to provide a method, program product, electronic device and storage medium for processing intracardiac electrophysiological examination data based on artificial intelligence technology, so as to solve the technical problem in the prior art of low efficiency and accuracy in detecting arrhythmias based on intracardiac electrophysiological examination data.

[0005] In a first aspect, an embodiment of the present application provides a method for processing intracardiac electrophysiological examination data based on artificial intelligence technology, including: obtaining intracardiac electrophysiological examination data corresponding to an object to be detected; dynamically tracking the intracardiac electrophysiological examination data to obtain a first detection result corresponding to the object to be detected, wherein the first detection result represents arrhythmia information corresponding to the object to be detected.

[0006] In the above scheme, because intracardiac electrophysiological examination data is dynamic data obtained by recording cardiac electrical activity over a long period of time, arrhythmia information can be determined by dynamically tracking the intracardiac electrophysiological examination data. Compared with manual detection in the prior art, the artificial intelligence technology-based intracardiac electrophysiological examination data processing method provided in the embodiment of the present application can not only improve the efficiency of arrhythmia detection, but also avoid subjective misjudgment, thereby improving the accuracy of arrhythmia detection.

[0007] In an optional embodiment, the dynamic tracking of the intracardiac electrophysiological examination data to obtain the first detection result corresponding to the object to be detected includes: obtaining electrophysiological characteristics corresponding to the intracardiac electrophysiological examination data by dynamically tracking the intracardiac electrophysiological examination data, wherein the electrophysiological characteristics include at least one of the following: the interval circumference between multiple waveforms, the order between multiple reference electrodes, and the difference between the entrainment interval and the arrhythmia circumference; and determining the first detection result based on the electrophysiological characteristics. In the above scheme, since the intracardiac electrophysiological examination data is dynamic data obtained by recording cardiac electrical activity for a long time, arrhythmia information can be determined by detecting the electrophysiological characteristics corresponding to the intracardiac electrophysiological examination data.

[0008] In an optional embodiment, dynamically tracking the intracardiac electrophysiological examination data to obtain a first detection result corresponding to the subject to be detected includes: inputting the intracardiac electrophysiological examination data into a dynamic tracking model to obtain the first detection result output by the dynamic tracking model. In the above scheme, the dynamic tracking model can be used to automatically detect the intracardiac electrophysiological examination data. Compared with manual detection in the prior art, this not only improves the efficiency of arrhythmia detection, but also avoids subjective misjudgments, thereby improving the accuracy of arrhythmia detection.

[0009] In an optional embodiment, the dynamic tracking model includes a bidirectional long short-term memory network, and the bidirectional long short-term memory network includes a channel attention module. In the above scheme, a bidirectional long short-term memory network can be used to simultaneously capture the contextual feature information and the contextual feature information of the input data, so that the global features of the input data can be obtained more comprehensively; at the same time, a channel attention module is introduced to capture global information, so that the features between the input data can be obtained more comprehensively. Therefore, the detection accuracy of arrhythmias can be improved by using the dynamic tracking model provided in this application to detect intracardiac electrophysiological examination data.

[0010] In an optional embodiment, the method further includes: inputting the intracardiac electrophysiological examination data into an arrhythmia detection model to obtain a second detection result output by the detection model, wherein the second detection result indicates whether the subject to be detected has an arrhythmia. In the above scheme, the arrhythmia detection model can be used to automatically detect the intracardiac electrophysiological examination data. Compared with manual detection in the prior art, this not only improves the efficiency of arrhythmia detection, but also avoids subjective misjudgments, thereby improving the accuracy of arrhythmia detection.

[0011] In an optional embodiment, the arrhythmia detection model includes a convolutional neural network and a long short-term memory network. The convolutional neural network includes an input layer, a convolutional layer, and a pooling layer, and the long short-term memory network includes a long short-term memory layer. In this solution, the convolutional neural network can be used to extract local features of the input data, and the long short-term memory network can be used to capture long-term temporal dependencies between heartbeats in the input image, thereby improving the accuracy of arrhythmia detection.

[0012] In an optional embodiment, the inputting of the intracardiac electrophysiological examination data into the arrhythmia detection model to obtain the second detection result output by the arrhythmia detection model includes: utilizing the input layer to receive the intracardiac electrophysiological examination data; utilizing the convolutional layer to extract the local morphological features of the intracardiac electrophysiological examination data to obtain a corresponding feature map; utilizing the pooling layer to reduce the resolution of the feature map; utilizing the long short-term memory layer to capture the long-term temporal dependencies between heartbeats in the feature map to obtain the second detection result. In the above scheme, the convolutional neural network can be used to extract the local features of the input data, and the long short-term memory network can be used to capture the long-term temporal dependencies between heartbeats in the input image, thereby improving the accuracy of arrhythmia detection.

[0013] In an optional embodiment, the arrhythmia detection model includes an input layer and residual blocks, each of which includes a channel attention module. In this solution, the residual blocks can be used to extract multi-scale features of the input data, and the channel attention module can adaptively recalibrate channel-wise feature responses by explicitly modeling the interdependencies between channels, thereby improving the accuracy of arrhythmia detection.

[0014] In an optional embodiment, the step of inputting the intracardiac electrophysiological examination data into an arrhythmia detection model to obtain a second detection result output by the arrhythmia detection model includes: utilizing the input layer to receive the intracardiac electrophysiological examination data; and utilizing the residual block to extract multi-scale features of the intracardiac electrophysiological examination data to obtain the second detection result. In the above scheme, the residual block can be used to extract multi-scale features of the input data, and the channel attention module can adaptively recalibrate the channel-based feature response by explicitly modeling the interdependence between channels, thereby improving the accuracy of arrhythmia detection.

[0015] In an optional embodiment, the intracardiac electrophysiological examination method further includes: training the initial model using the following steps to obtain the arrhythmia detection model: acquiring sample data and sample labels corresponding to the sample data; inputting the sample data into the initial model to obtain sample prediction results output by the initial model; calculating label loss values ​​based on the sample labels and the sample prediction results, and optimizing the initial model using the label loss values ​​to obtain the arrhythmia detection model. In the above scheme, the sample data and sample labels can be used to train the deep learning network to obtain a trained arrhythmia detection model, which can then be used in the above arrhythmia detection model to detect intracardiac electrophysiological examination data.

[0016] In an optional embodiment, obtaining sample data and sample labels corresponding to the sample data includes: obtaining initial electrical signal data; inputting the initial electrical signal data into a generative adversarial network to obtain simulated electrical signal data output by the generative adversarial network; and determining the sample data and sample labels based on the initial electrical signal data and the simulated electrical signal data. In this solution, a generative adversarial network can be used to generate simulated electrical signal data for a specific disease, thereby resolving the problem of sample scarcity.

[0017] In an optional embodiment, the generative adversarial network is trained using cross-entropy loss and maximum mean difference loss. In the above scheme, the introduction of maximum mean difference loss to train the generative adversarial network can align the distribution differences of different data sources, thereby improving the cross-domain generalization ability of the generative adversarial network.

[0018] In an optional embodiment, the obtaining of intracardiac electrophysiological examination data corresponding to the object to be detected includes: reading a saved file corresponding to the intracardiac electrophysiological signal data; parsing the saved file according to the format definition of the saved file to obtain the intracardiac electrophysiological signal data; and performing vectorized processing on the intracardiac electrophysiological signal data to obtain the intracardiac electrophysiological examination data.

[0019] In an optional embodiment, the obtaining of intracardiac electrophysiological examination data corresponding to the object to be detected includes: reading intracardiac electrophysiological image data, wherein the intracardiac electrophysiological image data is generated based on intracardiac electrophysiological signal data; and performing vectorization processing on the intracardiac electrophysiological image data to obtain the intracardiac electrophysiological examination data.

[0020] In the second aspect, an embodiment of the present application provides a device for processing intracardiac electrophysiological examination data based on artificial intelligence technology, including: an acquisition module for acquiring intracardiac electrophysiological examination data corresponding to an object to be detected; a tracking module for dynamically tracking the intracardiac electrophysiological examination data to obtain a first detection result corresponding to the object to be detected, wherein the first detection result represents arrhythmia information corresponding to the object to be detected.

[0021] In the above scheme, because intracardiac electrophysiological examination data is dynamic data obtained by recording cardiac electrical activity over a long period of time, arrhythmia information can be determined by dynamically tracking the intracardiac electrophysiological examination data. Compared with manual detection in the prior art, the artificial intelligence technology-based intracardiac electrophysiological examination data processing method provided in the embodiment of the present application can not only improve the efficiency of arrhythmia detection, but also avoid subjective misjudgment, thereby improving the accuracy of arrhythmia detection.

[0022] In an optional embodiment, the tracking module is specifically configured to: dynamically track the intracardiac electrophysiological examination data to obtain electrophysiological characteristics corresponding to the intracardiac electrophysiological examination data, wherein the electrophysiological characteristics include at least one of the following: the period length of the interval between multiple waveforms, the order between multiple reference electrodes, and the difference between the post-entrainment interval and the arrhythmia period; and determine the first detection result based on the electrophysiological characteristics. In the above embodiment, since the intracardiac electrophysiological examination data is dynamic data obtained by recording cardiac electrical activity over a long period of time, arrhythmia information can be determined by detecting the electrophysiological characteristics corresponding to the intracardiac electrophysiological examination data.

[0023] In an optional embodiment, the tracking module is specifically configured to input the intracardiac electrophysiological examination data into a dynamic tracking model to obtain the first detection result output by the dynamic tracking model. In the above embodiment, the dynamic tracking model can be used to automatically detect the intracardiac electrophysiological examination data. Compared with manual detection in the prior art, this not only improves the efficiency of arrhythmia detection, but also avoids subjective misjudgments, thereby improving the accuracy of arrhythmia detection.

[0024] In an optional embodiment, the dynamic tracking model includes a bidirectional long short-term memory network, and the bidirectional long short-term memory network includes a channel attention module. In the above scheme, a bidirectional long short-term memory network can be used to simultaneously capture the contextual feature information and the contextual feature information of the input data, so that the global features of the input data can be obtained more comprehensively; at the same time, a channel attention module is introduced to capture global information, so that the features between the input data can be obtained more comprehensively. Therefore, the detection accuracy of arrhythmias can be improved by using the dynamic tracking model provided in this application to detect intracardiac electrophysiological examination data.

[0025] In an optional embodiment, the artificial intelligence-based intracardiac electrophysiological examination data processing device further includes an input module for inputting the intracardiac electrophysiological examination data into an arrhythmia detection model to obtain a second detection result output by the detection model, wherein the second detection result indicates whether the subject to be detected has an arrhythmia. In the above scheme, the arrhythmia detection model can be used to automatically detect the intracardiac electrophysiological examination data. Compared with manual detection in the prior art, this not only improves the efficiency of arrhythmia detection, but also avoids subjective misjudgments, thereby improving the accuracy of arrhythmia detection.

[0026] In an optional embodiment, the arrhythmia detection model includes a convolutional neural network and a long short-term memory network. The convolutional neural network includes an input layer, a convolutional layer, and a pooling layer, and the long short-term memory network includes a long short-term memory layer. In this solution, the convolutional neural network can be used to extract local features of the input data, and the long short-term memory network can be used to capture long-term temporal dependencies between heartbeats in the input image, thereby improving the accuracy of arrhythmia detection.

[0027] In an optional embodiment, the input module is specifically configured to: utilize the input layer to receive the intracardiac electrophysiological examination data; utilize the convolutional layer to extract local morphological features of the intracardiac electrophysiological examination data to obtain a corresponding feature map; utilize the pooling layer to reduce the resolution of the feature map; utilize the long short-term memory layer to capture the long-term temporal dependencies between heartbeats in the feature map to obtain the second detection result. In the above scheme, the convolutional neural network can be used to extract local features of the input data, and the long short-term memory network can be used to capture the long-term temporal dependencies between heartbeats in the input image, thereby improving the accuracy of arrhythmia detection.

[0028] In an optional embodiment, the arrhythmia detection model includes an input layer and residual blocks, each of which includes a channel attention module. In this solution, the residual blocks can be used to extract multi-scale features of the input data, and the channel attention module can adaptively recalibrate channel-wise feature responses by explicitly modeling the interdependencies between channels, thereby improving the accuracy of arrhythmia detection.

[0029] In an optional embodiment, the input module is specifically configured to: receive the intracardiac electrophysiological examination data using the input layer; and extract multi-scale features of the intracardiac electrophysiological examination data using the residual block to obtain the second detection result. In the above scheme, the residual block can be used to extract multi-scale features of the input data, and the channel attention module can improve the accuracy of arrhythmia detection by adaptively recalibrating channel-based feature responses by explicitly modeling the interdependencies between channels.

[0030] In an optional embodiment, the processing device for intracardiac electrophysiological examination data based on artificial intelligence technology further includes: a training module for training the initial model using the following steps to obtain the arrhythmia detection model: obtaining sample data and sample labels corresponding to the sample data; inputting the sample data into the initial model to obtain sample prediction results output by the initial model; calculating label loss values ​​based on the sample labels and the sample prediction results, and optimizing the initial model using the label loss values ​​to obtain the arrhythmia detection model. In the above scheme, the sample data and sample labels can be used to train the deep learning network to obtain a trained arrhythmia detection model, which can then be used for the above arrhythmia detection model to detect intracardiac electrophysiological examination data.

[0031] In an optional embodiment, the training module is further configured to: obtain initial electrical signal data; input the initial electrical signal data into a generative adversarial network to obtain simulated electrical signal data output by the generative adversarial network; and determine the sample data and the sample label based on the initial electrical signal data and the simulated electrical signal data. In this solution, a generative adversarial network can be used to generate simulated electrical signal data for specific diseases, thereby resolving the problem of sample scarcity.

[0032] In an optional embodiment, the generative adversarial network is trained using cross-entropy loss and maximum mean difference loss. In the above scheme, the introduction of maximum mean difference loss to train the generative adversarial network can align the distribution differences of different data sources, thereby improving the cross-domain generalization ability of the generative adversarial network.

[0033] In an optional embodiment, the acquisition module is specifically used to: read the saved file corresponding to the intracardiac electrophysiological signal data; parse the saved file according to the format definition of the saved file to obtain the intracardiac electrophysiological signal data; and perform vectorization processing on the intracardiac electrophysiological signal data to obtain the intracardiac electrophysiological examination data.

[0034] In an optional embodiment, the acquisition module is specifically used to: read intracardiac electrophysiological image data, wherein the intracardiac electrophysiological image data is generated based on intracardiac electrophysiological signal data; and perform vectorization processing on the intracardiac electrophysiological image data to obtain the intracardiac electrophysiological examination data.

[0035] In a third aspect, an embodiment of the present application provides a computer program product, comprising computer program instructions, which, when read and executed by a processor, execute the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology as described in the first aspect.

[0036] In a fourth aspect, an embodiment of the present application provides an electronic device comprising: a processor, a memory, and a bus; the processor and the memory communicate with each other through the bus; the memory stores computer program instructions that can be executed by the processor, and the processor calls the computer program instructions to execute the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology as described in the first aspect.

[0037] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores computer program instructions. When the computer program instructions are executed by a computer, the computer executes the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology as described in the first aspect.

[0038] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the following specifically cites the embodiments of the present application and provides a detailed description in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 A flowchart of a method for processing intracardiac electrophysiological examination data based on artificial intelligence technology provided in an embodiment of the present application;

[0041] Figure 2 A schematic diagram of a first intracardiac electrical examination image provided in an embodiment of the present application;

[0042] Figure 3 A schematic diagram of a second intracardiac electrical examination image provided in an embodiment of the present application;

[0043] Figure 4 A schematic diagram of a third intracardiac electrical examination image provided in an embodiment of the present application;

[0044] Figure 5 A schematic diagram of a fourth intracardiac electrical examination image provided in an embodiment of the present application;

[0045] Figure 6 A schematic diagram of a fifth intracardiac electrical examination image provided in an embodiment of the present application;

[0046] Figure 7 A schematic diagram of a sixth type of intracardiac electrical examination image provided in an embodiment of the present application;

[0047] Figure 8 A schematic diagram of a seventh intracardiac electrical examination image provided in an embodiment of the present application;

[0048] Figure 9 A schematic diagram of an eighth type of intracardiac electrical examination image provided in an embodiment of the present application;

[0049] Figure 10 This is a structural block diagram of a device for processing intracardiac electrophysiological examination data based on artificial intelligence technology provided in an embodiment of the present application;

[0050] Figure 11 This is a structural block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0051] In the prior art, doctors generally perform manual detection of arrhythmias based on intracardiac electrophysiological examination data, resulting in low detection efficiency and accuracy. Therefore, an embodiment of the present application provides a method for processing intracardiac electrophysiological examination data based on artificial intelligence technology. This method uses artificial intelligence technology to automatically measure, analyze and compare intracardiac electrophysiological examination data, and automatically determine the diagnosis based on the test results, thereby providing a basis for further treatment.

[0052] Intracardiac electrophysiology (ECP) examinations are inherently difficult and repetitive, requiring the AI ​​foundation needed to automate, precisely read and write data, and perform measurements. Research on AI for ECP not only fills a gap in this field both domestically and internationally, but also offers practical implications for improving the effectiveness and safety of arrhythmia treatment in clinical practice.

[0053] It should be noted that the electrophysiological examinations currently recognized in medicine cover a wide range of content and involve numerous related electrophysiological characteristics and phenomena. The embodiments of this application only introduce some of the electrophysiological detection characteristics by way of example. Other electrophysiological characteristics not introduced are also applicable to the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology provided in the embodiments of this application.

[0054] The following will describe the technical solutions in the embodiments of the present application in conjunction with the accompanying drawings. Figure 1 , Figure 1 This is a flowchart of a method for processing intracardiac electrophysiological examination data based on artificial intelligence technology provided in an embodiment of the present application. The method can be, but is not limited to, executed by an electronic device. Figure 11 The possible structure of the electronic device is shown. For details, please refer to the following Figure 11 The above-mentioned method for processing intracardiac electrophysiological examination data based on artificial intelligence technology may specifically include the following steps:

[0055] Step S101: Acquire intracardiac electrophysiological examination data corresponding to the subject to be detected.

[0056] Step S102: Dynamically track the intracardiac electrophysiological examination data to obtain a first detection result corresponding to the object to be detected.

[0057] Specifically, in the above step S101, the subject to be detected refers to the subject that needs to be detected. It is understandable that the subject can be a confirmed patient, a suspected patient, a healthy person, etc., and the embodiments of the present application do not specifically limit this. In addition, the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology provided in the embodiments of the present application can be applied not only when the subject to be detected has an arrhythmia attack, but also when the subject to be detected has not had an arrhythmia attack. Among them, the above-mentioned arrhythmia can include ventricular tachycardia and ventricular bradycardia.

[0058] Intracardiac electrophysiological examination data is various information about cardiac electrical activity obtained through intracardiac electrophysiological examination. For example, intracardiac electrophysiological examination data may include: electrical signals of various parts of the heart (such as: electrical activity signals of the atria and ventricles, electrical signals of the cardiac conduction system, etc.), cardiac electrophysiological parameters (such as: heart rate, refractory period, conduction velocity, etc.), etc.

[0059] Among them, the embodiment of the present application only briefly introduces the specific execution method of intracardiac electrophysiological examination: first, blood vessel puncture and electrode catheter placement are performed, so that the electrode catheter can be used to record the electrical signals of various parts of the heart; then different forms of stimulation signals are issued through the electrode catheter to try to induce arrhythmia in the subject to be tested; finally, the intracardiac electrophysiological examination data of the subject to be tested is recorded, such as: the frequency of arrhythmia, atrial wave morphology, ventricular wave morphology, morphology of the stimulation signal, etc.

[0060] It should be noted that the embodiments of this application do not impose specific limitations on the specific implementation methods for obtaining intracardiac electrophysiological examination data corresponding to the subject to be detected, and those skilled in the art may make appropriate adjustments based on actual circumstances. For example, intracardiac electrophysiological examination data sent by an external device may be received; or intracardiac electrophysiological examination data pre-stored in the cloud or locally may be read; or intracardiac electrophysiological electrical signals may be processed to obtain corresponding intracardiac electrophysiological examination data, etc.

[0061] Please refer to Figure 2 , Figure 2 This is a schematic diagram of the first intracardiac electrical examination image provided in an embodiment of the present application. Figure 2 The waveform of the stimulation wave (S) corresponding to the stimulation signal, the waveform of the atrial wave signal (A) corresponding to the atrium (Atrium), and the waveform of the ventricular wave signal (V) corresponding to the ventricle (Ventricle) are shown in FIG. Figure 2 The middle left image represents the intracardiac electrical examination image without stimulation. Figure 2 The middle image represents the intracardiac electrical examination image under atrial stimulation. Figure 2 The middle right figure shows the intracardiac electrical examination image of the ventricle under this condition.

[0062] It's important to note that, unlike simple surface ECG waveform detection (which identifies individual, independent waveforms), intracardiac electrical signal tracking focuses on maintaining continuity, ensuring the same waveform is consistently identified over time. Furthermore, during tracking, multiple objects can be detected, assigned unique numbers, and tracked throughout a dynamic, real-time sequence.

[0063] In step S102, the intracardiac electrophysiological examination data can be dynamically tracked. Dynamic tracking is used to capture the electrophysiological characteristics of the intracardiac electrophysiological signals over a certain period of time and automatically calculate index parameters such as peak values ​​and specific waveform intervals to obtain a first detection result corresponding to the subject to be detected. The first detection result represents arrhythmia information corresponding to the subject to be detected.

[0064] It should be noted that the embodiments of this application do not specifically limit the specific implementation of dynamic tracking, and those skilled in the art may make appropriate adjustments based on actual circumstances. For example, a neural network model may be used to dynamically track intracardiac electrophysiological examination data; or a filtering algorithm may be used to dynamically track intracardiac electrophysiological examination data; or a morphological analysis algorithm may be used to dynamically track intracardiac electrophysiological examination data, etc.

[0065] In addition, the embodiments of the present application do not specifically limit the specific implementation of the arrhythmia information, and those skilled in the art may make appropriate adjustments based on actual conditions. For example, the arrhythmia information may include arrhythmia type, arrhythmia nature, arrhythmia location, etc.

[0066] In the above scheme, because intracardiac electrophysiological examination data is dynamic data obtained by recording cardiac electrical activity over a long period of time, arrhythmia information can be determined by dynamically tracking the intracardiac electrophysiological examination data. Compared with manual detection in the prior art, the artificial intelligence technology-based intracardiac electrophysiological examination data processing method provided in the embodiment of the present application can not only improve the efficiency of arrhythmia detection, but also avoid subjective misjudgment, thereby improving the accuracy of arrhythmia detection.

[0067] Furthermore, based on the above embodiments, the embodiments of the present application can also track and process intracardiac electrophysiological examination data in real time, including but not limited to identifying newly generated waveforms, calculating the spacing between different interval circumferences of each waveform in real time, and combining the electrode sequence and various electrophysiological characteristics to provide multiple evidences for arrhythmia diagnosis.

[0068] Among them, compared with the processing of surface ECG data, the intracardiac electrophysiological examination data in the embodiment of the present application requires static data comparison when there is no arrhythmia or dynamic tracking analysis when an arrhythmia occurs under different electrophysiological stimulation parameters. The core is the measurement of different interval cycle lengths and the comparison of multiple reference electrode sequences, while the processing of surface ECG is simply the detection of waveform curves.

[0069] Based on the above content, as an implementation method, the above step S103 may specifically include the following steps:

[0070] Step 1) Dynamically tracking the intracardiac electrophysiological examination data to obtain electrophysiological features corresponding to the intracardiac electrophysiological examination data.

[0071] Step 2) Determine a first detection result based on the electrophysiological characteristics.

[0072] Specifically, in step 1) above, the electrophysiological characteristics may include at least one of the following: the circumference of the intervals between multiple waveforms, the order between multiple reference electrodes, and the difference between the post-entrainment interval and the circumference of the arrhythmia. It is understood that the above three electrophysiological characteristics are currently medically recognized electrophysiological characteristics related to intracardiac electrophysiological examination methods. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology in the embodiments of this application is not limited to the above three electrophysiological characteristics. Other electrophysiological characteristics related to intracardiac electrophysiological examination methods are also applicable to the embodiments of this application.

[0073] The multiple waveforms may include the waveforms of the stimulation wave, atrial wave signal, and ventricular wave signal corresponding to the stimulation signal. Interval length refers to the time interval between two specific waveforms on the electrocardiogram (ECG). Specifically, the interval length between the multiple waveforms mentioned above refers to the time interval between the stimulation signal, atrial wave signal, and ventricular wave signal. For example, in the case of S1S2 atrial program with incremental stimulation, the jump between the attack signal and the ventricular wave signal can be measured to determine whether it is greater than 50ms.

[0074] The multi-reference electrode sequence refers to the order or arrangement of multiple reference electrodes, which can be demonstrated through the relationship between waveforms. For example, during S1S2 programmed ventricular stimulation, it can be used to determine whether the atrial and ventricular waves are gradually prolonged or relatively fixed.

[0075] The post-pacing interval (PPI) is the time interval from the first arrhythmia-related electrical activity recorded after the termination of the pacing stimulation to the same electrical activity marker in the next arrhythmia cycle. The tachycardia cycle length (TCL) is the time interval between two consecutive arrhythmia electrical activities during an arrhythmia episode. For example, the success of pacing can be determined by determining whether the difference between the post-pacing interval and the tachycardia cycle length is greater than 110 ms.

[0076] The above electrophysiological characteristics are introduced using paroxysmal supraventricular tachycardia (PSVT) as an example. Paroxysmal supraventricular tachycardia is a paroxysmal, rapid arrhythmia, which can be divided into narrow and broad definitions. Broadly speaking, paroxysmal supraventricular tachycardia refers to arrhythmias caused by etiologies above the His bundle, including atrial fibrillation, atrial tachycardia, and other large categories of tachycardias; while narrowly speaking, paroxysmal supraventricular tachycardia is divided into two types: atrioventricular nodal reentrant tachycardia (AVNRT) (referred to as dual pathway) and atrioventricular reentrant tachycardia (AVRT) (referred to as bypass pathway).

[0077] In step 2) above, a first test result can be determined based on the electrophysiological characteristics determined in step 1) above. As an implementation method, the present application embodiment can combine artificial intelligence technology to automatically measure, analyze, and compare intracardiac electrophysiological examination data to automatically determine a diagnosis based on the first test result, thereby providing a basis for further treatment. The following example introduces an implementation method for determining the first test result based on electrophysiological characteristics.

[0078] Please refer to Figure 3 , Figure 3 This is a schematic diagram of the second intracardiac electrical examination image provided in the embodiment of the present application. When the subject to be detected is not experiencing tachycardia and is stimulated by atrial S1S2 stimulation, if Figure 3 As shown in the figure, if the difference between adjacent A2-H2 is greater than 50ms (that is, the jump phenomenon occurs), it can be considered that Figure 3 The corresponding object to be detected may have dual atrioventricular node pathways.

[0079] Atrial S1S2 stimulation involves administering a baseline stimulus, S1, followed by an additional stimulus, S2, at an appropriate interval. The time interval between S1 and S2 is varied to observe the cardiac electrophysiological response. A2 refers to the atrial depolarization wave induced by S2 stimulation, while H2 refers to the His bundle depolarization wave induced by S2 stimulation. The A2-H2 difference is the time interval difference between these two waves. As an implementation, the A2-H2 difference between adjacent atrial and ventricular waves can be used to characterize the interval between adjacent atrial and ventricular waves.

[0080] in, Figure 3 The middle left image shows the intracardiac electrical examination image corresponding to the atrial S1S2 stimulation of 360 degrees, at which time A2-H2 is 285; Figure 3The right side of the middle image is the corresponding intracardiac electrical examination image when the atrial S1S2 stimulation is 350, and the A2-H2 is 370. It can be seen that the difference between adjacent A2-H2 (A2-H2 on the left and A2-H2 on the right) is 85. Therefore, the difference between adjacent A2-H2 is greater than 50ms, which can be considered Figure 3 The corresponding object to be detected may have dual atrioventricular node pathways.

[0081] Please refer to Figure 4 , Figure 4 This is a schematic diagram of the third intracardiac electrical examination image provided in the embodiment of the present application. When continuous stimulation is used to induce tachycardia in the subject to be tested, if Figure 4 As shown in the figure, VA overlaps and takes the shape of a central vertical line (i.e., it changes centrally), then it can be considered that Figure 4 The corresponding object to be detected may have dual atrioventricular node pathways.

[0082] In normal cardiac electrophysiological activity, the depolarization of the ventricles and atria occurs sequentially, each generating independent waves of electrical activity. However, in certain circumstances, the conduction of electrical impulses between two different pathways within the atrioventricular node can cause the ventricles and atria to depolarize almost simultaneously, manifesting as VA coincidence on an electrocardiogram (ECG) or cardiac electrophysiological recording. This occurs when the depolarization waves of the ventricles and atria merge together, forming a graph that looks like a vertical line.

[0083] Central changes here refer to the relatively uniform and symmetrical nature of VA overlap, suggesting that electrical impulses are uniformly conducted from the heart's center (primarily the atrioventricular node region) to the periphery, rather than through eccentric abnormal conduction pathways (such as atrioventricular accessory pathways). This is a typical electrophysiological feature of tachycardia caused by dual atrioventricular node pathways and distinguishes it from eccentric conduction changes caused by atrioventricular accessory pathways.

[0084] Please refer to Figure 5 , Figure 5 This is a schematic diagram of the fourth intracardiac electrical examination image provided in the embodiment of the present application. When the subject to be detected has a tachycardia attack and is stimulated by ventricular S1S1 entrainment stimulation, if Figure 5 As shown in the figure, when the difference between the calculated entrainment interval and the tachycardia cycle is greater than 110ms, it can be considered that Figure 5 The corresponding object to be detected may have dual atrioventricular node pathways.

[0085] Among them, ventricular S1S1 entrainment stimulation refers to the continuous administration of two identical electrical stimulations to the ventricles, and the stimulation frequency is usually slightly faster than the frequency of tachycardia; through this entrainment stimulation, the rhythm of tachycardia can be temporarily reorganized to better observe and analyze the characteristics of cardiac electrical activity.

[0086] Please refer to Figure 6 , Figure 6 This is a schematic diagram of the fifth intracardiac electrical examination image provided in the embodiment of the present application. When the subject to be detected is not experiencing tachycardia and is stimulated by ventricular S1S2 stimulation, if Figure 6 As shown, VA does not prolong conduction, so it can be considered that Figure 6 The corresponding subject to be detected may have an atrioventricular accessory pathway.

[0087] Among them, no decremental prolonged conduction means that under the stimulation of ventricle S1S2, the conduction from ventricle to atrium does not show the decremental characteristics of the normal atrioventricular conduction system, and the conduction time and speed do not extend or slow down accordingly with the change of stimulation.

[0088] Please refer to Figure 7 , Figure 7 This is a schematic diagram of the sixth intracardiac electrical examination image provided in the embodiment of the present application. When the subject to be detected is not experiencing tachycardia and is stimulated by transventricular S1S2 stimulation, if Figure 7 As shown, VA conducts eccentrically, and the A wave mapping sequence of CS1-2 electrodes is earlier than that of other electrodes, so it can be considered that Figure 7 The corresponding subject to be detected may have a left free wall atrioventricular accessory pathway.

[0089] Among them, under normal circumstances, the heart's electrical conduction is from the atria to the ventricles, but in certain pathological conditions, electrical impulses can be conducted reversely from the ventricles to the atria, that is, VA conduction; eccentric conduction means that this VA conduction does not spread evenly to various parts of the atria along the normal atrioventricular conduction system, but presents an eccentric, non-uniform conduction pattern, suggesting that there may be an abnormal conduction path.

[0090] The coronary sinus (CS) is a venous structure located behind the heart, surrounded by numerous electrodes that record cardiac electrical activity. By comparing the temporal order of the A waves recorded by different electrodes, the origin and conduction direction of atrial electrical activity can be determined. When the A wave mapping sequence recorded by the CS1-2 electrodes is earlier than that of other electrodes, it indicates that the atrial tissue near the coronary sinus ostium is activated first after ventricular stimulation, suggesting that the atrioventricular accessory pathway may be located in the left free wall. This is because the atrioventricular accessory pathway in the left free wall rapidly conducts electrical impulses from the ventricles to the atria near the coronary sinus, resulting in the premature appearance of the A wave in this area.

[0091] Please refer to Figure 8 , Figure 8 This is a schematic diagram of the seventh intracardiac electrical examination image provided in the embodiment of the present application. When the subject to be detected has a tachycardia attack and is stimulated by ventricular S1S1 entrainment stimulation and CS9-10 electrode entrainment, if Figure 8As shown in Figure 2, when the difference between the calculated entrainment interval and the tachycardia cycle is less than 30ms, it can be considered that CS9-10 is on the reentrant circuit. Figure 8 The corresponding object to be detected may have right atrial tachycardia.

[0092] Please refer to Figure 9 , Figure 9 This is a schematic diagram of the eighth intracardiac electrical examination image provided in the embodiment of the present application. Figure 9 As shown in the figure, when the sinus node recovery time (SNRT) exceeds the normal range, it can be considered Figure 9 The corresponding subject to be detected may have sick sinus syndrome.

[0093] Among them, the sinus node recovery time refers to the method of overspeeding the inhibition of the sinus node to temporarily stop the sinus node from working, and then measuring the time required from the cessation of stimulation to the resumption of autonomous pacing activity of the sinus node; generally speaking, the sinus node recovery time of normal adults is less than 1500 milliseconds.

[0094] It should be understood that, first, the first detection result can be determined based on one electrophysiological feature or multiple electrophysiological features. When the first detection result is determined based on multiple electrophysiological features, the first detection result can be determined as an abnormal condition when any one electrophysiological feature meets the abnormal condition; alternatively, the first detection result can be determined as an abnormal condition when a large number of electrophysiological features meet the abnormal condition, etc. This embodiment of the application does not specifically limit this.

[0095] For example, based on Figure 3 ,or Figure 4 ,or Figure 5 Any of the following situations is used to judge the dual pathway of the atrioventricular node. Figure 3 ,or Figure 4 ,or Figure 5 The electrophysiological characteristics shown in FIG1 can determine that the first detection result indicates that the subject to be detected has a dual atrioventricular node pathway; or, based on the electrophysiological characteristics shown in FIG1 Figure 3 、 Figure 4 as well as Figure 5 The situation shown is used to judge the dual pathway of the atrioventricular node, that is, if there are Figure 3 、 Figure 4 as well as Figure 5 The electrophysiological characteristics shown can determine that the first test result indicates that the subject to be tested has a dual atrioventricular node pathway. Figure 3 、 Figure 4 as well as Figure 5A comprehensive judgment is made to further confirm whether the subject to be tested has a dual atrioventricular node pathway, thereby improving the reliability of the diagnosis of the dual atrioventricular node pathway.

[0096] Secondly, the various electrophysiological characteristics proposed in the embodiments of this application are merely examples. Other electrophysiological characteristics that can be obtained during intracardiac electrophysiological examinations can all be used in the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology provided in this application, and the embodiments of this application do not make any specific limitations on this.

[0097] In the above scheme, since the intracardiac electrophysiological examination data is dynamic data obtained by recording cardiac electrical activity over a long period of time, arrhythmia information can be determined by detecting the electrophysiological characteristics corresponding to the intracardiac electrophysiological examination data.

[0098] Furthermore, based on the above embodiment, as another implementation, the above step S103 may specifically include the following steps:

[0099] The intracardiac electrophysiological examination data is input into the dynamic tracking model to obtain a first detection result output by the dynamic tracking model.

[0100] Specifically, the dynamic tracking model is used to determine arrhythmia information in a patient based on intracardiac electrophysiological examination data. As an implementation, the dynamic tracking model can utilize a time-series deep learning algorithm to capture waveform changes in the stimulation signal in real time, automatically calculating key parameters such as peak spacing and waveform slope, and correlating these with disease symptom diagnosis.

[0101] It should be noted that the embodiments of this application do not specifically limit the specific implementation of the dynamic tracking model, and those skilled in the art may make appropriate adjustments based on actual circumstances. For example, the dynamic tracking model may adopt the Deep Simple Online and Realtime Tracking (Deep SORT) model, the Multi-Domain Convolutional Neural Network (MDNet), etc.

[0102] The Deep SORT model is an advanced version of the SORT algorithm, which originally relied on bounding box association and Kalman filtering for tracking. The Deep SORT model improves on this by integrating deep appearance features, enabling it to distinguish between visually similar objects. This capability enables it to track objects even after temporary occlusion or sudden changes in trajectory.

[0103] MDNet is a deep learning-based object tracking algorithm inspired by the R-CNN object detection network. It leverages multiple domain-specific networks to adapt to diverse tracking environments. MDNet uses convolutional neural networks to extract object appearance features and classify them across different tracking domains. During initialization, MDNet samples multiple candidate regions and fine-tunes its neural network for the specific object being tracked. The tracker continuously updates itself using domain adaptation techniques, making it robust to appearance changes and occlusions.

[0104] It is understandable that the dynamic tracking model in the embodiment of the present application can be applied to all currently recognized intracardiac electrophysiological examination methods and contents, and related electrophysiological characteristic phenomena in medicine.

[0105] By inputting the intracardiac electrophysiological examination data obtained in step S101 into the dynamic tracking model, a first detection result output by the dynamic tracking model can be obtained, wherein the first detection result represents the arrhythmia information corresponding to the subject to be detected.

[0106] In the above scheme, a dynamic tracking model can be used to automatically detect intracardiac electrophysiological examination data. Compared with manual detection in the existing technology, it can not only improve the detection efficiency of arrhythmias, but also avoid subjective misjudgment and thus improve the detection accuracy of arrhythmias.

[0107] Furthermore, based on the above embodiment, the dynamic tracking model may include a bidirectional long short-term memory network, and the bidirectional long short-term memory network may include a channel attention module.

[0108] Specifically, the Long Short-Term Memory (LSTM) network is an extension of the Recurrent Neural Network (RNN) and is used to address the problem of long-term dependency loss in time series. The key to the LSTM network lies in the cell state of the LSTM unit, which adds or deletes information to the cell state through a gate structure, which selectively allows information to pass through.

[0109] The bidirectional long short-term memory network adopted in the embodiment of the present application can solve the problem of being unable to process subsequent context information; the bidirectional long short-term memory network has both a forward LSTM and a reverse LSTM in the hidden layer. The forward LSTM captures the feature information of the previous context, while the reverse LSTM captures the feature information of the following context, and then finally obtains the global context information by fusing the captured previous feature information and the following feature information.

[0110] In addition, since the channel attention module can capture global information in one step and obtain the features between input data more comprehensively, the embodiment of the present application introduces the attention mechanism.

[0111] In the above scheme, a bidirectional long short-term memory network can be used to simultaneously capture the contextual feature information of the input data and the contextual feature information, thereby more comprehensively obtaining the global features of the input data; at the same time, a channel attention module is introduced to capture global information, thereby more comprehensively obtaining the features between the input data. Therefore, using the dynamic tracking model provided by this application to detect intracardiac electrophysiological examination data can improve the accuracy of arrhythmia detection.

[0112] Furthermore, based on the above embodiment, the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology provided in the embodiment of the present application may further include the following steps:

[0113] The intracardiac electrophysiological examination data is input into the arrhythmia detection model to obtain a second detection result output by the detection model.

[0114] Specifically, in the above steps, the arrhythmia detection model is used to classify whether the subject to be detected has arrhythmia based on the intracardiac electrophysiological examination data. It should be noted that the embodiment of the present application does not specifically limit the specific implementation of the arrhythmia detection model, and those skilled in the art can make appropriate adjustments according to actual conditions. For example, the arrhythmia detection model can adopt a convolutional neural network model (CNN), a long short-term memory network (LSTM), a ResNet, etc.

[0115] It is understandable that the arrhythmia detection model in the embodiment of the present application can be applicable to all currently recognized medical intracardiac electrophysiological examination methods and contents, and related electrophysiological characteristic phenomena.

[0116] By inputting the intracardiac electrophysiological examination data obtained in the above steps into the above arrhythmia detection model, a second detection result output by the arrhythmia detection model can be obtained. The second detection result indicates whether the subject to be tested has an arrhythmia. In one embodiment, the arrhythmia may include ventricular and supraventricular arrhythmias.

[0117] As an implementation method, the above steps can be performed before step S102; through step S101, the above steps and step S102, a method for processing intracardiac electrophysiological examination data based on artificial intelligence technology is established according to relevant principles of computer engineering, and the effects of intelligent analysis, automatic diagnosis, automatic warning or prompting are achieved through automatic tracking, automatic identification, and automatic measurement methods, so as to improve the efficiency of arrhythmia detection, avoid subjective misjudgment, reduce manual burden, and thereby assist diagnosis, increase surgical safety, and improve surgical prognosis.

[0118] In the above scheme, the arrhythmia detection model can be used to automatically detect the intracardiac electrophysiological examination data. Compared with the manual detection in the existing technology, it can not only improve the detection efficiency of arrhythmia, but also avoid subjective misjudgment and thus improve the detection accuracy of arrhythmia.

[0119] Furthermore, based on the above embodiments, as an implementation method, the arrhythmia detection model may include a convolutional neural network and a long short-term memory network, wherein the convolutional neural network may include an input layer, a convolution layer and a pooling layer, and the long short-term memory network includes a long short-term memory layer.

[0120] Specifically, the input layer is used to receive preprocessed trace data signals.

[0121] The convolutional layer is used to extract local morphological features (such as QRS waves, T waves, and other key electrophysiological features). The number of convolution kernels can be set to 64, 128, or 256, and the kernel size can be set to 3 or 5. Smaller kernels can capture high-frequency details, and nonlinear activation functions such as ReLU can be used.

[0122] The pooling layer is used to reduce the resolution of the temporal dimension and enhance translation invariance. The pooling window size can be set to 2, and the step size can be set to 2.

[0123] The LSTM layer is used to capture long-term temporal dependencies between heartbeats (such as RR interval variations). When selecting parameters, one can first try a single-layer, 16-unit LSTM model, then gradually increase the number of hidden layers and units until performance no longer improves. As an implementation, a bidirectional LSTM layer can be used to enhance contextual information. Alternatively, techniques such as cross-validation can be used to determine optimal parameters and ultimately select the model with the best performance.

[0124] In the above scheme, convolutional neural networks can be used to extract local features of input data, and long short-term memory networks can be used to capture the long-term temporal dependencies between heartbeats in the input image, thereby improving the accuracy of arrhythmia detection.

[0125] Furthermore, based on the above embodiment, the step of inputting the intracardiac electrophysiological examination data into the arrhythmia detection model to obtain the second detection result output by the arrhythmia detection model may specifically include the following steps:

[0126] Step 1) Utilize the input layer to receive intracardiac electrophysiological examination data.

[0127] Step 2) The convolutional layer is used to extract the local morphological features of the intracardiac electrophysiological examination data and obtain the corresponding feature map.

[0128] Step 3) Use the pooling layer to reduce the resolution of the feature map.

[0129] Step 4) Use the long short-term memory layer to capture the long-term temporal dependency between heartbeats in the feature map to obtain the second detection result.

[0130] In this arrhythmia detection model, the convolutional layer extracts image features, while the long short-term memory layer processes these feature sequences. Ultimately, the model can classify the input layer vector data. As an implementation, an attention mechanism can be introduced into this arrhythmia detection model to help the long short-term memory layer better focus on important parts of the image.

[0131] During the training process, the arrhythmia detection model can calculate the loss function based on the input image and the corresponding label, and update the parameters through the back propagation algorithm.

[0132] In the above scheme, convolutional neural networks can be used to extract local features of input data, and long short-term memory networks can be used to capture the long-term temporal dependencies between heartbeats in the input image, thereby improving the accuracy of arrhythmia detection.

[0133] Furthermore, based on the above embodiment, as another implementation method, the arrhythmia detection model may include an input layer and a residual block, and each residual block may include a channel attention module.

[0134] Specifically, the input layer is used to receive preprocessed trace data signals.

[0135] Residual blocks serve as the backbone of ResNet. These can be implemented using basic residual blocks (with a kernel size of 3 in the convolutional layer) and bottleneck residual blocks (1x1 convolution → 3x3 convolution → 1x1 convolution), followed by a normalization layer (also known as a batch normalization layer) with a ReLU activation function. The residual network in this arrhythmia detection model consists of multiple stages, each stacked with multiple residual blocks. This deep network extracts multi-scale features while maintaining training stability.

[0136] Within each residual block, a channel attention module can be inserted after the convolution operation to adaptively recalibrate channel-wise feature responses by explicitly modeling the interdependencies between channels. Specifically, this module automatically learns the importance of each channel and then uses this importance to enhance useful features and suppress features that are less useful for the task at hand.

[0137] In the above scheme, the residual block can be used to extract multi-scale features of the input data, and the channel attention module can adaptively recalibrate the channel-wise feature responses by explicitly modeling the interdependence between channels, thereby improving the detection accuracy of arrhythmias.

[0138] Furthermore, based on the above embodiment, the step of inputting the intracardiac electrophysiological examination data into the arrhythmia detection model to obtain the second detection result output by the arrhythmia detection model may specifically include the following steps:

[0139] Step 1) Utilize the input layer to receive intracardiac electrophysiological examination data.

[0140] Step 2) The residual block is used to extract multi-scale features of the intracardiac electrophysiological examination data to obtain a second detection result.

[0141] In the above scheme, the residual block can be used to extract multi-scale features of the input data, and the channel attention module can adaptively recalibrate the channel-wise feature responses by explicitly modeling the interdependence between channels, thereby improving the detection accuracy of arrhythmias.

[0142] Furthermore, based on the above embodiment, the intracardiac electrophysiological examination method provided in the embodiment of the present application can also use the following steps to train the initial model to obtain an arrhythmia detection model:

[0143] Step 1) Obtain sample data and sample labels corresponding to the sample data.

[0144] Step 2) Input the sample data into the initial model to obtain the sample prediction results output by the initial model.

[0145] Step 3) Calculate the label loss value based on the sample label and the sample prediction result, and use the label loss value to optimize the initial model to obtain the arrhythmia detection model.

[0146] In the above scheme, the sample data and sample labels can be used to train the deep learning network to obtain a trained arrhythmia detection model, which can then be used to detect intracardiac electrophysiological examination data using the above arrhythmia detection model.

[0147] Furthermore, based on the above embodiment, the step of obtaining sample data and sample labels corresponding to the sample data may specifically include the following steps:

[0148] Step 1) Obtain initial electrical signal data.

[0149] Step 2) Input the initial electrical signal data into the generative adversarial network to obtain the simulated electrical signal data output by the generative adversarial network.

[0150] Step 3) Determine sample data and sample labels based on the initial electrical signal data and the simulated electrical signal data.

[0151] Specifically, in step 1) above, as an implementation, data from various wearable health monitoring devices, such as smart bracelets and smart watches, can be collected and combined with clinical diagnostic data to pre-train the deep learning model. Because data from diverse sources has varying data distributions, a domain adaptation approach can be employed. This involves pre-processing the data from multiple sources and then inputting the data, adapted to account for these data distribution differences, into the model.

[0152] In step 2) above, as an implementation method, a Wasserstein Generative Adversarial Network (GAN) can be used. The core of the Wasserstein GAN is to replace the loss function of the traditional GAN ​​with a function that minimizes the Wasserstein distance. The Wasserstein distance, also known as the Earth-Mover (EM) distance, is used to measure the difference between the data distribution and the distribution created by the generator. The EM distance can provide a smoother gradient signal for the generator because it can measure how much "mass" needs to be moved to transform one distribution into another and the distance moved. This distance is more effective when the two distributions do not overlap or only overlap slightly.

[0153] In the above scheme, a generative adversarial network can be used to generate simulated electrical signal data of specific diseases, thereby solving the problem of sample scarcity.

[0154] Furthermore, based on the above embodiment, the generative adversarial network can be trained using cross entropy loss and maximum mean difference loss.

[0155] Specifically, for the above-mentioned Maximum Mean Discrepancy (MMD) loss, the GAN synthetic data can be regarded as the source domain and the real clinical data as the target domain. Since the two types of data have different distributions but the same classification tasks, the labeled data of the source domain can be used to train the model, so that it performs well on unlabeled clinical data.

[0156] The maximum mean difference loss mitigates domain shift by minimizing the difference in feature distributions between the source and target domains, mapping them to the same feature space. The "maximum" in maximum mean difference refers to maximizing the mean difference within a specific function space. The goal of maximizing mean difference is to find a function f within a function space that best distinguishes the two distributions P and Q. This means finding a function that maximizes the expected difference between the two distributions.

[0157] As an implementation method, the present application provides a domain adaptation model, which may include a feature extractor, a task classifier, and an MMD alignment module. The feature extractor inputs the original vector data and outputs a high-level feature representation, with the goal of extracting domain-invariant features; the task classifier inputs the feature representation f and outputs the task prediction result, with the goal of minimizing the classification loss of the source domain; the MMD alignment module inputs the features of the source and target domains, and outputs the MMD distance value, with the goal of minimizing the MMD distance of the feature distributions of the two domains.

[0158] In the above scheme, the maximum mean difference loss is introduced to train the generative adversarial network, which can align the distribution differences of different data sources, thereby improving the cross-domain generalization ability of the generative adversarial network.

[0159] Furthermore, based on the above embodiment, as an implementation method, the above step S101 may specifically include the following steps:

[0160] Step 1) Read the saved file corresponding to the intracardiac electrophysiological signal data.

[0161] Step 2) Parse the saved file according to the format definition of the saved file to obtain intracardiac electrophysiological signal data.

[0162] Step 3) Perform vector quantization on the intracardiac electrophysiological signal data to obtain intracardiac electrophysiological examination data.

[0163] Furthermore, based on the above embodiment, as another implementation method, the above step S101 may specifically include the following steps:

[0164] Step 1) reading intracardiac electrophysiological image data, wherein the intracardiac electrophysiological image data is generated based on intracardiac electrophysiological signal data.

[0165] Step 2) vectorizes the intracardiac electrophysiological image data to obtain intracardiac electrophysiological examination data.

[0166] In an embodiment of the present application, first, deep learning image classification technology is used to construct a pattern recognition and precise classification model for static images of electrophysiological detection signals, and target tracking technology for time series data is used to dynamically track and capture the electrophysiological characteristics of stimulation signals within a certain time period, and indicator parameters such as peak values ​​and distances between specific waveforms are automatically measured and calculated, and automatic diagnosis of symptoms is performed based on the electrophysiological characteristics and indicator parameters.

[0167] Secondly, in the medical field, there are factors that hinder deep learning models from achieving good results, such as the scarcity and difficulty in collecting case data, the difficulty in manually labeling data features, the uneven distribution of data on different diseases, the diverse sources of data, and the differences in the distribution of data from different sources and people of different age groups. Therefore, generative adversarial networks are used to generate case data and specific disease data to provide more sufficient data for model training; at the same time, domain adaptive algorithms are introduced to align the distribution differences between data from different sources and different age groups, thereby improving the performance of the overall model.

[0168] Therefore, the embodiment of the present application provides a method for processing intracardiac electrophysiological examination data based on artificial intelligence technology, which processes the intracardiac electrophysiological examination data through artificial intelligence technology to obtain detection results, thereby improving the detection efficiency and accuracy of arrhythmias.

[0169] In addition, the application of artificial intelligence technology in electrophysiological examinations can also include assisting in radiofrequency ablation operation reminders. For example, during the slow pathway ablation of atrioventricular nodal reentrant tachycardia, if different proportions of atrioventricular wave AV information are found to be transmitted downward, an early warning signal can be issued to remind the operator to avoid further complications of atrioventricular conduction block in advance.

[0170] Please refer to Figure 10 , Figure 10A structural block diagram of a device for processing intracardiac electrophysiological examination data based on artificial intelligence technology provided in an embodiment of the present application, the device 1000 for processing intracardiac electrophysiological examination data based on artificial intelligence technology includes: an acquisition module 1001, used to obtain intracardiac electrophysiological examination data corresponding to an object to be detected; a tracking module 1002, used to dynamically track the intracardiac electrophysiological examination data if the second detection result indicates that the object to be detected has the arrhythmia, so as to obtain a first detection result corresponding to the object to be detected, wherein the first detection result indicates arrhythmia information corresponding to the object to be detected.

[0171] In the above scheme, because intracardiac electrophysiological examination data is dynamic data obtained by recording cardiac electrical activity over a long period of time, arrhythmia information can be determined by dynamically tracking the intracardiac electrophysiological examination data. Compared with manual detection in the prior art, the artificial intelligence technology-based intracardiac electrophysiological examination data processing method provided in the embodiment of the present application can not only improve the efficiency of arrhythmia detection, but also avoid subjective misjudgment, thereby improving the accuracy of arrhythmia detection.

[0172] Furthermore, based on the above embodiment, the tracking module 1002 is specifically used to: obtain electrophysiological characteristics corresponding to the intracardiac electrophysiological examination data by dynamically tracking the intracardiac electrophysiological examination data, wherein the electrophysiological characteristics include at least one of the following: the interval circumference between multiple waveforms, the order between multiple reference electrodes, and the difference between the post-entrainment interval and the arrhythmia circumference; determine the first detection result based on the electrophysiological characteristics.

[0173] In the above scheme, since the intracardiac electrophysiological examination data is dynamic data obtained by recording cardiac electrical activity over a long period of time, arrhythmia information can be determined by detecting the electrophysiological characteristics corresponding to the intracardiac electrophysiological examination data.

[0174] Furthermore, based on the above embodiment, the tracking module 1002 is specifically configured to: input the intracardiac electrophysiological examination data into a dynamic tracking model to obtain the first detection result output by the dynamic tracking model.

[0175] In the above scheme, a dynamic tracking model can be used to automatically detect intracardiac electrophysiological examination data. Compared with manual detection in the existing technology, it can not only improve the detection efficiency of arrhythmias, but also avoid subjective misjudgment and thus improve the detection accuracy of arrhythmias.

[0176] Furthermore, based on the above embodiment, the dynamic tracking model includes a bidirectional long short-term memory network, and the bidirectional long short-term memory network includes a channel attention module.

[0177] In the above scheme, a bidirectional long short-term memory network can be used to simultaneously capture the contextual feature information of the input data and the contextual feature information, thereby more comprehensively obtaining the global features of the input data; at the same time, a channel attention module is introduced to capture global information, thereby more comprehensively obtaining the features between the input data. Therefore, using the dynamic tracking model provided by this application to detect intracardiac electrophysiological examination data can improve the accuracy of arrhythmia detection.

[0178] Furthermore, based on the above embodiments, the device 1000 for processing intracardiac electrophysiological examination data based on artificial intelligence technology also includes: an input module for inputting the intracardiac electrophysiological examination data into an arrhythmia detection model to obtain a second detection result output by the detection model, wherein the second detection result represents whether the object to be detected has arrhythmia.

[0179] In the above scheme, the arrhythmia detection model can be used to automatically detect the intracardiac electrophysiological examination data. Compared with the manual detection in the existing technology, it can not only improve the detection efficiency of arrhythmia, but also avoid subjective misjudgment and thus improve the detection accuracy of arrhythmia.

[0180] Furthermore, based on the above embodiment, the arrhythmia detection model includes a convolutional neural network and a long short-term memory network, the convolutional neural network includes an input layer, a convolution layer and a pooling layer, and the long short-term memory network includes a long short-term memory layer.

[0181] In the above scheme, convolutional neural networks can be used to extract local features of input data, and long short-term memory networks can be used to capture the long-term temporal dependencies between heartbeats in the input image, thereby improving the accuracy of arrhythmia detection.

[0182] Furthermore, based on the above embodiments, the input module is specifically used to: use the input layer to receive the intracardiac electrophysiological examination data; use the convolutional layer to extract the local morphological features of the intracardiac electrophysiological examination data to obtain a corresponding feature map; use the pooling layer to reduce the resolution of the feature map; use the long short-term memory layer to capture the long-term timing dependency between heartbeats in the feature map to obtain the second detection result.

[0183] In the above scheme, convolutional neural networks can be used to extract local features of input data, and long short-term memory networks can be used to capture the long-term temporal dependencies between heartbeats in the input image, thereby improving the accuracy of arrhythmia detection.

[0184] Furthermore, based on the above embodiment, the arrhythmia detection model includes an input layer and a residual block, and each residual block includes a channel attention module.

[0185] In the above scheme, the residual block can be used to extract multi-scale features of the input data, and the channel attention module can adaptively recalibrate the channel-wise feature responses by explicitly modeling the interdependence between channels, thereby improving the detection accuracy of arrhythmias.

[0186] Furthermore, based on the above embodiment, the input module is specifically used to: use the input layer to receive the intracardiac electrophysiological examination data; use the residual block to extract multi-scale features of the intracardiac electrophysiological examination data to obtain the second detection result.

[0187] In the above scheme, the residual block can be used to extract multi-scale features of the input data, and the channel attention module can adaptively recalibrate the channel-wise feature responses by explicitly modeling the interdependence between channels, thereby improving the detection accuracy of arrhythmias.

[0188] Furthermore, on the basis of the above-mentioned embodiment, the processing device 1000 for intracardiac electrophysiological examination data based on artificial intelligence technology also includes: a training module, which is used to train the initial model using the following steps to obtain the arrhythmia detection model: obtaining sample data and sample labels corresponding to the sample data; inputting the sample data into the initial model to obtain the sample prediction results output by the initial model; calculating the label loss value based on the sample label and the sample prediction result, and using the label loss value to optimize the initial model to obtain the arrhythmia detection model.

[0189] In the above scheme, the sample data and sample labels can be used to train the deep learning network to obtain a trained arrhythmia detection model, which can then be used to detect intracardiac electrophysiological examination data using the above arrhythmia detection model.

[0190] Furthermore, based on the above embodiments, the training module is also used to: obtain initial electrical signal data; input the initial electrical signal data into the generative adversarial network to obtain simulated electrical signal data output by the generative adversarial network; and determine the sample data and the sample label based on the initial electrical signal data and the simulated electrical signal data.

[0191] In the above scheme, a generative adversarial network can be used to generate simulated electrical signal data of specific diseases, thereby solving the problem of sample scarcity.

[0192] Furthermore, based on the above embodiment, the generative adversarial network is trained using cross entropy loss and maximum mean difference loss.

[0193] In the above scheme, the maximum mean difference loss is introduced to train the generative adversarial network, which can align the distribution differences of different data sources, thereby improving the cross-domain generalization ability of the generative adversarial network.

[0194] Furthermore, based on the above embodiments, the acquisition module 1001 is specifically used to: read the saved file corresponding to the intracardiac electrophysiological signal data; parse the saved file according to the format definition of the saved file to obtain the intracardiac electrophysiological signal data; and perform vectorization processing on the intracardiac electrophysiological signal data to obtain the intracardiac electrophysiological examination data.

[0195] Furthermore, based on the above embodiments, the acquisition module 1001 is specifically used to: read intracardiac electrophysiological image data, wherein the intracardiac electrophysiological image data is generated based on intracardiac electrophysiological signal data; and perform vectorization processing on the intracardiac electrophysiological image data to obtain the intracardiac electrophysiological examination data.

[0196] Please refer to Figure 11 , Figure 11 This is a structural block diagram of an electronic device provided in an embodiment of the present application. The electronic device 1100 includes: at least one processor 1101, at least one communication interface 1102, at least one memory 1103 and at least one communication bus 1104. Among them, the communication bus 1104 is used to realize direct connection and communication between these components, the communication interface 1102 is used to communicate signaling or data with other node devices, and the memory 1103 stores machine-readable instructions executable by the processor 1101. When the electronic device 1100 is running, the processor 1101 communicates with the memory 1103 through the communication bus 1104, and when the machine-readable instructions are called by the processor 1101, the above-mentioned method for processing intracardiac electrophysiological examination data based on artificial intelligence technology is executed.

[0197] For example, the processor 1101 of an embodiment of the present application reads a computer program from the memory 1103 through the communication bus 1104 and executes the computer program to implement the following method: obtaining intracardiac electrophysiological examination data corresponding to the object to be detected; dynamically tracking the intracardiac electrophysiological examination data to obtain a first detection result corresponding to the object to be detected, wherein the first detection result represents the arrhythmia information corresponding to the object to be detected.

[0198] Among them, the processor 1101 includes one or more, which can be an integrated circuit chip with signal processing capabilities. The above-mentioned processor 1101 can be a general-purpose processor, including a central processing unit (CPU), a micro control unit (MCU), a network processor (NP) or other conventional processors; it can also be a special-purpose processor, including a neural network processor (NPU), a graphics processing unit (GPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. Moreover, when there are multiple processors 1101, some of them can be general-purpose processors and the other part can be special-purpose processors.

[0199] The memory 1103 includes one or more, which may be, but is not limited to, random access memory (RAM), read only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0200] I understand. Figure 11 The structure shown is for illustration only. The electronic device 1100 may further include Figure 11 More or fewer components than shown, or with Figure 11 Different configurations shown. Figure 11Each component shown in the figure can be implemented using hardware, software, or a combination thereof. In the embodiments of the present application, the electronic device 1100 can be, but is not limited to, a physical device such as a desktop computer, a laptop computer, a smartphone, a smart wearable device, an in-vehicle device, or a virtual device such as a virtual machine. In addition, the electronic device 1100 does not necessarily have to be a single device, but can also be a combination of multiple devices, such as a server cluster, etc.

[0201] The present application also provides a computer program product, including a computer program stored on a computer-readable storage medium. The computer program includes computer program instructions. When the computer program instructions are executed by a computer, the computer can perform the steps of the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology in the above-mentioned embodiment, for example, including: Step S101: Acquire intracardiac electrophysiological examination data corresponding to a subject to be detected. Step S102: Dynamically track the intracardiac electrophysiological examination data to obtain a first detection result corresponding to the subject to be detected.

[0202] An embodiment of the present application also provides a computer-readable storage medium, which stores computer program instructions. When the computer program instructions are executed by a computer, the computer executes the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology as described in the aforementioned method embodiment.

[0203] In the embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0204] In addition, the units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0205] Furthermore, the functional modules in each embodiment of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0206] It should be noted that if the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0207] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any actual relationship or order between these entities or operations.

[0208] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for processing intracardiac electrophysiological examination data based on artificial intelligence technology, characterized in that: include: Acquiring intracardiac electrophysiological examination data corresponding to the object to be detected; The intracardiac electrophysiological examination data is dynamically tracked to obtain a first detection result corresponding to the object to be detected, wherein the first detection result represents arrhythmia information corresponding to the object to be detected.

2. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to claim 1, characterized in that: The dynamically tracking the intracardiac electrophysiological examination data to obtain a first detection result corresponding to the object to be detected includes: By dynamically tracking the intracardiac electrophysiological examination data, electrophysiological characteristics corresponding to the intracardiac electrophysiological examination data are obtained, wherein the electrophysiological characteristics include: the period length between multiple waveforms, the order between multiple reference electrodes, and the difference between the post-entrainment interval and the arrhythmia period; The first detection result is determined according to the electrophysiological characteristic.

3. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to claim 1, characterized in that: The dynamically tracking the intracardiac electrophysiological examination data to obtain a first detection result corresponding to the object to be detected includes: The intracardiac electrophysiological examination data is input into a dynamic tracking model to obtain the first detection result output by the dynamic tracking model.

4. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to any one of claims 1 to 3, characterized in that: The method further comprises: The intracardiac electrophysiological examination data is input into an arrhythmia detection model to obtain a second detection result output by the detection model, wherein the second detection result indicates whether the subject to be detected has arrhythmia.

5. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to claim 4, characterized in that: The arrhythmia detection model includes a convolutional neural network and a long short-term memory network, the convolutional neural network includes an input layer, a convolution layer and a pooling layer, and the long short-term memory network includes a long short-term memory layer; and / or, The arrhythmia detection model includes an input layer and a residual block, each residual block includes a channel attention module; and / or, The intracardiac electrophysiological examination method further comprises: The initial model is trained using the following steps to obtain the arrhythmia detection model: Obtaining sample data and sample labels corresponding to the sample data; Inputting the sample data into the initial model to obtain a sample prediction result output by the initial model; A label loss value is calculated based on the sample label and the sample prediction result, and the initial model is optimized using the label loss value to obtain the arrhythmia detection model.

6. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to claim 5, characterized in that: If the arrhythmia detection model includes the convolutional neural network and the long short-term memory network, the convolutional neural network includes the input layer, the convolutional layer, and the pooling layer, and the long short-term memory network includes the long short-term memory layer, then inputting the intracardiac electrophysiological examination data into the arrhythmia detection model to obtain a second detection result output by the arrhythmia detection model includes: Utilizing the input layer to receive the intracardiac electrophysiological examination data; Extracting local morphological features of the intracardiac electrophysiological examination data using the convolutional layer to obtain a corresponding feature map; Reducing the resolution of the feature map using the pooling layer; Using the long short-term memory layer to capture the long-term temporal dependency between heartbeats in the feature map, to obtain the second detection result; and / or, If the arrhythmia detection model includes the input layer and the residual block, and each residual block includes the channel attention module, then inputting the intracardiac electrophysiological examination data into the arrhythmia detection model to obtain a second detection result output by the arrhythmia detection model includes: Utilizing the input layer to receive the intracardiac electrophysiological examination data; The residual block is used to extract multi-scale features of the intracardiac electrophysiological examination data to obtain the second detection result.

7. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to claim 4, characterized in that: The acquiring of sample data and sample labels corresponding to the sample data includes: acquiring initial electrical signal data; Inputting the initial electrical signal data into a generative adversarial network to obtain simulated electrical signal data output by the generative adversarial network; The sample data and the sample label are determined according to the initial electrical signal data and the simulated electrical signal data.

8. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to claim 7, characterized in that: The generative adversarial network is trained using cross entropy loss and maximum mean difference loss.

9. The method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to any one of claims 1 to 3, characterized in that: The step of obtaining intracardiac electrophysiological examination data corresponding to the subject to be detected includes: Read the saved file corresponding to the intracardiac electrophysiological signal data; Parsing the saved file according to the format definition of the saved file to obtain the intracardiac electrophysiological signal data; Performing vector quantization processing on the intracardiac electrophysiological signal data to obtain the intracardiac electrophysiological examination data; and / or, The step of obtaining intracardiac electrophysiological examination data corresponding to the subject to be detected includes: reading intracardiac electrophysiological image data, wherein the intracardiac electrophysiological image data is generated based on intracardiac electrophysiological signal data; Vectorization processing is performed on the intracardiac electrophysiological image data to obtain the intracardiac electrophysiological examination data.

10. A computer program product, characterized in that The method comprises computer program instructions, which, when read and executed by a processor, executes the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology as described in any one of claims 1 to 9.

11. An electronic device, characterized in that: include: processor, memory, and bus; The processor and the memory communicate with each other via the bus; The memory stores computer program instructions that can be executed by the processor, and the processor calls the computer program instructions to execute the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology as described in any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer program instructions, which, when executed by a computer, enable the computer to execute the method for processing intracardiac electrophysiological examination data based on artificial intelligence technology according to any one of claims 1 to 9.

Images