Heart electrophysiological analysis method, system and device and storage medium
By integrating the multimodal attention mechanism and ViT model of electrocardiogram, hypercardiography and clinical index data, the problems of insufficient accuracy and insufficient interpretation of cardiac electrophysiological analysis are solved, and higher analysis accuracy and interpretability are achieved .
Patent Information
- Application Number
- CN202411565580.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-05
- Publication Date
- 2025-02-07
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art is not accurate enough in cardiac electrophysiological analysis, and it is difficult to effectively integrate electrocardiogram, hypercardiac image and clinical index data, and the deep learning model is insufficiently interpretable, making it difficult to gain clinical trust.
By obtaining electrocardiogram data, hyperoscillation image data and clinical index data, preprocessing and feature extraction are performed, feature fusion is fusion using ViT model and multimodal attention mechanism, fusion features are generated, and analysis is performed through the decoder, and interpretability is designed. Modules to highlight key features.
It improves the accuracy of cardiac electrophysiological analysis, enhances the interpretability of the model, can learn cardiac electrophysiological data more comprehensively, provide more accurate and reliable analysis results, and gain the trust of clinicians.
Smart Images
Figure CN119385573A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of smart medical technology, and in particular to a method, system, device and computer-readable storage medium for cardiac electrophysiological analysis. Background Art
[0002] Arrhythmia is a common cardiovascular disease that seriously threatens human health. Cardiac electrophysiological examination is the "gold standard" for diagnosing and treating arrhythmia. By recording and analyzing the heart's electrical activity data, it can accurately locate the lesion, identify the cause, evaluate the condition, and guide the treatment plan.
[0003] In traditional technology, ECG analysis mainly relies on manual interpretation by doctors, which has the following shortcomings:
[0004] 1. Highly subjective and dependent on doctor’s experience: Arrhythmia types are complex and diverse, and manual interpretation is easily affected by the doctor’s subjective factors, making it difficult to ensure the consistency and accuracy of the diagnostic results.
[0005] 2. Inefficiency and difficulty in processing massive data: With the popularization of wearable devices, cardiac electrophysiological data has exploded, and traditional manual analysis methods are unable to cope with the processing needs of massive data.
[0006] 3. Low information utilization and neglect of multimodal data: Traditional methods mainly focus on the ECG signal itself, ignoring other physiological signals (such as blood pressure, blood oxygen, etc.) and patient clinical information, making it impossible to conduct a comprehensive integrated analysis.
[0007] In recent years, deep learning has made breakthrough progress in the field of medical image analysis, such as image recognition, object detection, and semantic segmentation. However, the application of deep learning in the field of cardiac electrophysiology still has the following problems:
[0008] 1. Insufficient accuracy: Cardiac electrophysiological signals have the characteristics of high sampling rate, high noise, and large individual differences. The existing analysis models are limited in scale and it is difficult to fully utilize massive data. The performance in effectively extracting features and learning rules is unsatisfactory, resulting in insufficient accuracy of the analysis results. In addition, the existing models only analyze the ECG signal itself, without effectively integrating other physiological signals and patient clinical information, and lack the ability to integrate multimodal data. The accuracy and reliability of the analysis results still need to be improved.
[0009] 2. Lack of explainability and difficulty in gaining clinical trust: Deep learning models are often considered “black boxes” and their decision-making process is difficult to explain and gain the trust of clinicians, which hinders their application in clinical diagnosis. Summary of the invention
[0010] The embodiments of the present application solve the problem of insufficient accuracy of cardiac electrophysiological analysis methods in the prior art by providing a method, system, device and computer-readable storage medium for cardiac electrophysiological analysis.
[0011] In a first aspect, an embodiment of the present application provides a cardiac electrophysiological analysis method, characterized in that the method comprises the following steps:
[0012] Acquire electrocardiogram data, echocardiogram data and clinical index data, and pre-process the acquired data;
[0013] Extracting features of pre-processed electrocardiogram data, echocardiogram data and clinical index data;
[0014] Fusing the features of the electrocardiogram data, echocardiogram data, and clinical index data to achieve interaction of different modality information and generate fusion features;
[0015] Decoding the fused features into fused output data;
[0016] The cardiac electrophysiological performance is analyzed based on the fused output data.
[0017] Preferably, the method further comprises:
[0018] Analyze the weight distribution of the electrocardiogram data, echocardiogram data, and clinical indicator data features to identify key features;
[0019] Visualize the key features to generate intuitive heat maps or importance rankings;
[0020] Key features are highlighted through weight distribution to generate structured analysis reports.
[0021] Preferably, the clinical indicator data include the patient's age, gender, blood pressure, heart rate and / or blood oxygen saturation.
[0022] Preferably, the features of the pre-processed electrocardiogram data, echocardiogram data and clinical index data are specifically:
[0023] The ViT model is used as a feature extractor. The ViT model includes an electrocardiogram encoder, an echocardiogram image encoder and a clinical indicator encoder, which are respectively used to encode the preprocessed electrocardiogram data, echocardiogram image data and clinical indicator data into vectors and extract features from the vectors.
[0024] Preferably, the features of the electrocardiogram data, echocardiogram data and clinical indicator data are fused to achieve the interaction of different modal information and generate fusion features, specifically:
[0025] The self-attention mechanism is used to learn the feature relationships within the electrocardiogram data and echocardiogram image data respectively, and the fully connected layer is used to learn the feature relationships within the clinical indicator data. The cross-modal attention mechanism is used to calculate the correlation between the data feature relationships of each modality, and attention weights are assigned to weight the features of different modalities to generate fused features containing the most relevant information in each modality.
[0026] Preferably, the analysis of cardiac electrophysiological performance includes identifying the type of arrhythmia, identifying the myocardial infarction area and / or predicting cardiac function indicators.
[0027] In a second aspect, an embodiment of the present application provides a cardiac electrophysiological analysis system, the system comprising:
[0028] A data acquisition and preprocessing module is used to acquire electrocardiogram data, echocardiogram image data and clinical index data, and preprocess the acquired data;
[0029] A feature extraction module is used to extract features of pre-processed electrocardiogram data, echocardiogram data and clinical index data;
[0030] A feature fusion module, used to fuse the features of the electrocardiogram data, echocardiogram data and clinical index data, realize the interaction of different modal information, and generate fusion features;
[0031] A decoder, used for decoding the fused features into fused output data;
[0032] a task-specific head for analyzing cardiac electrophysiological performance based on the fused output data;
[0033] The task-specific header includes any one or more of the following:
[0034] a classification head, used to identify the type of arrhythmia;
[0035] Segmentation head, used to identify the myocardial infarction area;
[0036] Regression heads are used to predict cardiac function indices.
[0037] Preferably, the system further comprises:
[0038] The explainability module is used to analyze the weight distribution of each feature in the feature fusion module and identify key features; visualize the key features to generate an intuitive heat map or importance ranking; highlight the key features through weight distribution and generate a structured analysis report.
[0039] In a third aspect, an embodiment of the present application provides a cardiac electrophysiological analysis device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned cardiac electrophysiological analysis method when executing the program.
[0040] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the steps of the above-mentioned cardiac electrophysiological analysis method are implemented.
[0041] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:
[0042] 1. High accuracy: Thanks to the large-scale Transformer architecture and multimodal learning mechanism, the method proposed in this invention can learn cardiac electrophysiological data more comprehensively and deeply, and then accurately analyze cardiac electrophysiological performance.
[0043] 2. Strong interpretability: By designing an interpretability module, it can highlight key features and generate analysis and explanation reports of visual sources, providing a reliable basis for doctors' clinical decision-making and improving doctors' trust and acceptance.
[0044] 3. Superior comprehensive performance: By making full use of massive learning data, it can effectively extract features and learning rules, and at the same time integrate multimodal data. It has comprehensive functions such as identifying the type of arrhythmia, identifying the area of myocardial infarction, and predicting cardiac function indicators, and the analysis results are accurate and reliable.
[0045] 4. Wide application: The method of the present invention can be applied to a variety of clinical scenarios, such as wearable devices, telemedicine platforms, hospital information systems, etc., to provide auxiliary decision support for diagnosis, treatment and prognosis in the field of heart. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.
[0047] Figure 1 This is a flow chart of the cardiac electrophysiological analysis method in Example 1 of the present application;
[0048] Figure 2 This is a schematic diagram of the feature fusion process in Example 1 of the present application;
[0049] Figure 3 This is a schematic diagram of the explainability module in Example 1 of the present application;
[0050] Figure 4This is a structural block diagram of the cardiac electrophysiological analysis system in the second embodiment of the present application;
[0051] Figure 5 is the ROC curve of the model obtained in Example 3 of the present application;
[0052] Figure 6 This is a structural block diagram of the cardiac electrophysiological analysis device in Example 7 of the present application.
[0053] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0054] The embodiments of the present application solve the problem of insufficient accuracy of cardiac electrophysiological analysis methods in the prior art by providing a method, system, device and computer-readable storage medium for cardiac electrophysiological analysis.
[0055] In order to better understand the above technical solution, exemplary embodiments will be described in detail here, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are only examples of devices consistent with some aspects of the present application as detailed in the attached claims.
[0056] Embodiment 1
[0057] Figure 1 This is a flow chart of the cardiac electrophysiological analysis method provided in Example 1 of the present application. The cardiac electrophysiological analysis method comprises the following steps:
[0058] Step S1: acquiring electrocardiogram data, echocardiogram data and clinical index data, and preprocessing the acquired data;
[0059] Among them, the electrocardiogram is ECG (electrocardiogram) data from medical institutions such as hospitals and clinics, including different types of electrocardiograms, such as standard electrocardiograms, electrocardiogram signals, electrocardiogram waveforms, etc.
[0060] An echocardiogram, i.e., a cardiac ultrasound image, is image data from an echocardiogram device of a medical institution, including a cardiac ultrasound image, a cardiac ultrasound image signal, and the like.
[0061] Clinical indicators are clinical indicator data from patients, including age, gender, blood pressure, heart rate, blood oxygen saturation, etc.
[0062] In a preferred embodiment, the data preprocessing includes data standardization, denoising, and segmentation processing. Data preprocessing can improve the training efficiency and generalization ability of the model.
[0063] Standardization processing: Standardize the ECG and echocardiogram image data from different sources, such as adjusting the image size to a uniform size, converting the data to a uniform format, and normalizing the data.
[0064] Denoising: removing noise from data, such as using filters to remove noise from images and using statistical methods to remove outliers from data.
[0065] Segmentation processing: divide the data into segments or image blocks of a set size, such as dividing the electrocardiogram data into segments of a fixed length, using image segmentation technology to divide the echocardiogram data into image blocks of a fixed size, and extracting the heart area.
[0066] Step S2: extracting features of the pre-processed electrocardiogram data, echocardiogram data and clinical index data;
[0067] Specifically, a Vision Transformer (ViT) model is used as a feature extractor. The ViT model includes an electrocardiogram encoder, an echocardiogram image encoder, and a clinical indicator encoder, which are respectively used to encode the preprocessed electrocardiogram data, echocardiogram image data, and clinical indicator data into vectors and extract features from the vectors.
[0068] This embodiment uses the ultra-large-scale Transformer architecture as the model basis, and adopts a multi-head attention mechanism through an encoder module with billions of parameters to capture long-range dependencies and complex patterns in cardiac electrophysiological signals and fully mine the potential information of massive cardiac electrophysiological data.
[0069] Step S3: fusing the features of the electrocardiogram data, echocardiogram data and clinical index data to achieve interaction of different modal information and generate fusion features;
[0070] The specific process is as follows: the feature relationships within the ECG data and echocardiogram data are learned through the self-attention mechanism, the feature relationships within the clinical indicator data are learned through the fully connected layer, the correlation between the data feature relationships of each modality is calculated through the cross-modal attention mechanism, and attention weights are assigned to weight the features of different modalities to generate fusion features containing the most relevant information in each modality, such as Figure 2 shown.
[0071] By introducing a multimodal learning mechanism, multi-source heterogeneous data such as electrocardiograms, echocardiograms, and clinical indicators are mapped to a unified feature space for joint training and analysis. The relationship between features of different modalities is learned through a cross-modal attention mechanism to achieve information complementarity and fusion.
[0072] Step S4: decoding the fused features into fused output data;
[0073] Specifically, the ViT model includes a decoder, and the decoder is used to decode the fused features into fused output data.
[0074] Step S5: analyzing the cardiac electrophysiological performance based on the fused output data, wherein the analysis of the cardiac electrophysiological performance includes identifying the type of arrhythmia, identifying the myocardial infarction area and / or predicting cardiac function indicators.
[0075] Specifically, a task-specific header is designed and connected to the decoder. The task-specific header includes any one or more of the following:
[0076] Classification Head: Used to identify the type of arrhythmia, such as tachycardia, bradycardia, arrhythmia, etc.
[0077] Segmentation head: used to identify myocardial infarction areas, such as myocardial infarction area, myocardial infarction boundary, etc.
[0078] Regression head: used to predict cardiac function indicators, such as cardiac function index, cardiac function score, etc.
[0079] In a preferred embodiment, the cardiac electrophysiological analysis method further includes: designing an interpretability module based on an attention mechanism to enhance the transparency and credibility of the model.
[0080] Specifically, if Figure 3 As shown, the specific method is as follows:
[0081] Analyze the weight distribution of the cross-modal attention mechanism and identify key features; the key features refer to ECG signal segments, echocardiogram regions and / or clinical indicators that play a key role (have a large weight) in model decision-making;
[0082] The key features are visualized to generate intuitive heat maps or importance rankings to help doctors understand the reasoning process of the model. For example, Grad-CAM technology can be used to generate heat maps of electrocardiograms and echocardiograms, and the key areas of concern of the model can be highlighted through visualization results, such as abnormal electrocardiogram waveforms, abnormal heart structures, etc.
[0083] By highlighting key features through weight distribution, a structured diagnostic analysis report is generated to provide a reliable basis for clinical decision-making.
[0084] In addition, the above-mentioned ultra-large-scale Transformer model can be pre-trained using a two-stage strategy. Specifically:
[0085] Phase 1: Self-supervised pre-training
[0086] Data enhancement: Use data enhancement techniques, such as random cropping, random rotation, random scaling, etc., to enhance the generalization ability of the model;
[0087] Self-supervised learning: Use self-supervised learning techniques such as image generation, image classification, etc. to train the model.
[0088] Stage 2: Supervised Fine-tuning
[0089] Supervised learning: Use supervised learning techniques, such as image classification, image segmentation, etc. to train the model;
[0090] Multimodal data: Use multimodal data for fine-tuning, such as electrocardiogram, echocardiogram images, clinical indicators, etc.
[0091] Embodiment 2
[0092] Based on the same inventive concept as the cardiac electrophysiological analysis method in the aforementioned embodiment 1, this embodiment also provides a cardiac electrophysiological analysis system, such as Figure 4 As shown, the system comprises:
[0093] Data acquisition and preprocessing module: used to acquire electrocardiogram data, echocardiogram image data and clinical index data, and preprocess the acquired data;
[0094] Feature extraction module: used to extract features of pre-processed electrocardiogram data, echocardiogram data and clinical index data;
[0095] Feature fusion module: fuses the features of the electrocardiogram data, echocardiogram data and clinical index data to achieve interaction of different modal information and generate fusion features;
[0096] Decoder: used for decoding the fused features into fused output data;
[0097] Task-specific head: used to analyze cardiac electrophysiological performance based on the fused output data, including identifying the type of arrhythmia, identifying the myocardial infarction area and / or predicting cardiac function indicators, etc.
[0098] Furthermore, the feature extraction module includes an electrocardiogram encoder, an echocardiogram image encoder and a clinical indicator encoder, which are respectively used to encode the preprocessed electrocardiogram data, echocardiogram image data and clinical indicator data into vectors and extract features from the vectors.
[0099] The task-specific header includes any one or more of the following:
[0100] Classification head: used to identify the type of arrhythmia, such as tachycardia, bradycardia, arrhythmia, etc.
[0101] Segmentation head: used to identify the myocardial infarction area, such as the size of the myocardial infarction area, the myocardial infarction boundary, etc.
[0102] Regression head: used to predict cardiac function indicators, such as cardiac function index, cardiac function score, etc.
[0103] In a preferred embodiment, the cardiac electrophysiological analysis system further comprises:
[0104] The explainability module is used to analyze the weight distribution of each feature in the feature fusion module and identify key features; visualize the key features to generate an intuitive heat map or importance ranking; highlight the key features through weight distribution and generate a structured analysis report.
[0105] The key features refer to ECG signal segments, echocardiogram regions and / or clinical indicators that play a key role (have a larger weight) in model decision-making.
[0106] The explainability module can help doctors understand the reasoning process of the model, enhance doctors' trust, and provide a reliable basis for clinical decision-making.
[0107] The various specific processes and specific examples of the cardiac electrophysiological analysis method in the aforementioned embodiment 1 are also applicable to the cardiac electrophysiological analysis system of this embodiment. Through the detailed description of embodiment 1, those skilled in the art can clearly know the implementation method of the cardiac electrophysiological analysis system in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here.
[0108] Embodiment 3
[0109] Application of cardiac electrophysiological analysis system in arrhythmia classification.
[0110] Based on the cardiac electrophysiological analysis system provided in Example 2, the model input is electrocardiogram, echocardiogram and clinical index data; the task-specific head is a classification head, which is used to output the probability distribution of different arrhythmia types. Arrhythmia types include: tachycardia, bradycardia, and arrhythmia.
[0111] During model training, the first stage uses a large-scale training sample set to perform self-supervised pre-training on the ViT model; the second stage uses a multimodal arrhythmia dataset to perform supervised fine-tuning on the model to optimize classification performance.
[0112] The trained model is then tested using an independent validation sample set.
[0113] Figure 5 The ROC curve of the model obtained in this implementation is shown. The ROC curve is a common tool used to evaluate the performance of a binary classification model. Figure 5 In the figure, the horizontal axis False Positive Rate represents the false positive rate (FPR), and the vertical axis True Positive Rate represents the true positive rate (TPR). The area under the curve (AUC) is an important indicator of the ROC curve, which indicates the ability of the model to distinguish between positive and negative samples.
[0114] from Figure 5 It can be seen that the AUC value (Training AUC) of the model in this implementation on the training sample set is 0.866, and the AUC value (Validation AUC) on the validation sample set is 0.731. This shows that the model obtained in this implementation has a high discrimination ability on the training sample set and still maintains good performance on the validation sample set.
[0115] Although the AUC value on the validation sample set is slightly lower than that on the training sample set, this may be due to the difference in data distribution between the validation sample set and the training sample set. However, the AUC value on the validation sample set is still higher than 0.7, indicating that the model obtained in this implementation has good generalization ability.
[0116] In summary, the cardiac electrophysiological analysis system provided in this embodiment has excellent performance and effect in arrhythmia classification.
[0117] Embodiment 4
[0118] Application of cardiac electrophysiological analysis system in the prediction of myocardial infarction.
[0119] Based on the cardiac electrophysiological analysis system provided in Example 2, the model input is electrocardiogram, echocardiogram image and clinical index data; the task-specific head is a segmentation head, which is used to output the predicted probability of the boundary of the myocardial infarction area.
[0120] Use interpretable models such as decision trees to learn the model's decision boundaries and convert them into human-understandable rules, such as "If the patient is older than 60 years old and the ST segment of the electrocardiogram is significantly elevated, the risk of myocardial infarction is higher."
[0121] Experiments show that the cardiac electrophysiological analysis system provided in this embodiment has excellent performance and effect in predicting myocardial infarction, and can be used to predict the risk of acute myocardial infarction in patients in the future, providing auxiliary support for doctors' clinical decision-making.
[0122] Embodiment 5
[0123] Application of cardiac electrophysiological analysis system in the prediction of cardiac function indicators.
[0124] Based on the cardiac electrophysiological analysis system provided in Example 2, the model input is electrocardiogram, echocardiogram image and clinical index data; the task-specific head is a regression head, which is used to output the prediction results of cardiac function indicators.
[0125] Cardiac function indicators include ejection fraction, stroke volume, cardiac output, cardiac index, myocardial strain, etc.
[0126] Experiments show that the cardiac electrophysiological analysis system provided in this embodiment has good performance and effect in predicting cardiac function indicators, and provides auxiliary support for doctors' clinical decision-making.
[0127] Embodiment 6
[0128] Comprehensive system for cardiac electrophysiological analysis.
[0129] Based on the cardiac electrophysiological analysis system provided in Example 2, the model input is electrocardiogram, echocardiogram image and clinical indicator data; the task-specific heads include three, namely: a classification head, used to output the probability distribution of different arrhythmia types; a segmentation head, used to output the predicted probability of the boundary of the myocardial infarction area; a regression head, used to output the predicted results of cardiac function indicators.
[0130] The system provided in this embodiment has the performance of the systems described in Embodiments 3, 4 and 5. It has strong comprehensive capabilities.
[0131] Embodiment 7
[0132] Based on the same inventive concept as the cardiac electrophysiological analysis method in the aforementioned embodiment 1, this embodiment also provides a cardiac electrophysiological analysis device, such as Figure 6 As shown, a computer program is stored thereon, and when the program is executed by a processor, the steps of the cardiac electrophysiological analysis method described in Example 1 are implemented.
[0133] Among them, Figure 6 In the embodiment of the present invention, a bus architecture (represented by bus 300) is shown, which may include any number of interconnected buses and bridges, and bus 300 links various circuits including one or more processors represented by processor 302 and memory represented by memory 304. Bus 300 may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are not further described herein. Bus interface 306 provides an interface between bus 300 and receiver 301 and transmitter 303. Receiver 301 and transmitter 303 may be the same element, namely a transceiver, which provides a unit for communicating with various other devices over a transmission medium.
[0134] The processor 302 is responsible for managing the bus 300 and general processing, while the memory 304 may be used to store data used by the processor 302 when performing operations.
[0135] Embodiment 8
[0136] Based on the same inventive concept as the cardiac electrophysiological analysis method in the aforementioned embodiment 1, this embodiment further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the cardiac electrophysiological analysis method described in embodiment 1 are implemented.
[0137] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0138] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0139] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0141] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.
[0142] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
Claims
1. A cardiac electrophysiological analysis method, characterized in that: The method comprises the following steps: Acquire electrocardiogram data, echocardiogram data and clinical index data, and pre-process the acquired data; Extracting features of pre-processed electrocardiogram data, echocardiogram data and clinical index data; Fusing the features of the electrocardiogram data, echocardiogram data, and clinical index data to achieve interaction of different modality information and generate fusion features; Decoding the fused features into fused output data; The cardiac electrophysiological performance is analyzed based on the fused output data.
2. The cardiac electrophysiological analysis method according to claim 1, characterized in that: The method further comprises: Analyze the weight distribution of the electrocardiogram data, echocardiogram data, and clinical indicator data features to identify key features; Visualize the key features to generate intuitive heat maps or importance rankings; Key features are highlighted through weight distribution to generate structured analysis reports.
3. The cardiac electrophysiological analysis method according to claim 1, characterized in that: The clinical indicator data include the patient's age, gender, blood pressure, heart rate and / or blood oxygen saturation.
4. The cardiac electrophysiological analysis method according to claim 1, characterized in that: The features of the extracted pre-processed electrocardiogram data, echocardiogram data and clinical index data are specifically: The ViT model is used as a feature extractor. The ViT model includes an electrocardiogram encoder, an echocardiogram image encoder and a clinical indicator encoder, which are respectively used to encode the preprocessed electrocardiogram data, echocardiogram image data and clinical indicator data into vectors and extract features from the vectors.
5. The cardiac electrophysiological analysis method according to claim 1, characterized in that: The features of the electrocardiogram data, echocardiogram data and clinical index data are fused to achieve the interaction of different modal information and generate fusion features, specifically: The self-attention mechanism is used to learn the feature relationships within the electrocardiogram data and echocardiogram image data respectively, and the fully connected layer is used to learn the feature relationships within the clinical indicator data. The cross-modal attention mechanism is used to calculate the correlation between the data feature relationships of each modality, and attention weights are assigned to weight the features of different modalities to generate fused features containing the most relevant information in each modality.
6. The cardiac electrophysiological analysis method according to claim 1, characterized in that: The analysis of cardiac electrophysiological performance includes identifying the type of arrhythmia, identifying the myocardial infarction area and / or predicting cardiac function indicators.
7. A cardiac electrophysiological analysis system, characterized in that: The system comprises: A data acquisition and preprocessing module is used to acquire electrocardiogram data, echocardiogram image data and clinical index data, and preprocess the acquired data; A feature extraction module is used to extract features of pre-processed electrocardiogram data, echocardiogram data and clinical index data; A feature fusion module, used to fuse the features of the electrocardiogram data, echocardiogram data and clinical index data, realize the interaction of different modal information, and generate fusion features; A decoder, used for decoding the fused features into fused output data; a task-specific head for analyzing cardiac electrophysiological performance based on the fused output data; The task-specific header includes any one or more of the following: a classification head, used to identify the type of arrhythmia; Segmentation head, used to identify the myocardial infarction area; Regression heads are used to predict cardiac function indices.
8. The cardiac electrophysiological analysis system according to claim 7, characterized in that: The system further comprises: The explainability module is used to analyze the weight distribution of each feature in the feature fusion module and identify key features; visualize the key features to generate an intuitive heat map or importance ranking; highlight the key features through weight distribution and generate a structured analysis report.
9. A cardiac electrophysiological analysis device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the cardiac electrophysiological analysis method according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the cardiac electrophysiological analysis method according to any one of claims 1 to 7 are implemented.