Cardiovascular disease recognition system and device based on electrocardio fusion and medium

Abstract

The invention relates to the technical field of cardiovascular disease recognition, in particular to a cardiovascular disease recognition system and device based on electrocardio fusion and a medium. The system divides preprocessed electrocardio time-domain signals to obtain electrocardio depolarization / repolarization time-domain signals; performing dynamic modeling on the electrocardio depolarization / repolarization time domain signal to obtain a depolarization / repolarization dynamic domain signal; determining a depolarization time domain non-threshold recurrence plot based on the electrocardio depolarization time domain signal; based on the ECG depolarization dynamic domain signal, determining a depolarization dynamic domain non-threshold recurrence plot; based on the electrocardio repolarization time-domain signal, determining a repolarization time-domain non-threshold recurrence plot; determining a repolarization dynamic domain non-threshold recurrence plot based on the electrocardio repolarization dynamic domain signal; determining a non-threshold joint recurrence plot based on the depolarized / repolarized non-threshold recurrence plot; determining a fusion recurrence plot based on the upper / lower triangular matrix of the non-threshold joint recurrence plot; the recognition result of the cardiovascular disease category is obtained based on the fused recurrence plot, and the accuracy of cardiovascular disease recognition is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cardiovascular disease identification, and in particular to a cardiovascular disease identification system, equipment and medium based on electrocardiogram fusion. Background Art

[0002] Cardiovascular disease has become one of the leading causes of death worldwide, and its early and accurate diagnosis is crucial for improving patient outcomes. As a noninvasive and easily accessible test, the electrocardiogram (ECG) plays a vital role in the diagnosis of cardiovascular disease. However, clinical ECG analysis relies on manual interpretation by physicians, which is susceptible to subjective factors and inefficient, limiting diagnostic accuracy and consistency.

[0003] In recent years, with the rapid development of artificial intelligence technology, ECG-based automatic diagnosis methods have gradually become a research hotspot. Existing automatic diagnosis methods mainly focus on traditional time-domain signal processing of raw ECG signals. However, in the early stages of a disease, the ECG may only show subtle changes, or even appear to be a normal ECG, making it difficult to fully capture subtle abnormalities related to the disease by relying solely on traditional ECG feature extraction methods. In addition, existing methods are highly dependent on manual experience in feature engineering, and the quality of feature selection directly affects diagnostic performance. More importantly, existing technologies often ignore the intrinsic connection between the time domain and dynamic domain of ECG signals, as well as the interaction between the depolarization and repolarization processes, resulting in limited comprehensiveness and depth of diagnosis. Summary of the Invention

[0004] In order to address the deficiencies of the existing technology, the present invention provides a cardiovascular disease identification system, device and medium based on electrocardiogram fusion; the present invention analyzes ECG signals more comprehensively and deeply, providing an effective new paradigm for cardiovascular disease diagnosis.

[0005] On the one hand, a cardiovascular disease identification system based on electrocardiogram fusion is provided, including: An acquisition module is configured to: acquire an original ECG time domain signal, preprocess the original ECG time domain signal, and obtain a preprocessed ECG time domain signal; a segmentation module configured to segment the preprocessed ECG time domain signal to obtain an ECG depolarization time domain signal and an ECG repolarization time domain signal; A dynamic modeling module is configured to: use deterministic learning theory to perform dynamic modeling on the ECG depolarization time domain signal to obtain the ECG depolarization dynamic domain signal; and perform dynamic modeling on the ECG repolarization time domain signal to obtain the ECG repolarization dynamic domain signal; A fusion recursive module is configured to: determine a depolarization time-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG depolarization time-domain signal; determine a depolarization dynamic-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG depolarization dynamic-domain signal; determine a repolarization time-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG repolarization time-domain signal; and determine a repolarization dynamic-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG repolarization dynamic-domain signal; Determine a non-threshold joint recursive graph based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph, and the repolarization dynamic domain non-threshold recursive graph; determine a fused recursive graph based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph; The disease recognition module is configured to: input the fused recursive graph into the trained neural network model to obtain the recognition result of the cardiovascular disease category.

[0006] On the other hand, an electronic device is provided, comprising: a memory for non-temporarily storing computer-readable instructions; and a processor for executing the computer-readable instructions, wherein when the computer-readable instructions are executed by the processor, the following steps are performed: obtaining an original ECG time-domain signal, pre-processing the original ECG time-domain signal, and obtaining a pre-processed ECG time-domain signal; segmenting the pre-processed ECG time-domain signal, and obtaining an ECG depolarization time-domain signal and an ECG repolarization time-domain signal; using deterministic learning theory, dynamically modeling the ECG depolarization time-domain signal, and obtaining an ECG depolarization dynamic domain signal; and dynamically modeling the ECG repolarization time-domain signal, and obtaining an ECG repolarization dynamic domain signal; determining based on the Euclidean distance between any two time states of the ECG depolarization time-domain signal Depolarization time domain non-threshold recursive graph; determine the depolarization dynamic domain non-threshold recursive graph based on the Euclidean distance between any two time states of the ECG depolarization dynamic domain signal; determine the repolarization time domain non-threshold recursive graph based on the Euclidean distance between any two time states of the ECG repolarization time domain signal; determine the repolarization dynamic domain non-threshold recursive graph based on the Euclidean distance between any two time states of the ECG repolarization dynamic domain signal; determine the non-threshold joint recursive graph based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph and the repolarization dynamic domain non-threshold recursive graph; determine the fused recursive graph based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph; input the fused recursive graph into the trained neural network model to obtain the recognition result of cardiovascular disease category.

[0007] In another aspect, a storage medium is provided that non-transitory stores computer-readable instructions, wherein when the non-transitory computer-readable instructions are executed by a computer, the following steps are performed: Acquire an original ECG time domain signal, preprocess the original ECG time domain signal, and obtain a preprocessed ECG time domain signal; Segment the preprocessed ECG time domain signal to obtain ECG depolarization time domain signal and ECG repolarization time domain signal; Using deterministic learning theory, dynamic modeling is performed on the ECG depolarization time domain signal to obtain the ECG depolarization dynamic domain signal; and dynamic modeling is performed on the ECG repolarization time domain signal to obtain the ECG repolarization dynamic domain signal; Based on the Euclidean distance between any two states of the ECG depolarization time domain signal at any moment, a depolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG depolarization dynamic domain signal at any moment, a depolarization dynamic domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization time domain signal at any moment, a repolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization dynamic domain signal at any moment, a repolarization dynamic domain non-threshold recursive graph is determined; Determine a non-threshold joint recursive graph based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph, and the repolarization dynamic domain non-threshold recursive graph; determine a fused recursive graph based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph; The fused recursive graph is input into the trained neural network model to obtain the recognition results of cardiovascular disease categories.

[0008] The above technical solution has the following advantages or beneficial effects: 1. Clinical knowledge-driven automatic decomposition of ECG signals into depolarization and repolarization time domain signals. Integrating the physiological and pathological mechanisms of cardiovascular diseases, and based on the independence and complementarity of the QRS wave and ST-T segment, this method automatically segments the ECG signal into depolarization and repolarization time domain signals using a multi-scale decomposition method, avoiding the inaccurate positioning of key points in traditional QRS wave and ST-T segment segmentation.

[0009] 2. Mining dynamic information of time-domain signals. Dynamic modeling of ECG depolarization and repolarization time-domain signals is performed to obtain more comprehensive and interpretable dynamic domain signals, which can be used to reveal the underlying laws of ECG signals and expand the information dimension of ECG signals.

[0010] 3. Recursive visualization and fusion. Using a threshold-free joint recursive analysis method, time series are converted into two-dimensional images. The ECG signal is then fused with dynamic domain signals, as well as ECG depolarization and repolarization time domain signals. This enhances the ability to characterize subtle ECG anomalies and improves cardiovascular disease diagnosis. This threshold-free joint recursive analysis method avoids over- or under-representation of the system due to inappropriate threshold selection.

[0011] 4. Automated diagnosis of cardiovascular disease: Automatically extract and fuse features of recursive graphs based on neural networks to achieve automatic identification of cardiovascular disease. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0013] Figure 1 2 is a diagram of system function modules in an embodiment of the present invention. DETAILED DESCRIPTION

[0014] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0015] Example 1 This embodiment provides a cardiovascular disease identification system based on electrocardiogram fusion; like Figure 1 As shown in FIG, the cardiovascular disease identification system based on electrocardiogram fusion includes: An acquisition module is configured to: acquire an original ECG time domain signal, preprocess the original ECG time domain signal, and obtain a preprocessed ECG time domain signal; a segmentation module configured to segment the preprocessed ECG time domain signal to obtain an ECG depolarization time domain signal and an ECG repolarization time domain signal; A dynamic modeling module is configured to: use deterministic learning theory to perform dynamic modeling on the ECG depolarization time domain signal to obtain the ECG depolarization dynamic domain signal; and perform dynamic modeling on the ECG repolarization time domain signal to obtain the ECG repolarization dynamic domain signal; A fusion recursive module is configured to: determine a depolarization time-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG depolarization time-domain signal; determine a depolarization dynamic-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG depolarization dynamic-domain signal; determine a repolarization time-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG repolarization time-domain signal; and determine a repolarization dynamic-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG repolarization dynamic-domain signal; Determine a non-threshold joint recursive graph based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph, and the repolarization dynamic domain non-threshold recursive graph; determine a fused recursive graph based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph; The disease recognition module is configured to: input the fused recursive graph into the trained neural network model to obtain the recognition result of the cardiovascular disease category.

[0016] Furthermore, an original ECG time domain signal is acquired and preprocessed to obtain a preprocessed ECG time domain signal, wherein the preprocessing includes filtering and baseline correction.

[0017] Exemplarily, the preprocessing includes: using a bandpass filter with a cutoff frequency of 0.5-40 Hz to remove power frequency interference and high-frequency noise, and using a 0.5 Hz Butterworth filter to eliminate baseline drift caused by breathing, etc.

[0018] Furthermore, the preprocessed ECG time domain signal is segmented to obtain ECG depolarization time domain signal and ECG repolarization time domain signal, specifically including: First, the preprocessed ECG time domain signal is linearly transformed using the Kros matrix. The Kros matrix is ​​as follows: ; ; in, is the Kros matrix, The preprocessed ECG time domain signal, Indicates the first and second leads of a bipolar lead, Represents 6 chest leads, Refers to the ECG signal after dimensionality reduction, They represent the three leads of the ECG signal after dimensionality reduction.

[0019] Perform wavelet packet decomposition on the three-lead ECG signal after dimensionality reduction: select the wavelet basis function (such as the orthogonal wavelet basis Daubechies series, etc.) and the number of decomposition layers ,Through an iterative downsampling process, the signal is decomposed into sub-bands with different frequencies.

[0020] For any node wavelet packet coefficient (or signal) , through a low-pass filter and high-pass filter Decompose it and get The wavelet packet coefficients of the two child nodes of the layer and : ; ; in, represents the number of decomposition levels, Represents the node index in the current layer, Indicates the sampling point index of the current layer signal, is the sampling point index of the next decomposition layer, and It is defined by wavelet basis functions.

[0021] For sampling rates The ECG time domain signal, its wavelet packet decomposition layer number is , a total of Sub-bands. Any node The corresponding frequency range is ,Right now:

[0022] The STT and QRS wavelet packet coefficient sets are defined as the sets of node coefficients that satisfy the following conditions:

[0023]

[0024] in, .

[0025] The collected wavelet packet coefficients and Reconstructed by inverse wavelet packet transform, the ST-T band signals were obtained by iterative application of reconstruction filters and upsampling. and QRS complex signals , the inverse wavelet packet transform formula is as follows: ; in, and are low-pass and high-pass filters for wavelet reconstruction, respectively. and and wavelet decomposition filters and The relationship is related to the specific wavelet basis function. Taking the orthogonal wavelet basis as an example, the wavelet reconstruction filter is the time reversal of the wavelet decomposition filter, that is, , .

[0026] The QRS complex signal is an ECG depolarization time domain signal; the ST-T band signal is an ECG repolarization time domain signal.

[0027] The beneficial effect of the above technical solution is that the two signals are segmented according to the independence and complementarity between the QRS complex signal and the ST-T band signal in the pathological mechanism of cardiovascular disease.

[0028] Furthermore, the deterministic learning theory is used to perform dynamic modeling on the ECG depolarization time domain signal to obtain the ECG depolarization dynamic domain signal; and the ECG repolarization time domain signal is also performed dynamic modeling to obtain the ECG repolarization dynamic domain signal, specifically including: The depolarization and repolarization time domain signals are regarded as three-dimensional nonlinear dynamic systems The depolarization and repolarization time domain signals are sampled to obtain Based on deterministic learning theory, a neural network identifier is used to perform locally accurate neural network modeling of the intrinsic dynamics of depolarization and repolarization.

[0029]

[0030] in, It is an RBF (Radial Basis Function) neural network; It's an electrocardiogram signal The estimated amount, is the estimation of the RBF neural network weights, is the regression vector of the RBF neural network, is the learning rate, is the sampling time of the ECG signal; Using the weight update algorithm to : ; in, is the state estimation error, represents the learning gain, where for Maximum two norm, is the sampling time of the ECG signal.

[0031] Furthermore, the intrinsic nonlinear dynamics of depolarization and repolarization are obtained :

[0032] in, , , Indicates that the intrinsic ECG dynamics information is obtained through deterministic learning. Get dynamic domain information of ECG depolarization ,when Get ECG repolarization dynamic domain information .

[0033] The beneficial effect of the above technical solution is that by determining the learning theory for dynamic modeling, a more comprehensive and interpretable dynamic signal can be obtained, and the dynamic signal here is the dynamic domain signal.

[0034] Furthermore, based on the Euclidean distance between any two time-point states of the ECG depolarization time-domain signal, a depolarization time-domain non-threshold recursive graph is determined; based on the Euclidean distance between any two time-point states of the ECG depolarization dynamic-domain signal, a depolarization dynamic-domain non-threshold recursive graph is determined; based on the Euclidean distance between any two time-point states of the ECG repolarization time-domain signal, a repolarization time-domain non-threshold recursive graph is determined; based on the Euclidean distance between any two time-point states of the ECG repolarization dynamic-domain signal, a repolarization dynamic-domain non-threshold recursive graph is determined; specifically including: Based on the time domain signal of ECG depolarization calculate and The Euclidean distance between two states at a time, determining the non-threshold recursive graph in the depolarization domain ; Based on the dynamic domain signal of cardiac depolarization calculate and The Euclidean distance between two time states determines the non-threshold recursive graph of the extreme dynamic domain ; Based on ECG repolarization time domain signal calculate and The Euclidean distance between two time states determines the repolarization time domain non-threshold recurrence diagram ; Based on ECG repolarization dynamic domain signal calculate and The Euclidean distance between two time states determines the non-threshold recursive graph of the repolarization dynamic domain ; Specifically include: ; ; in, is the time series of ECG depolarization time domain signal and ECG repolarization time domain signal after dimensionality reduction, are the depolarization dynamic domain signal and the repolarization dynamic domain signal, represents the Euclidean norm, that is, the Euclidean distance. Represents the non-threshold recursive graph of ECG time domain signal; Represents the non-threshold recursive graph of dynamic domain signals; when , we get the depolarization time domain non-threshold recursive graph ;when , we get the depolarization time domain non-threshold recursive graph ;when , we get the depolarization time domain non-threshold recursive graph ;when , we get the depolarization time domain non-threshold recursive graph .

[0035] The beneficial effect of the above technical solution is that the time series can be converted into a two-dimensional image to enhance the representation of subtle abnormalities in the electrocardiogram signal.

[0036] Furthermore, based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph, and the repolarization dynamic domain non-threshold recursive graph, a non-threshold joint recursive graph is determined; specifically, the method includes: In order to enhance the similarity of the two time series while reducing the irrelevance, the Hadamard product is calculated to obtain the non-threshold joint recursive graph : ; in, Represents the time domain signal Moment and Moment Euclidean distance, Indicates the dynamic domain signal Moment and Moment Euclidean distance, ; when , , we get the depolarization non-threshold joint recursive graph ; when , , and obtain the repolarization non-threshold joint recurrence diagram .

[0037] Furthermore, based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph, a fused recursive graph is determined, specifically including: The non-threshold joint recurrence diagram of depolarization and repolarization was calculated using the Matlab software Function and The function transforms the upper triangular matrix and the lower triangular matrix and adds them together to obtain a fused recursive graph. : ; in, The function is used to extract the upper triangular part of the matrix. The function is used to extract the lower triangular part of a matrix.

[0038] The beneficial effect of the above technical solution is: in order to effectively reflect the correlation between the depolarization and repolarization processes and solve the problem that the recursive graph contains general repeated information, the threshold-free joint recursive graphs of different subsystems are converted into upper triangular matrices and lower triangular matrices, and added together to obtain a fused recursive graph containing complete information in the time domain and dynamic domain of the depolarization and repolarization subsystems.

[0039] Furthermore, the fused recursive graph is input into the trained neural network model to obtain the recognition results of cardiovascular disease categories, including: The fused recursive graph is input into the trained convolutional neural network to obtain the recognition results of cardiovascular disease categories.

[0040] The training process of the trained convolutional neural network includes: Constructing a training set, wherein the training set is a fused recursive graph of real clinical labels of known cardiovascular diseases (such as myocardial ischemia); The training set is input into the convolutional neural network, and the network is trained to obtain the trained convolutional neural network.

[0041] Specifically, the fused recurrent graph is converted into a format suitable for convolutional neural network input. Since the fused recurrent graph is color-coded, the number of channels is selected as 3.

[0042] The convolutional neural network model structure is designed, and the convolutional neural network parameters are determined based on the size of the fused recurrent graph. The features of the fused recurrent graph are automatically extracted. Finally, a Softmax function or a Sigmoid function is used to output the probability values ​​of different cardiovascular disease categories, achieving automatic recognition of cardiovascular disease.

[0043] The beneficial effect of the above technical solution is that a neural network model (optionally, a convolutional neural network) is trained using the fused recurrence graph as input. The convolutional neural network utilizes its inherent ability to automatically extract image features to extract discriminative features from the fused recurrence graph. Ultimately, the trained model outputs recognition results corresponding to cardiovascular diseases.

[0044] The present invention relates to the field of cardiovascular disease diagnosis technology and specifically includes the following steps: preprocessing collected ECG signals to obtain standardized, high-quality ECG signals; segmenting the preprocessed ECG signals into a depolarization subsystem (corresponding to the QRS complex) and a complex subsystem (corresponding to the ST-T segment) using a multiscale decomposition method; applying deterministic learning theory to dynamic modeling of these two subsystems to obtain dynamic domain signals; recursively analyzing the subsystem time and dynamic domain signals to generate a threshold-free joint recursive graph. This graph is then transformed and summed up using upper and lower triangular matrices to obtain a fused recursive graph containing complete information about the depolarization and complex subsystems. Finally, the fused recursive graph is input into a trained neural network model (e.g., a convolutional neural network) to automatically extract features and perform disease identification, thereby enabling automatic diagnosis of cardiovascular disease. This invention provides an effective new paradigm for the automatic identification of cardiovascular disease and has significant clinical application value.

[0045] Example 2 This embodiment further provides an electronic device, comprising: one or more processors, one or more memories, and one or more computer programs; wherein the processor is connected to the memory, and the one or more computer programs are stored in the memory. When the electronic device is running, the processor executes the one or more computer programs stored in the memory, so that the electronic device performs the following steps: Acquire an original ECG time domain signal, preprocess the original ECG time domain signal, and obtain a preprocessed ECG time domain signal; Segment the preprocessed ECG time domain signal to obtain ECG depolarization time domain signal and ECG repolarization time domain signal; Using deterministic learning theory, dynamic modeling is performed on the ECG depolarization time domain signal to obtain the ECG depolarization dynamic domain signal; and dynamic modeling is performed on the ECG repolarization time domain signal to obtain the ECG repolarization dynamic domain signal; Based on the Euclidean distance between any two states of the ECG depolarization time domain signal at any moment, a depolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG depolarization dynamic domain signal at any moment, a depolarization dynamic domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization time domain signal at any moment, a repolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization dynamic domain signal at any moment, a repolarization dynamic domain non-threshold recursive graph is determined; Determine a non-threshold joint recursive graph based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph, and the repolarization dynamic domain non-threshold recursive graph; determine a fused recursive graph based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph; The fused recursive graph is input into the trained neural network model to obtain the recognition results of cardiovascular disease categories.

[0046] Example 3 This embodiment further provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the following steps are performed: Acquire the original ECG time domain signal, preprocess the original ECG time domain signal to obtain a preprocessed ECG time domain signal; segment the preprocessed ECG time domain signal to obtain an ECG depolarization time domain signal and an ECG repolarization time domain signal; Using deterministic learning theory, dynamic modeling is performed on the ECG depolarization time domain signal to obtain the ECG depolarization dynamic domain signal; and dynamic modeling is performed on the ECG repolarization time domain signal to obtain the ECG repolarization dynamic domain signal; Based on the Euclidean distance between any two states of the ECG depolarization time domain signal at any moment, a depolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG depolarization dynamic domain signal at any moment, a depolarization dynamic domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization time domain signal at any moment, a repolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization dynamic domain signal at any moment, a repolarization dynamic domain non-threshold recursive graph is determined; Determine a non-threshold joint recursive graph based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph, and the repolarization dynamic domain non-threshold recursive graph; determine a fused recursive graph based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph; The fused recursive graph is input into the trained neural network model to obtain the recognition results of cardiovascular disease categories.

[0047] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. Cardiovascular disease identification system based on electrocardiogram fusion, characterized by: include: An acquisition module is configured to: acquire an original ECG time domain signal, preprocess the original ECG time domain signal, and obtain a preprocessed ECG time domain signal; a segmentation module configured to segment the preprocessed ECG time domain signal to obtain an ECG depolarization time domain signal and an ECG repolarization time domain signal; A dynamic modeling module is configured to: use deterministic learning theory to perform dynamic modeling on the ECG depolarization time domain signal to obtain the ECG depolarization dynamic domain signal; and perform dynamic modeling on the ECG repolarization time domain signal to obtain the ECG repolarization dynamic domain signal; A fusion recursive module is configured to: determine a depolarization time-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG depolarization time-domain signal; determine a depolarization dynamic-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG depolarization dynamic-domain signal; determine a repolarization time-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG repolarization time-domain signal; determine a repolarization dynamic-domain non-threshold recursive graph based on the Euclidean distance between any two time-point states of the ECG repolarization dynamic-domain signal; determine a non-threshold joint recursive graph based on the depolarization time-domain non-threshold recursive graph, the depolarization dynamic-domain non-threshold recursive graph, the repolarization time-domain non-threshold recursive graph, and the repolarization dynamic-domain non-threshold recursive graph; and determine a fusion recursive graph based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph; The disease recognition module is configured to: input the fused recursive graph into the trained neural network model to obtain the recognition result of the cardiovascular disease category.

2. The cardiovascular disease identification system based on electrocardiogram fusion according to claim 1, characterized in that: The preprocessed ECG time domain signal is segmented to obtain ECG depolarization time domain signal and ECG repolarization time domain signal, specifically including: First, the preprocessed ECG time domain signal is linearly transformed using the Kros matrix. The Kros matrix is ​​as follows: ； ； in, is the Kros matrix, The preprocessed ECG time domain signal, Indicates the first and second leads of a bipolar lead, Represents 6 chest leads, Refers to the ECG signal after dimensionality reduction, Represent the three leads of the ECG signal after dimensionality reduction; Perform wavelet packet decomposition on the three-lead ECG signal after dimensionality reduction: select the wavelet basis function and the number of decomposition layers ,Through an iterative downsampling process, the signal is decomposed into sub-bands with different frequencies; For any node wavelet packet coefficient , through a low-pass filter and high-pass filter Decompose it and get The wavelet packet coefficients of the two child nodes of the layer and : ； ； in, represents the number of decomposition levels, Represents the node index in the current layer, Indicates the sampling point index of the current layer signal, The sampling point index for the next decomposition layer; For sampling rates The ECG time domain signal, its wavelet packet decomposition layer number is , a total of sub-bands; any node The corresponding frequency range is : ； The STT and QRS wavelet packet coefficient sets are defined as the sets of node coefficients that satisfy the following conditions: ； ； in, ; The collected wavelet packet coefficients and Reconstructed by inverse wavelet packet transform, the ST-T band signals were obtained by iterative application of reconstruction filters and upsampling. and QRS complex signals , the inverse wavelet packet transform formula is as follows: ； in, and are low-pass and high-pass filters for wavelet reconstruction, respectively, where , ; The QRS complex signal is an ECG depolarization time domain signal; the ST-T band signal is an ECG repolarization time domain signal.

3. The cardiovascular disease identification system based on electrocardiogram fusion according to claim 1, characterized in that: Based on the Euclidean distance between any two states of the ECG depolarization time domain signal, a depolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG depolarization dynamic domain signal, a depolarization dynamic domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization time domain signal, a repolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization dynamic domain signal, a repolarization dynamic domain non-threshold recursive graph is determined; specifically including: Based on the time domain signal of ECG depolarization calculate and The Euclidean distance between two states at a time, determining the non-threshold recursive graph in the depolarization domain ; Based on the dynamic domain signal of cardiac depolarization calculate and The Euclidean distance between two time states determines the non-threshold recursive graph of the extreme dynamic domain ; Based on ECG repolarization time domain signal calculate and The Euclidean distance between two time states determines the repolarization time domain non-threshold recurrence diagram ; Based on ECG repolarization dynamic domain signal calculate and The Euclidean distance between two time states determines the non-threshold recursive graph of the repolarization dynamic domain ; Specifically include: ； ； in, is the time series of ECG depolarization time domain signal and ECG repolarization time domain signal after dimensionality reduction, are the depolarization dynamic domain signal and the repolarization dynamic domain signal, represents the Euclidean norm; Represents the non-threshold recursive graph of ECG time domain signal; Represents the non-threshold recursive graph of dynamic domain signals; when , we get the depolarization time domain non-threshold recursive graph ;when , we get the depolarization time domain non-threshold recursive graph ;when , we get the depolarization time domain non-threshold recursive graph ;when , we get the depolarization time domain non-threshold recursive graph .

4. The cardiovascular disease identification system based on electrocardiogram fusion according to claim 1, characterized in that: Based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph, and the repolarization dynamic domain non-threshold recursive graph, a non-threshold joint recursive graph is determined; specifically, the non-threshold joint recursive graph includes: In order to enhance the similarity of the two time series while reducing the irrelevance, the Hadamard product is calculated to obtain the non-threshold joint recursive graph : ； in, Represents the time domain signal Moment and Moment Euclidean distance, Indicates the dynamic domain signal Moment and Moment Euclidean distance, ; when , , we get the depolarization non-threshold joint recursive graph ; when , , and obtain the repolarization non-threshold joint recurrence diagram .

5. The cardiovascular disease identification system based on electrocardiogram fusion according to claim 1, characterized in that: Based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph, a fused recursive graph is determined, specifically including: The non-threshold joint recurrence diagram of depolarization and repolarization is used Function and The function transforms the upper triangular matrix and the lower triangular matrix and adds them together to obtain a fused recursive graph. : ； in, The function is used to extract the upper triangular part of the matrix. The function is used to extract the lower triangular part of a matrix.

6. The cardiovascular disease identification system based on electrocardiogram fusion according to claim 1, characterized in that: The fused recursive graph is input into the trained neural network model to obtain the recognition results of cardiovascular disease categories, including: The fused recursive graph is input into a trained convolutional neural network to obtain recognition results of cardiovascular disease categories; wherein the training process of the trained convolutional neural network includes: constructing a training set, which is a fused recursive graph of real clinical labels of known cardiovascular diseases; inputting the training set into the convolutional neural network, training the network, and obtaining a trained convolutional neural network.

7. The cardiovascular disease identification system based on electrocardiogram fusion according to claim 1, characterized in that: The original ECG time domain signal is acquired and preprocessed to obtain a preprocessed ECG time domain signal, wherein the preprocessing includes filtering and baseline correction.

8. The cardiovascular disease identification system based on electrocardiogram fusion according to claim 1, characterized in that: The preprocessing includes: using a bandpass filter to select a cutoff frequency of 0.5-40 Hz to remove power frequency interference and high-frequency noise; and simultaneously, using a 0.5 Hz Butterworth filter to eliminate baseline drift caused by breathing.

9. An electronic device, characterized in that: include: A memory for non-temporarily storing computer-readable instructions; and a processor for executing the computer-readable instructions, wherein when the computer-readable instructions are executed by the processor, the processor performs the following steps: acquiring an original ECG time-domain signal, preprocessing the original ECG time-domain signal, and obtaining a preprocessed ECG time-domain signal; The preprocessed ECG time domain signal is segmented to obtain ECG depolarization time domain signal and ECG repolarization time domain signal; the ECG depolarization time domain signal is dynamically modeled using deterministic learning theory to obtain ECG depolarization dynamic domain signal; and the ECG repolarization time domain signal is dynamically modeled to obtain ECG repolarization dynamic domain signal; based on the Euclidean distance between any two time states of the ECG depolarization time domain signal, the depolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two time states of the ECG depolarization dynamic domain signal, the depolarization dynamic domain non-threshold recursive graph is determined; based on the ECG repolarization time domain signal The Euclidean distance between any two time-domain states is used to determine the non-threshold recursive graph of repolarization in the time domain. The Euclidean distance between any two time-domain states of the ECG repolarization dynamic domain signal is used to determine the non-threshold recursive graph of repolarization in the dynamic domain. The non-threshold joint recursive graph is determined based on the non-threshold recursive graph of depolarization in the time domain, the non-threshold recursive graph of depolarization in the dynamic domain, the non-threshold recursive graph of repolarization in the time domain, and the non-threshold recursive graph of repolarization in the dynamic domain. The fused recursive graph is determined based on the upper triangular matrix and lower triangular matrix of the non-threshold joint recursive graph. The fused recursive graph is input into the trained neural network model to obtain the recognition results of cardiovascular disease categories.

10. A storage medium characterized by being non-transitory storing computer-readable instructions, wherein when the non-transitory computer-readable instructions are executed by a computer, performing the following steps: Acquire an original ECG time domain signal, preprocess the original ECG time domain signal, and obtain a preprocessed ECG time domain signal; Segment the preprocessed ECG time domain signal to obtain ECG depolarization time domain signal and ECG repolarization time domain signal; Using deterministic learning theory, dynamic modeling is performed on the ECG depolarization time domain signal to obtain the ECG depolarization dynamic domain signal; and dynamic modeling is performed on the ECG repolarization time domain signal to obtain the ECG repolarization dynamic domain signal; Based on the Euclidean distance between any two states of the ECG depolarization time domain signal at any moment, a depolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG depolarization dynamic domain signal at any moment, a depolarization dynamic domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization time domain signal at any moment, a repolarization time domain non-threshold recursive graph is determined; based on the Euclidean distance between any two states of the ECG repolarization dynamic domain signal at any moment, a repolarization dynamic domain non-threshold recursive graph is determined; Determine a non-threshold joint recursive graph based on the depolarization time domain non-threshold recursive graph, the depolarization dynamic domain non-threshold recursive graph, the repolarization time domain non-threshold recursive graph, and the repolarization dynamic domain non-threshold recursive graph; determine a fused recursive graph based on the upper triangular matrix and the lower triangular matrix of the non-threshold joint recursive graph; The fused recursive graph is input into the trained neural network model to obtain the recognition results of cardiovascular disease categories.

Images