Feature extraction method and system based on electrocardiogram signals

Abstract

The invention discloses a feature extraction method and system based on an electrocardiogram signal. The method comprises the steps that S1, the electrocardiogram signal of an electrocardiogram image is collected; s2, pre-processing the electrocardiogram signal to obtain a pre-processed signal, and decomposing the pre-processed signal to obtain a preliminary decomposition feature of a signal waveform; s3, based on the preliminary decomposition characteristics, obtaining a separation matrix and reconstruction parameters of electrocardiogram wave-form components by using a waveform independence judgment model; s4, obtaining a waveform boundary positioning result based on the separation matrix and the reconstruction parameters; and S5, performing feature extraction on the electrocardiogram signal based on the waveform boundary positioning result to obtain electrocardiogram signal features.

Description

Technical Field

[0001] This invention relates to the field of medical image feature extraction technology, specifically to a feature extraction method and system based on electrocardiogram (ECG) signals. Background Art

[0002] Electrocardiograms (ECGs), as vital biosignals reflecting cardiac electrophysiological activity, play an irreplaceable role in the diagnosis of cardiovascular diseases. With the continuous rise in the incidence of cardiovascular diseases, accurate and rapid ECG analysis has become an urgent need in clinical medicine. Traditional ECG interpretation relies primarily on the professional experience and visual judgment of physicians. This manual analysis method is not only time-consuming and labor-intensive but also easily influenced by subjective factors, making it difficult to guarantee consistency and accuracy when dealing with large amounts of ECG data.

[0003] Current automated electrocardiogram (ECG) analysis methods generally suffer from insufficient precision in feature extraction. Existing technologies often employ simple filtering and threshold detection methods, which struggle to effectively handle noise interference and individual differences in ECG signals, resulting in low accuracy in identifying key waveforms. Furthermore, these methods perform poorly when dealing with complex arrhythmias and subtle pathological features, failing to meet the precision requirements of clinical diagnosis.

[0004] Electrocardiogram (ECG) signals contain rich pathological information, but this information is often hidden within complex waveform variations. Key components of an ECG, such as the P wave, QRS complex, and T wave, overlap in both the time and frequency domains, making effective waveform separation difficult with traditional signal processing techniques. This waveform aliasing directly affects the accurate extraction of feature parameters for each band, including the measurement accuracy of key indicators such as amplitude, duration, and morphology. Inaccurate feature extraction severely impacts subsequent pattern recognition and disease classification. Adding to the complexity, significant individual differences exist in ECG signals among different patients; the same heart disease may exhibit different waveform characteristics in different patients, making the establishment of a unified and effective feature extraction model extremely difficult. Summary of the Invention

[0005] This invention provides a feature extraction method based on electrocardiogram (ECG) signals, mainly including:

[0006] Step S1: Acquire electrocardiogram (ECG) signals to form an ECG image;

[0007] Step S2: Preprocess the electrocardiogram signal to obtain a preprocessed signal, and decompose the preprocessed signal to obtain preliminary decomposition features of the signal waveform;

[0008] Step S3: Based on the preliminary decomposition features, use the waveform independence judgment model to obtain the separation matrix and reconstruction parameters of the electrocardiogram waveform components;

[0009] Step S4: Obtain the waveform boundary localization result based on the separation matrix and reconstruction parameters;

[0010] Step S5: Based on the waveform boundary positioning results, perform feature extraction on the electrocardiogram signal to obtain electrocardiogram signal features.

[0011] Optionally, in step S2, the preprocessing includes:

[0012] After denoising the electrocardiogram signal, data reconstruction is performed to obtain the preprocessed signal:

[0013] Variational mode decomposition of electrocardiogram signals:

[0014] Initial IMF components are obtained based on electrocardiogram signals and signal envelopes;

[0015] The updated electrocardiogram signal is obtained based on the electrocardiogram signal and the IMF component;

[0016] The final IMF component is obtained based on the updated electrocardiogram signal;

[0017] Based on the final IMF components, reconstructed data is obtained, resulting in a preprocessed signal.

[0018] Optionally, in step S2, obtaining the preliminary decomposition features of the signal waveform specifically involves:

[0019]

[0020] Among them, STFTP xy (t, f) represents the short-time Fourier cross-power spectrum data; STFTP x (t, f) represents the short-time Fourier autopower spectrum data of the random signal x(t); STFTP y (t, f) represents the short-time Fourier autopower spectrum data of the random signal y(t).

[0021] Optionally, step S3 specifically includes:

[0022] A statistical independence model of electrocardiogram (ECG) waveforms is constructed using the independent component analysis (ICA) algorithm. If the mutual information between two waveform components is lower than the independence threshold, the separation is considered successful, and the accurate separation matrix and reconstruction parameters of each ECG waveform component are obtained.

[0023] This invention also discloses a feature extraction system based on electrocardiogram signals, the system comprising:

[0024] The data acquisition module is used to acquire electrocardiogram (ECG) signals for ECG images;

[0025] The preliminary decomposition module is used to preprocess the electrocardiogram signal to obtain a preprocessed signal, and to decompose the preprocessed signal to obtain preliminary decomposition features of the signal waveform.

[0026] The waveform separation module is used to obtain the separation matrix and reconstruction parameters of the electrocardiogram waveform components based on the preliminary decomposition features and using the waveform independence judgment model.

[0027] A boundary definition module is used to obtain waveform boundary positioning results based on the separation matrix and reconstruction parameters;

[0028] The feature extraction module is used to extract features from the electrocardiogram signal based on the waveform boundary positioning results to obtain electrocardiogram signal features.

[0029] Optional, preprocessing includes:

[0030] After denoising the electrocardiogram signal, data reconstruction is performed to obtain the preprocessed signal:

[0031] Variational mode decomposition of electrocardiogram signals:

[0032] Initial IMF components are obtained based on electrocardiogram signals and signal envelopes;

[0033] The updated electrocardiogram signal is obtained based on the electrocardiogram signal and the IMF component;

[0034] The final IMF component is obtained based on the updated electrocardiogram signal;

[0035] Based on the final IMF components, reconstructed data is obtained, resulting in a preprocessed signal.

[0036] Optionally, the preliminary decomposition features of the signal waveform are obtained as follows:

[0037]

[0038] Among them, STFTP xy (t, f) represents the short-time Fourier cross-power spectrum data; STFTP x (t, f) represents the short-time Fourier autopower spectrum data of the random signal x(t); STFTP y (t, f) represents the short-time Fourier autopower spectrum data of the random signal y(t).

[0039] Optionally, the specific workflow of the waveform separation module is as follows:

[0040] A statistical independence model of electrocardiogram (ECG) waveforms is constructed using the independent component analysis (ICA) algorithm. If the mutual information between two waveform components is lower than the independence threshold, the separation is considered successful, and the accurate separation matrix and reconstruction parameters of each ECG waveform component are obtained.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] This invention reconstructs electrocardiogram (ECG) signals by calculating the average value of the signal envelope, further smoothing the signal, reducing noise interference, and improving signal stability and reliability, thus providing a clearer signal foundation for subsequent feature extraction. Based on waveform boundary localization results, this invention can calculate quantitative parameters such as amplitude, duration, area, and slope of each waveform component, providing a comprehensive and quantitative basis for ECG signal analysis. These parameters can more accurately reflect the strength, time process, and waveform morphological changes of cardiac electrical activity, helping physicians to make more detailed judgments about cardiac function. Attached Figure Description

[0043] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart illustrating the method steps of the feature extraction method based on electrocardiogram signals according to an embodiment of the present invention;

[0045] Figure 2 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention;

[0046] Explanation of reference numerals in the attached figures:

[0047] 1010, Processor; 1020, Memory; 1030, Input / Output Interface; 1040, Communication Interface; 1050, Bus. Detailed Implementation

[0048] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] Example 1

[0050] A feature extraction method based on electrocardiogram signals, such as Figure 1 As shown, the method includes:

[0051] Step S1: Acquire electrocardiogram (ECG) signals to form an ECG image.

[0052] The raw electrocardiogram signal is acquired by the acquisition device and stored as digital sequence data to obtain the initial signal data.

[0053] A professional medical 12-lead electrocardiograph (ECG) is used as the acquisition device, equipped with a high-quality bioelectric amplifier, filter, and analog-to-digital converter. First, disposable ECG electrodes are attached to designated locations on the patient's body, including the limbs (both arms and both legs) and specific areas of the chest wall (V1 to V6 chest leads). The electrodes are connected to the ECG's input ports via lead wires, ensuring a secure and reliable connection to avoid interference and signal loss during transmission.

[0054] Once the electrocardiograph (ECG) machine starts its acquisition program, it begins acquiring the raw ECG signal generated by the patient's heart. A bioelectric amplifier amplifies these weak electrical signals, bringing them within the amplitude range that the analog-to-digital converter (ADC) can effectively process. A filter removes environmental interference signals, such as 50Hz power line interference and low-frequency baseline drift noise, ensuring signal purity and accuracy. The ADC then converts the continuous analog ECG signal into discrete digital sequence data at a specific sampling frequency (e.g., 500Hz or 1000Hz).

[0055] The acquired digital sequence data is stored in real time on the hard drive of a computer connected to the electrocardiograph (ECG) machine, saved in a specific file format (such as EDF). This initial signal data contains information such as the time and amplitude of various ECG waveforms (such as P waves, QRS complexes, T waves, etc.), reflecting the electrical activity state of the patient's heart. In the initial processing stage, simple filtering and noise reduction may be performed on the data to further improve signal quality and provide accurate basic data for subsequent ECG image generation and analysis.

[0056] Step S2: Preprocess the electrocardiogram signal to obtain a preprocessed signal, and decompose the preprocessed signal to obtain preliminary decomposition features of the signal waveform.

[0057] Preprocessing includes:

[0058] After denoising the electrocardiogram signal, data reconstruction is performed to obtain the preprocessed signal:

[0059] Variational mode decomposition of electrocardiogram signals:

[0060] Initial IMF components are obtained based on electrocardiogram signals and signal envelopes;

[0061] Calculate the average value of the signal envelope:

[0062]

[0063] The electrocardiogram signal is x(t), and the maximum envelope e is fitted from the maximum points. + (t), minimum envelope e - (t).

[0064] The sequence difference is calculated as follows:

[0065]

[0066] The IMF component is calculated as follows:

[0067]

[0068] The updated electrocardiogram signal is obtained based on the electrocardiogram signal and the IMF component;

[0069] r1(t) = x(t) - c1(t)

[0070] The final IMF component is obtained based on the updated electrocardiogram signal;

[0071] Based on the final IMF components, reconstructed data is obtained, resulting in a preprocessed signal.

[0072] The preliminary decomposition characteristics of the signal waveform are as follows:

[0073]

[0074] Among them, STFTP xy (t, f) represents the short-time Fourier cross-power spectrum data; STFTP x (t, f) represents the short-time Fourier autopower spectrum data of the random signal x(t); STFTP y (t, f) represents the short-time Fourier autopower spectrum data of the random signal y(t).

[0075] STFTP x The calculation method for (t, f) is as follows:

[0076]

[0077] The same calculation method is used for STFTP. y The calculation is performed using (t, f).

[0078] Step S3: Based on the preliminary decomposition features, use the waveform independence judgment model to obtain the separation matrix and reconstruction parameters of the electrocardiogram waveform components.

[0079] A statistical independence model of electrocardiogram (ECG) waveforms is constructed using the independent component analysis (ICA) algorithm. If the mutual information between two waveform components is lower than the independence threshold, the separation is considered successful, and the accurate separation matrix and reconstruction parameters of each ECG waveform component are obtained.

[0080] Independent Component Analysis (ICA) is used to preprocess the input electrocardiogram (ECG) signal, decomposing it into multiple waveform components to obtain an initial separation matrix. Feature vectors for each waveform component are extracted from the initial separation matrix, and the mutual information between each pair of waveform components is calculated. If the mutual information is below a preset independence threshold, the corresponding waveform component is determined to be statistically independent, resulting in a set of independent waveform components. Based on this set, the initial separation matrix is ​​optimized to generate an accurate separation matrix. A linear transformation is performed on the original ECG signal using the accurate separation matrix to obtain the reconstructed independent waveform components. Time-series features are extracted from the reconstructed independent waveform components to generate corresponding reconstruction parameters. The reconstructed parameters are used to verify the separated waveform components, determining the signal integrity of each component and obtaining the final separation result.

[0081] Step S4: Obtain the waveform boundary positioning result based on the separation matrix and reconstruction parameters.

[0082] Based on the separation matrix, the waveform components are reconstructed in the time domain. The precise time position of the start point, peak point and end point of each waveform is determined by the peak detection algorithm. If the waveform amplitude change rate exceeds the preset gradient threshold, it is marked as a key feature point, and the complete waveform boundary positioning result is obtained.

[0083] The separation matrix is ​​obtained from the input signal, and independent component analysis (ICA) is used to decompose each independent waveform component, resulting in a time-domain signal set. Gaussian filtering is then applied to smooth each waveform using this time-domain signal set, yielding a smoothed time-domain signal. For the smoothed time-domain signal, a peak detection algorithm is applied to identify the start, peak, and end points of each waveform, determining a set of time positions. The amplitude changes between adjacent points are obtained from the time position set, and the rate of change of amplitude is calculated, resulting in an amplitude change rate sequence. If the rate of change of amplitude exceeds a preset gradient threshold, the corresponding point is marked as a key feature point, obtaining a set of key feature points. Based on the set of key feature points and the time position set, the boundaries of each waveform are defined, generating waveform boundary localization results. The consistency between the time positions and key feature points of each waveform is verified using the waveform boundary localization results, yielding the final waveform boundary localization result.

[0084] Step S5: Based on the waveform boundary positioning results, perform feature extraction on the electrocardiogram signal to obtain electrocardiogram signal features.

[0085] Based on the waveform boundary location results, quantitative parameters such as amplitude, duration, area, and slope of each waveform component are calculated. The influence of individual differences is eliminated through parameter normalization. If the parameter value deviates from the normal range by more than two standard deviations, it is marked as an abnormal feature, and a standardized feature parameter vector is obtained.

[0086] The waveform boundary localization results are obtained from the waveform data. Signal processing algorithms are used to determine the start and end points of each waveform component, thus obtaining the waveform boundary localization. Based on the waveform boundary localization, the amplitude, duration, area, and slope of each waveform component are calculated. Numerical integration is used to calculate the area, resulting in a waveform parameter set. For this parameter set, z-score normalization is applied to map the amplitude, duration, area, and slope parameters to a standard normal distribution, eliminating individual differences and obtaining a normalized parameter set. From the normalized parameter set, the mean and standard deviation of each parameter are obtained. A pre-defined statistical model is used to calculate the normal range. If any parameter value exceeds the mean plus or minus two standard deviations, it is marked as an anomalous feature, resulting in an anomalous label set. Based on the anomalous label set, k-means clustering is used to classify the waveform components into normal and anomalous categories, resulting in a classification feature set. From the classification feature set, the amplitude, duration, area, and slope of the anomalous features are extracted and combined into a standardized feature parameter vector, resulting in the final feature vector. For the final feature vector, principal component analysis is used to reduce dimensionality to retain key information, resulting in a dimensionality-reduced feature vector.

[0087] When calculating waveform parameters, amplitude, duration, area, and slope are key indicators. Taking the QRS complex in an electrocardiogram as an example, assuming its amplitude is 1.5 mV and its duration is 0.08 seconds, the area of ​​the QRS complex is calculated using numerical integration, yielding approximately 0.06 mV·s. The slope is calculated by dividing the amplitude difference between adjacent sampling points by the time interval, resulting in approximately 18.75 mV / s. These parameters reflect the morphological characteristics of the waveform, providing a quantitative basis for subsequent analysis. Numerical integration preserves the complete information of the waveform by accumulating the product of amplitude and time interval between sampling points.

[0088] Example 2

[0089] A feature extraction system based on electrocardiogram (ECG) signals, the system comprising:

[0090] The data acquisition module is used to acquire electrocardiogram (ECG) signals for ECG images.

[0091] The raw electrocardiogram signal is acquired by the acquisition device and stored as digital sequence data to obtain the initial signal data.

[0092] A professional medical 12-lead electrocardiograph (ECG) is used as the acquisition device, equipped with a high-quality bioelectric amplifier, filter, and analog-to-digital converter. First, disposable ECG electrodes are attached to designated locations on the patient's body, including the limbs (both arms and both legs) and specific areas of the chest wall (V1 to V6 chest leads). The electrodes are connected to the ECG's input ports via lead wires, ensuring a secure and reliable connection to avoid interference and signal loss during transmission.

[0093] Once the electrocardiograph (ECG) machine starts its acquisition program, it begins acquiring the raw ECG signal generated by the patient's heart. A bioelectric amplifier amplifies these weak electrical signals, bringing them within the amplitude range that the analog-to-digital converter (ADC) can effectively process. A filter removes environmental interference signals, such as 50Hz power line interference and low-frequency baseline drift noise, ensuring signal purity and accuracy. The ADC then converts the continuous analog ECG signal into discrete digital sequence data at a specific sampling frequency (e.g., 500Hz or 1000Hz).

[0094] The acquired digital sequence data is stored in real time on the hard drive of a computer connected to the electrocardiograph (ECG) machine, saved in a specific file format (such as EDF). This initial signal data contains information such as the time and amplitude of various ECG waveforms (such as P waves, QRS complexes, T waves, etc.), reflecting the electrical activity state of the patient's heart. In the initial processing stage, simple filtering and noise reduction may be performed on the data to further improve signal quality and provide accurate basic data for subsequent ECG image generation and analysis.

[0095] The preliminary decomposition module is used to preprocess the electrocardiogram signal to obtain a preprocessed signal, and to decompose the preprocessed signal to obtain preliminary decomposition features of the signal waveform.

[0096] Preprocessing includes:

[0097] After denoising the electrocardiogram signal, data reconstruction is performed to obtain the preprocessed signal:

[0098] Variational mode decomposition of electrocardiogram signals:

[0099] Initial IMF components are obtained based on electrocardiogram signals and signal envelopes;

[0100] Calculate the average value of the signal envelope:

[0101]

[0102] The electrocardiogram signal is x(t), and the maximum envelope e is fitted from the maximum points. + (t), minimum envelope e - (t).

[0103] The sequence difference is calculated as follows:

[0104]

[0105] The IMF component is calculated as follows:

[0106]

[0107] The updated electrocardiogram signal is obtained based on the electrocardiogram signal and the IMF component;

[0108] r1(t) = x(t) - c1(t)

[0109] The final IMF component is obtained based on the updated electrocardiogram signal;

[0110] Based on the final IMF components, reconstructed data is obtained, resulting in a preprocessed signal.

[0111] The preliminary decomposition characteristics of the signal waveform are as follows:

[0112]

[0113] Among them, STFTP xy (t, f) represents the short-time Fourier cross-power spectrum data; STFTP x (t, f) represents the short-time Fourier autopower spectrum data of the random signal x(t); STFTP y (t,f) represents the short-time Fourier autopower spectrum data of the random signal y(t).

[0114] STFTP x The calculation method for (t,f) is as follows:

[0115]

[0116] The same calculation method is used for STFTP. y The calculation is performed using (t,f).

[0117] The waveform separation module is used to obtain the separation matrix and reconstruction parameters of the electrocardiogram waveform components based on the preliminary decomposition features and using the waveform independence judgment model.

[0118] A statistical independence model of electrocardiogram (ECG) waveforms is constructed using the independent component analysis (ICA) algorithm. If the mutual information between two waveform components is lower than the independence threshold, the separation is considered successful, and the accurate separation matrix and reconstruction parameters of each ECG waveform component are obtained.

[0119] Independent Component Analysis (ICA) is used to preprocess the input electrocardiogram (ECG) signal, decomposing it into multiple waveform components to obtain an initial separation matrix. Feature vectors for each waveform component are extracted from the initial separation matrix, and the mutual information between each pair of waveform components is calculated. If the mutual information is below a preset independence threshold, the corresponding waveform component is determined to be statistically independent, resulting in a set of independent waveform components. Based on this set, the initial separation matrix is ​​optimized to generate an accurate separation matrix. A linear transformation is performed on the original ECG signal using the accurate separation matrix to obtain the reconstructed independent waveform components. Time-series features are extracted from the reconstructed independent waveform components to generate corresponding reconstruction parameters. The reconstructed parameters are used to verify the separated waveform components, determining the signal integrity of each component and obtaining the final separation result.

[0120] The boundary definition module is used to obtain waveform boundary positioning results based on the separation matrix and reconstruction parameters.

[0121] Based on the separation matrix, the waveform components are reconstructed in the time domain. The precise time position of the start point, peak point and end point of each waveform is determined by the peak detection algorithm. If the waveform amplitude change rate exceeds the preset gradient threshold, it is marked as a key feature point, and the complete waveform boundary positioning result is obtained.

[0122] The separation matrix is ​​obtained from the input signal, and independent component analysis (ICA) is used to decompose each independent waveform component, resulting in a time-domain signal set. Gaussian filtering is then applied to smooth each waveform using this time-domain signal set, yielding a smoothed time-domain signal. For the smoothed time-domain signal, a peak detection algorithm is applied to identify the start, peak, and end points of each waveform, determining a set of time positions. The amplitude changes between adjacent points are obtained from the time position set, and the rate of change of amplitude is calculated, resulting in an amplitude change rate sequence. If the rate of change of amplitude exceeds a preset gradient threshold, the corresponding point is marked as a key feature point, obtaining a set of key feature points. Based on the set of key feature points and the time position set, the boundaries of each waveform are defined, generating waveform boundary localization results. The consistency between the time positions and key feature points of each waveform is verified using the waveform boundary localization results, yielding the final waveform boundary localization result.

[0123] The feature extraction module is used to extract features from the electrocardiogram signal based on the waveform boundary positioning results to obtain electrocardiogram signal features.

[0124] Based on the waveform boundary location results, quantitative parameters such as amplitude, duration, area, and slope of each waveform component are calculated. The influence of individual differences is eliminated through parameter normalization. If the parameter value deviates from the normal range by more than two standard deviations, it is marked as an abnormal feature, and a standardized feature parameter vector is obtained.

[0125] The waveform boundary localization results are obtained from the waveform data. Signal processing algorithms are used to determine the start and end points of each waveform component, thus obtaining the waveform boundary localization. Based on the waveform boundary localization, the amplitude, duration, area, and slope of each waveform component are calculated. Numerical integration is used to calculate the area, resulting in a waveform parameter set. For this parameter set, z-score normalization is applied to map the amplitude, duration, area, and slope parameters to a standard normal distribution, eliminating individual differences and obtaining a normalized parameter set. From the normalized parameter set, the mean and standard deviation of each parameter are obtained. A pre-defined statistical model is used to calculate the normal range. If any parameter value exceeds the mean plus or minus two standard deviations, it is marked as an anomalous feature, resulting in an anomalous label set. Based on the anomalous label set, k-means clustering is used to classify the waveform components into normal and anomalous categories, resulting in a classification feature set. From the classification feature set, the amplitude, duration, area, and slope of the anomalous features are extracted and combined into a standardized feature parameter vector, resulting in the final feature vector. For the final feature vector, principal component analysis is used to reduce dimensionality to retain key information, resulting in a dimensionality-reduced feature vector.

[0126] When calculating waveform parameters, amplitude, duration, area, and slope are key indicators. Taking the QRS complex in an electrocardiogram as an example, assuming its amplitude is 1.5 mV and its duration is 0.08 seconds, the area of ​​the QRS complex is calculated using numerical integration, yielding approximately 0.06 mV·s. The slope is calculated by dividing the amplitude difference between adjacent sampling points by the time interval, resulting in approximately 18.75 mV / s. These parameters reflect the morphological characteristics of the waveform, providing a quantitative basis for subsequent analysis. Numerical integration preserves the complete information of the waveform by accumulating the product of amplitude and time interval between sampling points.

[0127] Example 3

[0128] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the feature extraction method based on electrocardiogram signals described in any of the above embodiments.

[0129] Figure 2 This embodiment illustrates a more specific hardware structure of an electronic device, which may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0130] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0131] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0132] The input / output interface 1030 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components within the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touchscreens, microphones, various sensors, etc., while output devices may include displays, speakers, vibrators, indicator lights, etc.

[0133] The communication interface 1040 is used to connect the communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB (Universal Serial Bus), network cable, etc.) or wireless means (such as mobile network, WIFI (Wireless Fidelity), Bluetooth, etc.).

[0134] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0135] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0136] The system described in the above embodiments is used to implement the corresponding feature extraction method based on electrocardiogram signals in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0137] Example 4

[0138] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this disclosure also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the feature extraction method based on electrocardiogram signals as described in any of the above embodiments.

[0139] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0140] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the feature extraction method based on electrocardiogram signals as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0141] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this disclosure (including the claims) is limited to these examples; within the framework of this disclosure, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this disclosure as described above, which are not provided in detail for the sake of brevity.

[0142] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this disclosure, the well-known power / ground connections to the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this disclosure, and this also takes into account the fact that the details of implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this disclosure will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this disclosure, it will be apparent to those skilled in the art that the embodiments of this disclosure may be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0143] Although this disclosure has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0144] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A feature extraction method based on electrocardiogram (ECG) signals, characterized in that, The method comprises: Step S1: Acquire electrocardiogram (ECG) signals to form an ECG image; Step S2: Preprocess the electrocardiogram signal to obtain a preprocessed signal, and decompose the preprocessed signal to obtain preliminary decomposition features of the signal waveform; Step S3: Based on the preliminary decomposition features, use the waveform independence judgment model to obtain the separation matrix and reconstruction parameters of the electrocardiogram waveform components; Step S4: Obtain the waveform boundary localization result based on the separation matrix and reconstruction parameters; Step S5: Based on the waveform boundary positioning results, perform feature extraction on the electrocardiogram signal to obtain electrocardiogram signal features.

2. The feature extraction method based on electrocardiogram signals according to claim 1, characterized in that, In step S2, the preprocessing includes: After denoising the electrocardiogram signal, data reconstruction is performed to obtain the preprocessed signal: Variational mode decomposition of electrocardiogram signals: Initial IMF components are obtained based on electrocardiogram signals and signal envelopes; The updated electrocardiogram signal is obtained based on the electrocardiogram signal and the IMF component; The final IMF component is obtained based on the updated electrocardiogram signal; Based on the final IMF components, reconstructed data is obtained, resulting in a preprocessed signal.

3. The feature extraction method based on electrocardiogram signals according to claim 1, characterized in that, In step S2, the preliminary decomposition features of the signal waveform are obtained as follows: Among them, STFTP xy (t,f) represents the short-time Fourier cross-power spectrum data; STFTP x (t,f) represents the short-time Fourier autopower spectrum of the random signal x(t); STFTP y (t,f) represents the short-time Fourier autopower spectrum data of the random signal y(t).

4. The feature extraction method based on electrocardiogram signals according to claim 3, characterized in that, Step S3 specifically involves: A statistical independence model of electrocardiogram (ECG) waveforms is constructed using the independent component analysis (ICA) algorithm. If the mutual information between two waveform components is lower than the independence threshold, the separation is considered successful, and the accurate separation matrix and reconstruction parameters of each ECG waveform component are obtained.

5. A feature extraction system based on electrocardiogram signals, the system being used to implement the method according to any one of claims 1-4, characterized in that, The system includes: The data acquisition module is used to acquire electrocardiogram (ECG) signals for ECG images; The preliminary decomposition module is used to preprocess the electrocardiogram signal to obtain a preprocessed signal, and to decompose the preprocessed signal to obtain preliminary decomposition features of the signal waveform. The waveform separation module is used to obtain the separation matrix and reconstruction parameters of the electrocardiogram waveform components based on the preliminary decomposition features and using the waveform independence judgment model. A boundary definition module is used to obtain waveform boundary positioning results based on the separation matrix and reconstruction parameters; The feature extraction module is used to extract features from the electrocardiogram signal based on the waveform boundary positioning results to obtain electrocardiogram signal features.

6. The feature extraction system based on electrocardiogram signals according to claim 5, characterized in that, Preprocessing includes: After denoising the electrocardiogram signal, data reconstruction is performed to obtain the preprocessed signal: Variational mode decomposition of electrocardiogram signals: Initial IMF components are obtained based on electrocardiogram signals and signal envelopes; The updated electrocardiogram signal is obtained based on the electrocardiogram signal and the IMF component; The final IMF component is obtained based on the updated electrocardiogram signal; Based on the final IMF components, reconstructed data is obtained, resulting in a preprocessed signal.

7. The feature extraction system based on electrocardiogram signals according to claim 5, characterized in that, The preliminary decomposition characteristics of the signal waveform are as follows: Among them, STFTP xy (t, f) represents the short-time Fourier cross-power spectrum data; STFTP x (t, f) represents the short-time Fourier autopower spectrum data of the random signal x(t); STFTP y (t,f) represents the short-time Fourier autopower spectrum data of the random signal y(t).

8. The feature extraction system based on electrocardiogram signals according to claim 7, characterized in that, The specific workflow of the waveform separation module is as follows: A statistical independence model of electrocardiogram (ECG) waveforms is constructed using the independent component analysis (ICA) algorithm. If the mutual information between two waveform components is lower than the independence threshold, the separation is considered successful, and the accurate separation matrix and reconstruction parameters of each ECG waveform component are obtained.

Images