A feature extraction method and system for detecting coronary heart disease

Abstract

The present invention proposes a feature extraction method and system for detecting coronary heart disease, which relates to the field of biotechnology analysis. The method comprises: collecting synchronous electrocardiogram, heart sound and limb pulse signals of a test subject; preprocessing the collected electrocardiogram, heart sound and pulse signals; constructing an analytical function through Hilbert transform, and obtaining the instantaneous phase sequences of the preprocessed electrocardiogram, heart sound and pulse signals respectively based on the analytical function; extracting the independent phase features of a single signal and the joint phase features of two signals based on the instantaneous phase sequences of the electrocardiogram, heart sound and pulse signals; the present invention extracts the instantaneous phase sequences of synchronous electrocardiogram, heart sound and limb pressure pulse signals, extracts the independent phase features of a single signal and the joint phase features of multiple signals from multiple directions, and can significantly improve the detection accuracy.

Description

Technical Field

[0001] The invention belongs to the field of biotechnology analysis, and in particular relates to a feature extraction method and system for detecting coronary heart disease. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Coronary heart disease is a common cardiovascular disease and one of the main causes of rising global mortality. Plaque accumulation in the coronary arteries causes coronary artery stenosis, which reduces blood flow to the myocardium and leads to the occurrence of coronary heart disease. In severe cases, plaque rupture completely blocks the coronary artery lumen, causing acute myocardial infarction and even death.

[0004] Currently, coronary angiography is considered the "gold standard" for detecting coronary heart disease, but this test is an invasive technology with certain risks, requires professional surgical operations, and is expensive. Therefore, this test is usually only used for patients suspected of having severe coronary artery blockage and is not suitable for early detection and large-scale screening.

[0005] ECG and heart sound signals reflect the electrical and mechanical activities of the heart, respectively, and the pulse signal records the status of the cardiovascular system from the heart to the peripheral arteries. Studies have shown that in the resting ECG of patients with coronary heart disease, there may be symptoms such as T wave inversion, ST-T segment abnormalities, left ventricular hypertrophy and ventricular premature beats. In the heart sound signals of patients with coronary heart disease, coronary artery stenosis will lead to an increase in high-frequency components in the diastolic period. In the resting pulse signals of patients with coronary heart disease, there are characteristics such as changes in time domain intervals, changes in waveform morphology, shortened pulse wave propagation time and faster propagation speed. Therefore, ECG, heart sound and pulse signals are non-invasive, efficient, easy to operate and low-cost, and can be used as a promising tool for non-invasive screening of coronary heart disease.

[0006] There have been many studies on the use of ECG, heart sounds and pulse signals to detect coronary heart disease. In the research on ECG and heart sound signals, the focus is mainly on the traditional feature extraction and classification process of time series, which has a low accuracy rate; in the research on pulse signals, features such as propagation time, propagation speed, waveform characteristics, sclerosis index and cardiac output estimation are used to analyze coronary heart disease, but these studies are based on the extraction of pulse signal feature points and the estimation of arterial blood vessel length, and there is a certain error in accuracy; in fact, it is far from enough to use only one of the ECG, heart sounds or pulse signals to detect coronary heart disease, because some patients do not show abnormalities in ECG signals, but they are reflected in heart sounds and pulse signals, and vice versa.

[0007] At present, the research on coronary heart disease detection using electrocardiogram, heart sounds and pulse mainly extracts the time domain, frequency domain, time-frequency domain and nonlinear domain characteristics of time series for analysis. When the amplitude of the time series changes slightly, its characteristics do not change significantly. Therefore, there are limitations in analyzing the time series characteristics alone. Studies have shown that the phase characteristics of physiological signals contain rich physiological and pathological information, which can be used to determine the location of signal feature points and the synchronization of signals. The T wave of the electrocardiogram of patients with coronary heart disease is inverted, and its phase changes in the opposite direction. The increase in high-frequency components of heart sounds is weak in the time series characteristics, but changes significantly in the phase characteristics. The high correlation of time domain sequences between pulse signals reduces the correlation in the phase characteristics. Therefore, how to combine the respective advantages of electrocardiogram, heart sounds and pulse to extract high-quality features is a topic worthy of study. Summary of the invention

[0008] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a feature extraction method and system for detecting coronary heart disease, which extracts the instantaneous phase sequence of synchronous electrocardiogram, heart sound and limb pressure pulse signals, extracts the independent phase features of a single signal and the joint phase features of multiple signals from multiple directions, and can significantly improve the detection accuracy.

[0009] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0010] A first aspect of the present invention provides a feature extraction method for detecting coronary heart disease;

[0011] A feature extraction method for detecting coronary heart disease, comprising:

[0012] Collect the synchronous ECG, heart sounds and limb pulse signals of the tested person;

[0013] Pre-process the collected ECG, heart sound and pulse signals;

[0014] An analytical function is constructed through Hilbert transform, and the instantaneous phase sequences of the preprocessed electrocardiogram, heart sound and pulse signals are obtained based on the analytical function;

[0015] Based on the instantaneous phase sequences of ECG, heart sound and pulse signals, the independent phase features of single signals and the joint phase features of two signals are extracted.

[0016] Furthermore, in the preprocessing, a Butterworth bandpass filter is used to remove noise interference in the signal, an IIR notch filter is used to remove 50 Hz power frequency interference in the signal, and a Z-score method is used to normalize the filtered signal and divide it into 10 s segments.

[0017] Furthermore, the instantaneous phase sequence is obtained by:

[0018] The preprocessed ECG, heart sound and pulse signals are subjected to Hilbert transform, and the analytical functions of the ECG, heart sound and pulse signals are constructed respectively. The arc tangent of the ratio of the imaginary part to the real part of the analytical function is calculated and rounded to obtain the instantaneous phase of the ECG, heart sound and pulse signals.

[0019] Furthermore, the independent phase characteristics of the single signal include independent phase statistical characteristics, independent phase distribution entropy and independent phase arrangement entropy.

[0020] Furthermore, the joint phase characteristics of the pairwise signals include joint phase distribution entropy, joint phase arrangement entropy, phase transfer entropy and joint phase linear correlation characteristics.

[0021] Furthermore, the steps of obtaining the joint phase arrangement entropy are:

[0022] Reconstruct the instantaneous phases of the two signals in phase space respectively;

[0023] Construct a reconstruction matrix based on the index value sorted by the size of each component;

[0024] The instantaneous phase θ a (n) and θ b (n) Each reconstruction component in the reconstruction matrix is ​​arranged according to the numerical value, and the obtained sorted index values ​​constitute the reconstruction matrix of the k components of the two groups of symbol sequences;

[0025] According to the number of times the state synchronously appears in the reconstruction matrix, a joint state distribution matrix is ​​constructed;

[0026] Calculate the joint phase alignment entropy.

[0027] Furthermore, the pairwise signals include electrocardiogram and heart sound, electrocardiogram and pulse, heart sound and pulse, and pulse and pulse.

[0028] A second aspect of the present invention provides a feature extraction system for detecting coronary heart disease.

[0029] A feature extraction system for detecting coronary heart disease, comprising a signal acquisition module, a signal preprocessing module, a phase extraction module and a feature extraction module:

[0030] The signal acquisition module is configured to: acquire synchronized ECG, heart sound and limb pulse signals of the tested person;

[0031] The signal preprocessing module is configured to: preprocess the collected electrocardiogram, heart sound and pulse signals;

[0032] The phase extraction module is configured to: construct an analytical function through Hilbert transform, and obtain instantaneous phase sequences of the preprocessed electrocardiogram, heart sound and pulse signals respectively based on the analytical function;

[0033] The feature extraction module is configured to extract the independent phase features of a single signal and the joint phase features of two signals based on the instantaneous phase sequences of the electrocardiogram, heart sound and pulse signals.

[0034] The third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in a feature extraction method for detecting coronary heart disease as described in the first aspect of the present invention.

[0035] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in a feature extraction method for detecting coronary heart disease as described in the first aspect of the present invention are implemented.

[0036] One or more of the above technical solutions have the following beneficial effects:

[0037] The present invention extracts the instantaneous phase sequence of synchronous electrocardiogram, heart sound and limb pressure pulse signals, extracts the independent phase features of a single signal from multiple directions, and extracts the joint phase features of two signals through the joint distribution of the instantaneous phases of two signals. The independent phase features of multiple signals can capture possible useful information from multiple directions, and the joint phase features of two signals provide synchronization information. The extracted phase features can significantly improve the detection accuracy.

[0038] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings in the specification, 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.

[0040] Figure 1 This is a flow chart of the method of the first embodiment.

[0041] Figure 2 This is a preprocessing flow chart in the first embodiment.

[0042] Figure 3 Schematic diagram of nine modes of phase arrangement entropy in the first embodiment.

[0043] Figure 4 It is a system structure diagram of the second embodiment. DETAILED DESCRIPTION

[0044] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0045] It should be noted that the following detailed descriptions are all exemplary and are intended to provide further explanation of the present invention; unless otherwise specified, all technical and scientific terms used herein have the same meanings as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0046] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention; as used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations.

[0047] Embodiment 1

[0048] This embodiment discloses a feature extraction method for detecting coronary heart disease;

[0049] like Figure 1 As shown, a feature extraction method for detecting coronary heart disease comprises:

[0050] Step S1: collecting the synchronous ECG, heart sounds and limb pulse signals of the tested person;

[0051] The cardiovascular function detection device CVFD-II developed by Jinan Huiyirongong Technology Co., Ltd. is used to collect the synchronous ECG, heart sounds and limb pulse signals of the test subject, and to synchronously inflate and pressurize the limb arteries. After the limb pulse signals are stable, the ECG, heart sounds and limb pulse signals are synchronously intercepted.

[0052] Step S2: pre-processing the collected ECG, heart sound and pulse signals, including filtering, denoising, A / D conversion, normalization and segmentation, such as Figure 2 As shown;

[0053] A Butterworth bandpass filter (1-60 Hz) was used to remove the baseline drift and high-frequency noise interference of the ECG signal, an IIR notch filter was used to remove the 50 Hz power frequency interference of the ECG signal, and the Z-score method was used to normalize the filtered ECG signal and divide it into 10 s segments.

[0054] A Butterworth high-pass filter (20 Hz) was used to remove the low-frequency noise of the heart sound signal, an IIR notch filter was used to remove the 50 Hz power frequency interference of the heart sound signal, and the Z-score method was used to normalize the filtered heart sound signal and divide it into 10 s segments.

[0055] A Butterworth bandpass filter (0.3-43 Hz) was used to remove the baseline drift and high-frequency noise interference of the pulse signal, and the filtered pulse signal was normalized and segmented into 10 s segments using the Z-score method.

[0056] Step S3: construct an analytical function through Hilbert transform, and obtain the instantaneous phase sequences of the preprocessed electrocardiogram, heart sound and pulse signals respectively based on the analytical function;

[0057] The preprocessed ECG, heart sound and pulse signals are subjected to Hilbert transform, and the analytical functions of the ECG, heart sound and pulse signals are constructed respectively. The arc tangent of the ratio of the imaginary part to the real part of the analytical function is calculated and rounded to obtain the instantaneous phase of the ECG, heart sound and pulse signals.

[0058] Signal x a The Hilbert transform of (t) is:

[0059]

[0060] Among them, t is the time variable and τ is the integral term time variable.

[0061] The constructed parsing function is:

[0062]

[0063] Here, j is a complex number symbol.

[0064] Then x a The instantaneous phase of (t) is:

[0065]

[0066] Among them, x a (t) is the ECG, heart sound or pulse signal, θ a (t) is the instantaneous phase of the ECG, heart sound or pulse signal.

[0067] The instantaneous phase θ a (t) is rounded to get the instantaneous phase sequence θ a (n), extract phase features.

[0068] Step S4: Based on the instantaneous phase sequences of the ECG, heart sound and pulse signals, the independent phase features of the single signals and the joint phase features of the two signals are extracted.

[0069] Step S401: Extraction of independent phase features

[0070] The independent phase characteristics of a single signal include independent phase statistical characteristics, independent phase distribution entropy and independent phase arrangement entropy:

[0071] (1) Independent phase statistical characteristics

[0072] Equal interval statistical processing is performed on the instantaneous phase of the electrocardiogram, heart sound, pulse and the instantaneous phase difference of the pulse, and 10 statistical features are extracted from each signal, for a total of 120 features.

[0073] The instantaneous phase difference of the pulse is the difference between the instantaneous phases of the pulse signals of the four limbs.

[0074] (2) Independent phase distribution entropy

[0075] The instantaneous phase sequence θ is obtained a The (n) value is windowed and divided into 180 windows with a width of 2°, namely (-180°,-178°], (-178°,-176°], ..., (178°, 180°), and the number of phases in each window is counted to construct an independent phase distribution and calculate the independent phase distribution entropy SphasedistrEn:

[0076]

[0077] Among them, p i is the distribution probability of the phase in each window.

[0078] (3) Independent phase arrangement entropy

[0079] The instantaneous phase θ a (n) Reconstruct the phase space, embedding dimension is m = 3, delay time is τ = 1, and the phase space is based on Figure 2 As shown, it is divided into 9 states, and the distribution probabilities of the 9 states are counted. The calculation steps of the independent phase arrangement entropy SphasepermuEn are:

[0080] a. Phase space reconstruction: instantaneous phase θ a (n) Sequentially extract three adjacent sample points with a delay time of τ = 1, and construct a k×3-dimensional phase space matrix:

[0081]

[0082] b. Extract symbol sequence

[0083] Each reconstructed row vector is arranged according to the numerical value, and the obtained sorted index values ​​constitute a set of symbol sequence k×3-dimensional reconstruction matrices.

[0084] The rows of the matrix are used as reconstruction components, and each instantaneous phase sequence constitutes a total of k reconstruction components; after the reconstruction is completed, the three values ​​of each reconstruction component are compared, and a total of Figure 2 9 states.

[0085] c. Count the k components in Figure 3 The number of times a state appears in the set 9 states is used to calculate the probability of the 9 states. The ratio of the number of times each state appears to the total number of times is the probability corresponding to each state.

[0086] d. Calculate the entropy of independent phase arrangement:

[0087]

[0088] Among them, p i is the probability of each of the 9 states arranged in the instantaneous phase sequence.

[0089] Step S402: Joint phase feature extraction

[0090] The joint phase features of pairwise signals include joint phase distribution entropy, joint phase arrangement entropy, phase transfer entropy and joint phase linear correlation features.

[0091] (1) Joint phase distribution entropy: reflects the synchronization of two instantaneous phase sequences.

[0092] The two instantaneous phases θ are obtained a (n) and θ b (n) is divided into 180 windows with a width of 2°, respectively (-180°, -178°], (-178°, -176°], ..., (178°, 180°), and constructs (θ a (n),θ b (n))Joint phase probability distribution, calculate the joint phase distribution entropy MphasedistrEn:

[0093]

[0094] Among them, p i,j is the instantaneous phase θ a (n) and θ b (n) Joint distribution probability.

[0095] (2) Joint phase permutation entropy: characterizes the consistency of the modes of two phase sequences.

[0096] The instantaneous phase θ a (n) and θ b (n) Reconstruct the phase space respectively, embedding dimension m = 3, delay time τ = 1, and each phase space is based on Figure 2 Divided into 9 states, the joint phase arrangement probability of the two instantaneous phase sequences is counted, and the joint phase arrangement entropy MphasepermuEn is calculated:

[0097] a. Instantaneous phase θ a (n) and θ b (n) Phase space reconstruction: instantaneous phase θa (n) and θ b (n) Three adjacent sample points are sequentially extracted with a delay time of τ=1, and two sets of k×3 dimensional phase space matrices are constructed accordingly.

[0098]

[0099] b. Extract symbol sequence

[0100] The instantaneous phase sequence θ a (n) and θ b (n) Each reconstruction component in the reconstruction matrix is ​​arranged according to the numerical value, and the obtained sorted index values ​​constitute 2 groups of symbol sequence k×3 reconstruction matrices.

[0101] c. Construct the corresponding positions of the two sets of symbol sequences as follows Figure 3 The number of times the 9 states appear at the same time is used to construct the joint state distribution matrix, and the probability of the joint state in the joint distribution matrix is ​​calculated. The ratio of the number of joint state times to the total number of times is the probability corresponding to the joint state distribution.

[0102] d. Calculate the joint phase permutation entropy MphasepermuEn:

[0103]

[0104] Among them, p i,j is the instantaneous phase θ a (n) and θ b The joint distribution probability of the 9 states in (n).

[0105] (3) Phase transfer entropy: It reflects the coupling degree and directionality of the two phase sequences using conditional probability distribution.

[0106] The two instantaneous phase sequences obtained are reconstructed in phase space respectively, with a delay time of τ = 10, and three phase spaces θ are constructed. a (n) = [θ a (1),θ a (2),...,θ a (n-τ)], θ b (n) = [θ b (1),θ b (2),...,θ b (n-τ)] and θ a (n+τ)=[θ a (τ+1),θ a (τ+2),...,θ a (n)], each phase space is divided into 180 windows with a width of 2°, the phase space joint phase distribution matrix is ​​constructed, and θ is calculated. a (n) and θb Phase transfer entropy PhasetransfEn of (n):

[0107] PhasetransfEn(θ b (n)→θ a (n))

[0108] =H(θ a (n+τ),θ a (n))+H(θ a (n),θ b (n))-H(θ a (n)-H(θ a (n+τ),θ a (n),θ b (n))

[0109]

[0110]

[0111]

[0112]

[0113] Among them, p(θ a (n+τ),θ a (n))、p(θ a (n),θ b (n))、p(θ a (n))、p(θ a (n+τ),θ a (n),θ b (n)) respectively represent the corresponding (θ a (n+τ),θ a (n))、(θ a (n),θ b (n))、θ a (n), (θ a (n+τ),θ a (n),θ b (n))'s joint phase distribution probability.

[0114] (4) Joint phase linear correlation feature: The mutual correlation coefficient and mean square error of two instantaneous phases reflect the synchronization and consistency of two signals.

[0115] Cross Correlation Coefficient (CC):

[0116]

[0117] Where N is the number of signal sampling points. and The instantaneous phase θ is a (n) and θ b (n)The mean and variance of

[0118] Root Mean Square Error (RMSE):

[0119]

[0120] Where N is the number of signal sampling points.

[0121] This embodiment uses data from 189 subjects collected clinically, including 129 patients with coronary heart disease (blockage greater than 50%) and 60 patients without coronary heart disease (blockage less than or equal to 50%); at the same time, the coronary artery blockage is divided into different degrees: 111 patients with severe coronary heart disease (blockage greater than or equal to 75%), 46 patients with mild coronary heart disease (blockage 30% to 75%) and 32 patients without coronary heart disease; the data of each subject includes electrocardiogram, heart sound and limb pressure pulse signals, which are preprocessed and instantaneous phase extracted according to steps S1 and S2, and independent phase features and joint phase features are extracted from the instantaneous phases of the electrocardiogram, heart sound and limb pressure pulse signals according to step S3, and the extracted phase features are input into a nonlinear classification model to obtain the coronary heart disease detection results.

[0122] This embodiment extracts phase features from instantaneous signals of ECG, heart sounds and limb pressure pulse signals for classification, and the average classification accuracy of non-coronary heart disease and coronary heart disease is 94.8%, sensitivity is 96.4%, and specificity is 86.9%. At the same time, the classification accuracy of non-coronary heart disease, mild coronary heart disease and severe coronary heart disease is 89.3%; the results show that phase features have a good effect on analyzing coronary heart disease, and phase features can be used as a quantitative indicator for coronary heart disease risk assessment.

[0123] Embodiment 2

[0124] This embodiment discloses a feature extraction system for detecting coronary heart disease;

[0125] like Figure 4 As shown, a feature extraction system for detecting coronary heart disease includes a signal acquisition module, a signal preprocessing module, a sequence extraction module and a feature extraction module:

[0126] The signal acquisition module is configured to: acquire synchronized ECG, heart sound and limb pulse signals of the tested person;

[0127] The signal preprocessing module is configured to: preprocess the collected ECG, heart sound and pulse signals by filtering, denoising, A / D conversion, normalization and segmentation;

[0128] The phase extraction module is configured to: construct an analytical function through Hilbert transform, and obtain instantaneous phase sequences of the preprocessed electrocardiogram, heart sound and pulse signals respectively based on the analytical function;

[0129] The feature extraction module is configured to extract the independent phase features of a single signal and the joint phase features of two signals based on the instantaneous phase sequences of the electrocardiogram, heart sound and pulse signals.

[0130] Embodiment 3

[0131] The purpose of this embodiment is to provide a computer-readable storage medium.

[0132] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in a feature extraction method for detecting coronary heart disease as described in Example 1 of the present disclosure.

[0133] Embodiment 4

[0134] The purpose of this embodiment is to provide an electronic device.

[0135] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps in a feature extraction method for detecting coronary heart disease as described in Example 1 of the present disclosure are implemented.

[0136] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A feature extraction method for detecting coronary heart disease, characterized in that: include: Collect the synchronous ECG, heart sounds and limb pulse signals of the tested person; Pre-process the collected ECG, heart sound and pulse signals; The preprocessing includes filtering, denoising, A / D conversion, normalization and segmentation; An analytical function is constructed through Hilbert transform, and the instantaneous phase sequences of the preprocessed electrocardiogram, heart sound and pulse signals are obtained based on the analytical function; The method for acquiring the instantaneous phase sequence is: Perform Hilbert transform on the preprocessed ECG, heart sound and pulse signals, construct analytical functions of the ECG, heart sound and pulse signals respectively, calculate the arc tangent of the ratio of the imaginary part to the real part of the analytical function and round it to obtain the instantaneous phase of the ECG, heart sound and pulse signals; Based on the instantaneous phase sequence of ECG, heart sound and pulse signals, the independent phase features of single signals and the joint phase features of two signals are extracted; The joint phase characteristics of the pairwise signals include joint phase distribution entropy, joint phase arrangement entropy, phase transfer entropy and joint phase linear correlation characteristics; The pairwise signals include electrocardiogram and heart sound, electrocardiogram and pulse, heart sound and pulse, and pulse and pulse; The joint phase linear correlation feature reflects the synchronization and consistency of the two signals with the mutual correlation coefficient and mean square error of the two instantaneous phases; The steps for obtaining the joint phase arrangement entropy are: Reconstruct the instantaneous phases of the two signals in phase space respectively; Construct a reconstruction matrix based on the index value sorted by the size of each component; The instantaneous phase θ a (n) and θ b (n) Each reconstruction component in the reconstruction matrix is ​​arranged according to the numerical value, and the obtained sorted index values ​​constitute the reconstruction matrix of the k components of the two groups of symbol sequences; According to the number of times the state synchronously appears in the reconstruction matrix, a joint state distribution matrix is ​​constructed; Calculate the joint phase alignment entropy.

2. A feature extraction method for detecting coronary heart disease according to claim 1, characterized in that: The preprocessing uses a Butterworth bandpass filter to remove noise interference in the signal, uses an IIR notch filter to remove 50 Hz power frequency interference in the signal, and uses a Z-score method to normalize the filtered signal and divide it into 10 s segments.

3. A feature extraction method for detecting coronary heart disease as claimed in claim 1, characterized in that: The independent phase characteristics of the single signal include independent phase statistical characteristics, independent phase distribution entropy and independent phase arrangement entropy.

4. A feature extraction system for detecting coronary heart disease, characterized in that: Including signal acquisition module, signal preprocessing module, phase extraction module and feature extraction module: The signal acquisition module is configured to: acquire synchronized ECG, heart sound and limb pulse signals of the tested person; The signal preprocessing module is configured to: preprocess the collected electrocardiogram, heart sound and pulse signals; The preprocessing includes filtering, denoising, A / D conversion, normalization and segmentation; The phase extraction module is configured to: construct an analytical function through Hilbert transform, and obtain instantaneous phase sequences of the preprocessed electrocardiogram, heart sound and pulse signals respectively based on the analytical function; The method for acquiring the instantaneous phase sequence is: Perform Hilbert transform on the preprocessed ECG, heart sound and pulse signals, construct analytical functions of the ECG, heart sound and pulse signals respectively, calculate the arc tangent of the ratio of the imaginary part to the real part of the analytical function and round it to obtain the instantaneous phase of the ECG, heart sound and pulse signals; The feature extraction module is configured to: extract the independent phase feature of a single signal and the joint phase feature of two signals based on the instantaneous phase sequence of the electrocardiogram, heart sound and pulse signal; The joint phase characteristics of the pairwise signals include joint phase distribution entropy, joint phase arrangement entropy, phase transfer entropy and joint phase linear correlation characteristics; The pairwise signals include electrocardiogram and heart sound, electrocardiogram and pulse, heart sound and pulse, and pulse and pulse; The joint phase linear correlation feature reflects the synchronization and consistency of the two signals with the mutual correlation coefficient and mean square error of the two instantaneous phases; The steps for obtaining the joint phase arrangement entropy are: Reconstruct the instantaneous phases of the two signals in phase space respectively; Construct a reconstruction matrix based on the index value sorted by the size of each component; The instantaneous phase θ a (n) and θ b (n) Each reconstruction component in the reconstruction matrix is ​​arranged according to the numerical value, and the obtained sorted index values ​​constitute the reconstruction matrix of the k components of the two groups of symbol sequences; According to the number of times the state synchronously appears in the reconstruction matrix, a joint state distribution matrix is ​​constructed; Calculate the joint phase alignment entropy.

5. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps in the feature extraction method for detecting coronary heart disease as described in any one of claims 1 to 3 are implemented.

6. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the feature extraction method for detecting coronary heart disease as described in any one of claims 1 to 3 are implemented.

Images