阅读说明：本技术 一种基于因子同步性分析的多导联胎儿心电提取装置 (Multi-lead fetal electrocardiogram extraction device based on factor synchronism analysis ) 是由 谭波 李凌 于 2021-09-16 设计创作，主要内容包括：该发明公开了一种基于因子同步性分析的多导联胎儿心电提取装置,属于心电信号识别领域。该发明首先通过心电测试系统同步记录孕妇腹部和胸部多导联心电混合信号并完成预处理；将胸部多导联心电混合信号输入主成分分析模块完成主成分分解；随后将胸导分解的主成分作为参考信号,分别计算腹部分解的各因子与参考信号间的相位同步值并排序,再结合母亲和胎儿心电频率及波形特征进行分类筛选并排除母亲心电因子；利用经验模态分解排除非胎儿心电分量来完善胎儿心电的局部特征并重构出胎儿心电因子；最后利用因子分析的逆运算重构出清晰的胎儿心电信号。本发明相对现有技术所提取的胎儿心电信号具有更多胎儿心电局部特征,信号也更平滑,对胎儿心电信号及围产医学研究都具有重要意义。(The invention discloses a multi-lead fetal electrocardio extraction device based on factor synchronism analysis, and belongs to the field of electrocardio signal identification. Firstly, synchronously recording multi-lead electrocardio mixed signals of the abdomen and the chest of a pregnant woman through an electrocardio testing system and completing pretreatment; inputting chest multi-lead electrocardio mixed signals into a principal component analysis module to complete principal component decomposition; then, taking the principal component of the chest lead decomposition as a reference signal, respectively calculating and sequencing phase synchronization values between each factor of the abdominal lead decomposition and the reference signal, and then classifying and screening by combining the maternal and fetal electrocardio frequency and waveform characteristics and eliminating maternal electrocardio factors; using empirical mode decomposition to eliminate non-fetal electrocardio components to perfect local characteristics of fetal electrocardio and reconstruct fetal electrocardio factors; and finally, reconstructing a clear fetal electrocardiosignal by utilizing the inverse operation of factor analysis. Compared with the fetal electrocardiosignals extracted by the prior art, the fetal electrocardiosignals extracted by the method have more local fetal electrocardiosignals and smoother signals, and have important significance for fetal electrocardiosignals and perinatal medical research.)

1. A multi-lead fetal electrocardiogram extraction apparatus based on factor synchronicity analysis, the apparatus comprising: the device comprises an electrocardiosignal acquisition and preprocessing module, a decomposition analysis module, a phase synchronism analysis module, a fetal electrocardio factor component screening module, a fetal electrocardio factor purification module and a reconstructed fetal electrocardiosignal module;

the electrocardiosignal acquisition and preprocessing module comprises an electrocardiosignal testing system and a preprocessing module, the electrocardiosignal testing system acquires abdominal multi-lead electrocardiosignals and chest multi-lead electrocardiosignals of the pregnant woman, then the abdominal multi-lead electrocardiosignals and the chest multi-lead electrocardiosignals are transmitted to the preprocessing module, the preprocessing module respectively carries out power frequency interference removal, band-pass filtering and mean value removal preprocessing on the abdominal multi-lead electrocardiosignals and the chest multi-lead electrocardiosignals, and then the preprocessed abdominal multi-lead electrocardiosignals and the chest multi-lead electrocardiosignals are transmitted to the decomposition analysis module;

the decomposition analysis module includes: the main component analysis module receives the chest multi-lead electrocardiosignals and performs main component analysis on the chest multi-lead electrocardiosignals to obtain mother electrocardio main component signals; the factor analysis module receives the abdomen multi-lead electrocardiosignals, carries out factor analysis decomposition on the abdomen multi-lead electrocardiosignals, and decomposes the abdomen multi-lead electrocardiosignals into three factors, wherein the three factors comprise: fetal, maternal, other factors; then the mother electrocardio principal component signal and the decomposed abdomen multi-lead electrocardio signal are transmitted to a phase synchronism analysis module;

the phase synchronism analysis module takes the maternal electrocardiogram principal component signal as a reference signal, respectively calculates a phase synchronism value between each factor obtained by decomposing the abdomen multi-lead electrocardiogram signal and the reference signal, and the factor with the highest synchronism value is determined as the maternal electrocardiogram factor; the abdomen multi-lead electrocardiosignals and maternal electrocardiofactors are transmitted to a fetal electrocardio factor component screening module;

the fetal electrocardio factor component screening module discharges maternal electrocardio factors and other factors from the abdomen multi-lead electrocardiosignals, and then the remaining signals are transmitted to the fetal electrocardio factor purification module;

decomposing the acquired signals into a plurality of groups of eigenmode functions (IMFs) in the fetal electrocardio factor purification module, and removing non-fetal electrocardio IMF components by combining local characteristics of each IMF component under different scales; reconstructing a clean fetal electrocardio main factor and other factors containing fetal electrocardio information after screening, and transmitting the clean fetal electrocardio main factor to a reconstructed fetal electrocardio signal module;

and the reconstructed fetus electrocardiosignal module carries out inverse operation according to the clean fetus electrocardio main factor to obtain clear fetus electrocardiosignals.

2. The multi-lead fetal electrocardiogram extraction apparatus based on factor synchronism analysis as claimed in claim 1, wherein the signal processing method of the principal component analysis module in the decomposition analysis module is:

step a 1: averaging and decentralizing the chest multi-lead electrocardiosignals, namely subtracting respective average values from each pair of signals to obtain mean value-removed signals;

step a 2: calculating the covariance matrix of each connected electrocardiosignal after the decentralization obtained in the step 1;

step a 3: calculating the characteristic and the eigenvector of each covariance matrix through SVD decomposition or eigenvalue decomposition;

step a 4: sorting the eigenvalues from big to small, and selecting the largest k of the eigenvalues; then, respectively taking the k eigenvectors corresponding to the k eigenvectors as column vectors to form an eigenvector matrix;

step a 5: converting the data into a new space constructed by k eigenvectors to obtain corresponding k principal components;

the signal processing method in the factor analysis module comprises the following steps:

step b 1: calculating a correlation coefficient matrix of the abdominal multi-lead electrocardiosignals;

step b 2: calculating the eigenvalue lambda and the eigenvector U of the correlation coefficient matrix obtained in the step b1 to obtain a load matrix of the corresponding factor

Step b 3: according to F ═ A^-1X_ECGThe decomposition obtaining a plurality of decomposed factors, X_ECGRepresenting abdominal multi-lead cardiac electrical signals.

3. The multi-lead fetal electrocardiogram extraction apparatus based on factor synchronism analysis as claimed in claim 1, wherein the signal processing method in the phase synchronism analysis module is:

step 1: extraction of phase values phi of principal components of chest multi-lead electrocardiosignal by Fourier transform₁[f]And a phase value phi of a factor of an abdominal multi-lead electrocardiographic signal₂[pc])；

Step 2: the following formula is sampled to calculate a phase synchronization value;

wherein, N represents the length of the single-channel electrocardiosignal, i (-) represents the complex transformation of the phase difference of the two electrocardiosignals;

and step 3: and repeating the step 1 and the step 2 to obtain phase synchronization values between each principal component and each factor.

Technical Field

The invention belongs to the field of electrocardiosignal identification, and particularly relates to a device for analyzing fetal electrocardio extraction based on factor phase synchronism.

Background

Fetal Electrocardiogram (FECG) is a physiological signal generated by the heart activity of a fetus, can reflect the real growth health condition of the fetus in the perinatal period in a pregnant woman, and is an important index for the obstetric examination of the pregnant woman. However, since the perinatal fetus is located in the mother, it is difficult to detect and study the FECG. Currently, there are two main ways to detect fetal electrocardiograms: one is invasive fetal scalp electrode method, which can obtain relatively clean FECG but will cause certain damage to both mother and fetus; the other method is a noninvasive abdominal electrode method, namely electrodes are arranged at different positions of the abdomen of the mother to detect body surface signals of the mother, and the FECG is extracted by combining later data analysis. Although the method can obtain the electrocardiosignals more easily, the electrocardiosignals of the fetus are often covered by physiological signals of the electrocardio, the breath and the like of the mother, which puts high requirements on the specific method and the device for extracting. Therefore, the method for effectively extracting the fetal electrocardiosignals has important theoretical value and certain clinical application value.

At present, many researches on FECG extraction methods, such as coherent averaging, autocorrelation and cross-correlation, adaptive filtering, wavelet transform methods, singular value decomposition-based, independent quantity analysis, neural networks and the like, are carried out at home and abroad. Although the methods can extract fetal electrocardiosignals to a certain degree, the obtained signals are too strict in data acquisition and low in accuracy, the extraction or calculation process is too complex and difficult to realize, more manual intervention is needed, and the like, so that the applicability of various methods is limited, and a particularly suitable noninvasive fetal electrocardio-extraction method is not provided at present. The invention relates to a multi-lead fetal electrocardiogram extraction method and a device based on factor synchronism analysis, which organically integrates factor analysis, principal component analysis and phase synchronism analysis and is applied to fetal electrocardiogram extraction. Compared with other traditional methods, the device provided by the invention has the advantages that the extraction quality is obviously improved, the overlapping part of the maternal electrocardiogram and the fetal electrocardiogram is greatly improved, and the stability, the robustness and the practicability are better.

Disclosure of Invention

In order to overcome some defects of the existing fetal electrocardiogram extraction, the extraction quality and the practicability of the fetal electrocardiogram are improved. The invention provides a multi-lead fetal electrocardio extraction device based on factor synchronism analysis. Inputting the abdomen multi-lead electrocardio mixed signal into a factor analysis module to complete factor decomposition, and inputting the chest multi-lead electrocardio mixed signal into a principal component analysis module to complete principal component decomposition; then, all main components decomposed by the chest lead and all factors decomposed by the abdomen are subjected to synchronism analysis and screened to eliminate maternal electrocardio factors, and factors containing fetal electrocardio information are reserved; and decomposing the fetal electrocardio-factor into a plurality of Intrinsic Mode Functions (IMFs) with different scales by using Empirical Mode Decomposition (EMD), further correcting and perfecting the local characteristics of the fetal electrocardio-factor, and finally reconstructing a clear fetal electrocardiosignal by using inverse operation of factor analysis.

The invention is realized by the following technical scheme: a multi-lead fetal electrocardiogram extraction apparatus based on factor synchronicity analysis, the apparatus comprising: the device comprises an electrocardiosignal acquisition and preprocessing module, a decomposition analysis module, a phase synchronism analysis module, a fetal electrocardio factor component screening module, a fetal electrocardio factor purification module and a reconstructed fetal electrocardiosignal module;

the electrocardiosignal acquisition and preprocessing module comprises an electrocardiosignal testing system and a preprocessing module, the electrocardiosignal testing system acquires abdominal multi-lead electrocardiosignals and chest multi-lead electrocardiosignals of the pregnant woman, then the abdominal multi-lead electrocardiosignals and the chest multi-lead electrocardiosignals are transmitted to the preprocessing module, the preprocessing module respectively carries out power frequency interference removal, band-pass filtering and mean value removal preprocessing on the abdominal multi-lead electrocardiosignals and the chest multi-lead electrocardiosignals, and then the preprocessed abdominal multi-lead electrocardiosignals and the chest multi-lead electrocardiosignals are transmitted to the decomposition analysis module;

the decomposition analysis module includes: the main component analysis module receives the chest multi-lead electrocardiosignals and performs main component analysis on the chest multi-lead electrocardiosignals to obtain mother electrocardio main component signals; the factor analysis module receives the abdomen multi-lead electrocardiosignals, carries out factor analysis decomposition on the abdomen multi-lead electrocardiosignals, and decomposes the abdomen multi-lead electrocardiosignals into three factors, wherein the three factors comprise: fetal, maternal, other factors; then the mother electrocardio principal component signal and the decomposed abdomen multi-lead electrocardio signal are transmitted to a phase synchronism analysis module;

the phase synchronism analysis module takes the maternal electrocardiogram principal component signal as a reference signal, respectively calculates a phase synchronism value between each factor obtained by decomposing the abdomen multi-lead electrocardiogram signal and the reference signal, and the factor with the highest synchronism value is determined as the maternal electrocardiogram factor; the abdomen multi-lead electrocardiosignals and maternal electrocardiofactors are transmitted to a fetal electrocardio factor component screening module;

the fetal electrocardio factor component screening module discharges maternal electrocardio factors and other factors from the abdomen multi-lead electrocardiosignals, and then the remaining signals are transmitted to the fetal electrocardio factor purification module;

decomposing the acquired signals into a plurality of groups of eigenmode functions (IMFs) in the fetal electrocardio factor purification module, and removing non-fetal electrocardio IMF components by combining local characteristics of each IMF component under different scales; reconstructing a clean fetal electrocardio main factor and other factors containing fetal electrocardio information after screening, and transmitting the clean fetal electrocardio main factor to a reconstructed fetal electrocardio signal module;

and the reconstructed fetus electrocardiosignal module carries out inverse operation according to the clean fetus electrocardio main factor to obtain clear fetus electrocardiosignals.

Further, the signal processing method of the principal component analysis module in the decomposition analysis module is as follows:

step a 1: averaging and decentralizing the chest multi-lead electrocardiosignals, namely subtracting respective average values from each pair of signals to obtain mean value-removed signals;

step a 2: calculating the covariance matrix of each connected electrocardiosignal after the decentralization obtained in the step 1;

step a 3: calculating the characteristic and the eigenvector of each covariance matrix through SVD decomposition or eigenvalue decomposition;

step a 4: sorting the eigenvalues from big to small, and selecting the largest k of the eigenvalues; then, respectively taking the k eigenvectors corresponding to the k eigenvectors as column vectors to form an eigenvector matrix;

step a 5: converting the data into a new space constructed by k eigenvectors to obtain corresponding k principal components;

the signal processing method in the factor analysis module comprises the following steps:

step b 1: calculating a correlation coefficient matrix of the abdominal multi-lead electrocardiosignals;

step b 2: calculating the eigenvalue lambda and the eigenvector U of the correlation coefficient matrix obtained in the step b1 to obtain a load matrix of the corresponding factor

Step b 3: according to F ═ A^-1X_ECGThe decomposition obtaining a plurality of decomposed factors, X_ECGRepresenting abdominal multi-lead cardiac electrical signals.

Further, the signal processing method in the phase synchronism analysis module is as follows:

step 1: extraction of phase values phi of principal components of chest multi-lead electrocardiosignal by Fourier transform₁[f]And a phase value phi of a factor of an abdominal multi-lead electrocardiographic signal₂[pc])；

Step 2: the following formula is sampled to calculate a phase synchronization value;

wherein, N represents the length of a single-channel electrocardiosignal, and i () represents the complex transformation of the phase difference of two electrocardiosignals;

and step 3: and repeating the step 1 and the step 2 to obtain phase synchronization values between each principal component and each factor.

The invention provides a multi-lead fetal electrocardiogram extraction method and a device based on factor synchronism analysis, which fully combine the advantages of principal component analysis and factor analysis to obtain the principal component of maternal chest electrocardiogram (reference signal) and mixed abdominal electrocardiogram factors (separation target signals), then combine the phase information and waveform characteristics of maternal and fetal electrocardiogram signals to screen and remove maternal electrocardiogram factors, and reserve factors containing fetal electrocardiogram information; then, EMD is used for decomposing the fetal electrocardio-factors in a self-adaptive manner to further correct and perfect the local characteristics of the fetal electrocardio-factors, and finally, the clear fetal electrocardiosignals are accurately and quickly reconstructed by utilizing the inverse operation of factor analysis. Compared with the fetal electrocardiosignals extracted by the conventional ICA, the fetal electrocardiosignals extracted by the method have more fetal electrocardio waveform characteristics, the accuracy and robustness of the extracted signals are greatly improved, and the signals are relatively smoother, so that the method and the device have important significance for preventing fetal heart diseases and research on perinatal medicine.

Drawings

FIG. 1 is a block diagram illustrating the detailed operation and structure of the method and apparatus of the present invention.

Fig. 2 is the original multi-lead electrocardio-hybrid signal of the abdomen and the chest of the pregnant woman in the embodiment.

Fig. 3 shows the factors and principal components extracted from the ecg mixed signal in the embodiment.

Fig. 4 shows the phase synchronization values between the factors and the principal components in the embodiment.

FIG. 5 shows IMFs components obtained by decomposing the fetal ECG main factor by EMD in the embodiment.

FIG. 6 is a graph comparing the present invention with the extraction of fetal electrocardiograms for independent component analysis in the examples.

Detailed Description

To facilitate a better understanding and an implementation of the present invention for those skilled in the art, the following description and the accompanying drawings illustrate specific embodiments, structures, features, and functions of the present invention, but the present invention is not limited thereto. The implementation steps included in the invention are as follows:

1. electrocardio mixed signal recording and preprocessing module

A. Recording pregnant woman abdomen through electrocardiogram test systemThe multi-lead electrocardio mixed signal of the chest and the abdomen is preprocessed by removing power frequency interference, band-pass filtering, mean value removing and the like to obtain the preprocessed multi-lead electrocardio mixed signal C of the abdomen and the chest_ECGAnd S_ECG；

Firstly, synchronously recording 4-lead electrocardio mixed signals of the abdomen and 2-lead electrocardio mixed signals of the chest of a pregnant woman (the sampling rate is 250Hz) by an electrocardio test system, wherein the recording time is 10 s; meanwhile, preprocessing the recorded electrocardio data such as power frequency interference (50Hz), band-pass filtering, mean value removing and the like, and further obtaining a multi-lead electrocardio mixed signal (lead number time point) after preprocessing the abdomen and the chest;

2. decomposition analysis module for electrocardio mixed signal

B. Utilizing principal component analysis to preprocess chest electrocardio mixed signal S_ECGDecomposing into a plurality of principal components as reference signals (namely, the mother electrocardio principal component PC); simultaneously utilizes factor analysis to analyze the preprocessed abdomen electrocardio mixed signal C_ECGDecomposing the signal into a plurality of factors F (namely target analysis signals) including fetal electrocardiofactors, maternal electrocardiofactors and other factors;

for chest electrocardio mixed signal S_ECG＝{S₁,S₂,S₃,…S_nThe steps of extracting k principal components for principal component analysis are as follows: 1) averaging and decentralizing, that is, subtracting the respective average value from each feature to obtain a de-averaged signal M _ S ═ M _ S₁,m_S₂,m_S₃,…m_S_n}；

2) Calculating a covariance matrix

3) And calculating an eigenvalue lambda and an eigenvector U of the covariance matrix through SVD decomposition or eigenvalue decomposition.

4) The eigenvalues λ are sorted from large to small, and the largest k of them are selected. Then, the corresponding k eigenvectors are respectively used as column vectors to form an eigenvector matrix.

5) Converting the data into a new space constructed by k eigenvectors, i.e.Corresponding k principal components PC ═ PC are obtained₁,pc₂,…pc_k}. For abdomen electrocardio mixed signal X_ECG＝{X₁,X₂,X₃,…X_nThe factorization procedure is as follows:

1) solving electrocardio mixed signal X_ECGIs given by the correlation coefficient matrix R_ECG；

2) Solving the eigenvalue lambda and the eigenvector U of the correlation coefficient matrix to obtain the load matrix of the corresponding factor

3) And then according to F ═ A^-1X_ECGThe factor F ═ F for multiple decompositions can be obtained₁,f₂,f₃,…f_n}。

The 4-lead electrocardio mixed signal of the preprocessed abdomen is decomposed into 4 factor components (F) by utilizing factor analysis, and the preprocessed chest electrocardio mixed signal is decomposed into 2 Principal Components (PC) by principal component analysis.

3. Phase synchronism analysis module

C. Calculating phase synchronization values among all factors of the abdominal lead decomposition and all main components of the chest lead decomposition, and sequencing and storing the obtained phase synchronization values from high to low;

the synchronous calculation method for the same factors and main components of the electrocardio mixed signal is as follows:

1) using Fourier transform S (n) | e^Φ(n)Extracting phase value phi of chest electrocardio principal component pc and abdomen electrocardio factor f₁(f) And phi₁(pc)；

2) Method for calculating coupling value of reuse phase and phaseObtaining the phase synchronization value of the principal component pc and the electrocardio factor f;

3) repeating the above process to obtain the phase synchronization value between each factor and each main component.

Phase synchronization values between 4 factor components of the abdominal lead decomposition and 2 main components of the chest lead decomposition are calculated by using phase synchronization analysis, and are sorted from high to low and stored (see fig. 4 for phase synchronization values of the embodiment).

4. A fetal electrocardio factor component screening module:

D. based on characteristics of the electrocardio frequencies, the waveforms and the like of the mother and the fetus, classifying and screening all factors into three classes of mother electrocardio factors, fetus electrocardio factors and other factors by combining a phase synchronization value, removing the mother electrocardio factors and other factors, and reserving the factors containing the electrocardio information of the fetus;

establishing a classifier by combining the characteristics of fetal electrocardio, such as frequency, waveform amplitude and the like, and screening and removing maternal electrocardio factors and other factors by combining a phase synchronization value; in the embodiment, a screened factor 3 is a main factor of fetal electrocardiogram, and a screened factor 4 is a factor containing a small amount of fetal electrocardiogram information;

5. purification module of fetal electrocardiofactors

E. Decomposing factors including fetal electrocardiogram information into a plurality of groups of eigenmode functions (IMFs) by using an Empirical Mode Decomposition (EMD), and excluding non-fetal electrocardiogram IMF components by combining local characteristics of each IMF component under different scales; and reconstructing clean fetal electrocardiogram main factors and other factors containing fetal electrocardiogram information after screening.

Decomposing the screened main factors of the fetal electrocardiogram into IMF components by using an EMD method (the embodiment has 13 IMF components, see the attached figure 5, and removing non-fetal electrocardiogram components by combining local characteristics of IMF components with different scales and reconstructing clean factor components containing fetal electrocardiogram;

6. reconstruction module for clear fetal electrocardiosignals

F. Finally, reconstructing the fetal electrocardiosignals from the screened and purified fetal factors through the inverse operation of factor analysis, namely the clear fetal electrocardiosignals extracted by the invention (see attached figure 6). In addition, the invention can obtain more stable fetal electrocardiosignals by repeatedly measuring and circulating the analysis process to superpose and average for a plurality of times.

To further illustrate the beneficial effects of the present invention, we compared the present invention with the conventional ICA method to extract fetal electrocardiograms. Firstly, referring to fig. 2 and fig. 6, it can be understood that the method and the device of the present invention can indeed extract clear fetal electrocardiosignals from the electrocardio-mixed signals. Secondly, as can be seen from the attached figure 4, the maternal electrocardiogram factors and the chest signal principal components which are analyzed and screened in a phase synchronization manner have higher synchronization values (the factor 1 is about 0.84), while the screened fetal electrocardiogram factors and the chest signal principal component phase synchronization values are very small (the factor 3 is about 0.03), so that the reconstructed fetal electrocardiogram eliminates the maternal electrocardiogram factors with high synchronization values, and the retained fetal electrocardiogram factors with low synchronization values are effective; finally, as can be seen from fig. 6, compared with the ICA method, the signal extracted by the present invention has more fetal electrocardiogram features (such as QRS waves), and is relatively smoother, and meanwhile, the present invention has higher extraction efficiency and accuracy, which will play an important role in monitoring the healthy growth of the perinatal fetus.

It should be emphasized that the embodiments described herein are illustrative and not restrictive, and thus the present invention includes, but is not limited to, the embodiments described in the detailed description, and that other embodiments derived from the teachings of the present invention by those skilled in the art are within the scope of the present invention.