Non-invasive Fetal ECG Extraction Method Based On Template Matching And Blind Source Separation

The morphology of Fetal Electrocardiogram(FECG)signals can directly reflect the health status of the fetal heart,help to understand the problems of the fetal heart,and can be used for medical diagnosis of fetal hypoxia and congenital heart defects.Non-invasive FECG extraction has the advantages of non-invasive,low bacterial infection,and simple operation,but it is easily affected by Maternal Electrocardiogram(MECG)and other noise.In this paper,the method of FECG extraction combined with template matching and blind source separation is used to improve the clarity and accuracy of FECG signals by improving the signal processing effect of FECG separation and FECG noise reduction.It has certain practical application value.First of all,according to the differences in Baseline Wander(BW),Muscle Artifact(MA),and power line interference(PLI)in Electrocardiogram(ECG)signals,three different denoising methods are used.The low-pass filter is used to remove the MA.The median filter is used to suppress the BW.The zero-phase shift Kaiser window digital filter removes PLI.Experiments have proved that classifying and processing different noise interference can effectively solve the phase distortion problem of the signal to obtain a relatively noisy Common Electrocardiogram(CECG)signal.Secondly,using the Pan-Tompkins algorithm to perform the MECG R wave detection on the pre-processed ECG signals,and using the coherent averaging method to construct the MECG template.Using the QRS signal segment of the CECG signal to replace the QRS signal segment of the MECG template can reduce the probability that the peak formed by the MECG QRS wave error is misidentified as the FECG R wave.The pre-processed CECG signal is different from the MECG template to complete the FECG separation to obtain a FECG signal containing noise.Then,a FECG noise reduction method based on blind source separation is used to pass the FECG signal containing noise through a low-pass filter to smooth the signal,and high-pass filtering is used to filter out the remaining MECG P and T waves.The improved JADE algorithm restricts the orthogonal matrix by introducing a double-constrained optimization criterion.Compared with the JADE algorithm,it can isolate FECG signals with fewer bottom glitches,and has better anti-noise performance and separation performance.The separated FECG signal is detected by double threshold to obtain the FECG R wave position.Finally,the simulation and clinical data in the Cardiac Computing Challenge Database(PCDB)were tested and experimentally analyzed to verify the reliability of the algorithm.The results show that compared with the existing FECG extraction methods,the method in this paper improves the signal quality of FECG and is helpful for the medical diagnosis of FECG waveform.