Algorithm Research For Extracting Fetal ECG

At present,fetus electrocardiogram is the most effective method for doctors to diagnose the fetal health status.By analyzing the characteristic waveforms of fetal electrocardiogram(ECG)signals,they can obtain relevant abnormal information including fetal congenital heart disease,intrauterine hypoxia,intrauterine distress,umbilical cord entanglement and other perinatal information and provide a reliable guarantee for the health of the fetus.Now,abdominal ECG acquisition method is the highest recognition of fetal ECG acquisition method,which detects the fetal ECG signal by placing the electrode at the maternal abdominal surface.The resulting fetal ECG contains a variety of noise,especially the mother’s ECG interference signal.Therefore,to obtain a pure fetal ECG is clinically significant.In this thesis,fetal ECG extraction was studied in detail,aiming at the limitation of independent component analysis(ICA)to separate fetal ECG,a wavelet transform and improved matched filtering method are proposed to extract fetal ECG.The main work of this paper is as follows:Firstly,this thesis described the significance of fetal ECG testing.Summarizes the current development and research status of fetal ECG test at home and abroad,focuses on describing the classification of fetal ECG test methods and indicating the basic principles of each test method.Secondly,the fast independent component analysis(Fast ICA)algorithm was used to separate the fetal ECG signal from synthetic abdominal ECG signal and real abdominal ECG signal respectively.The results showed that the method is feasible for separating synthesize maternal abdominal electrocardiogram and the real eight-channal ECG signal in Daisy database,meanwhile,it also proves the limitation of separating the real four lead maternal abdominal electrocardiogram in the MIT Physio Net / Cin C2013 database.Furthermore,this thesis proposes to extract fetal ECG by using wavelet transform and improved matched filtering.The fetal ECG is preprocessed by stationary wavelet transform.After selecting the appropriate wavelet coefficients,the R peak of maternal ECG is further positioned by wavelet threshold method.Based on this algorithm,the GUI interface of maternal ECG R peak detection is designed.Finally,an improved matched filtering method is proposed,which can accurately extract fetal ECG signal when the maternal abdominal ECG signal has the coincidence wave of maternal ECG signal and fetal ECG signal in the time domain.Then the GUI interface of fetal ECG extraction is built based on this algorithm,and two sets of data were used to test the validity of the algorithm and the interface.At present,fetus electrocardiogram is the most effective method for doctors to diagnose the fetal health status.By analyzing the characteristic waveforms of fetal ECG signals,they can obtain relevant abnormal information including fetal congenital heart disease,intrauterine hypoxia,intrauterine distress,umbilical cord entanglement and other perinatal information and provide a reliable guarantee for the health of the fetus.