Research On Analysis Method Of Fetal Heart Sound Signal

With the deepening of the national policy of prenatal and postnatal care, fetal monitoring has become an important means to ensure the health of the fetus. Most of the fetus and the intrauterine birth defects are hypoxic distress caused by ischemia. Perinatal fetal monitoring has a very important practical significance. Electronic fetal heart rate monitoring is the most commonly used method for fetal monitoring. Fetal heart rate and contraction signal can be detected by fetal monitor and the data can be transmitted to a central monitoring station to keep, analyze and manage the maternal record. How to quickly and accurately detect and calculate fetal heart rate has become a research direction in recent years.This dissertation summarizes the analysis methods of fetal heart rate signal commonly used, proposes fetal heart rate analysis algorithms based on linear discriminant analysis and based on blind signal separation technique, which based on independent component analysis and non-negative matrix factorization, to detect the fetal heart rate and to try section analysis method to analyze the signal. Section analysis method is employed to detect fetal heart rate, which can be more quickly and accurately to help doctors to understand the status of fetal development and provide better clinical care to the fetus.The main work of this dissertation includes the following several aspects:Firstly, this dissertation introduces the research background and current situation of cardiac sound signal analysis, then provides a brief introduction to some relevant theoretical knowledge, such as linear discriminant analysis, wavelet transform method, blind source separation technique based on independent component analysis and nonnegative blind source separation technology.Secondly, this dissertation employs a fetal heart rate detection algorithm based on linear discriminant analysis is proposed, the basic steps are:collecting and pretreating the clinical fetal heart sound signals and obtaining the time-frequency graph by short-time Fourier transform, choosing the fetal heart sound signals template and the noise signal template, determining the linear discriminant equation and discriminant constant, using linear discriminant to position fetal heart sound signal and calculate the instantaneous fetal heart rate.Thirdly, this dissertation proposes an analysis method of fetal heart sound signal based on wavelet transform and independent component analysis, the basic steps are: applying ultrasonic Doppler sensor to extract fetal heart sound signals, going filtering and noise reduction with stationary wavelet transform, employing independent component analysis technique to extract segmentation component, which can be used to extract the instantaneous heart rate.Fourthly, this dissertation presents a non-negative blind separation of fetal heart sound segmentation analysis, the basic steps are:using wavelet transform to filter and reduce noise of the fetal heart sound signals collected, seeking wigner-ville distribution of the fetal heartbeat signal to transfer these signal into the time frequency domain, calculating the absolute value and obtain a non-negative spectrum, applying non-negative matrix factorization to obtain the non-negative fetal heartbeat signal frequency component, using the normalized average high order Shannon entropy method to draw an envelope curve, setting dual-threshold to realize the envelope section and calculate the instantaneous heart rate.Through the cases of clinical data in MATLAB simulation, the proposed methods can quickly and accurately calculate the instantaneous fetal heart rate and fetal heart sound signal segment, which is convenient for doctors to take more scientific clinical diagnosis and suitable for the field of fetal heart rate monitoring.