Research About BSS Suitable For Maternal ECG

In the field of biomedicine, fetal ECG signal has a profound impact on the health of fetus. ECG signal acquired by abdominal electrodes is multi-source mixed signal and mixed signal contains strong noise of skin, power-line interference and non-biological noise. Fetal ECG is very weak in many strong interference signals. Fetal ECG contains strong interference signals. The extracted algorithm of complex environment has been plagued by many scholars at home and abroad, which is a difficulty in the field of signal processing. Therefore, the extraction of fetal ECG signal can be used as a typical extraction of weak signal in strong interference environment. Separated algorithm based on independent component analysis (Independent Component Analysis, ICA) is the algorithm source signal of independent approximation. Affection on strong interference signal is small, which is applicable on fetal ECG extraction. ICA algorithm processes observed signals directly. ICA algorithm has good real-time and anti-interference ability, which becomes a hot spot in the scholar research.The characteristic of biomedical signals in independent statistic of each other is the basic premise of independent component analysis in this paper. In recent years, the scholars of various countries put forward a lot of excellent ICA algorithms based on framework for the ICA model. Corresponding to the requirements, ICA algorithm has higher separation accuracy, anti-noise performance and real-time.From theory to application, application fields of ICA algorithm are expanding rapidly. At present, researches on ICA algorithm for electrocardiogram is poorly understood, both ICA algorithm of high separation precision and good anti noise becomes the research hotspot of fetal ECG extraction.The existing ICA algorithm has little fetal ECG in face of the complex environment. Few ICA algorithms have both separated accurate and noisy properties.Aiming at the problems, the following work is done:1. This paper introduces the characteristics of fetal ECG signal, noise of ECG signal and contrast with intensity of maternal and fetal ECG signal;2. This paper introduces the basic model and mathematical theory of independent component analysis. We mainly discuss the ICA algorithm on the basis of judgment, which include maximum thought, minimum thought and Non-Gauss measure;3. In the mainstream of ICA algorithm, high order statistic algorithms are selected according to the characteristics of fetal ECG extraction, we introduce two order statistic algorithm and higher-order statistic algorithm for the selected focus. This paper introduces two second-order blind indentified algorithm (SOBI) and joint approximate diagonalization of four-order cumulant algorithm (JADE), which analyzes the characteristics of each algorithm and structure. The objective function of current single constraint criteria, DC-JADE algorithm is introduced in double constraint criteria. DC-JADE algorithm is objective while the maximization and minimization of ideological thoughts constrained in information on theory;4. In separated experiments, mixed signal obtained by electrodes is selected as the observed signal. We study the separation of SOBI algorithm, JADE algorithm on the fetal ECG signal and discuss the separated accuracy and anti-noise performance. Then, we contrast the DC-JADE algorithm to SOBI algorithm, JADE algorithm in separated effect and compared anti-noise performance. Under the optimal parameters, we select typical metrics in blind source separation and evaluate the performance of the algorithm. This algorithm can not only extract clear fetal ECG This algorithm has more excellent separated accuracy and anti-noise performance in ECG separation compared with other algorithms, which can be used as a reference for weak signal under strong jamming recovery reference;5. Finally, prospects and summarization are made on fetal ECG This paper describes the lack of possible research direction in the future;...