Study On FECG Extraction Algorithm Based On Autocorrelation And Kurtosis

Fetal ECG contents important information about the health condition of the fetus. It is an important index of clinical diagnosis of fetal development and the fetal heart disease during pregnancy. The research on FECG extraction is important in academic field and in practical. In this paper, FECG extraction based on autocorrelation and kurtosis is researched. The major contents and results are as follows:①Two famous ICA algorithms for FECG extraction are simulated and analysized, and their performance are compared. Extended-Infomax algorithm is an online algorithm based on gradient searching, and extracts FECG at the same time extracting MECG; Fixed-pointed extraction algorithm always firstly extracts MECG with large kurtosis, then extracts FECG with small kurtosis. Both algorithms have been applied to the problem of FECG extraction②An algorithm based on kurtosis and autocorrelation for FECG extraction is proposed. It fully exploits the difference between FECG's temporal property and the MECG's temporal property, gets the autocorrelation of the mixed signal and the delayed mixed signal, choose proper delay-time, can get FECG extracted firstly. Then use the algorithm based on kurtosis to get FECG.③Experiments of FECG extraction using the algorithm based on kurtosis and autocorrelation are done. Experiments show that the proposed algorithm convergences fast and needs less calculation, less time. Moreover it can extract clearer FECG, compared to existing algorithms.