Investigation of blind source separation methods for extraction of fetal ECG

Abstract

Extraction of fetal electrocardiogram (FECG) from maternal skin electrode measurements needs a robust technique. This problem can be modeled from the perspective of blind source separation (BSS) and so the problem reduces to the estimation of the independent sources of fetal and maternal cardiac activity. In this paper, analysis and study of some major algorithms like Bell and Sejnowski's infomax algorithm, Cardoso's joint approximate diagonalization of eigen matrices (JADE) algorithm, Hyvarinen's fixed-point algorithm and Comon's algorithm, was made for this important biomedical application. For robustness, two scenarios, i.e, (a) different amplitude ratios of simulated maternal and fetal ECG and (b) different values of additive white Gaussian noise, were investigated. It was observed that if the ratio of the amplitude of maternal to fetal ECG is 10:1 with an input SNR of 2 dB, all four algorithms were able to extract the fetal ECG. The signal-to-error (SER) ratios of the extracted maternal and fetal ECG were around 3 dB and 1 dB, respectively.