Extraction of FECG Based on Time Frequency Blind Source Separation and Wavelet De-Noising

Extraction of the fetal electrocardiogram (FECG) from the mixture of maternal electrocardiogram (MECG) and noise is an important but hard task. The electrocardiograph (ECG) signal presents non-stationary. In this work, a two stage method is proposed based on time frequency blind source separation and wavelet de-noising to extract FECG. In the first stage the recorded signals from the electrodes of pregnant woman are prewhitened, then the spatial time frequency matrix are computed using Cohen´s class spatial time-frequency distributions, joint diagonalization of the sets of spatial time frequency matrixes, the mixing matrix is estimated and the FECG corrupted by noise is extracted. In the second stage, the corrupted FECG is performed by wavelet transform to eliminate the affect of noise. By the two stages, a clear FECG signal is obtained and experiments on real world data show the validation of the proposed method.