A robust framework for noninvasive extraction of fetal electrocardiogram signals

In this paper, a robust framework is presented for fECG extraction from maternal abdomen recordings. The idea is based on extracting the fECG from contaminated signals using a multistage interference and noise cancelation method, designed specifically according to the time, space and frequency characteristics of the fECG and its interferences. The suitable combination of different denoising methods has both the advantages of multichannel and single channel denoising schemes. The proposed framework has also exploited the benefits of both temporal and statistical properties of ECG signals. Moreover, the ranking property of the algorithm, in comparison to permutation ambiguity of the independent component analysis (ICA) methods, helps reliable automatic detection of fECG signal in long recordings. The results have shown that the proposed framework outperforms conventional ICA and has effectively detected the fetal QRS in presence of full-rank maternal interference.