Semi-Blind Fetal Electrocardiogram Extraction by Eliminating the Cross-Terms of the Wigner-Ville Representations

Blind source separation methods on time-frequency representations usually use kernel functions, like Cohen class time-frequency representations, to suppress the useless cross terms of the sources produced by the mixture operation before further processing. However, for sparse signals like maternal ECG and fetal ECG, cross-terms are not thought to be useless and interference. In this context, by introducing the cross-term of spatial Wigner-Ville distributions to the contrast function, a new semi-blind source separation method is proposed to extract fECG from the mixed abdominal signals. Experiments on real world clinical data show that the proposed method is comparable to the FastlCA.