Description

Fetal ECG (FECG) is the electrical activity of fetal heart muscles and it can reveal vital information of fetal heart status. This signal can be recorded non-invasively during pregnancy from the abdomen of a pregnant woman. However, such a recording of the FECG would be useless since it is largely overwhelmed by the maternal ECG (MECG) and other sources of noise. Nonetheless, if the abdominal recording is properly processed then a clean FECG can be extracted, FECG can be used to assess the fetal heart functionality. This paper introduces a novel two-tier technique for extracting fetal electrocardiogram (ECG) from a single abdominal record. The proposed method in its first tier extracts an estimate of the maternal ECG by processing the abdominal signal through a smoothing filter. The estimated maternal ECG is then nonlinearly aligned with the abdominal signal using polynomial networks to extract the fetal ECG signal results on synthetic and real abdominal ECG data show that the proposed method can extract fetal ECG with signal quality comparable or better than that extracted by multichannel based methods.