A Hybrid Technique for Non-Invasive Fetal ECG Extraction and Heart Rate Estimation From the Mother’s Abdomen Signal

Abstract

A non invasive fetal electrocardiogram is used to keep track of fetal cardiac status during pregnancy from the 20th week of gestation until delivery. Obtaining of fetal ECG signal is critical, as ECG electrodes can not be placed on a direct fetal heart. Fetal ECG signal has 2–8 times less amplitude compared to the maternal ECG signal, as well it is influenced by motion artifacts, baseline wandering, and uterine contractions. So it is a challenging task to extract an accurate fetal ECG signal in noninvasive way. We proposed a hybrid method consisting of independent component analysis with adaptive filtering to extract fetal ECG signals from mixed signals and to filter out the noise. Pan-Tompkins algorithm measures fetal heartbeats and mother heartbeats. Proposed algorithm is verified by using physionet abdominal and direct fetal ECG (ADFECGDB) database. We obtained a reference fetal heartbeat from direct fetal scalp recording to compare with the extracted fetal heart rate. We have achieved 18.09 dB SNR value. The implemented method gives the advantage of knowing fetal cardiac status in the early phase.