Spatial filtering and adaptive rule based fetal heart rate extraction from abdominal fetal ECG recordings

Despite advances in adult electrocardiography (ECG) and signal processing techniques, the analysis of fetal ECGs (fECG) is still in its infancy. The clinical potential of abdominal fECG monitoring by placing electrodes over mother's abdomen in antepartum (prior to labor) has been hampered by difficulties in obtaining a reliable fECG. We propose an algorithm to extract fetal heart rate from abdominal fECG based on spatial filtering and adaptive rule-based fetal QRS detection. The algorithm was trained and validated on 75 and 100 fECG datasets respectively, all obtained as part of PhysionNet Challenge 2013. Two metrics were used by the Challenge to assess the algorithm's performance: (Event4) the mean square error of fetal heart rate (HR) and (Event5) root mean square error of fetal RR interval between the HR obtained via the proposed approach and the HR obtained via the fetal scalp electrode. The proposed algorithm achieved mean scores of 52.49 and 10.61 and (Event 4 and 5 respectively) in the validation dataset. These results suggest the robustness of the proposed algorithm and its potential to advance fECG monitoring in antepartum surveillance.