A wavelet-based method for assessing fetal cardiac rhythms from abdominal ECGs

Fetal heart rate monitoring is widespread all over the world. However, despite recent advances in analysis methods, there are still inherent technical limitations. One of the main challenges is to extract accurate and useful information from the external fetal ECG, which may provide a better non-invasive characterization of the fetal cardiovascular system during the third trimester of pregnancy. In this work, maternal ECG waves were first located using a wavelet transform based system previously developed and validated. Then, a similar strategy with adapted parameters for fetal physiology was applied to detect fetal QRS, excluding signal singularities associated to maternal heartbeats. Single lead based annotations were combined in a single annotator from which RR and HRV fetal measures could be taken. Data provided from PhysioNet 2013 Challenge was considered. The average scores were: 521.43 / 401,50 for event 4 and 32.99 / 29,70 for event 5 on set B /A respectively. The median rate of true positives, false positives and false negatives, normalized by number of fetal beats, found in the (training) set A was, respectively 78%, 17% and 22%. Median FHR estimation error was 2.0 bpm. Correlation between reference and estimated median FHR was 0.65 for estimated FHR>120bpm. The proposed approach seems promising for assessing fetal cardiac rhythms from abdominal ECGs.