Fetal heart rate discovery: Algorithm for detection of fetal heart rate from noisy, noninvasive fetal ECG recordings

Fetal heart rate variability is known to be of a great meaning in assessing fetal health status. The simplest way of measuring fetal heart rate is to the non-invasive fetal ECG (fECG). A novel and efficient algorithm for detection of fetal ECG is needed. We analyzed 75 FECG recordings from the PhysioNet Challenge 2013 database. The detected RR interval peaks were compared with fetal scalp electrode measurements. Our algorithm focuses on detecting the most prominent part of the fetal QRS complex i.e. the RS slope. First, we remove long-range trends and find the two channels with the best quality fetal ECG. Then, we localize the repolarisations having the required characteristics (adequate amplitude and slope). Note, that the algorithm is adaptive and finds by itself the optimal RS slope characteristics for every recording. These steps allowed us to obtain accurate and reliable results of fetal R peak detection, even in the case of very noisy data. The preliminary test score of the PhysioNet Challenge were 132.664 (event 4) and 11.961 (event 5). The phase 3 score of the PhysioNet Challenge were 118.221 (event 4) and 10.663 (event 5). This is an opensource algorithm available at the PhysioNet library.