Maternal signal estimation by Kalman filtering and Template Adaptation for fetal heart rate extraction

Abstract

The fetal ECG derived from abdominal leads is an interesting alternative to standard fetal heart monitoring methods. Due to the temporal and spectral overlap of maternal and fetal signals, the usage of abdominal leads requires elaborated signal processing routines. In this work a modular combination of processing techniques is presented. The core of the evaluated scheme consists of two maternal ECG estimation techniques, namely the Extended Kalman Smoother (EKS) and Template Adaption (TA). Different method combinations are compared using the Computing in Cardiology Challenge 2013's scoring system. The obtained results prove that the proposed methods produce reliable fetal heart rate estimates. For training Set-A we obtained scores of 9.90/2.98 (Event 4/5) for TA, 196/6.78 for EKS and 15.7/3.25 combined versions of each method. For Set-B the best scores (18.1/4.38) were produced by TA in combination with statistical postprocessing.