Proceedings of the Institution of Mechanical Engineers. Part H, Journal of engineering in medicine
-
Proc Inst Mech Eng H · Nov 2008
Blind adaptive filtering for non-invasive extraction of the fetal electrocardiogram and its non-stationarities.
The objective is to extract automatically a beat-to-beat fetal electrocardiogram (fECG) from a maternal electrocardiogram (mECG) using surface electrodes placed on the maternal abdomen and to derive fetal PR, QT, QTc, and QS durations to allow early diagnosis and monitoring treatment of certain fetal cardiac disorders. mECG and abdominal noise in abdominal maternal recordings can be orders of magnitude stronger than the fECG signal and the P and T waves that are embedded in them. A two-stage blind adaptive filtering algorithm was used for fECG extraction, the first stage using frequency-domain electrocardiogram features and the second considering time-domain features. ⋯ The combined filter allowed identification of diagnostically important PR, QT, and RR durations. Comparison with synthetic data is also included.