Identification of fetal QRS complexes in low density non-invasive biopotential recordings

Non-invasive fetal Electrocardiogram (ECG) is currently a missing diagnostic tool. Despite the technology advancements and the improvements of the signal processing techniques, the possibility of extracting this signal from recordings of biopotentials gathered on the maternal abdomen is still unexploited in the clinical practice. The 2013 Physionet/Computing in Cardiology Challenge proposes to address this specific problem, making available a dateset of annotated abdominal signals, with a reduced number of channels, taken with different instruments and protocols. In this paper a novel algorithm based on template matching for maternal QRS subtraction and fetal ECG detection is presented and evaluated on the available dataset. The algorithm achieves a score of 639.465 and 23.821 on dataset B and of 684.158 and 47.990 on dataset C.