Ambulatory fetal heart monitoring with QRS detection employing independent component analysis

This paper covers the development of a fetal heart monitor (fHM) on an ARM Cortex-M3 based platform, and its application in extracting fetal heart rate (fHR) non-invasively from the maternal abdomen. The fHM is a low-power, inexpensive ambulatory device connected to smart-phones via Bluetooth. This monitor uses active-electrodes for primary signal-conditioning, and has a digital back end for running compute-intensive signal processing algorithms. The output, can be displayed on a smart-phone, can generate audible alarms during critical conditions or can be stored on-board for further processing. The fHM implements multiple algorithms to monitor the fetal heart rate. First, a tumbling-window based implementation of Independent Component Analysis (ICA) is employed on abdominal signals in order to simultaneously extract fetal and maternal cardiac components. Thereafter, the Pan-Tompkins algorithm (PTA) is used to detect QRS complexes. Preliminary tests show that this combination of algorithms can be employed to successfully estimate fetal and maternal heart rates, and has a low rate of spurious detections.