Computer algorithm for adaptive extraction of fetal cardiac electrical signal

Abstract

A microcomputer-controlled data acquisition system was designed to detect the fetal cardiac electrical signal in maternal transabdominal recordings. Data were collected in 3-min blocks over rive channels at a rate of 833 Hz per channel. A two-queue sampling algorithm allowed for data to be collected in one queue while the other queue was writing to disk. All five channels were visualized simultaneously with the aid of an interactive graphical display. Periodic Doppler transducer noise appeared in the raw signal and was removed with a digital algorithm. To remove the maternal cardiac signal, a moving template was constructed by averaging the index maternal R-wave and the four immediately preceding and immediately following maternal R-waves. This template was then subtracted from the raw signal, yielding a residual signal which contained only the fetal cardiac complex. The location of each fetal R-wave was determined using an interactive graphical interface combined with a digital search routine. Application of adaptive signal processing techniques resulted in almost 100% recovery of the fetal signal. We conclude that transabdominal fetal R-wave detection was possible with very little signal loss using a microcomputer-controlled data acquisition system with multichannel visualization.