Extraction Of Fetal Electrocardiogram Using An Adaptive Neuro-Fuzzy System

Abstract

In this paper, adaptive neuro fuzzy inference system (ANFIS) was used for the cancellation of maternal electrocardiogram (MECG) in fetal electrocardiogram extraction (FECG) from the composite abdominal electrocardiogram (AECG). This technique is used to estimate the MECG present in the abdominal signal of a pregnant woman. The FECG is then extracted by subtracting the estimated MECG from the abdominal signal.

In the furtherance of extraction, MATLAB (version 7.6) was used to code the system in order to generate the maternal heartbeat signal and the fetal heartbeat signal which were added to form the measured signal. For the fetal heartbeat signal to be recovered from the interference (maternal heartbeat) signal, a reference signal (which is a clean version of the original maternal heartbeat signal) was introduced in the system. It is this signal that cancelled the maternal heartbeat signal in the measured signal, thereby leaving the fetal heartbeat signal as an error signal.

However, though the recovered signal still contained some traces of the maternal heartbeat signal, performance of the soft computing technique applied is in terms of the capability of adaptive neuro fuzzy inference system in removing the overlapping between the MECG and the FECG signals. The results obtained show that this method is a simple and powerful means for the extraction of Fetal Electrocardiogram.

Keywords: Fetal Electrocardiogram Extraction (FECG), Neuro-fuzzy system, Noise Cancellation