Fetal Electrocardiogram Signal Extraction by ANFIS Trained with PSO Method

Abstract

Studies indicate that the primary source of distress in pregnent mothers is their concerns about fetus’s condition and health. One way to know about condition of fetus is non-invasive fetal electrocardiogram signal extraction through which the components of fetal electrocardiogram signal are extracted from a signal recorded at abdominal area of mother which is a combination of fetal and maternal electrocardiogram signal and noise source components. The purpose of this study is to propose an algorithm to boost this extraction. To this end, we decomposed electrocardiogram signal to its Intrinsic Mode Functions (IMFs) thruogh Empirical Mode Decomposition algorithm; then, we removed the last and collected the other IMFs to reconstruct electrocardiogram signal without Baseline. Afterwards, we used Particle Swarm Optimization to train and adjust the parameters of Adaptive Neuro-Fuzzy Inference System to model the path that maternal electrocardiogram signal travel to reach abdominal area. Accordingly, we were able to distinguish and remove maternal electrocardiogram signal components from the recorded signal and hence we obtained a good approximation of fetal electrocardiogram signal. We implemented our algorithm and other algorithms on simulated and real signals and found out that, in most cases, the proposed algorithm improved the extraction of fetal electrocardiogram signal.

DOI:http://dx.doi.org/10.11591/ijece.v2i2.231