Least Mean Squares Adaptive Algorithms Optimization for Fetal Phonocardiogram Extraction

Abstract

The paper deals with extraction of fetal phonocardiogram (fPCG) from the abdominal signal mixture by using adaptive filters based on Least Mean Squares (LMS) algorithms. Optimal setting of the filter parameters is a vital condition to ensure efficiency of adaptive filters. In this paper, we suggest ideal parameters setting for this purpose. The filtration quality was assessed by means of objective parameters, namely Signal to Noise Ratio (SNR) and Positive Predictive Value (PPV). The results showed that if optimized, LMS-based adaptive algorithms are efficient in suppressing maternal component in the composed abdominal signal.