Stationary Wavelet and Recursive Least Square Filtering Based Fetal ECG Data Extraction from Composite Abdominal Signal

Abstract

This paper introduces a fast methodology for fetal ECG extraction based on stationary wavelet and noise canceler adaptive filter with the recursive least square filter. Firstly, signals are preprocessed by moving averaging filter for removing baseline wander. The stationary wavelet and the recursive least square filter is applied on preprocessed signal that effectively separates the maternal ECG and extracts fetal ECG (fECG) from the abdominal ECG. Finally, fetal ECG(fECG) is extracted with R peak detection using the Pan-Tompkins algorithm. The experiment has been conducted at MIT-BIH abdominal and direct fetal ECG database (adfecgdb) available on physionet. The obtained results are evaluated by the parameters of accuracy (Acc), sensitivity (Se), positive diagnostic value (PDV) and cross-correlation. The achieved results Acc=100%, Se=100%, PDV=100% based on counting the number of R-peak and cross correlation is 98.93% between direct fetal scalp recording and extracted fECG. Moreover, the results outperform the existing state of art works on fetal ECG extraction.

Keywords: Fetal electrocardiogram, fetal heart rate, adaptive filter, recursive least square filter (RLS), abdomen ECG, wavelet analysis

Cite this Article
Priya Gupta, Brajinder Singh Saini, R.K. Sunkaria et al. Stationary Wavelet and Recursive Least Square Filtering Based Fetal ECG Data Extraction from Composite Abdominal Signal. Current Trends in Signal Processing. 2017; 7(2): 46–58p.