Title: Foetal ECG extraction using BPN and UWT

Authors: R. Parameshwari; C. Emlyn Gloria Ponrani; S. Shenbaga Devi

Addresses: Department of ECE, College of Engineering, Guindy, Anna University, Chennai 600025, India ' Department of ECE, College of Engineering, Guindy, Anna University, Chennai 600025, India ' Department of ECE, College of Engineering, Guindy, Anna University, Chennai 600025, India

Abstract: Foetal electrocardiogram (FECG) signal could assist the clinicians in making appropriate and timely decisions during pregnancy and labour since it contains information about the status of the foetus. The extraction and detection of the FECG signal from the composite abdominal signals of the mother with powerful and advance methodology is becoming a very important requirement in foetal monitoring. This paper illustrates the algorithm developed based on neural network approach to provide efficient and effective way of separating FECG signal from the abdominal ECG (AECG) signals. The FECG signal is isolated from the abdominal signal by back propagation neural network approach. The input signal is considered as maternal ECG (MECG) obtained from the thorax of the mother, and the target signal is abdominal signal (AECG) which has both MECG and FECG. The output of the network is subtracted from the target input (AECG). To reduce the difference between the input and target signal the weights have been updated every step, so that acceptable FECG signal can be obtained. To enhance the FECG signal, post-processing steps are followed which involve undecimated wavelet transform (UWT), thresholding and filtering technique.

Keywords: neural networks; foetal ECG; electrocardiograms; UWT; undecimated wavelet transform; thresholding; filtering; maternal abdominal ECG; heart rate variability; ECG signals; pregnancy; labour; foetal monitoring.

DOI: 10.1504/IJBET.2016.078980

International Journal of Biomedical Engineering and Technology, 2016 Vol.22 No.1, pp.1 - 13

Received: 16 Jun 2015
Accepted: 02 Nov 2015
Published online: 08 Sep 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article