Extracting Fetal ECG Signals Through a Hybrid Technique Utilizing Two Wavelet-Based Denoising Algorithms

It is essential to create a clever method for detecting fetal heartbeats in order to track the fetus's cardiac health during the first few months of pregnancy. This study proposes two hybrid algorithms for fetal ECG extraction that combine the stationary wavelet transform (SWT) with the recursive least square approach (RLS). By using the threshold-based denoising methodology or the improved spatially selective noise filtering (ISSNF) method in the wavelet domain, the research aims to improve the fetal ECG signal, lower noise and artifact, and precisely identify the R-peaks. Fetal heart problems can be diagnosed and treated with the help of precise fetal R-peak identification, which can also yield crucial clinical information. The principal objective is to obtain a distinct fetal electrocardiogram signal from the mixed abdomen signal. Using SWT, the abdominal signal is split into multiscale components, with the wavelet decomposition scale being determined by varying noise levels. The maternal ECG components are then eliminated using the RLS method, and denoising in the wavelet domain is accomplished using either ISSNF or threshold-based algorithms. We assess our suggested method's efficacy using both clinical and synthetic data. Both qualitative and quantitative methods are used in our investigation, such as visual inspection, QRS complex detection, and signal-to-noise ratio (SNR) calculation. Our results show that, in comparison to traditional adaptive filtering methods, the suggested approach performs better. According to the experimental findings, the suggested method may be able to extract clear fetal ECG signals with low disruptions and good SNR values.

Keywords: ECG extraction, Fetal ECG, Improved Spatially Selective Noise Filtration, Recursive Least Square Algorithm, Stationary Wavelet Transforms, Threshold-Based Algorithm.