Improved Fetal Electrocardiogram Detection Using a Hybrid Wavelet-Based Denoising Method

This study shows a mixed way to get fetal electrocardiogram (fECG) signals utilizing the recursive least squares (RLS) algorithm and the stationary wavelet transform (SWT). The goal is to improve fECG data by getting rid of noise and artifacts and finding R-peaks accurately using either a threshold-based denoising method or the improved spatially selective noise filtering (ISSNF) method in the wavelet domain. Finding the R-peak accurately is important for detecting fetal heart problems and getting important clinical information. After the fECG is taken, a convolutional neural network (CNN) is used to figure out the fetal heart rate (FHR). This makes it possible to keep an eye on the health of the fetus. SWT breaks down the abdominal electrocardiogram (ECG) into parts at different scales, with the scales selected based on the amount of noise. The RLS approach gets rid of the mother’s ECG, and then ISSNF or threshold-based methods are used to eliminate noise. Both real and fake datasets are utilized to test the suggested method. We check performance by looking at the results, finding the QRS complex, and measuring the signal-to-noise ratio (SNR). The proposed method works better than standard adaptive filtering methods, as shown by experimental findings. It achieves a high SNR, low distortion, and clearer signals.