Robust Fetal ECG Signal Denoising via Continuous Wavelet Transform and Highly Accurate Classification Using XGBoost

This paper outlines a powerful method for studying fetal electrocardiogram (FECG) signals through the use of superior signal denoising and machine learning strategies. FECG signals are, by nature, very weak, and they get mixed with the maternal ECG and other noise, which makes them very difficult to process. As a solution, Continuous Wavelet Transform (CWT) is utilized for noise removal in the high-resolution time–frequency domain, which enables the cardiac features to be clearly seen. The XGBoost classifier is then used to train on the set of features that are extracted in both statistical and frequency domains. Experimental evaluation proves that the suggested approach attains classification performance almost perfect on four abdominal ECG datasets, highlighting its capability for dependable real-time prenatal monitoring.