ECG Signal Denoising Method Based on CEEMDAN Algorithm and Strong Adaptive Wavelet Threshold

This article presents a novel denoising approach aimed at mitigating power frequency interference, EMG interference, and baseline drift in ECG signals. The method leverages the CEEMDAN algorithm and a strong adaptive wavelet threshold for effective noise reduction. Initially, the ECG signal is decomposed into a series of IMF components across a spectrum of frequencies using the CEEMDAN algorithm. Subsequently, the wavelet threshold method, fortified with a decomposition layer adaptive threshold and a parameter self-tuning threshold function, is employed to denoise the high-frequency IMF components. Low-frequency IMF components, identified as baseline drift signals, are then eliminated. Finally, the denoised IMF components are merged with the retained ones to reconstruct the signal and yield a clean ECG output. Experimental findings underscore the method's superior adaptability and stability compared to EMD-like algorithms combined with conventional wavelet thresholds. The approach showcases enhanced denoising performance across diverse noise conditions.