Adaptive ECG Signal Denoising Algorithm Based on the Improved Whale Optimization Algorithm

• Published in: IEEE sensors journal 2024-11, Vol.24 (21), p.34788-34797
• Main Authors: Zhao, Zhi-Hao, Yin, Yi-Fan, Wang, Yu-Ke, Qin, Kai-Rong, Xue, Chun-Dong
• Format: Article
• Language: English
• Subjects: Adaptive algorithms; Adaptive filter; Adaptive filters; Correlation coefficients; Decomposition; electrocardiogram (ECG); Electrocardiography; Feature extraction; indicator S; Noise; Noise reduction; Optimization; Optimization algorithms; Physiological effects; Reference signals; Sensors; Signal to noise ratio; variational modal decomposition (VMD); whale optimization algorithm (WOA); Whale optimization algorithms; Whales
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2024.3422995

Electrocardiogram (ECG) contains rich physiological information. During acquisition, various noises are inevitably introduced due to its weak amplitude. Therefore, noise reduction is required after acquisition. When using the decomposition algorithm for denoising, the traditional algorithm has problems such as mode aliasing, which makes it challenging to achieve the ideal decomposition effect. At the same time, the signal reconstructed using the components obtained after decomposition still has a certain amount of noise. To address these situations, this article proposes an adaptive ECG signal denoising algorithm based on the improved whale optimization algorithm (IWOA). The IWOA algorithm selects the combination of parameters in the variational modal decomposition (VMD) algorithm to enhance its decomposition. After the decomposition, each component's correlation coefficient (CC) with the original signal is calculated. At the same time, the baseline wander (BW) in the low-frequency components is removed using the wavelet algorithm. Finally, the ECG signal with completed denoising is obtained by selecting appropriate components to reconstruct the output reference signal for the adaptive filtering algorithm. To verify the effectiveness of the proposed algorithm, we select various ECG signals for denoising experiments. Several sets of experimental results show that common evaluation indicators, such as the signal-to-noise ratio of the proposed algorithm, are better than those of other denoising algorithms and better preserve the morphological features of the signal, which is conducive to the subsequent extraction of the deep features.