Generative Adversarial Networks in ECG Signal Denoising: A Survey

In this survey, we present a thorough analysis of denoising ECG signal approaches using Generative Adversarial Networks (GANs). Our aim with this survey is to examine the most recent studies that have utilized different types of GANs architecture for removing various types of noise from ECG signals, ranging from real-world to synthetic noise. This review paper offers experts a clear understanding of ECG denoising methods, particularly those utilizing the GANs architecture, and enables them to explore unresolved issues. We have identified several limitations and recognized the need for improvement to enhance the performance of these approaches in handling different types of noise with varying SNR levels, while also minimizing preprocessing and computational costs. Ultimately, our intention is to address the raised concerns and limitations to further enhance the performance of ECG signal denoising approaches, especially those utilizing GAN architecture, thereby increasing their potential for practical clinical diagnosis.