A Comprehensive Survey of ECG Signal Denoising Techniques, Challenges and Novel Trends

Abstract

The accurate analysis of electrocardiogram (ECG) signals is crucial for cardiovascular diagnosis, but these signals are frequently corrupted by various forms of noise during collection and preprocessing. This survey presents a comprehensive overview of the primary types of noise that affect ECG signals, such as baseline wander, muscle noise, power line interference, and motion artifacts. These sources of noise significantly impair the effectiveness of ECG diagnosis systems. While conventional filtering methods can address some types of noise, they often fall short when it comes to dealing with non-stationary and complex noise patterns. Recent advancements in denoising techniques, including wavelet transforms, empirical mode decomposition (EMD), and deep learning models, demonstrate enhanced performance in reducing noise and preserving signal quality. This review underscores the increasing significance of hybrid approaches that combine traditional and modern techniques, highlighting their potential for real-time applications and improved diagnostic accuracy.