Akademik Veri Yönetim Sistemi
Karadeniz Fen Bilimleri Dergisi, cilt.14, sa.4, ss.2115-2127, 2024 (Hakemli Dergi)
Electrocardiogram (ECG) signals provide information about heart functions and some cardiac diseases. However, various interferences distort the ECG waveforms during its measurement and transmission can cause inaccurate analysis and diagnosis. So, this unwanted disturbance signals must be eliminated and an acceptable ECG signal must be extracted the noisy ECG recordings. Researchers developed several methods to overcome the undesired noises and interferences contaminated to the ECG recordings. The adaptive filtering techniques have attracted the attention of scientists due to their adaptation mechanism to time-varying nature of undesired signals. Most of the presented adaptive filtering algorithms are gradient-based and have the advantage of simple implementation, but are affected negatively by disturbance signals; for example, they can have slow convergence rates and poor steady-state properties. Least squares-based algorithms are advantageous due to their faster convergence rates and better steady-state properties. In this paper, Recursive Gauss-Seidel (RGS) algorithm, which is an alternative least squares-based method to Recursive Least Squares (RLS) algorithm with less computational complexity, is presented to obtain an acceptable waveform from noisy ECG recordings. The denoising performance of the RGS algorithm is studied and compared to the widely used gradient-based algorithms and the popular RLS algorithm.