ECG signal denoising using time-frequency based filtering approach

Electrocardiogram (ECG) has become a primary tool for proper diagnosis of cardiovascular disease. During the acquisition of ECG signal, it is embedded with noise. Presence of noise in the signal may lead to variation in the amplitude and time intervals, which in turn falsely represents the abnormalities associated with the heart. This may lead to false diagnosis of arrhythmia. Hence to facilitate proper analysis, this paper presents a novel ECG signal denoising method based on S-transform based time-frequency filtering approach. The performance of the proposed method is evaluated by taking different normal and abnormal ECG signals from the MIT/BIH arrhythmia database. Two different types of noises are added with the data, white Gaussian noise and muscle noise. The results show that the proposed method is promising in terms of better signal to noise ratio (SNR), minimum mean square error (MSE) and percent root mean square difference (PRD).