Retrieving and smoothing fundamental waves from noise corrupted ECG beat using Gaussian functions

ECG denoising is a prerequisite for modeling, detecting various artifacts, measuring component waveforms and for almost all computational analysis of ECG signal. Gaussian waveform is a good candidate for ECG modeling due to its similarity with the characteristic waves of ECG beat. This paper presents a novel method for denoising and smoothing ECG signal using Gaussian functions. The advantage of Gaussian modeling is that by fitting Gaussian waveforms it inherently extracts the fundamental waveforms from the noisy signal, removes wrinkles and makes the signal sufficiently smooth for accurate detection and measurement of the on-set and off-set of the characteristic waves. Sixteen Gaussian waves are used to model each noisy ECG beat. The proposed algorithm has been applied on ECG signals with different SNR values. The results show significant improvement in the output SNR.