Electrocardiogram signal denoising by a new noise variation estimate

Separating or eliminating the noise from a biomedical signal is what allows the accuracy of a diagnosis. In particular, in the case of an electrocardiogram (ECG), it is necessary to reduce the distortions caused by several sources of noise. In this paper, we propose a new ECG denoising method called by noise reduction by genetic algorithm minimization of a new noise variation estimate (GAMNVE). The GAMNVE method applies the discrete wavelet transform (DWT) in the noisy ECG signal and processes the wavelet coefficients by the minimization of a new noise variance estimate. This minimization was made by genetic algorithm. For the simulations, we consider eight real ECG signal corrupted by additive white Gaussian noise (AWGN), power line interference (PLI), and muscle artifact (MA). We compare the GAMNVE method with five well-known denoising methods. The simulations results show that the GAMNVE method presents a better performance for the considered cases. Simulations have demonstrated that the GAMNVE method can be applied in noisy ECG signals with superior performance than other methods established in the literature.