Neural-Network-based Adaptive ECG Denoising

M.H. Sedaaghi and E. Ajami (Iran)


Neural networks, adaptive filtering, ECG denoising.


A novel approach in suppressing Hz noise, baseline drift and EMG noise from electrocardiogram (ECG) sig nals is presented in this paper using artificial neural net works (ANNs). In the newly introduced approach, the ma jor objective is to design an adaptive noise canceller with appropriate forms of various neural networks (NNs), which are recurrent, feed-forward and radial basis, and applying them for ECG denoising. Experimental results prove the dominance of NN-based adaptive filtering over other con ventional adaptive algorithms such as LMS (least mean square), RLS (recursive least square) and KLM (Kalman) in both single and multiple noisy environments.

Important Links:

Go Back