A NLMS Based Approach for Artifacts Removal in Multichannel EEG Signals with ICA and Double Density Wavelet Transform

Abstract

Presence of artifacts in electroencephalogram (EEG) signals is significant hurdles in analysis of spectral behavior. These artifacts are the low amplitude signals from unconscious ocular activity and muscles activity of human body. Since the source and noise in received signals originate from different sources, ICA method has been extensively revised for proper filtering. It involves the generating a set of individual components of given signal followed by rejection of unwanted artifacts. The results of this research show that considerable artifacts components persist in clean EEG signals. In this paper, we propose Double-Density DWT algorithm as the overhead computation with ICA for further filtering the signals. ICA segments the artifact peaks and DWT decompose them for suitable signal value. The Wavelet ICA suppression not only remove artifacts but also preserves the spectral (amplitude) and coherence (phase) characteristics of neural activity. In addition to this, NLMS filter is used at output of DWT to discard any trace of artifacts left in signal. The comparison of proposed scheme and conventional ICA indicates that NLMS filtered DWT-ICA outperforms the previous methods.