Relative wavelet energy as a tool to select suitable wavelet for artifact removal in EEG

Abstract

We proposed a novel approach to select a wavelet for artifact removal in electroencephalogram (EEG) by comparing their relative wavelet energies before and after thresholding. Relative wavelet energy (RWE) gives information about the relative energy associated with different frequency bands and can be considered as a time-scale density. RWE can be used as a tool to detect and characterize a specific phenomenon in time and frequency planes. We used Lifting Wavelet Transform to remove the common artifacts exist in EEG i.e. blink and eye movements. Three basic steps involved are to transform the EEG, hard thresholding the wavelet coefficients and the corrected EEG is obtained by inverse transform these threshold coefficients. It is of paramount important to select a suitable wavelet and threshold to accomplish this task. From this study, we concluded that cdf4.4 outperformed db4 and haar wavelets by removing the artifacts at the correct times and frequency bands.