Artifact reduction in multichannel pervasive EEG using hybrid WPT-ICA and WPT-EMD signal decomposition techniques

TL;DR

This paper introduces and compares two hybrid signal decomposition algorithms, WPT-ICA and WPT-EMD, for reducing muscle artifacts in multi-channel pervasive EEG signals, demonstrating improved artifact removal performance.

Contribution

The study proposes novel hybrid algorithms combining WPT with ICA and EMD for EEG artifact reduction, providing a comparative analysis of their effectiveness.

Findings

WPT-ICA and WPT-EMD outperform traditional methods in artifact suppression.

Both algorithms effectively reduce muscle artifacts in EEG signals.

WPT-ICA shows slightly better performance in SNR improvement.

Abstract

In order to reduce the muscle artifacts in multi-channel pervasive Electroencephalogram (EEG) signals, we here propose and compare two hybrid algorithms by combining the concept of wavelet packet transform (WPT), empirical mode decomposition (EMD) and Independent Component Analysis (ICA). The signal cleaning performances of WPT-EMD and WPT-ICA algorithms have been compared using a signal-to-noise ratio (SNR)-like criterion for artifacts. The algorithms have been tested on multiple trials of four different artifact cases viz. eye-blinking and muscle artifacts including left and right hand movement and head-shaking.