A Novel Method to Remove Eye-Blink Artifacts Based on Correlation Using ICA
Hengsong Sheng, Hongjun Tian · 2013
In this study a novel method is proposed to automatically remove eye-blink artifacts from EEG signals. EEG signals were first decomposed into independent components (ICs) by independent component analysis (ICA). Using temporal correlation theory, the sum of the correlation values between each IC and the EEG signals at some special electrodes was calculated, respectively. Consequently, all the correlation values were sorted by descending order. The ICs which have significantly bigger correlation than the other ICs were picked to set as eye-blink artifact components, which were then set as zero to reconstruct clean EEG signals. Finally, the proposed method was evaluated on some contaminated EEG signals. The experimental results show that it is able to effectively remove eye-blink artifacts with little distortion of underlying brain signals.