ICA-based muscle artefact correction of EEG data: what is muscle and what is brain? Comment on McMenamin et al

Abstract

Independent component analysis (ICA)-based muscle artefact correction has become a popular tool within electroencephalographic (EEG) research. As a comment on the article by McMenamin et al. (2010), we want to address three issues concerning the claimed lack of sensitivity and specificity of this method. The under- or overestimation of myogenic and neurogenic signals after ICA-based muscle artefact correction reported by McMenamin et al. might be explainable in part by a) insufficient temporal independence of myogenic and neurogenic components when exploring more than one condition, b) wrong classification of myogenic or neurogenic components by human raters and c) differences of neuronal mass activity during tensed or relaxed-muscle conditions. Our own data show only significant differences regarding intracortical alpha band EEG-source estimates for contrasts between clean EEG data and artificially contaminated EEG data at group-analysis level but not between clean data and data after ICA-based correction. ICA-based artefact correction already provides a powerful tool for muscle artefact rejection. More research is needed for determining reliable criteria to delineate myogenic from neurogenic components.