Accounting for microsaccadic artifacts in the EEG using independent component analysis and beamforming

Area of Science:

• Neuroscience
• Cognitive Science
• Biomedical Engineering

Background:

• Gamma-band neuronal activity in electroencephalography (EEG) was historically linked to object representation.
• Scalp-recorded EEG gamma-band signals are often contaminated by miniature saccade-related muscle artifacts.
• Independent Component Analysis (ICA) is a common method for artifact removal in EEG data.

Purpose of the Study:

• To evaluate the effectiveness of ICA-based artifact correction on EEG gamma-band signals.
• To utilize beamforming to assess the impact of ICA correction on genuine neuronal activity.
• To determine if ICA correction successfully removes artifactual signals while preserving neural signals.

Main Methods:

• Applied ICA-based correction to a previously published EEG dataset.
• Employed beamforming, a source analysis technique using adaptive spatial filters, to analyze EEG signals.
• Compared beamforming results on EEG data before and after ICA correction.

Main Results:

• Before ICA correction, beamforming identified significant gamma-band activity originating from deep frontal sources, attributed to eye muscle artifacts.
• Beamforming confirmed that ICA predominantly removes these artifactual signals.
• Post-ICA correction, remaining gamma-band activity was plausibly localized to the visual cortex, indicating successful artifact removal and preservation of neural signals.

Conclusions:

• ICA-based correction significantly improves the signal-to-noise ratio in EEG gamma-band data.
• Beamforming serves as a valuable tool to validate the efficacy of artifact removal procedures like ICA.
• ICA correction should be used in conjunction with, rather than be replaced by, beamforming for comprehensive EEG analysis.