Independent component analysis as a tool to eliminate artifacts in EEG: a quantitative study

Area of Science:

• Neuroscience
• Signal Processing
• Biomedical Engineering

Background:

• Electroencephalogram (EEG) signals often contain artifacts from various sources.
• Artifacts can obscure important neurological information, hindering accurate diagnosis.
• Existing artifact removal methods, like digital filters, can introduce signal distortion.

Purpose of the Study:

• To apply Independent Component Analysis (ICA) for artifact removal in EEG recordings.
• To objectively quantify the efficacy of ICA in eliminating common EEG artifacts.
• To compare ICA's performance against traditional artifact removal techniques.

Main Methods:

• Eighty EEG samples with artifacts (EKG, eye movements, 50-Hz interference, muscle, electrode) were analyzed.
• The Joint Approximate Diagonalization of Eigen-matrices (JADE) algorithm was used to compute ICA components.
• Signals were reconstructed by excluding artifact-related components.
• Normalized correlation coefficients assessed signal changes post-artifact removal.

Main Results:

• ICA successfully cleared artifacts in all analyzed EEG samples.
• Signal reconstruction using ICA showed minimal distortion of interictal activity.
• Correlation analysis confirmed that the underlying EEG signal remained largely unchanged.
• Independent assessment by two examiners corroborated the effective removal of artifacts.

Conclusions:

• ICA is a valuable tool for cleaning artifacts in short EEG samples.
• ICA offers an effective alternative to digital filters, avoiding their disadvantages.
• The technique preserves the integrity of important EEG signal components while removing noise.