Automatic correction of ocular artifacts in the EEG: a comparison of regression-based and component-based methods

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Ocular artifacts, such as electrooculogram (EOG) activity, significantly contaminate electroencephalogram (EEG) recordings.
• Several methods, including regression, principal component analysis (PCA), and independent component analysis (ICA), exist for EEG artifact correction.

Purpose of the Study:

• To compare the efficacy of regression, PCA, and ICA methods for ocular artifact correction in EEG.
• To evaluate a modified regression approach incorporating Bayesian adaptive regression splines for EOG filtering.
• To assess the impact of these correction methods and varying epoch lengths on EEG spectral parameters.

Main Methods:

• Comparison of regression, PCA, and ICA techniques for ocular artifact removal.
• Application of a modified regression approach with Bayesian adaptive regression splines for EOG filtering.
• Analysis of real and simulated EEG data with different epoch lengths.
• Quantification of the impact of artifact correction on EEG spectral parameters.

Main Results:

• The adaptive regression filter enhanced regression-based artifact correction.
• Automated PCA effectively reduced ocular artifacts with minimal spectral distortion.
• ICA correction showed potential distortion in EEG power between 5 and 20 Hz.
• Shorter epoch lengths improved alpha and beta band spectral accuracy but worsened theta band accuracy and distorted time-domain features.

Conclusions:

• Regression-based and PCA-based methods are supported for ocular artifact correction in EEG.
• Further research is needed to investigate potential spectral distortions caused by ICA-based correction methods.
• Epoch length selection impacts spectral accuracy and time-domain features, requiring careful consideration.