Isolating Eye-Movement Artifacts from EEG Signals

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalogram (EEG) is crucial for measuring brain activity but often suffers from eye-movement artifacts.
• Current artifact removal methods like Independent Component Analysis (ICA) risk removing genuine neural signals.
• Accurate artifact removal is vital to prevent bias in EEG analysis.

Purpose of the Study:

• To develop and validate a novel framework for deconfounding eye-movement artifacts from EEG data.
• To differentiate between eye-movement-related artifacts and neural activity.
• To enhance the reliability of EEG analysis by ensuring neurogenic activity is preserved.

Main Methods:

• Developed complementary deep learning and biophysical modeling approaches.
• Leveraged eye-tracking (ET) data to inform artifact removal.
• Trained a deep learning model to predict EEG components from ET data.
• Utilized a realistic head model to simulate eye-movement artifacts.

Main Results:

• Successfully distinguished neural and non-neural correlates of eye movements.
• Enabled the separation of eye-movement artifacts from non-artifactual neural activity.
• Provided a framework to evaluate the sensitivity and specificity of artifact removal techniques.

Conclusions:

• The developed ET-informed framework accurately removes eye-movement artifacts without compromising neural data.
• This approach offers a robust solution to a significant challenge in EEG analysis.
• Paves the way for more refined artifact removal strategies and future research in disentangling neural components.