EyeCatch: data-mining over half a million EEG independent components to construct a fully-automated eye-component...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Independent Component Analysis (ICA) is crucial for distinguishing brain activity from artifacts in electroencephalographic (EEG) data.
• Accurate identification of non-brain independent components (ICs), such as eye or muscle activity, is essential for robust EEG analysis.

Purpose of the Study:

• To develop a fully automated method for identifying eye-movement related EEG components.
• To improve the efficiency and objectivity of EEG data preprocessing.

Main Methods:

• The EyeCatch method analyzes the spatial distribution of independent component (IC) scalp projections.
• It compares input scalp maps against a large database of pre-identified eye-related IC scalp maps.
• The database was constructed by data-mining over 80,006 EEG datasets, encompassing half a million IC scalp maps.

Main Results:

• EyeCatch demonstrates comparable performance to the semi-automated CORRMAP method for identifying eye-related EEG components.
• The method successfully automates the distinction between brain and non-brain ICs related to eye movements.
• This represents the largest sample of IC scalp maps analyzed for automated artifact detection.

Conclusions:

• The EyeCatch method provides a fully automated and effective solution for detecting eye-movement artifacts in EEG data.
• It significantly reduces the need for manual inspection in EEG preprocessing pipelines.
• This advancement facilitates more efficient and reliable analysis of large-scale EEG datasets.