Automated pipeline for EEG artifact reduction (APPEAR) recorded during fMRI

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Simultaneous electroencephalography-functional magnetic resonance imaging (EEG-fMRI) offers high spatiotemporal resolution for brain research.
• EEG data acquired during fMRI are heavily contaminated by MRI gradients and physiological artifacts.
• Manual artifact reduction is time-consuming and can introduce bias due to subjective processing steps.

Purpose of the Study:

• To introduce APPEAR, an open-access, fully automatic toolbox for comprehensive artifact reduction in simultaneous EEG-fMRI data.
• To develop a pipeline that integrates average template subtraction and independent component analysis for artifact suppression.
• To validate the automated pipeline against expert manual review for both resting-state and task-based EEG-fMRI data.

Main Methods:

• Developed a fully automatic processing pipeline integrating average template subtraction and independent component analysis.
• Tested the APPEAR toolbox on EEG data from 48 healthy controls during resting-state and 8 subjects during event-related potentials (ERPs) tasks.
• Validated automated artifact correction by comparing results with expert manual review using frequency analysis, continuous wavelet transformation, and ERP measures.

Main Results:

• No significant differences were found between manually and automatically corrected resting-state EEG data when analyzed with frequency analysis and continuous wavelet transformation.
• Comparisons of event-related potential (ERP) data (amplitude measures, signal-to-noise ratio) showed no significant differences between manual and automated fMRI-EEG correction.
• The APPEAR toolbox demonstrated effective suppression of both MRI-related and physiological artifacts.

Conclusions:

• APPEAR provides the first comprehensive, open-source solution for automatic artifact reduction in simultaneous EEG-fMRI.
• This toolbox significantly speeds up EEG analysis, enhances reproducibility by removing experimenter subjectivity, and facilitates large-scale EEG-fMRI cohort studies.
• The automated approach allows for efficient processing of large datasets, reducing researcher time and effort while maintaining data integrity.