EEG-fMRI: Ballistocardiogram Artifact Reduction by Surrogate Method for Improved Source Localization

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Simultaneous electroencephalography-functional magnetic resonance imaging (EEG-fMRI) offers rich insights into brain activity but is challenged by artifacts.
• The ballistocardiogram (BCG) artifact is particularly difficult to remove without compromising neural signal integrity.
• Effective artifact removal is crucial for accurate analysis of simultaneous EEG-fMRI data.

Purpose of the Study:

• To evaluate surrogate source models for separating BCG artifacts from neural signals in EEG-fMRI recordings.
• To compare the efficacy of principal components analysis (PCA-S) and independent components analysis (ICA-S) against established BCG removal methods.
• To assess the impact of artifact removal methods on electroencephalography (EEG) event-related potential (ERP) and source localization analyses.

Main Methods:

• Developed and applied surrogate source models (PCA-S and ICA-S) to separate artifactual signals.
• Compared PCA-S and ICA-S with Blind Source Separation (BSS), Optimal Basis Set (OBS), and OBS-ICA using resting-state EEG-fMRI data.
• Evaluated methods based on artifact threshold survival, signal-to-noise ratio (SNR), source localization error, and signal variance explained.

Main Results:

• PCA-S and ICA-S demonstrated superior performance in BCG artifact removal compared to established methods.
• Significant improvements were observed in source localization accuracy using surrogate source models.
• The PCA-S approach was successfully applied to a Berger experiment, confirming its practical utility.

Conclusions:

• Surrogate source models (PCA-S and ICA-S) offer a substantial advancement for analyzing simultaneous EEG-fMRI data.
• These methods effectively remove BCG artifacts with minimal distortion of neural signals, particularly benefiting source analysis.
• The findings suggest a new standard for artifact removal in combined EEG-fMRI research.