Enhancing Signal and Network Integrity: Evaluating BCG Artifact Removal Techniques in Simultaneous EEG-fMRI Data

Area of Science:

• Neuroscience
• Biomedical Engineering
• Data Science

Background:

• Simultaneous Electroencephalography (EEG) and functional Magnetic Resonance Imaging (fMRI) are crucial for studying brain dynamics.
• Ballistocardiogram (BCG) artifacts in EEG data degrade signal quality, hindering accurate brain connectivity assessment.
• Existing artifact removal methods require comprehensive evaluation within a holistic framework.

Purpose of the Study:

• To evaluate the efficacy of different EEG artifact removal techniques (AAS, OBS, ICA, and hybrid methods) in the context of simultaneous EEG-fMRI.
• To assess the impact of these methods on both signal quality and graph-theoretical measures of brain connectivity.
• To investigate the frequency-specific effects of artifact removal on static and dynamic brain network structures.

Main Methods:

• Comparison of Average Artifact Subtraction (AAS), Optimal Basis Set (OBS), Independent Component Analysis (ICA), and hybrid approaches (AAS + ICA, OBS + ICA).
• Evaluation using a combination of signal quality metrics and graph-theoretical network analysis (static and dynamic).
• Analysis of topological properties and frequency-specific patterns in EEG-fMRI connectivity.

Main Results:

• AAS yielded the best signal quality, while OBS preserved structural similarity.
• ICA showed sensitivity to frequency-specific patterns in dynamic connectivity.
• The OBS + ICA hybrid method demonstrated significant improvements, particularly in dynamic graph analysis across various frequency band pairs.
• Artifact removal methods substantially altered network topology, with dynamic analyses revealing more pronounced frequency-specific effects.

Conclusions:

• The choice of artifact removal method critically influences EEG-fMRI connectivity results.
• Hybrid methods, especially OBS + ICA, offer advantages for dynamic brain network analysis.
• Multimodal and frequency-sensitive evaluation strategies are essential for robust EEG-fMRI research.
• Findings provide guidance for preprocessing decisions in EEG-fMRI studies, impacting the interpretation of brain connectivity.