Understanding the Influences of EEG Reference: A Large-Scale Brain Network Perspective

Area of Science:

• Neuroscience
• Computational Neuroscience
• Biomedical Engineering

Background:

• Reference selection significantly impacts electroencephalography (EEG) and event-related potentials (ERPs) research.
• Previous studies lacked systematic neuroscience-level analysis of reference influence on source localization.
• Understanding reference effects is crucial for accurate interpretation of brain activity.

Purpose of the Study:

• To systematically investigate the influence of EEG reference locations on brain network activity.
• To provide a systems-level perspective on reference selection in EEG/ERPs.
• To compare the performance of different reference schemes across large-scale brain networks.

Main Methods:

• Simulated scalp EEG data from vertices distributed across eight large-scale human brain networks.
• Calculation of electrode sensitivity and neutrality patterns using lead-field matrices.
• Comparison of reference schemes: FCz, Oz, mean mastoids (MM), average (AVE), and reference electrode standardization technique (REST).

Main Results:

• Reference electrode standardization technique (REST) and average (AVE) referencing demonstrated the lowest relative error.
• Relative error followed the pattern: REST < AVE < MM < (FCz, Oz).
• Network-specific patterns of sensitive electrodes were observed, while neutral electrodes were distributed across the scalp.

Conclusions:

• REST is a potentially superior reference for analyzing large-scale brain networks in EEG/ERPs.
• AVE referencing performs comparably to REST under various conditions.
• Findings offer valuable recommendations for selecting EEG references in clinical and basic neuroscience research.