EEG coherence and reference signals: experimental results and mathematical explanations

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalography (EEG) coherence analysis is vital for understanding functional brain connectivity.
• The choice of reference signal significantly impacts EEG coherence results, potentially leading to misinterpretation.
• Existing reference methods may introduce artifacts or distort essential signal components.

Purpose of the Study:

• To investigate the influence of different reference electrode placements on EEG coherence.
• To identify optimal reference strategies for reliable EEG coherence analysis.
• To introduce a mathematical model explaining the impact of reference signals on coherence values.

Main Methods:

• EEG signals were recorded from a single volunteer across 11 sessions using the international 10-20 system.
• Re-referencing was performed to evaluate various reference sites: FCz, averaged non-cephalic ([A1 + A2]/2), common average, Laplacian, and bipolar.
• A mathematical model based on signal-to-noise ratios was developed to interpret experimental findings.

Main Results:

• EEG coherence values are highly dependent on the chosen reference electrode site.
• Averaged non-cephalic reference signals ([A1 + A2]/2) yielded the most reliable coherence results.
• Common average reference, Laplacian, and bipolar methods produced significantly different results, likely due to signal cancellation effects.
• The mathematical model confirmed that noisy references increase coherence, while coherent references decrease it.

Conclusions:

• The selection of a reference site is a critical determinant of EEG coherence outcomes.
• Averaged non-cephalic reference signals are recommended for robust and interpretable EEG coherence analysis.
• Common average, Laplacian, and bipolar referencing techniques should be used with caution due to potential signal distortion.