Related Experiment Videos
EEG coherency II: experimental comparisons of multiple measures.
P L Nunez1, R B Silberstein, Z Shi
1Department of Biomedical Engineering, Tulane University, New Orleans, LA, USA.
Summary
Different electroencephalography (EEG) coherence methods offer unique insights into brain activity. Verifying theoretical models with EEG data shows that while large-scale brain state changes are consistent, specific electrode pair coherence varies by method, necessitating multiple estimates for comprehensive analysis.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Previous theoretical models using a concentric spheres model estimated the impact of volume conduction, reference electrodes, and spatial filtering on electroencephalography (EEG) coherence.
- This study aimed to empirically validate these theoretical predictions using real EEG data.
Purpose of the Study:
- To verify theoretical predictions regarding EEG coherence measures using empirical data.
- To assess the influence of different recording and processing methods on EEG coherence estimates.
- To understand how various EEG coherence metrics reflect large-scale and specific neural interactions during different brain states.
Main Methods:
- Analysis of three distinct EEG datasets: (1) 64-channel during resting and mental calculation, (2) 128-channel comparing eyes open/closed states, and (3) 128-channel during different sleep stages (deep sleep and REM).
- Comparison of different EEG coherence estimation methods, including average reference, digitally linked mastoids, close bipolar, Laplacian, and dura image methods.
- Evaluation of coherence changes across different brain states and frequencies.
Main Results:
- Large-scale (lobe-level) changes in EEG coherence between brain states were largely consistent across different coherence measures.
- Coherence estimates between specific electrode pairs demonstrated sensitivity to the chosen method and frequency band.
- Average reference and digitally linked mastoids provided semi-quantitative estimates of large-scale neocortical coherence.
- Methods like close bipolar, Laplacian, and dura imaging reduced distortions from reference electrodes and volume conduction but potentially underestimated coherence due to spatial filtering.
Conclusions:
- Each EEG coherence method possesses inherent sources of error and estimates coherence for neural populations of varying sizes and locations.
- To fully leverage the information within recorded EEG signals, studies examining coherence and brain states should employ multiple, diverse estimation techniques.
- The choice of EEG coherence method significantly impacts the interpretation of neural connectivity, particularly at finer spatial scales.