Related Experiment Videos
Spatial filtering and neocortical dynamics: estimates of EEG coherence
R Srinivasan1, P L Nunez, R B Silberstein
1Department of Psychology, University of Oregon, Eugene 97403-1227, USA. srinivasan@nsi.edu
IEEE Transactions on Bio-Medical Engineering
|June 30, 1998
Summary
Spatial statistics of electroencephalogram (EEG) coherence are affected by volume conduction. This study models head tissues to estimate artificial correlations, suggesting surface Laplacian analysis can reduce this effect for better neocortical function studies.
Area of Science:
- Neuroscience
- Biophysics
- Signal Processing
Background:
- Scalp electroencephalogram (EEG) spatial statistics often use coherence within frequency bands.
- Coherence arises from neocortical source correlations and volume conduction through head tissues.
- The skull's low conductivity acts as a spatial low-pass filter, creating artificial electrode correlations in EEG.
Purpose of the Study:
- To analytically estimate scalp EEG coherence caused by volume conduction of uncorrelated sources.
- To investigate the impact of head tissue conductivity on spatial EEG statistics.
- To evaluate the effectiveness of surface Laplacian estimation in mitigating volume conduction artifacts.
Main Methods:
- Utilized a four-concentric-spheres head model (brain, CSF, skull, scalp).
- Applied stochastic field theory to derive an analytic estimate of volume conduction effects.
- Compared coherence estimates from raw scalp EEG with surface Laplacian transformed potentials.
Main Results:
- The model predicted artificial correlations between electrodes within 10-12 cm due to volume conduction.
- Surface Laplacian estimation spatially bandpass filters scalp potentials.
- This filtering effectively reduces artificial coherence introduced by volume conduction.
Conclusions:
- Raw scalp EEG coherence is sensitive to broad spatial bandwidths, including volume conduction artifacts.
- Surface Laplacian estimates are sensitive to narrower spatial bandwidths, reducing volume conduction influence.
- Parallel use of both raw scalp EEG and Laplacian estimates is recommended for comprehensive neocortical dynamic function studies.