Adjusting EEG coherence for inter-electrode distance effects: an exploration in normal children

Abstract

Electroencephalographic (EEG) coherence between two points is strongly related to the distance between them, being inflated by volume conduction effects at short distances and reduced by signal phase differences at larger distances. This precludes simple comparison of coherence estimates involving different inter-electrode distances. We investigated adjusting coherence measures to remove such distance effects. After subtracting the estimated effects of random coherence due to volume conduction, exponential regression of the reduced coherence values against measured inter-electrode distance was used to estimate the remaining effects of inter-electrode distance. Residuals from this procedure were taken as coherences corrected for the systematic distance effects. These were adjusted to the mean reduced coherence level to avoid complexities in conceptualising negative residual coherences. It was found that systematic inter-electrode distance effects accounted for more than 50% of the variance remaining after removal of random coherence estimates. After these were also removed, substantial effects of EEG frequency band, different regions of the brain, and interhemispheric versus intrahemispheric values, as well as laterality effects within the latter, were obtained. Regional and frequency differences in adjusted coherence appear to reflect patterns expected from normal cortical development, but detailed understanding of more complex interactive effects is limited by the lack of relevant developmental data. Adjusting coherence values to remove systematic variability due to inter-electrode distances better represents cortico-cortical coupling and allows more efficient statistical analysis. This may contribute towards a better integration of coherence data in the EEG exploration of both normal and atypical brain functioning.