Abstracts

Rationale: Coherence and correlation of the intracranial electroencephalogram (iEEG) may be used to assess connectivity between brain areas. These measures correspond to some degree with noninvasive estimates using correlated fluctuations of the functional magnetic resonance imaging (fMRI) signal. A better understanding of this relationship might be revealed by examining the correspondence of the dynamics of these measures in situations where connectivity is disrupted in a predictable fashion. We examined resting intracranial EEG and resting fMRI data from three subjects obtained before and after anterior corpus callosotomy. For each subject, we compared different analyses for resting intracranial EEG functional connectivity – coherence, correlation in high gamma power, and correlation in slow oscillations of high gamma power – to functional connectivity measured by the resting BOLD signal. Methods: Three patients were implanted with bilateral symmetric depth electrodes in frontal, parietal, temporal and insular cortices for presurgical monitoring in preparation for epilepsy surgery. Electrodes were secured with skull bolts throughout monitoring and during surgery. For each patient, anterior corpus callosotomy was performed using stereotactic laser ablation. Resting intracranial EEG, resting fMRI, and diffusion tensor imaging (DTI) data were acquired before and after laser ablation of the anterior corpus callosum.We compared BOLD time courses extracted from seed regions of interest at the location of each implanted electrode with resting iEEG recorded from implanted electrodes, both before and after corpus callosotomy. For resting intracranial EEG, we computed raw coherence between electrode pairs, correlation between electrode pairs in band-limited high gamma power (50-150 Hz), and correlation in band-limited high gamma power filtered to extract slow (.1-1Hz) oscillations (Keller et al., 2013).  Results: Of the intracranial EEG analyses evaluated, slow oscillations in high gamma power of the intracranial EEG signal yielded the most similar map of functional connectivity to that obtained from resting fMRI. We found that slow oscillations in high gamma power also largely tracked disruption of white matter tracts after callosotomy, as measured by DTI. Slow oscillations in high gamma power of the intracranial EEG signal were more sensitive, compared to fMRI, to disruptions of interhemispheric functional connectivity following callosotomy. Conclusions: Our results indicate that changes in intracranial EEG metrics of resting functional connectivity correspond with changes in structural connectivity measured by DTI, and changes in functional connectivity as measured by resting-state fMRI. Compared to correlations in high gamma power or coherence, dynamics of the correlations of slow oscillations in high gamma power better correspond to changes in functional connectivity measured by resting fMRI after callosotomy. Funding: NIH 1R01-MH111439-01.