Abstracts

Rationale: Determining the extent of ‘eloquent' cortex as well as regions involved in early seizure activity will likely benefit epilepsy surgery. However, techniques used to measure these networks involve high frequency stimulation to observe behavioral arrests and recording of ictal activity, respectively. Localization of these networks during the interictal period would provide a complementary measure to the current gold standards and possibly identify regions not implicated during clinical workup. Methods: We combined single pulse stimulation, resting state fMRI, and resting state electrocorticography (ECoG) from 5 patients with intractable epilepsy undergoing surgical evaluation of seizure foci. Cortical regions involved in seizure initiation and early propagation (the ‘pathological network') were identified visually in the iEEG by a neurologist blind to the study. Functional networks were defined by behavioral changes in language, vision, movement, or sensation as a result of high frequency stimulation mapping. Single pulse stimulation (10mA, 100us/phase, 20 trials, 2s ISI) was applied to adjacent contacts on implanted grids and strips and the evoked potentials (CCEPs) were recorded at each location. Pre-operative resting state fMRI and post-implantation ‘resting state' ECoG were performed for 3-5 minutes while patients were instructed to close their eyes and let their mind wander. Correlated fluctuations of the BOLD signal (resting state functional connectivity - RSFC) as well as correlated fluctuations of the high gamma (70-150 Hz) power signal (HGP correlations) at each electrode pair were quantified. For each electrode, its network was identified (‘functional', ‘pathologic', or neither) and the connectivity at electrodes within or outside each network were compared. Additionally, graph theoretical measures were computed from each connectivity matrix (RSFC, HGP correlations, CCEPs). Results: For each subject, connectivity measures produced similar spatial maps. Functional and pathological networks were highly segregated, exhibiting higher connectivity in all three measures within each network than outside the network. Considering that most of these networks are located in close proximity to each other, we applied a distance correction factor and found that within-network measures were still significantly higher than between-network measures. Higher path length and lower degree nodes were present in pathological networks when compared to other nodes. Conclusions: We found functional and pathological networks to be segregated from other cortical regions by means of metabolic and neuronal connectivity measures. Localization of these networks using non-invasive methods during the interictal period would provide a complementary measure to the current gold standards and possibly identify regions not implicated during clinical workup.