Abstracts

Rationale: Voxel based measures of intrinsic connectivity contrast (ICC), based on network measures reflect how well each tissue element is connected to all the other tissues in the brain or within a hemisphere. We hypothesized that a novel ICC approach contrasting ipsi-contralateral connectivity could isolate epileptogenic tissue, due to altered connectivity patterns, in epilepsy patients when they are compared to normal healthy control subjects. This study compares the tissue regions localized with ICC, to regions isolated using simultaneous eeg-fMRI spike localization methods. Such an ICC approach could potentially provide a relatively easy approach for isolating the tissue nodes that lead to seizures in epilepsy patients. Methods: Simultaneous EEG and fMRI recordings were performed on nine intractable epilepsy patients (age 7-50 yo, mean 29.8 yo, 5 male). A gradient-echo fMRI sequence was used for the fMRI (3T scanner , TR=1550 ms, TE=60 ms, FA=80, 25 slices, FoV=240 mm, base resolution=128). Interictal spikes were read from the preprocessed EEG data, and the timing of spikes (events in an event related analysis) were used to model the fMRI data using the generalized linear model. Seventy five healthy control subjects ages between 18 to 65 were scanned in the resting state using 6 runs of BOLD data, also on a 3T scanner. The ICC measure used a voxel based calculation of the network property degree for each voxel in the brain within the same (ipsilateral) hemisphere and subtracted from the contralateral measure of degree on a voxel by voxel basis. This ICC difference map was then compared with the spike correlated fMRI map and also with the clinical consensus that was drawn from examining MRI, difference SPECT, PET, EEG, and EEG data that are available for each patient. Results: In four out of nine patients, the intrinsic connectivity maps had the largest clusters in the same regions as the spike correlated fMRI maps, which also agreed with the clinical impressions. In three other subjects the intrinsic connectivity maps had prominent clusters that coincided with the spike correlated fMRI maps, and which also were in agreement with clinical impressions. The remaining 2 subjects either had poor spike localization or no agreement with the ICC difference map. Conclusions: Overall, the intrinsic connectivity maps agreed with both the spike correlated fMRI maps and with the clinical impressions. The results suggest that this ICC measure may be a promising tool for epilepsy surgical planning, especially considering that it is non-invasive and possibly can be used for evaluating patients with infrequent or low-amplitude interictal spikes. Further studies with more patients are needed to confirm these promising results.