Abstracts

Rationale: Given the complexity in the identification of the epileptogenic tubers in patients with tuberous sclerosis complex (TSC), we continue our efforts in seeking additional non-invasive tools to improve the localization accuracy of surgical targets. Methods: Retrospective data from 9 children (ages 0.92-10.42 years old, mean 3.71 years old; 4M:5F) with intractable epilepsy from TSC were analyzed. All children underwent epileptogenic tuber resection in 2003-2010 with seizure freedom in all but 2 who became seizure free after a second operation (follow up 0.25-5.67 years). Their surgical evaluation included scalp EEGs (ictal and interictal), MRI, FDG-PET, and MSI. We used a dipole fitting model (BESA) and a distributed source model (LORETA) to study the pre-surgical digital EEGs (21 channels 10-20 T1-T2). Interictal spikes were automatically searched and classified according to morphology and topography, which were then visually screened, artifact excluded, and averaged, without prior knowledge of surgical evaluation localization results. The resultant localization maps were then compared to the eventual ECoG-guided surgical site. Results: Between 9 children, 11 interictal EEG studies were analyzed. The length of useful EEG segments for analysis ranged between 52 minutes and 8h 14 min. The total number of spikes captured for each patient ranges from 34 to 3273 with a mean of 1567. The number of spike clusters for each study ranges from 2 to 8. The average number of spikes in each cluster for each patient ranges from 17 to 1383. With BESA, the weighed proportion of clusters in the resection site ranges from 10% to 100% (overall 61.0%). With LORETA, the weighed proportion of clusters in the resection site ranges from 11% to 100 (overall 57.7%). For clinical relevance, we define 4 categories: A) All clusters overlap with the resection site. B) >=50% of the clusters overlap with the resection site. C) <50% of the clusters overlap with the resection site. D) No clusters overlap with the resection site. With BESA, the number of studies in category A/B/C/D is 3/7/1/0, while that for LORETA is 3/5/3/0. Conclusions: In the majority of our patients, either BESA or LORETA can accurately identify the more active lobe. Compared with the interictal or ictal EEG reports, the computer-assisted cluster analysis offers an objective and quantitative measure of the relative importance of each cluster using number of spikes. Moreover, BESA can be used to breakdown spikes into sequential time-dependent components. In one patient with modified hypsarrhythmia, BESA correctly identified the lobe of onset despite the more obvious expression on the opposite side with no clear asymmetry by visual interpretation. Overall, these tools are not yet validated to be clinically practical. However, given that they are non-invasive with relatively low-cost and no risk to the patients, they may be refined to become adjunctive tools in pre-surgical evaluation of children with epileptogenic tubers.