Abstracts

Rationale: Determining the most clinically valuable protocols for EEG source localization (ESL) is a critical step in the development of highly effective surgical treatment for intractable focal epilepsy. Clinical validation of ESL techniques could increase availability of non-invasive epilepsy localization data and increase the number of patients who should appropriately receive surgical treatment. This study evaluated the accuracy of ESL using 28-35 channel scalp EEG (nonuniform mixed 10-10 and 10-20 placement), a strategy that allows widespread implementation of ESL even in long-term recordings. Methods: Fourteen adult patients with drug resistant neocortical epilepsy were recruited while admitted to the hospital for long-term video EEG recordings. An average of 8 extra electrodes (10-10 system) were placed over the scalp region with suspected maximal change in voltage gradient, which was determined either by prior spike identification or lesion localization. Source analysis was performed on both interictal and early ictal spikes (when available), using spike averaging (when appropriate). MRI brain segmentation was used to create boundary element forward models (BEM) with an equivalent current dipole inverse model in the environment of Curry 7 (Compumedics?(R), Charlotte, NC) for image analysis and source computation. To provide a measure of diagnostic accuracy, localization was compared to either degree of overlap between intracranial EEG recordings (ICEEG) or surgical outcome when ICEEG was not available. For patients in whom ICEEG is pending, concordance was measured by agreement with diagnostically conclusive ictal SPECT or MEG results. Concordance between ICEEG interictal spikes (or imaging tests) and ESL was defined as localization to the same sublobar region(s) defined in the NINDS Common Data Elements, Imaging Diagnostics with added basal temporal and basal occipital regions [1]. For patients with only surgical outcome, concordance was present if post-resection Engel class outcome I, II occurred with complete resection (?-70% of dipole-indicated source regions were located within the resection) of ESL or if outcome III, IV occurred with incomplete resection of ESL. [1] https://commondataelements.ninds.nih.gov/epilepsy.aspx#tab=Data_Standards Results: Dipoles were calculated from averaged interictal spikes for 9 patients and from averaged ictal onset spikes for 2 patients (see Table 1). No spikes were captured in three patients. Of the 11 patients with ESL spike results, 10 were either concordant (n=8) or partially concordant (n=2). One case without ICEEG or surgical outcome was discordant with MEG. Conclusions: ESL is a promising clinical tool in the diagnosis and treatment of intractable focal epilepsy. Addition of 5-10 extra electrodes in the field of interest appears to allow accurate ESL. Since nearly all clinical EEG systems have at least a single 32 channel amplifier, further validation of this strategy will permit widespread application of ESL, offering a major advance over traditional EEG interpretation in the seizure monitoring unit. Funding: AAN Medical Student Summer Research Scholarship