Abstracts

Rationale: Localization of the source of interictal epileptiform activity from EEG recordings (ESI = Electric Source Imaging) is an important tool in presurgical epilepsy evaluations. There is increasing awareness of the importance of a sufficient number of recording electrodes for correct source localization results, and previous studies on epilepsy data have showed that increasing the number of electrodes to at least 64 would lead to improvement of the source localization results. New estimations of brain and scull conductivities, however, indicate that much higher electrode counts might be necessary for maximal source localization accuracy. The purpose of this study was to systematically evaluate the importance of number of electrodes on source localization precision, while implementing recent scull/brain conductivity estimations. Methods: Eight patients with partial complex epileptic seizures were subject to an EEG recording with 256 electrodes. Between 19 and 144 interictal spikes were identified, and each single spike was downsampled to 128, 64 and 32 electrodes. Then for each spike a source reconstruction was performed (source model LAURA, head model Finite Difference Model, scull/brain conductivity estimation 1/15). For each patient the 256 channel spikes were averaged. The source maximum of these averaged spikes, which in all cases corresponded well to the clinically estimated focus location, was considered to be the most correct focus location (CFL). Then the source maximum of each single spike for each electrode setup was calculated and the euclidean distance from each spike maximum to CFL was determined. Results: The distances from the source maximum of the individual spikes to CFL with 256 electrodes were statistically compared to the distances with the sparser electrode counts. With 32 channels the distances were significantly longer in 7/8 patients, with 64 channels in 6/8 patients, and with 128 channels in 3/8 patients. In no case was a significantly shorter distance found with a sparser electrode count. Conclusions: These results confirm that increasing the number of recording electrodes does indeed increase the ESI localization accuracy. To our knowledge no previous clinical studies have demonstrated significant source localization improvement with more than 64 electrodes. The assumption of a scull/brain conductivity relation of 1/15 instead of 1/80 is based on recent studies in this field, and this more correct assumption may explain why a source localization improvement is seen also between 128 and 256 channels in some of the patients.