Abstracts

Rationale: With Electric Source Imaging (ESI) the source of interictal epileptiform activity in EEG can be localized with high accuracy. There are two different aspects required, however, in order to achieve high accuracy ESI; 1) a sufficient number of recording electrodes, and 2) adequate spatial coverage of electrodes over the recording surface. Recently developed EEG systems now allow recording with up to 256 surface electrodes. The purpose of this study was to evaluate the localization accuracy of epileptic spikes recorded with 256 electrodes compared to sparser electrode arrays and compared to incomplete, or uneven, spatial electrode coverage on the head surface. Methods: In four patients with intracranially verified temporal lobe epilepsy, the 3D location of the ictal onset in the intracranial electrodes was determined. To look at the importance of the number of recording electrodes, interictal spikes recorded with 256 electrodes were averaged, and a regular downsampling of the averaged spike was performed to 128,64 and 32 electrodes. To look at the importance the spatial distribution of electrodes, another kind of electrode reduction was used where the electrodes covering the lower temporal areas were taken out. Then, a distributed source model (LAURA) was used to localize the source of the original 256 channel spike data sets and also of both types of downsampled spikes (decreased number of electrodes and more limited spatial coverage). The source maximum was compared to the intracranial ictal onset location for validation of results. Results: Both regular downsampling and reduction of temporal lobe coverage resulted in a clear deterioration of the ESI results. This difference was particularly clear when analyzing the early phase of the spike where the signal to noise ratio was poor. The degree to which the downsampling affected the results varied between the patients, but the most correct source localizations with respect to the intracranial ictal onset were always obtained with the 256 channel setup. Conclusions: Both a dense sampling on the head and a sufficient coverage of the lower temporal areas is necessary for correct source localization results. Improvement of the results can be obtained at least up to 256 electrodes