Abstracts

The improvement in EEG source imaging techniques raises the question regarding the adequate number of recording scalp EEG electrodes to accurately localize underlying cerebral sources with clinically relevant spatial resolution. Boundary Element Modeling (BEM) has been useful in localizing interictal and ictal discharges in Temporal Lobe Epilepsy (TLE). However, direct comparison of localization accuracy between low and high-density EEG modeling has not been determined. In addition, effects of electrode position determination have not been systematically investigated.
We performed BEM of interictal EEG spikes on five patients with intractable TLE udergoing epilepsy monitoring before epilepsy surgery. Scalp EEG was recorded from 64-electrodes placed according to the International 10/10 system. The electrodes were digitized to determine their position on the scalp using Polhemus Fastrack digitizer. We segmented EEG files of interest (BESA 2000) and performed single moving dipole modeling of early and later spike activity to evaluate spike origin and propagation pattern using the BEM (CURRY 4.5). We analyzed a total of ten interictal discharges in each patient and coregistered modeling dipoles to brain MRI. Dipole localization and orientation were compared in the following four arms of analyses per each interictal spike; high density EEG (64-channels) with digitization, high density EEG without digitization, low density EEG (27- channels according to 10/20 International system in addition to three subtemporal electroedes bilaterally) with digitization and low density EEG without digitization.
Realistic head modeling of both low and high density scalp EEG electrode recording with and without digitization demonstrated close dipole localization, orientation and evolution throughout the spike activity in the various temporal EEG spike patterns. Dipole localization in the four analyses arms was confined to the temporal lobes. Low density recording however, was vulnerable to localization errors or dipole localization instability in case of loss of individual temporal or subtemporal recording electrodes. Two of five patients underwent temporal lobectomy and became seizure free after surgery.
BEM of low density scalp EEG recording produces reliable single dipole modeling when it includes the regular 10/20 International system temporal electrodes in addition to subtemporal electrodes. This may be partially related to the resistance of the single dipole model to dipole location and orientation variability because it represents the center of the active cerebral source. Similar studies may still be needed to determine the scalp electrode number required to accurately delineate the active cerebral sources in other EEG source models in TLE.