Abstracts

Rationale: For children with medication resistant seizures, determining the cortical location of ictal onset can be difficult. Usually a non-invasive first phase is required to determine hemisphere, then lobar location. Scalp EEG with 21 electrodes has a sufficiently long time window for seizure capture, but poor spatial resolution can limit early detection of ictal onset occurring between scalp electrode locations and limits accuracy of quantitative methods to localize ictal onset. Denser scalp electrode arrays have become available. In this study we evaluated the ability of dense array EEG (dEEG) to detect and localize ictal onset, and compared to ictal onset localization by intracranial EEG (iEEG) and surgical outcome.Methods: Pediatric patients with medication resistant epilepsy were recorded overnight during sleep with video and 128 to 256 channel scalp EEG electrodes. Clinical seizure onset timing was determined by video review. Location of the electrodes was determined from digital pictures from an array of 11 synchronized, position-calibrated digital cameras with fiducial locations at nasion and left and right preauricular. Three dimensional head models were constructed from MRI scans, with the same 3 fiducial locations. Each record was reviewed to determine whether ictal EEG activity could be detected prior to clinical onset. Examination of broad bandwidth (eg, 1-100 Hz) spectral power in time frequency analysis (TFA) plots was followed by examination of narrower bandwidth (eg, 10-20, 15-60 Hz) spectral windows to enhance detection of low amplitude, spatially focal scalp signals. The time of earliest spectral change was used for source localization using sLORETA. Source localizations were compared to ictal onset detection by intracranial subdural electrodes and to surgical outcome. Congruence of scalp dEEG was compared by hemisphere and lobe.Results: 14 patients (4-18 yrs old) were recorded overnight for approximately 16 hrs; 39 clinical seizures were captured. One patient did not have a seizure. One patient’s electrode file was unavailable for processing. Of the remaining 12 patients, 2 patients’ seizures stopped after a medication change, have not yet recurred, and the patients have not had surgery. Seven of the 10 had iEEG for comparison. dEEG matched hemisphere 6/7; lobar 6/7. Scalp 10/20 matched hemisphere 5/7, lobar 3/7. Of the 10 patients, 1 patient had hemispherotomy and 1 callosotomy, leaving 8 patients with focal resections. For surgical resection site, dEEG matched 7/8 lobar, 7/7 matched are seizure free at 3 months to 2 years (ave 13 months); the 1 dEEG non-match continues with seizures post resection. Of the 14 patients, 11 were extratemporal ictal onsets. The 3 temporal lobe patients were 1 lateral, 1 mesial, 1 indeterminate.Conclusions: Source localization of seizure onsets in pediatric presurgical patients is possible with dense array EEG (dEEG) with reasonable results when quiet recordings are obtained and earliest onset time is sought with spectral analysis.