Abstracts

Rationale: The accurate delineation of the seizure onset zone (SOZ) is one major aim of the presurgical evaluation in epilepsy patients. Since long it is suspected that interictal epileptic bursts in beta and gamma frequency bands (IEBs) are closely related to the SOZ. In addition it has been suggested in basic neuroscience that the bandwidth of cerebral activity may be closer related to cortical function than the exact frequency band [Crone et al., Int J Psychophysiology 2011]. Therefore it was the aim of our study to map the distribution of bandwidth changes during beta and gamma bursts in relation to the SOZ. Methods: From 22 patients with at least one implanted 64 contacts subdural grid electrode we excluded 8/22 (no IEBs), and 1/22 (no habitual seizures). 13/22 patients with a total number of 1152 invasive EEG (iEEG) contacts (mean/patient: 89 ± 17 contacts) were included. 11/13 patients underwent subsequent epilepsy surgery. We visually identified IEBs and baselines from 1-6 hours iEEG. We performed baseline-correction, spectral power analysis of IEBs and included changes at a q-value < 0.0001 (Ranksum test, false discovery rate (FDR) corrected). The spectral peak with the highest power was detected for each iEEG contact and the frequency corrected bandwidth at full-width-half-maximum (FWHM) of these peaks was calculated. FWHM of the three contacts with the narrowest power peaks determined an individual frequency band of interest (IFB) per patient. For the IFB we finally created normalized (norm.) and power-weighted narrow-bandedness (PWN) indexes (norm. 1/bandwidth x norm. power) thresholded at 30% and compared these indexes to SOZ contacts. Results: IEEG contacts with valid PWNs identified SOZ contacts (positive predictive value (PPV): 0.69 ± 0.34, specificity: 0.98 ± 0.02. The peak frequency of the primary spectral power peaks varied among patients (62 Hz ± 27 Hz). In contrast, broadband changes occurred widespread. Histological analysis revealed focal cortical dysplasia in 10/11 patients (FCD type 2a/b 5/10, FCD type 1a/b in 5/10), 1/11 patient: histological assignment not possible. PPV and specificity showed the tendency to be higher for FCD type 2 (0.83 ± 0.22, 0.99 ± 0.02) compared to FCD type 1 (0.41 ± 0.42, 0.82 ± 0.02). For some contacts we observed secondary peaks in addition to high power narrow band primary peaks. Their exact origin should be further explored. Conclusions: Our findings indicate that narrowband high power spectral peaks during IEBs correlate closely with the SOZ regardless their exact peak frequency. In addition our results confirm that IEBs are suggestive for FCDs. Activity that is topographically widespread and broad in its frequency content may be an effect of averaging over multiple, narrow-banded neural populations. Newly developed high-density subdural electrodes that combine different contact sizes may allow further investigating this context. In addition our findings could be helpful to improve closed-loop stimulation systems. Funding: BrainLinks-BrainTools Cluster of Excellence funded by the German Research Foundation (DFG - EXC 1086).