Abstracts

Rationale: Ictal patterns on scalp EEG may be challenging and visible only after propagation of seizures as compared to simultaneously recorded intracranial EEG (icEEG) or blood oxygen level dependent (BOLD) changes on fMRI. We hypothesized that mapping the BOLD changes using video-EEG-fMRI will provide valuable information to improve the localization of seizure onset zone (SOZ) by revealing distinct preictal and ictal BOLD changes.Methods: 25 patients with daily focal seizures were studied who had previously undergone video EEG monitoring. Two 20min sessions of BOLD-sensitive images were acquired on a 3Tesla scanner. Simultaneous 64-channel EEG synchronised with video was recorded (Chaudhary et al. 2010). EEG was pre-processed using Brain Analyzer2 (Brain Products). Images were pre-processed and analysed using SPM5 by comparing two general linear models (GLM): GLM1: Interictal epileptiform discharges (IED), seizures and non-seizure related physiological movements were modelled as brief events and blocks. Seizure-related blocks were defined on the basis of evolution of ictal discharges and clinical semiology on video-EEG. All the events and blocks were convolved with the canonical hemodynamic response function and derivatives. Motion and cardiac effects were modelled. GLM2: In addition to all effects included in GLM1, we modelled a preictal event of 30s duration using a Fourier basis set. SPM [F] maps (p<0.05 corrected for family wise error and p<0.001 uncorrected) were generated and BOLD clusters were overlaid on co-registered anatomical scans. Time courses of the condition specific BOLD clusters were plotted. The spatial concordance of the BOLD clusters with SOZ was evaluated at the lobar level.Results: 10 patients had seizures (# range:1-15) during video-EEG-fMRI acquisition and significant BOLD changes were revealed in all cases. Preictal changes were seen from -5s to -30s before the seizure onset on EEG, which were BOLD increases in 4/10 patients and BOLD decreases in 6/10 patients. Ictal changes were BOLD increases in 6/10 patients and BOLD decreases in 3/10 patients. Ictal BOLD changes were not interpretable in one patient because of motion related artefact. Four patients underwent icEEG evaluation for the localization of SOZ (table 1). Preictal global maximum (GM) BOLD clusters were concordant with the SOZ (defined on icEEG) in 3 patients (#1,2,10). Early ictal (EI)/ictal GM BOLD clusters were concordant with SOZ (defined on icEEG) in 2 patients (#2,10). SOZ was defined on scalp EEG in 6/10 patients (table 1), out of which 3 patients (#3,4,5) had preictal, EI and generalized ictal GM BOLD clusters concordant with the SOZ whereas, 2 patient (#7,8) had a BOLD clusters other than GM concordant with the SOZ. The size of the preictal BOLD clusters was found to be smaller than ictal-related clusters.Conclusions: We have revealed preictal and ictal hemodynamic changes using video-EEG-fMRI in cases with very frequent seizures, providing localising information on the SOZ which can be used to guide the implantation of intracranial electrodes in future.