Abstracts

Rationale: Functional MRI (fMRI) and magnetoencephalography (MEG) are non-invasive methods used to understand seizure generation. Typically done with analysis of interictal epileptiform discharges assumed to represent the seizure onset zone, fMRI or MEG done in isolation is subject to inherent temporal (fMRI) or spatial (MEG) limitations. Our goal was to identify and compare ictal fMRI and MEG findings in subjects with drug naïve childhood absence epilepsy (CAE). Areas of high spatial correspondence between hemodynamic and neuromagnetic abnormalities was hypothesized to constitute a critical network for absence seizures.Methods: Subjects recruited at Cincinnati Children’s Hospital underwent simultaneous EEG and MEG using a 275 channel CTF magnetometer with data analyzed using Curry 7.0 (Compumedics Neuroscan) and filtering at bandwidths of 3-20Hz, 20-70Hz, and 70-150Hz. Source localization corresponding at the ictal onset was done using a sLORETA algorithm. EEG-fMRI was collected on a 3T Phillips MRI. Pre-processing with SPM8 included realignment, slice-timing correction, co-registration, normalization, and smoothing. A previously described event-related independent component analysis (eICA) was done using a finite impulse response (FIR) basis set and a tensor ICA method (FSL MELODIC). Areas of fMRI activation/deactivation were co-registered with MEG using 12 affine transformation parameters. Euclidean distances between the peak current density result and peak positive and negative fMRI voxels were measured using x, y, z coordinates.Results: 7 patients were included (average age 7.9 years (SD 2.0 years), 2M/5F) with 22 absence seizures during MEG and 33 seizures during fMRI. Group fMRI analysis showed typical areas of activation in thalamus, infero-lateral frontal lobe, and superior parietal regions and deactivation in biparietal and precuneus regions (Figure 1). Average distances between fMRI activations (for all components) and MEG source localizations at 3-20Hz, 20-70Hz, and 70-150Hz were 45 ± 19mm, 51 ± 23mm, and 55 ± 28mm, respectively. Distances between fMRI deactivations (all components) and MEG source localizations were 67 ± 26mm, 86 ± 26mm, and 86 ± 26mm, respectively. In general, fMRI activations in thalamus and frontal cortex corresponded well with MEG (most distances <20mm for thalamus and <50mm for frontal cortex) and fMRI deactivations corresponded less well (most distances >50mm) (Figure 2).Conclusions: This is the first study comparing ictal EEG-fMRI and MEG findings in the same subjects. We have demonstrated good spatial correspondence between fMRI activation and MEG within the thalamus and infero-lateral frontal lobe, but fMRI deactivations corresponded less well. Future studies will focus on extracting and comparing MEG waveforms from areas with high fMRI spatial correspondence to explore coherence and causality and further understand the spatiotemporal dynamics of generalized seizures.