Abstracts

Rationale: Simultaneous scalp electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) have the combined strength of both high spatial and temporal resolution. There is growing evidence suggesting that regions where interictal spikes originate correlate with the epileptogenic areas. The aim of the present investigation was to use simultaneous EEG-fMRI to study both the electrophysiological and hemodynamic activities of an epileptic brain in a noninvasive fashion. Methods: 64-channel EEG signals were recorded in- and outside of the MR scanner. Outside scanner EEG served as the baseline reference. Source localization of reference EEG during interictal events was performed using CURRY6. Independent Component Analysis (ICA) decomposition was applied to all EEG segments. Components of EEG from inside of the scanner that highly correlated with those of the reference EEG were chosen. Temporal dynamics of the selected components were used as event markers and convolved with a hemodynamic response function (HRF) to construct fMRI regressor. fMRI analysis was conducted using both SPM8 and Brainvoyager QX. Spatial ICA was also done on the fMRI images to achieve quality assurance. Results: We have completed data acquisition for 6 patients with partial epilepsy, out of whom 4 had positive interictal events. EEG source localization for patient 1, 2 and 3 showed bilateral temporal spike origins. Patient 4 had left frontal slowing. Take the result of EEG informed fMRI on patient 1 for example, both dipole source localization and ICA analysis of the reference EEG showed separate left and right activation on anterior part of the temporal lobes during interictal spikes. Temporal dynamics of 7 components with high spatial correlation to reference were chosen to fit a general linear model. It showed bilateral temporal lobe activation with p value <0.0001. The activated areas fell into the same sub-lobe regions of MR images as those from EEG source localization alone and well agreed with the patient s medical history. Spatial ICA on fMRI data also revealed one component with bi-temporal activation. But fMRI analysis was not able to separate interictal spike activities on the left side from the right. This may be due to the relatively low temporal resolution of fMRI. Conclusions: Simultaneous EEG and fMRI were employed to study brain generators of interictal spikes. The findings of EEG-informed fMRI agreed well with both EEG source localization and ICA decomposition of fMRI alone. One difficulty we experienced was that despite our effort in selecting patients with frequent spikes, 2 out of 6 were spike-negative on the day of scanning. A pre-screening using EEG immediately prior to MR scanning might be necessary to achieve more fruitful recordings. Further investigation is needed to more thoroughly examine the neurovascular coupling in epileptic brains on a broader population.