Abstracts

Rationale: Simultaneous scalp EEG-functional MRI (EEG-fMRI) in epilepsy is a well-established technique for the imaging of the Blood Oxygen Level Dependent (BOLD) response associated with interictal epileptiform discharges (IED). Essential to this analysis is the application of the canonical hemodynamic response function (HRF), which quantifies the BOLD signal changes in response to neural events. The HRF was first derived from an auditory response task performed by healthy subjects and is presently used widely for fMRI analysis [Glover, Neuroimage,1999. 9(4): p. 416-29]. Current literature suggests that the HRF following epileptic spikes typically matches the canonical HRF shape with a peak response at 4-6 seconds after the spike, though some variations have been published. Despite the success of scalp EEG-fMRI experiments, it has limited ability to detect discharges in small cortical areas (< 10 cm2) or from deep structures such as mesial temporal lobe. By combining intracranial EEG with fMRI (iEEG-fMRI) at 3 T, we sought to quantify the BOLD signal changes and HRF associated with epileptiform discharges recorded directly from the epileptogenic source.Methods: Four subjects with temporal lobe epilepsy undergoing intracranial video-EEG monitoring for seizure focus localization were recruited and underwent iEEG-fMRI. Subjects were connected to a MR compatible EEG system (Neuroscan, Charlotte, NC) and simultaneous fMRI was performed at 3 T using a GE Signa LX scanner (GE, Waukesha, WI). Epileptiform discharges were identified and marked by two experienced electroencephalographers (PF, YA). fMRI analysis was restricted to IEDs that were relatively isolated, defined as follows: IEDs with a minimum spike-free period of 3 sec prior and 10 sec following the discharge. fMRI data were processed using the FEAT (FMRI Expert Analysis Tool) component of FSL (FMRIB Software Library) to correct for subject motion and interleaved slice acquisition, and were spatial smoothed to 6mm. Regions of interest (ROI) encompassing 50 voxels nearest to the active electrode were created for each subject. BOLD time-series were generated for the isolated spikes and the data were averaged across runs and subjects to generate hemodynamic response functions characteristic of the interictal spiking activity for each subject. Results: Nine to 23 isolated IEDs were identified for each subject, generating the individual average hemodynamic responses seen in Figure 1. Unlike the canonical HRF used for most fMRI analysis, the BOLD data recorded from the epileptogenic tissue show a later time-to-peak ranging from approximately 7.5-10 seconds (Figure 2) compared to 4-6s for the canonical HRF. Conclusions: Based on these preliminary data, the hemodynamic response associated with IEDs recorded via iEEG-fMRI, has a longer time-to-peak than the canonical HRF. These results point to a new method of iEEG-fMRI data analysis, which may provide a better understanding of hemodynamic response after epileptic spikes.