Abstracts

Rationale: Most EEG-fMRI studies in epileptic patients are analyzed using the general linear model (GLM), which assumes a known hemodynamic response function (HRF) to epileptic spikes. In contrast, independent component analysis (ICA) can extract Blood-Oxygenation Level Dependent (BOLD) responses without imposing constraints on the HRF. ICA might therefore detect reponses that vary in time and shape, and that are not detected in the GLM analysis (LeVan et al. Hum Brain Mapp 2008 in press). In this study we compared the findings of ICA and GLM analyses in patients with idiopathic generalized epilepsy (IGE). Methods: 12 IGE patients with generalized spike and wave discharges (GSW) during an EEG-fMRI study at 3T were included in the study. GLM analysis: GSW identified on the EEG were convolved with a regressor with four hemodynamic response functions with peaks at 3, 5, 7 and 9s. Significant responses were defined with a |t|>3.1 and a spatial extent of 4 voxels (p=0.05, corrected). ICA analysis: Spatial ICA was used to extract independent components (IC) from the fMRI data. A deconvolution method identified component time courses significantly related to the GSW, without constraining the shape of the HRF. In a next step HRFs were calculated for all ICs having GSW related time courses. Only ICs with HRFs showing a peak to baseline ratio over 3 were retained The results from the ICA analysis were compared to those from the GLM analysis. Results: The mean duration of GSW per minute (defined as mean duration of GSW x number of GSW per minute) varied from 0.3s to 3.6s per minute across the patients. GLM and ICA analyses confirmed the known BOLD responses in this patient group. The GLM analysis revealed thalamic activation in 11 patients (93%), deactivation in the caudate nucleus in 7 (58%) and deactivation in default mode areas in all 12. The ICA analysis showed thalamic activation in 7 patients (58%) whereas deactivations in the caudate nucleus were detected in 5 (42%) and in default mode areas in all 12. In the 4 patients with the fewest GSW (<0.5s/min) thalamic activation was detected in 3 when using the GLM, but in none when applying the ICA. If present in both analyses, the distribution of BOLD signal changes in the thalamus, the caudate nucleus and the default mode areas was consistent for ICA and GLM in each patient. The HRFs calculated for GSW related ICs had shapes similar to the canonical HRF.