Abstracts

Rationale: Non-invasive approaches for the accurate characterization of different epilepsy subtypes are critically needed. While clinical distinctions can often be made, intracranial EEG remains the gold standard for characterization. We evaluated the use of interictal regional cerebral blood flow (CBF) measured with arterial spin labeling (ASL) MRI to differentiate mesial temporal lobe epilepsy (mTLE) from neocortical epilepsy (NE).

Methods: 12 mTLE subjects (8 left sided and 5 right sided), 7 NE subjects (5 left sided and 2 right sided), and 11 control subjects were included in this study. Resting state whole-brain ASL as well as T1 and T2-weighted structural images were acquired at 3T. CBF maps were calculated from ASL data using a custom MATLAB pipeline. Control-mTLE and control-NE voxelwise CBF group differences were assessed using non-parametric permutation-based randomization (1000 permutations and thresholded at p< 0.05). A subject averaged segmentation was used to define the temporal, parietal, occipital and frontal lobes, and the proportion of voxels found to have significantly different CBF values in the mTLE and NE groups, relative to the control group, were calculated in each region. A linear support vector classifier (SVC) was built to distinguish between mTLE and NE subjects using the mean CBF values in each of the previously mentioned regions in addition to a feature which consisted of the ratio between the mean temporal and mean parietal CBF for each subject. The model’s performance was assessed using an isolated dataset of 6 mTLE subjects, as well as with leave-one-out (LOO) cross-validation.