Abstracts

Rationale: Language function is frequently reorganized in epilepsy patients, particularly in left frontal lobe epilepsy (FLE). Not only is the right frontal lobe increasingly recruited, there is also a significant reorganization within the left hemisphere. The degree of functional MRI (fMRI) activation, or measures such as the laterality index are often not correlated with the neuropsychologically measured language capabilities of patients. We hypothesize, that the proximity of functional cortex to the underlying language pathways is crucial for the language performance in left frontal lobe epilepsy. We investigate this with a combined fMRI and diffusion tensor imaging (DTI) pilot study in 16 patients with FLE. Methods: Patients were examined on a 3T MRI scanner. They performed a verbal fluency fMRI paradigm, comprising 5 blocks of 30 seconds covert word generation, alternating with rest. fMRI was recorded using a whole brain acquisition of 50 axial slices, 2.5 mm slice thickness with a TR of 2.5 sec and was analyzed with SPM 5 software. DTI was acquired with a total of 52 diffusion directions, a slice thickness of 2.5 mm and 2.5 mm in plane resolution. DTI data was processed using FSL 4.1.9 software. DTI data was normalized to a template in standard space and the inverse normalization was used to project seed and mask regions for probabilistic tracking of the arcuate fasciculus in each subjects individual space. The resulting tracts were also normalized to MNI space and superimposed with the coregistered fMRI statistical parametric maps. Both images were smoothed and thresholded, and the overlapping volume within the left hemisphere was determined. Results: One patient failed to activate during the fMRI paradigm, two showed a right language dominance and were also excluded. For the remaining patients, the peak fMRI z-score ranged from 3.2 to 7.4. The volume of overlap between arcuate fasciculus and fMRI activation ranged from 0.8 to 31.4 ml, not normally distributed. The Verbal fluency score from neuropsycholgical assessment ranged from 4 to 22 words per minute. This was not correlation with peak fMRI activation (Spearman correlation 0.22, but it was correlated with the overlapping volume of fMRI and DTI (Spearmann correlation 0.81). Conclusions: We present a processing pipeline for combined analysis of functional activation in relation to underlying structural pathways. In patients with left FLE, this measure correlated with their neuropsychological verbal fluency scores, whereas the fMRI activation alone showed no relation to cognitive performance. This provides evidence, that the efficiency of cortical reorganization of language depends on the proximity to major structural pathways.