Abstracts

Rationale: The arcuate fasciculus is a pathway mediating the temporal binding of inferior frontal and temporo-parietal cortices during fluent language production. Given that it is the predominant language pathway and that damage to it may result in aphasia, several complex methods have been used to identify it using diffusion tensor imaging (DTI). Various heuristic strategies have been proposed for this purpose, comprised of placing regions of interest over imprecisely defined targets in white matter, or based on fMRI activation maps. We explore here an novel semi-automated strategy for localizing the arcuate fasciculus and its homolog, based on thresholding the functional anisotropy (FA) values obtained by diffusion weighted sequences.Methods: Functional, diffusion-weighted and hi-res anatomical MR images were collected in 13 humans (8 normals, 5 patients w/refractory epilepsy) during multiple language tasks. In each individual, areas with significant BOLD activation were identified in the fMRI data, across all language tasks, using AFNI software. Language laterality was determined using supra-threshold voxels (p<0.0001) in BA44 and BA45, considered together as Broca’s area (L-R/L+R). All of the 5 patients underwent Wada testing. Structural, fMRI and diffusion weighed images (during 32 diffusion-sensitizing gradients) were spatially normalized to the ICBM452 atlas. DTIQuery software was used to compute tensors, and tensor eigenvalues used to compute fractional anisotropy (FA) maps. In voxels with FA>