Abstracts

Rationale: Standard voxel-based diffusion tensor imaging (DTI) analyses enable identification of white matter abnormalities in limbic structures of mesial temporal lobe epilepsy (MTLE) patients. Probabilistic tractography (PT) allows further evaluation of white matter connectivity of brain structures involved in specific disease processes. Here we investigated PT differences derived from hippocampal head (HH) white matter connectivity ipsi- and contralaterally to the seizure focus in unilateral MTLE patients.Methods: Twenty subjects were studied: nine MTLE patients (38 17yo) and 11 healthy controls (39 18yo). All subjects signed an REB approved informed consent for participation. Images were acquired on a 3T Siemens Trio scanner, with a diffusion echo planar sequence including 99 directions (b=1000s/mm2; 10 b=0; 2mm isotropic voxels). Regions of interest (ROIs) comprising the right and left HH boundaries were manually drawn in T1 images for each subject using Display software. Right MTLE patients (n=5) were right-to-left flipped, so that left HH represented the epileptogenic side (EPI-HH) and the right HH represented the non-epileptogenic side (NONEPI-HH) in all patients. FA maps were generated using FMRIB's Software Library Diffusion Toolkit (FSL-FDT). HH ROIs were used to calculate HH-to-brain PT maps (HH-PT) using an in-house pipeline based on FSL-FDT. In brief: (1) DTI images were skull-striped and motion corrected; (2) at each voxel, a probability distribution of fiber direction was generated with a maximum of two fiber directions per voxel; (3) probabilistic maps were generated for every voxel of the ROI using FSL-probtrackx (~150 maps/ROI); (4) a regional probabilistic map was computed via a max function derived from the single voxel maps; (5) regional probability maps were then non-linearly resampled to the MNI 152 space; and (6) images were convolved using a 4 mm Gaussian kernel. Voxel-based t-statistics were obtained by comparing HH-PT from patients and controls, and differences were visualized at a t-value threshold of 5.0-10.0 over the icbm_152 template.Results: EPI-HH-PT group comparison revealed reduction of white matter connectivity between the HH and the rest of the hippocampal formation in MTLE patients. Group differences were maximal along the fimbria/fornix/stria terminalis bundles. This was not observed in NONEPI-HH-PT group comparison even at lower thresholds, indicating disrupted connectivity of the anterior segment of the epileptogenic hippocampus only.Conclusions: HHH-PT showed white matter connectivity declines along the main hippocampal limbic pathways ipsilateral to the seizure focus in this small group of MTLE patients. Fibers constructed from the HH revealed specific disruption extending anteroposteriorly in a unilateral pattern. This lateralization might be less evident in PT derived from the whole hippocampal structure or standard voxel-based analysis of fractional anisotropy data in similar patients. Acknowledgements: Study supported by the AES Early Career Physician-Scientist Award, Canadian Institutes of Heath Research MOP 93614 and Fonds de la Recherche en Sante Quebec.