Diffusion tensor imaging analysis with tract-based spatial statistics of the white matter abnormalities after epilepsy surgery

Abstract

Introduction: Diffusion tensor imaging (DTI) techniques demonstrated diffuse bilateral temporal and extra-temporal abnormalities of white matter in patients presenting mesial temporal lobe epilepsy with hippocampal sclerosis (HS). The aim of this study was to assess these diffusion changes following temporal lobe surgery, by applying a novel voxel-based tract-based spatial statistics (TBSS) technique for whole-brain analysis of fractional anisotropy (FA) and mean diffusivity (MD). Second, region-of-interest analysis (ROI) was performed to improve statistical power.

Material and methods: The study included 22 patients with unilateral HS. Twelve patients underwent temporal lobe surgery. Follow up MRI was done in a mean interval of 4 months. Voxelwise pre-operative FA asymmetry in all 22 patients was assessed within subjects between lesional and contralateral hemispheres. The whole-brain post-operative dataset of 10 seizure-free patients was compared with the corresponding pre-operative dataset using voxel-wise statistical analysis. Additionally, regional analysis at the fornices was done with skeleton-based region of interest (SROI).

Results: Within a mean interval time of 6.3 months after surgery, 10 of 12 patients were seizure free (83.3%). The voxelwise comparison between lesional and contralateral hemispheres was consistent with previous studies showing a more widespread diffusion alteration in the lesional hemisphere. Voxel-wise comparison between post and pre-operative dataset did not show supra-thresholded voxels. SROI statistical analysis showed significant decrease in FA and increase in MD in the ipsilateral fornix. Significant increase in FA was observed in the contralateral fornix after surgery.

Conclusion: The ipsi-lesional fornix showed decreased FA and increased MD after surgery, consistent with Wallerian degeneration. In contrast, contra-lesional fornix demonstrated increase in FA. This observation is important for our understanding of the fate of the remaining brain tissue following removal of an epileptic focus. Postoperative increase in FA may reflect structural reorganization in response to epilepsy surgery. The discrepancy between SROI and voxelwise statistics emphasizes the difference of statistical sensitivity between voxelwise and ROI analyses.