Abstracts

Voxel based morphometry (VBM) is a statistical automated approach to evaluate brain pathology in group comparisons. In epilepsy, it has detected various brain pathologies exceeding the visual detected pathology. The aim was to evaluate a high contrast MRI sequence (hcontMRI) with an increased signal to noise ratio compared to conventional MRI (convMRI) in VBM. 6 patients with histologically proofed Ammon[rsquo]s Horn Sclerosis (AHS) and temporal lobe epilepsy (TLE) and 12 age matched normal controls. HcontMRI sequence included a four-fold average. All patients and controls were scanned with both sequences. VBM was performed with an individual created template separately for both MRI sequences. We compared different smoothing kernels as well as different significance levels, with respect to voxel threshold and cluster size. Conditions for significance and specificity to detect AHS were defined. HcontMRI VBM showed with smoothing kernels of 3 to 5 mm results, which met both criteria sensitivity and specificity. All significant voxels were within the ipsilateral hippocampus. ConvMRI VBM showed significant voxels in various smoothing steps, mostly between 6 and 9 mm smoothing but never met both criteria sensitivity and specificity. HcontMRI enhances the statistical significance of VBM and increases spatial resolution as smoothing of 3 to 5 mm showed best significant results.