Voxel-based morphometry in the detection of dysplasia and neoplasia in childhood epilepsy: combined grey/white matter analysis augments detection.

PURPOSE: Analysis of grey matter on MRI utilising voxel-based morphometry (VBM) may have insufficient sensitivity for routine clinical application. The aim of this exploratory study was to evaluate combined analysis of grey and white matter using VBM for detecting focal lesions underlying childhood epilepsy, and to establish the optimal statistical parameters in this context. METHODS: The patients were 16 children (10 boys) aged 5.9-15.2 years (11.3+/-2.8 years; mean+/-S.D.) with epilepsy and focal cortical dysplasia (FCD) or neoplasia. The control group comprised 24 normal children (12 boys), age matched to the patients. VBM was used to spatially normalise MRI volumes to a custom template and segment them into grey matter (GM) and white matter (WM). The combined GM/WM segments from each patient were contrasted with the control group. Three different VBM post-processing techniques of combined GM/WM were evaluated along with GM-only analysis. Maps showing increased/decreased GM or GM/WM concentration were generated and compared with visually identified lesions. Rates of detection and true/false positives voxels were calculated. RESULTS: The GM-only lesion detection rate was equal for FCD and neoplasia at 5/8, whereas the best combined GM/WM technique detected 8/8 FCD and 6/8 neoplasia. The combined technique also produced a higher overall rate of true positives (87%) than GM-only (44%) with a similar low rate of false positives. CONCLUSIONS: These preliminary data suggest that VBM is ineffective for precise delineation of lesion margins, but could potentially be used to detect subtle dysplasia in MRI negative and equivocal cases.