Comparison of standard and optimized voxel-based morphometry for analysis of brain changes associated with temporal lobe epilepsy

Abstract

We compared statistical parametric maps (SPMs) of group-wise regional gray matter differences between temporal lobe epilepsy (TLE) patients with unilateral hippocampal atrophy (HA) determined by manual volumetric analysis relative to a healthy control population using standard and optimized voxel-based morphometry (VBM). We also investigated the impact of customized neuroanatomical templates on SPMs. Standard and optimized VBM analyses of gray matter concentration (GMC) and gray matter volume (GMV) correctly identified HA, regardless of the template used for normalization. The distribution of hippocampal and extrahippocampal abnormalities differed according to the technique (standard v optimized; GMC v GMV), but was not dependent on template type (default v customized) within each technique. In particular, hippocampal GMC reduction was confined to subregions of hippocampus, whereas GMV reduction was observed in the hippocampal head, body, and tail. Unlike standard and optimized GMC reduction, symmetrical GMV reduction was observed in bilateral thalamus, lenticular nuclei, cerebellum, and ipsilateral entorhinal cortex, perirhinal cortex, and fusiform gyrus in both left and right HA patients. These results show that group-wise SPMs of GMC (i.e., regional distribution of gray matter) and GMV (i.e., volume per se) reduction can identify focal atrophy that has been quantified with manual region of interest techniques, although effects are attenuated in analyses of GMC. Unlike SPMs of GMC, analyses of GMV revealed similar extrahippocampal abnormalities as previous region-of-interest volumetric and histopathological studies of intractable TLE. We suggest that in studies of neurological disorders, optimized VBM analyses of GMV may reveal subtle neuroanatomical changes that are not identified in analyses of GMC.