This paper examines recent work toward quantitative mapping of cross-sectional structural differences in gyral anatomy. Taking the approach of deformation based morphometry, it examines an alternative method of calibrating the spatial filtering used in the statistical analysis of maps of relative tissue volume. This differs fundamentally from many common studies based on voxel morphometry. Rather than applying spatial filtering to simply create a significant statistical effect, filtering can be applied in a way which optimizes the qualitative accuracy of the maps with respect to structures of interest, allowing inferences to be made about the effect of a degenerative disease on the amount of tissue on a given region of anatomy. Results using this approach are included from a recent cross sectional study of Semantic Dementia.