Automated Volumetrics and Methodological Pitfalls in Special Patient Populations

The last decade has seen great advances in the analysis of structural MR images. With the introduction of automated methods of tissue classification and spatial normalization, the user-independent processing of large datasets has become feasible. Instead of the error-prone and extremely time-consuming manual delineation of pre-chosen structures of interest, an automated and unbiased exploration of the whole brain dataset is now possible. Recent methodological advances include a more “brain-based“ normalizaton and segmentation scheme or have allowed us to investigate true tissue volume instead of the more abstract concept of tissue density. Also, while numerous studies have proceeded to do voxel-wise analyses (as in the case of voxel-based morphometry, VBM), regional expansion schemes can also be included. Here, we aim at giving an overview over a typical „optimized“ processing stream in a widely-used software environment (spm) and highlight some important pitfalls when working with special populations, such as children or elderly subjects. Such population-specific peculiarities can introduce a substantial bias into the ensuing analyses and thus need to be considered.