MRT-Based Morphometry: Confounding Parameters and Sensitivity of Methods

In neurobiological applications, morphometry aims at a description of brain structure from imaging modalities in terms of size, shape, and texture. Such measures may be evaluated by automated statistical techniques and analyzed together with other clinical and experimental parameters. Thereby, morphometry offers promising approaches for an in vivo characterization of many neurological or psychiatric pathologies. A survey of the recent publications only allusively reflects this attractiveness of MR-based morphometry: published findings are heterogeneous, partly contradictory and not always plausible in terms of known neuropathological correlates. Hence, it is necessary to question the sensitivity of the applied methods. Variations in morphometric measurements can reflect differences in image acquisition and analysis techniques, alterations in neuronal and non-neuronal tissue compartments, physiological alterations in brain tissue, and changes in other chemical constituents that make up the living brain. Furthermore, a sound knowledge about the morphological variability of the normal brain is vital for the assessment of volumetric findings. With this contribution we review the sources of variation that can rapidly produce changes in MR-based morphometric measurements, some of greater magnitude than those reported for a specific neuropathology. In particular, we focus on the error in morphometric measurements determined by the MR scanner, as well as on physiological confounds that create the necessity for a better subject control.