Computational Anatomy Going Beyond Brain Morphometry

In this chapter, we provide a comprehensive historical overview of algorithms for feature extraction of structural magnetic resonance imaging data and the corresponding analytical frameworks. We then focus on the use of T1-weighted images, which still represent the working horse of surface- and voxel-based morphometry to elaborate on the complex relationships between magnetic resonance imaging contrast and underlying brain tissue properties. This critical point is the motivation for embarking on novel structural imaging protocols based on biophysical models, which are sensitive to tissue myelin, iron, and water content. We expand on the concept of voxel-based quantification to demonstrate the added value to existing methods using T1-weighted data—both in terms of robust brain tissue classification and of straightforward neurobiological interpretation of the obtained results. We do not stop short of considering the unresolved issues in voxel-based quantification currently implemented in the data processing and analysis framework of Statistical Parametric Mapping (SPM12). We conclude with an outlook to the future developments and perspectives in computational anatomy, particularly the need for integration of the available magnetic resonance imaging contrasts at the level of statistical analysis without hampering the interpretability of our findings.