Shape and the Information in Medical Images: A Decade of the Morphometric Synthesis

This keynote address reviews several techniques f rom morphometries (the multivariate biometrics of shape) developed mainly in the context of medical image analysis over the last decade. T h e new techniques provide powerful tools f o r geometric tasks that arise in the course of most analyses of medical images i n groups. These tasks include standardizing against Euclidean similarity transformations or shear transformations, encoding informative prior knowledge about shape variation, and detecting, testing, and visualizing linear statzstical patterns of variation within or between groups. I review the features of the present toolkit, the standard underlying data models entailed, and some of the extensions that reach out to the additional information content of medical images f o r common clznical or scientific applications.