The Random Atlas Model in Computational Anatomy and High Throughput Neuroinformatics
July 12, 2012
Michael Miller, Herschel and Ruth Seder Professor of Biomedical Engineering, Professor of Electrical and Computer Engineering, Director, Center for Imaging Science, Gilman Scholar, Johns Hopkins University
Over the past decade Computational Anatomy has been the study of structure and function in registered atlas coordinates. Unlike Google Maps which has been based on the rigid motions with scale for aligning coordinate systems, the underlying "alignment" groups in CA are the infinite dimensional diffeomorphisms. For rigid motion angular momentum plays a parsimonious role; in diffeomorphic motion the analogous role is played by diffeomorphic shape momentum. We present results from computational codes for generating diffeomorphic correspondences between anatomical coordinate systems and their encoding via diffeomorphic shape momentum.
Neuroinformatics on structural and functional phenotypes in normal and diseased populations are presented in registered atlas coordinates. The shape phenotypes are quantified via the shape momentum encoding of the population correspondence to registered atlas coordinates. The functional phenotypes are quantified via diffeomorphic transfer to registered atlas coordinates. The high dimensional feature vectors we associate to these structural and functional phenotypes we term neuroinformatics.