The goal of this project is to continue and significantly expand our work on image analysis methods for brain magnetic resonance images, with emphasis on deformable registration and its application to morphometric analysis and spatial normalization of brain images in a longitudinal study of aging, and the use of these methods to develop an image-based early diagnostic tool for mild cognitive impairment and Alzheimer's Disease. Quantification of individual morphometric characteristics is achieved via a shape transformation, i.e. a spatial transformation that adapts a template of anatomy to the morphology of the individual under study. The shape transformation is a very detailed mathematical representation of anatomy, and is used for inter-individual comparisons and spatial normalization of structural and functional images. The overall goal of this project is to address three limitations of current technology, which are treated in the respective specific aims. Specifically we propose to 1) develop and validate a methodology for obtaining a rich image representation from MR images, which will allow for different brain regions to have distinctive morphological signatures, thereby facilitating automated algorithms for determining anatomically accurate shape transformations, 2) develop and validate a methodology for finding 4-dimensional shape transformations from longitudinal image data, with the fourth dimension representing time; this methodology will significantly reduce measurement error by incorporating temporal smoothness constrains into the estimation of the shape transformation at different time-points, 3) develop and validate a morphological representation based on the shape transformation of Aim 2, which will represent an individual's anatomy in terms of a shape transformation of an anatomical template, and a residual image that captures information that is not captured by the shape transformation, and 4) to apply these methods to the Baltimore Longitudinal Study of Aging, in order to test our hypothesis that sensitivity and specificity of early detection of cognitive decline using MR images will be significantly improved by the new technology, because of improved accuracy in morphologic measurements, and to develop a high-dimensionality image-based pattern classification method for early diagnosis of Alzheimer's Disease.
Showing the most recent 10 out of 130 publications