Automated 3D Analysis of Cortical Geometry in MR Images

点击翻译按钮获取中文摘要

Characterizing the relationship between the structure of the human brain and its function is one of the most important goals in neuroscience today. Medical imaging has been used to gain significant new insights into this relationship through the use of both anatomical and physiological imaging methods. Despite significant recent advances, the current methodology is still limited by the lack of automatic methods for the detailed segmentation, geometric analysis, and labeling of the cerebral cortex. The major goals of the proposed research are to develop fully automated methods to find and mathematically represent the cerebral cortex in volumetric magnetic resonance (MR) images and to automatically identify and label the major sulci and gyri on the cortex using a detailed statistical analysis of cortical geometry. Specifically, we propose to 1) develop and validate methods to find and mathematically represent the cerebral cortex from volumetric MR images; 2) develop and validate methods to calculate regional measures of cortical shape and volume; 3) develop and validate automatic labelling of sulci and gryi; and 4) conduct studies of cortical variability and volume changes in normal aging. All methods will be extensively validated using both computer phantoms and manual in vivo truth models. The methods we develop to automatically represent and label the cortex in large numbers of subjects should also be useful in 1) the development of a description of normal versus diseased cortical geometry, 2) automatic landmark generation for deformable atlas registration, 3) statistical correlation studies of structure/function relationships, and 4) the analysis of morphological changes in ontogenesis, phylogenesis, aging, and disease.