This consortium project has the specific goal of developing a digital probabilistic reference system for the human brain. Through a comprehensive approach, the use of populations of subjects and an iterative development scheme, this program will produce a population- based probabilistic atlas and data base that performs morphometric correlations in three dimensions allowing efficient statistical interactions and a framework for modeling complex structural and physiological responses on an anatomical substrate. Give key scientific goals will be achieved. First, data sets will be acquired, after instrument calibration with phantoms, of postmortem and in vivo human brain structure (MRI) as well as function (PET) in sufficient quantity, detail and homogeneity for use in concept development, methodologic testing and feasibility determinations. Second, image segmentation will be employed to divide brain structure into neuroanatomical labels using the appropriate algorithms to classify each voxel in each data volume in an objective and reproducible fashion. Third, brain data correlations will establish the conceptual framework and practical tools for the intra- and inter-modality comparisons of structural anatomy both within and between subjects. This will include subsets of subjects who travel between participating sites to determine the variability introduced by different equipment. Fourth, an operational space will be iteratively defined beginning with established stereotactic space and extending to a minimum variance voxel field where the maximum overlap across all neuroanatomical labels (weighted sum) for the human brain is achieved. Fifth, a data base will be developed as a structure-probability map designed to account for attributes of the human brain including the appropriate visualization, representation and search tools. This comprehensive project utilizes the experience, expertise and instrumentation of three primary laboratories with long tract records for exploring the above issues and developing tools to address these problems through neuroscience, informatics and computer science. The participating laboratories include UCLA, the Montreal Neurologic Institute/McGill University (MNI), and the University of Texas Health Sciences Center at San Antonio (UTHSCSA). The distribution of labor into parallel complimentary pathways utilizing established leaders in the field will be efficient in terms of cost and time as compared to individual proposals generated at each site competitively. An advisory board of key consultants has also been established to direct the consortium in a wide range of scientific areas. Two cores support the projects described in this consortium and include: Core 1- Computer/Network: responsible for testing, storing and distributing software and data sets maintenance electronic bulletin boards and digital communication within an outside the consortium; and Core 2-Data Set Acquisition: in which subjects will be identified, examined and imaged to produce the test data to determine the feasibility of the approaches described in the projects. Liaisons with three other consortia have also been established to determine the capabilities and feasibility of networking and transmitting large data volumes, exchanging software and linking data bases developed independently by each group.
Showing the most recent 10 out of 36 publications
