The overall goal of this project is the conceptual development and prototype implementation of a database methodology that supports the archiving and statistical investigation of large numbers and types of brain images.
The specific aims of the study are: 1) to develop a morphologically factored image representation (MFIR) system that allows improved comparison of brain images, 2) to develop a BRAin Image Database (BRAID) that supports novel statistical analyses of image datasets, and 3) to evaluate the database by applying it to both simulated data and to real data from 3 current brain imaging studies. The MFIR is based on a nonlinear registration of an image to a standard Atlas to create a morphologically normalized signal component and a morphological variation component, represented as a displacement vector field in Atlas coordinates. The BRAID will implement storage, query and statistical operations on the MFIR components. The BRAID will be validated by testing its ability to recover known correlations from simulated data, and applied to the analysis of data from several collaborating epidemiological studies. The applications will test the system's ability to identify brain structure/function correlations from lesion/deficit data derived from stroke and injury, and its ability to identify patterns of morphological change in brain anatomy with age, and correlate these with functional data. Stroke data will be provided by the Cardiovascular Health Study, and NHLBI sponsored project that is collecting extensive prospective demographic, functional, and brain MRI data on over 3,600 participants. Injury data will be collected by the Psychopathology of Frontal Lobe Injury in Childhood study, which is collecting brain MRI and extensive psychiatric/functional data on 100 children with traumatic brain injuries. Aging-related morphological and functional change data will be supplied by Baltimore Longitudinal Study on Aging, which follows 180 patients over a 9 year period and performs MRI and PET scans, along with neurofunctional evaluations, on an annual basis. The newly developed database is intended to be flexible in terms of acceptable data types, robust in its querying mechanisms and extendable to other laboratories; thus providing the basis of a future broad based, multi-institutional brain informatics network.
Showing the most recent 10 out of 30 publications
