CORE--NEUROIMAGING

University Of Southern California, Los Angeles CA

Investigators

Linked publications & trials

Abstract

The overall goal of the Neuroimaging core is to provide quantitative analysis of brain structures, measured by MRIs acquired at two sites in Southern and Northern California, using identical protocol which permit combined use of data. The Core receives data from both sites, transfers the data to a common format, archives the MRIs, processes the data in various ways and archives processed images, transfers all data to a data base which is heavily used by all Cores and Projects. During the previous funding period, this Core developed a number of image segmentation (gray matter, white matter, CSF, white matter lesions, lacunes) voluming (hippocampus and entorhinal cortex, and localizing tools (lacune location), all of which have been incorporated into a semiautomated, user friendly, software package. During the current funding period the total processing performed was: segmentation= 581, edited (WML and lacunes identified) = 569, hippocampaI voluming = 335, ERC voluming = 236, longitudinal change= 50. The results have been used in two major publications from this PPG, both which emphasized the importance of decreased cortical and hippocampal gray matter as determinants of cognition, and de-emphasized the significance of white matter lesions and lacunes. Additionally, most other publications from this PPG used Imaging Core data. During the proposed funding period this Core will continue to perform image analysis of MRls collected in both Northern and Southern California and make this data available to PPG investigators. In addition, during the current funding period this core has implemented several new MRI analysis techniques : 1) Non-linear registration techrdques. These were used for: lobar marking, identifying shape changes due to AD, SIVD, and presence of lacunes in controls, measuring longitudinal atrophy and effects of white matter lesions and lacunes on this process, correlating shape changes with clinical and cognitive data. We propose continued development of these tools and applying them to more refined analysis of clinical data. 2) Completely automated probabilistic segmentation of 3D T 1 weighted MRIs. We propose to extend this approach to a more automated method for classifying WMLS detected on T-2 weighted MRI. 3) Semiautomated hippocampal voluming. This has been validated and applied to measurement of cross sectional and longitudinal clinical data. 4) Manual measurement of entorhinal cortex. This was validated and used for cross sectional and longitudinal studies. In summary, this Core provides an invaluable service function to process MRIs and also develops state of the art, automated image processing methods aimed at achieving the goals of this PPG.