ANATOMIC MORPHOLOGIC ANALYSIS OF MR BRAIN IMAGES

Massachusetts General Hospital, Boston MA

Investigators

Linked publications & trials

Abstract

DESCRIPTION (Adapted from applicant's abstract): The morphology of the human brain is exceptionally complex; reflecting a myriad of inextricably intertwined systems of neuronal cell bodies, axons and other components. Groupings of neural components which share common structural or functional properties comprise the structural and functional neuroanatomic framework of the brain. Characterization of the morphologic properties of the brain and its component parts by state-of-the-art MRI is well suited to permit a quantitative study of the parameters relevant to the structural and functional makeup of the human brain in vivo. The goal of this project is to continue development of tools and methods for the precise quantitative analysis of brain morphology in health and disease, and to disseminate the tools and results of the application of these tools to the neuroscience community as a whole. Specifically, this project will: (1) extend previously developed pixel segmentation and morphological quantification methods to time-domain (functional MRI) and tensor-valued (diffusion-weighted MRI), continuing efforts to develop a unified neuroanatomic segmentation framework, and transition these tools to clinical applications on a routinely available software platform; (2) characterize shape and shape change metrics in normal subjects and pathological patient populations; and (3) dissemination of segmentation tools and comparison methods, as well as the results of image segmentation and volumetric analysis to the community as a whole using the World Wide Web. This proposal continues to take advantage of several unique aspects which distinguishes it from other related work. First, a unified framework for segmentation and classification in support of a neurologically-based anatomic morphology has begun to emerge. Second, this unified framework incorporates the multispectral nature of MRI data. Third, this framework intrinsically includes estimates of the resulting underlying uncertainty associated with the segmentation and classification process which supports a rational assessment of sensitivity of a given method. Fourth, this approach expands upon traditional "static" image analysis by incorporation of shape-based analysis for anomaly detection. In addition, several clinical application areas were identified which, in addition to fostering enhanced analytic capabilities to studies in these areas, permits optimization of operational efficiency in the resulting analysis. Specifically, the segmentation, classification and shape analysis of MRI data in patients with Huntington's Disease, stroke, and central nervous system neoplasms, as well as the appropriate normative subjects, provide a vital testbed for the evaluation of the clinical utility of these morphological analysis techniques.