? This application for competing renewal aims to continue ongoing work on automated segmentation of vessels and brain tumors from MR images. Previous funding has resulted in over 30 refereed publications, three commercial licensures/sales, and software used within multiple applications by multiple institutions. The current proposal aims to continue this work with focus upon a novel area ripe for development-the diagnosis and staging of disease based upon a statistical analysis of vessel shape. The driving clinical problem is that of assessing tumor malignancy. Despite advances in imaging technology, non-invasive differentiation of benign from malignant tumors remains difficult. An important observation made by those working from histological section is that vessels associated with malignancy possess distinctive shape abnormalities. Our preliminary data suggest that these vessel shape abnormalities are NOT restricted to tiny vessels at the capillary level, but also extend to larger vessels that can be delineated from high-resolution, in vivo imaging studies. The method involves a statistical comparison of regional measures of vessel shape in each test subject to that of the shapes of vessels in a similar region of an atlas of healthy subjects. The region of interest is that defined by tumor segmentation.
Specific aims are to: 1) Fully automate our current vessel and tumor segmentation methods. 2) Develop and evaluate a new method for fully deformable, diffeomorphic, multi-modal atlas formation. 3) Create an atlas of healthy vascular anatomy under the rules developed by Specific Aim 2. 4) Extend our methods of measuring vessel shape and to evaluate the approach for assessing tumor activity/malignancy via four clinical studies. Although this proposal focuses upon brain, the method is applicable to the assessment of malignancy in any anatomical location. Furthermore, although this proposal focuses upon cancer, the general method is applicable to the noninvasive diagnosis and staging of the multiple diseases that affect blood vessels, including hypertension, diabetes, the vasculopathies, and many other pathological states. ? ?
Showing the most recent 10 out of 33 publications
