Model-based segmentation and animative visualization for analysis of anatomical structures

This dissertation explores the use of shape models in a number of medical data processing areas. Particular focus is on processing involving two anatomical structures, the kidney and the cardiac left ventricle (LV). The dissertation introduces a kidney shape constraint model and a snake-based deformation strategy for segmentation of the kidney and a LV shape model and animative visualization method useful for analysis of LV function. The LV visualization is implemented in a cluster computing environment that allows end-display on a single-processor PC or workstation. The dissertation also introduces a new method for volume interpolation, the multiradial correspondence approach. Experimental results are presented to show the effectiveness of the techniques developed in the dissertation.