In Image Analysis (Task 2.2), previously developed model-based curve evolution techniques have been extended in several ways. They now incorporate multiple objects, using landmarks from one tissue class to guide the segmentation of nearby classes. This enables robust delineation of ensembles of organs of interest. These methods have been amplified with physics-based deformation of anatomical models, in addition to statistically registered observations in large samples of medical image scans. The MRI prostate segmenter is now being ported to our prostate intervention testbed in Task 2.1. In related work, hierarchical segmentation methods have been developed, which incorporate statistical methods, surface evolution methods, and atlas-driven template methods. These will be coupled with deformation models, and the entire hierarchy will be applied to challenging segmentation problems. The model-based evolution techniques will migrate toward clinical applications through retrospective analysis of actual clinical data, for which the initial application is prostate therapy with MRI guidance.