Specific Aims

1. Develop deformable registration techniques using deformable models to map voxels of anatomic structures obtained at different times onto corresponding voxels of a volume chosen as the reference volume. By exploiting information from both the region of interest and its boundary, deformable anatomic models will be generated and transformed using the acquired multi-dimensional image data.

2. Validate volume registration algorithms using both phantom studies and patient studies. Test objects embedded in deformable phantoms will be used to verify registration algorithms in controlled experiment. In patient studies, CT images of a selected group of patients who have identifiable internal markers will be used. Examples are patients previously treated with implanted radioactive seeds for prostate cancers etc.

3. Calculate radiation dose deposition in deformed organs. The non-linear transformation developed in Aim #1 will be implemented into a radiotherapy treatment planning system to map the dose to a reference treatment plan.

4. Estimate the tumor control probability (TCP) and normal tissue complications (NTCP) in accounting for internal organ variations, setup errors, and biological response to radiation doses in a course of radiation treatment.