Image registration is the process of computing a geometric transformation that aligns homologous points in two images. My interest lies in applying this process in radiation therapy. Translation of basic algorithms into clinically feasible ones requires the optimization of accuracy, robustness, and computational efficiency. Image Guided Radiation Therapy (IGRT). Many linear accelerators are capable of acquiring 3D images of the patient prior to delivering a fraction of radiation therapy. Such images reveal patient setup error and normal anatomical motion. I am currently working on software algorithms for modifying treatment parameters in response to such changes. Atlas-Based Treatment Planning. Complex treatment planning is often extremely patient-specific and therefore time consuming. This may lead to large variability in treatment quality across patients. I am interested in the feasibility of building a multi-subject, plan database and customizing this atlas to future patients.