ABSTRACT
As the demands for improved accuracy in radiotherapy increase, so does the need for better accounting of the dynamic nature of patients. Radiotherapy is a primary cancer treatment modality that benefits from integrating information acquired with multimodality and serial images at all stages of the treatment process. This wealth of imaging information creates a demand for deformable image registration (DIR) to model anatomical changes caused by physiological processes (e.g., breathing motion, rectal and bladder filling, and peristalsis), changes in patient pose (e.g., patient positioning at each treatment and between imaging sessions), and the responses of tissues to radiation. Unlike many other algorithms available, biomechanical-based DIR algorithms model the physical properties of tissues that govern many of these sources of motion. Radiotherapy plans are often designed on a single static computed tomography (CT) image of the patient acquired before treatment. To account for geometric uncertainties in the treatment processes, radiation dose is planned to tumor volumes plus an additional safety margin, at a cost of increased normal tissue irradiation. During treatment delivered over multiple fractions, spanning days-to-weeks, daily image guidance is employed to aid in targeting tumors and improve confidence that the planned dose is accurately delivered to the patients. Motion and deformation during delivery can, however, cause deviations in the intended doses exceeding 5% in magnitude. Reducing the uncertainty in these processes through the application of DIR can potentially improve the therapeutic ratio. This chapter initially focuses on briefly describing biomechanical-based DIR techniques developed in several anatomical sites. The second half illustrates how the application of biomechanical DIR can reduce geometric and dosimetric uncertainties in radiotherapy, improve the design of treatment, and augment our understanding of response to several clinical scenarios.
