Session Item

The use of deformable image registration (DIR) to perform dose accumulation has made significant advances of the past 2 decades, however many challenges still remain. In this teaching lecture, we will review the current status of algorithms and their use as well as the numerous challenges that remain to be solved. As in room image guidance has progressed from 2D x-rays to volumetric CT and cone-beam CT (CBCT) and now to include MRI and PET images, the demand for multi-modality image registration has expanded to enable dose accumulation using these various images. The continued advancement in the quality of in room imaging provides further recognition of the complexity of tissue changes over the course of therapy that DIR must accurately model. DIR algorithms are faced with challenges including accounting for volumetric changes in tissue, sliding interfaces, tumor dynamics, as well as intra- and inter-fraction motion. The increasing conformity of treatment plans mandates that DIR accuracy must continue to improve. Quantification of the uncertainty in DIR and its impact on applications, such as dose accumulation, has been the topic of numerous studies in a variety of organs. However, the translation of these uncertainties into clinical recommendations has remained challenging due to several limitations, including the lack of large cohorts of testing data that can be shared and compared between clinics. Examples of these challenges, and promising solutions, will be illustrated with various anatomical sites, imaging modalities, and DIR algorithms.