Fig. 1.

Schematic representation of three different registration scenarios in a MR-CT radiotherapy workflow. The anatomy of the patient as visualized with MR is given in red and visualized with CT in green. In the registration column registrations between MR and CT are visualized. Structure positions that should not be used when planning the treatment are dotted. Row B shows the situation where the registration has been focused on the anatomical structures used for positioning of the patient at treatment. In this situation all structures should be delineated on MR, and the plan will be based on the observation of the anatomy as given in the MR examination. Row C shows a situation where a very accurate registration of the target volume has been achieved, while the registration of the structures used for positioning of the patient are of low quality. In this scenario all structures except the target should be delineated on CT, and the plan will be based on the observation of the anatomy as given in the CT examination. The final row D, illustrates the scenario where the registration has resulted in errors for both the target volume and the structure used for patient positioning. In this case, it will be impossible to get an unambiguous representation of the patient anatomy unless the positioning of the patient is done using the MR information. The position of the target relative the positioning structure will be a mixture between the MR and CT observation and delineations of the OAR can be consistent with the target or the positioning structure, but not with both. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)