Commissioning of MRI-only simulation for radiotherapy planning in pelvis

Modern radiation therapy delivery techniques enable ever conformal delivery of the radiation increasing the likelihood for successful treatment and reducing complications in nearby healthy tissue. In order to improve the treatment outcomes, in addition to advanced radiation delivery techniques, more accurate knowledge about the location and spread of both disease and organs at risk (OAR) is needed. Thus, the use of magnetic resonance imaging (MRI) has increased substantially during recent years. In MRI, the contrast resolution for soft tissue is superior compared to other imaging modalities enabling precise target definition and contouring of the OARs. Currently, the use of MRI in radiation therapy is based on co-registration of the images facilitating the use of the information provided by MRI while computed tomography (CT) is used for dose computation and as a reference image for patient positioning. Unfortunately, the dual modality workflow is laborious and cost inefficient. In addition, the co-registration uncertainty propagates to treatment uncertainty causing systematic error. During recent years several research groups have published methods enabling the generation of so-called synthetic CT (sCT). It can be used like traditional CT for density information in dose computation and as positioning reference images. The use of sCT enables external beam radiation therapy workflow using only MR imaging. In this work we studied the commissioning and accuracy of MRI-only workflow for external beam radiation therapy (EBRT) of pelvic malignancies. The commissioning test shall cover all steps in the radiation therapy workflow where geometric or dosimetric accuracy is affected by the substitution of the CT by the sCT. In publications I and III, we assessed the dosimetric accuracy of sCT images in pelvis by comparing to dose distributions computed using CT images. In publications II and III, we studied the patient positioning accuracy when sCT images are used as reference images. In addition, in publications I and III we evaluated the impact of geometric distortions to the total accuracy of MRI-only workflow. According to our results, the use of studied sCT method is sufficiently accurate for clinical use for pelvic indications. In addition, image-guided radiation therapy based on MR images is accurate enough so that the total geometric accuracy improves compared to current CT based work-flow.