Fast and accurate contour propagation 53 Introduction External beam radiotherapy treatment is challenged by inter- and intrafraction anatomical changes in shape, volume, and location of the target and organs-at-risk (OARs).1–4 This can result in a lower dose to the target and/or higher dose to the OARs as compared to the pre-treatment plan.5 The clinical introduction of magnetic resonance imaging (MRI)-guided linear accelerators (MR-Linac) has significantly impacted radiotherapy workflows by enabling MR imaging prior to and during beamon together with fast planning tools.6–10 Currently, MR-Linac systems allow for non-rigid interfraction adaptation by daily imaging, re-contouring, and treatment planning.7,8 With this approach, the treatment plan is optimised for the daily anatomy just prior to beam-on. Intrafractional changes during radiotherapy delivery have become even more important with current interest in extremely-hypofractionated radiotherapy (i.e. £ 3 fractions) with larger fractional doses and therefore longer beam-on times.11–13 Previously, we have presented intrafraction motion results in prostate cancer.1,14 These results demonstrated that to guarantee target coverage with Planning Target Volume (PTV) margins below 5 mm, workflows that allow intrafraction adaptation are needed. Ultimately, fully automatic, online adaptive workflows may become clinically available, allowing continuous adaptation without operator intervention. Theoretically, the daily adaptive workflow could be repeated multiple times during a single treatment session, delivering the daily fraction in multiple virtual fractions (Virtual Fractionation [VF]). This would allow accounting for intrafraction changes. During MRI-guided workflows, there is a crucial role for an operator. The operator determines if the propagated contours are acceptable for treatment re-planning and remains responsible.15 Typically, contours should be manually adjusted after contour propagation before re-planning can be initiated, to obtain representative dose-volume histograms. Current online clinical contour adaptation times in MR-Linac workflows are substantial due to inaccurate propagated contours, with reported interfraction contour editing times of over 10 min.15–17 Manual contour editing is therefore the major delaying and limiting factor in such a workflow and limits the benefits that theoretically can be obtained. For workflows using repetitive MR imaging, deformable image registration (DIR), contour propagation, and re-planning to be clinically feasible, a fast and accurate auto-contouring solution is needed that reduces the need for manual adaptation and that limits operator interaction.18 The aim of this study was to explore the clinical quality of intrafraction propagated contours produced by a DIR algorithmwith respect to need for manual editing and feasibility of editing contours within a short time frame to allow for a fast, online adaptive workflow for MRI-guided prostate cancer radiotherapy. 3
RkJQdWJsaXNoZXIy MTk4NDMw