ESTRO 2021 Abstract Book

S63

ESTRO 2021

Considering the entire population, the margin was at most 0.8 mm larger for single-isocenter compared to multiple isocenters. The dose profile smoothing showed that the 65% dose level coincides with the symmetric point in the dose profile for AP and LR and is thus invariant to blurring (Figure1A-C), whereas the margin contribution of random errors was 0.25σ tot for SI. This is in agreement with the theoretical β of 0.39 for a 65% prescription level when the penumbra width is 5.2 mm (Figure1D).

Conclusion The population-based margin calculation showed that patients with larger inter-target distance and one lesion not attached to the pleura required a larger margin. Nevertheless, incorporating the relative inter-lesion motion in the margin calculation has a small influence on safety margins (maximum 1 mm). OC-0089 Impact of using uncorrected CT-based DIR-propagated autocontours on online ART for pancreatic SBRT A. Magallón Baro 1 , M. Milder 1 , P. Granton 1 , W. den Toom 1 , J. Nuyttens 1 , M. Hoogeman 1 1 Erasmus MC Cancer Institute, Radiotherapy, Rotterdam, The Netherlands Purpose or Objective Stereotactic Body Radiotherapy (SBRT) for Locally Advanced Pancreatic Cancer (LAPC) patients is limited by dose restrictions due to its location close to organs-at-risk (OAR) that are prone to daily anatomical variations. Online adaptive radiotherapy (ART) has emerged as potential method to compensate for anatomical changes (i.e. interfraction motion) while real time tracking compensates for intrafraction motion of the target. For many institutes, a major challenge in achieving fully online ART lies in organ delineation. Automatic definition of these structures (autocontouring) by propagating contours from the planning CT (pCT) to the fraction CT (FxCT) through deformable image registration (DIR) has been proposed. However, due to poor soft tissue contrast in the abdominal area, DIR is susceptible to errors. In this study, we have quantified geometric differences and the dosimetric impact of uncorrected autocontours through a commercially available DIR algorithm for pancreatic SBRT using CyberKnife with integrated CT-on-rails. Materials and Methods A total of 35 pCT and 98 FxCTs high-resolution CT scans acquired in treatment position were collected for 35 LAPC patients. All pCT and FxCTs were manually contoured (MAN) to define a ground truth contour set, including stomach, duodenum and bowel. Next, pCT contours were propagated to FxCTs using the Precision TP