Abstract Book

S159

ESTRO 37

In this in-silico study, treatment plans for the MRI-Linac and for a standard linac were compared for esophageal cancer. Most relevant parameters were statistically equivalent, V 5 and D mean in the lung and D mean in the heart showed a significantly increased dose in MRI-Linac treatment plans. However, all MRI-Linac plans were clinically acceptable. Differences may be compensated in the future by margin reduction due to MR-based daily adapted radiotherapy and implementation of VMAT. OC-0304 Using a single knowledge-based proton planning model to create automated plans for different centers A. Delaney 1 , L. Dong 2 , A. Mascia 3 , W. Zou 2 , Y. Zhang 3 , L. Yin 2 , J. Hrbacek 4 , A. Lomax 4 , B. Slotman 1 , M. Dahele 1 , W. Verbakel 1 1 VUMC, Cancer Center Amsterdam- Department of Radiation Oncology, Amsterdam, The Netherlands 2 University of Pennsylvania, Department of Radiation Oncology, Philadelphia, USA 3 University of Cincinnati Medical Center, Department of Radiation Oncology, Cincinnati, USA 4 Paul Scherrer Institute, Center for Proton Radiotherapy, Villigen, Switzerland Purpose or Objective While the number of proton centers globally is increasing, proton treatment planning quality is subject to variation, experienced proton planners are limited in number, and planning for complex cases is time consuming. Such circumstances create a potential role for automated planning/optimization solutions. We evaluated whether a pre-clinical automated optimization solution (RapidPlan for protons, Varian Medical Systems), comprising a model based on proton treatment plans (TPs) from one center, could generate knowledge-based proton plans (KBPs) of an acceptable quality for other proton centers. Material and Methods Fifty intensity-modulated proton therapy (IMPT) TPs for locally-advanced head and neck cancer were designed using a 3-field beam arrangement and used to populate the KBP model. Two proton institutions (A and B) provided 5 contoured CT scans, for which KBPs were created (data from 3 rd center will be available at meeting). KBP optimization objectives were generated automatically, excluding organs-at-risk (OARs) which were not included in the model. Occasionally a subsequent “continue optimization” was applied to refine PTV coverage. KBPs were normalized to cover 95% of the boost planning target volume (PTV B ) with 100%/99% of the prescribed dose for centers A/B, respectively. KBPs were compared to the clinical TPs using PTV B homogeneity index (HI B ; 100*((D2%-D98%)/D50%), mean dose to individual salivary and swallowing OARs. The time required to create KBPs was also investigated. Results KBP creation required <10 minutes. Based on PTV coverage and OAR dose, KBPs were generally of comparable quality to clinical TPs. On average, HI B differed by <2.5%. Some KBP mean OAR doses were lower, e.g. for center A patients larynx/esophagus (on average 6.7/6.0Gy lower) and for center B contralateral submandibular gland/pharynx (on average 27.3/10.6Gy lower) (Table). However, on average KBP parotid mean dose was higher for center A patients (0.1/2.1Gy higher in contralateral/ipsilateral parotid gland). Certain improvements to KBPs are required, including a higher minimum dose to PTV B , lower maximum dose, and control of hot-spots outside PTV/under skin. In addition the model did not include all OARs used in certain clinical TPs (e.g. mandible/optic system). We expect that performing a subsequent optimization, using a few additional structures, should address these issues.

Conclusion RapidPlan for protons was able to generate KBPs with adequate OAR sparing. This study shows that fast, automated IMPT planning is feasible and that a single model can be used to create plans for other centers. It also confirms that planning is prone to variation. Hot spots outside PTVs should be rectifiable by a subsequent optimization. A limitation of the study is that only one standard beam angle set-up was used for KBPs, while clinical TPs had patient-specific beam arrangements. Further refinements to the KBP model and the optimization algorithm are needed, however completely automated planning is within reach. OC-0305 Knowledge-based models for automated planning are strongly affected by inter-organ dependency Y. Wang 1 , B.J.M. Heijmen 1 , S.F. Petit 1 1 Erasmus MC Cancer Institute, Radiation Oncology, Rotterdam, The Netherlands Purpose or Objective Knowledge-based (KB) dose prediction models have gained in popularity as a method for automated treatment planning. KB models predict OAR doses based on their geometry and location relative to the PTV, using a plan database of prior patients. Most published models consider each OAR separately. But for complex treatment sites, such as head and neck (HN), the achievable OAR doses may strongly depend on their priority compared to competing OARs. In these cases, KB models yield prediction errors and therefore suboptimal plans. In this study, we systematically investigated the effect of inter- OAR dependency on the prediction accuracy of KB models In total 108 oropharyngeal cancer patients were included in the study. Our fully automated, multi-criterial treatment planning system (not knowledge based) was used to generate plans that are Pareto-optimal and have consistent prioritization between sparing different OARs. Therefore the plans can be considered as golden standard to train and evaluate KB models. For each patient, 15 VMAT treatment plans were generated (1620 in total). For plan P0 the left parotid gland had the highest priority and was therefore not influenced by inter-OAR dependency. For each of the remaining 14 plans per patient (labeled P1, P2,…,P14) the planning priority of sparing the parotid glands vs. the other OARs was systematically lowered. Next, for each of the 15 sets of plans a KB model was trained on 54 patients and evaluated on the other 54. The KB models were based on the overlap volume histogram and principal component analysis. The effect of inter-OAR dependency was determined by comparing the prediction errors of the KB models as function of the priority of sparing the parotid glands. Results For all 1620 plans, the PTV coverage met the clinical constraints. As expected, the achieved D mean of the for HN cancer patients. Material and Methods