ESTRO 2020 Abstract book

S814 ESTRO 2020

It is possible to create IMRT MRL partial breast plans which are of comparable robustness to C-arm linac VMAT plans in terms of external contour changes. The electron return effect induces a degree of non-robustness at the chest wall / lung interface which can be quantified during reference planning, but not explicitly accounted for during optimisation. Initial patients will be treated free- breathing, so the results presented here potentially over- estimate this effect. PO-1438 Automated proton treatment planning with robust optimization and dose volume constraints V. Taasti 1 , L. Hong 1 , J. Deasy 1 , M. Zarepisheh 1 1 Memorial Sloan Kettering Cancer Center, Medical Physics, New York, USA Purpose or Objective To present a fully automated approach to proton treatment planning including robust optimization and dose volume constraints (DVCs). To fill the gap between two common extreme robust approaches, i.e. stochastic and worst-case, the generalized equivalent uniform dose (gEUD) formalism is used whereby a single gEUD parameter, a , can be used to control the level of robustness in an intuitive way. Material and Methods A fully automated photon treatment planning using Expedited Constrained Hierarchical Optimization (ECHO) is implemented in our clinic. ECHO strictly enforces the clinical criteria as hard constraints (max and mean dose, and dose volume constraints (DVCs)). We adapt ECHO for proton therapy and integrate it with robust optimization. Thirteen scenarios accounting for setup and range uncertainties are included in the robust optimization, and the max/mean/DVC constraints on organs-at-risk (OARs) and target are fulfilled in all scenarios. We propose combining the objective functions of the individual scenarios using the gEUD-based function. The gEUD parameters a =1 and a =∞ result in the stochastic and the worst-case approaches, respectively, and an intermediate robustness level is obtained by using a -values in-between. While the worst-case approach only focuses on the worst- case scenario which could potentially result in a non- pareto optimal plan, the gEUD approach with a large value (e.g. a ≈20) resembles the worst-case approach without completely neglecting other scenarios. The proposed approach is evaluated on three head and neck (HN) patients and a water phantom. The results are compared against the stochastic and worst-case approaches as well as the non-robust approach. Results The proposed algorithm successfully generates automated robust proton plans on all cases. As opposed to the non- robust plan (optimized only for the nominal scenario) the robust plans have narrower dose volume histogram (DVH) bands and meet all dose constraints (max, mean and DVC) on OARs and target for all scenarios (Fig 1). Compared to the worst-case approach, the gEUD-based approach results in DVH bands for clinical target volume (CTV) which are closer to the prescription dose at a negligible cost in the worst-case scenario plan, improving the overall plan quality. On average, going from the worst-case approach to the gEUD-based approach with a =20, the median objective function value across all the scenarios is improved by 15% while the objective function value for the worst-case scenario is only degraded by 3% (Fig 2).

Conclusion An automated treatment planning approach to proton therapy is developed including robustness, DVCs, and ability to control the robustness level using gEUD parameter a . We found a =2 to provide a good balance for CTV coverage across all the thirteen scenarios, including the nominal and worst-case. However, the value of a can be adjusted to fit the priorities deemed most important. We showed the feasibility of incorporating DVCs into our robust hierarchical optimization approach.