Optimizing the Scenario Positions for Robust Radiation Therapy Treatment Planning

Complex external beam radiation therapy treatment plans are susceptible to errors. Robust optimization methods that take uncertainties into account during the optimization can alleviate the effects of the errors. For many robust methods, the possible error realizations are discretized into scenarios. The methods then aim at achieving high quality plans in all scenarios simultaneously. The choice of scenarios has high impact on the quality of the plans: when the optimization is aimed at high target coverage in scenarios that are incompatible with healthy tissue sparing in other scenarios, the plan quality in all scenarios may suffer. It can then be beneficial to reduce the range of errors the scenarios represent. To accomplish this, we propose a method that determines how large errors that can be accounted for in robust optimization of intensity-modulated photon and proton therapy. It does so by optimizing the scenario positions along predefined directions simultaneously with the direct machine parameter optimization, under constraints on target coverage and healthy tissue sparing. For both modalities, scenario position optimization was applied to a prostate case and a lung case. The optimization reduced the ranges of errors that were accounted for compared to the a priori ranges. The determined scenario positions were used in robust optimization with fixed scenarios. This resulted in plans that satisfied a substantially larger number of constraints in the determined scenarios and moreover a larger number of constraints in the a priori scenarios than the plans optimized with a priori scenario positions.