TH‐A‐BRA‐09: Statistical Assessment of Plan Robustness Under Uncertainties: IMRT Vs. Proton Therapy

Abstract

Purpose: Fair comparisons between x‐ray IMRT and Proton therapy plans should include assessment of plan robustness. However, this is difficult because the dose distribution can be altered in the presence of uncertainties in different ways for each modality. In order to fairly evaluate plan's robustness for both arms, we propose the use of statistical parameters to assess the probabilistic distribution of dose and DVHs under setup and range uncertainties. Methods: A lung cancer patient currently undergoing randomized trial was selected for this study. Physician approved x‐ray IMRT and passively scattered proton plans were created under identical planning objectives. For each of these plans, a total of 500 dose calculations were performed that simulated 50 virtual treatments with each treatment having 10 fractions. We found random error quickly converged within 10 fractions. A random setup error was introduced for each of the multiple fractions while a static systematic setup and range errors were introduced over simulated treatments. In calculating the doses, we assumed a static dose distribution for the IMRT plan. For Proton plans, a new dose calculation was performed for every simulation using a fast dose calculation method. Based on these simulations, the expectation value of DVH (E[DVH]) and standard deviation (SD[DVH]) were calculated. Both SD[DVH] and the difference between E[DVH] and the original DVH characterize the plan's robustness. Results: The E[DVH] showed that the V100% target coverage was 0.5% and 4% less than the original coverage for IMRT and Proton plans, respectively. The SD[DVH] near the prescription dose bin were 1% and 3% for IMRT and Proton plans respectively. Conclusions: Our results showed that the E[DVH] and SD[DVH] are viable options for evaluating plan robustness for both treatment modalities. Clinical prescription and dose objectives should be evaluated under uncertainty to assure treatment goals can be realized against uncertainties considered. Supported in part by NCI P01CA021239‐29A1.