ESTRO 2021 Abstract Book

S463

ESTRO 2021

plan with one beam was optimized to deliver 1.8 Gy(RBE) using a particle extension of the research treatment planning system CERR. The dose distribution was recalculated 1.6∙10 4 times while simultaneously sampling RU and SU from their assigned distributions (setup in X,Y,Z directions: normal distributed (ND) with standard deviation (STD) 1mm, absolute range shift: ND with STD 1mm, relative range shift: ND with STD 3%) and from the 4 observer treatment plans (probability p=0.25 for each). Variance-based formalisms rank the parameters according to their impact on the overall variance of DVH quantiles resulting in sensitivity indices (SI) (“S=0: no impact, S=1: only influential input”). For all patients, mean values, 68% and 95% confidence intervals (CI) and SI for the DVH quantiles of CTV s were calculated. Results The mean time for 1.6∙10 4 dose calculations was 59 minutes (range 11 - 194). As example, the D 95% results of the CTV s are shown in table 1. For 6 patients, the variability of the D 95% was small in presence of SU, RU and IOV (width of 95% CI < 0.08 Gy). For the remaining 4 patients, the width of the 95% CI was 0.37, 0.22, 0.12 and 0.23 Gy, respectively. DVHs with the corresponding CIs and the relative impact of SU, RU and IOV on the D 95% uncertainty for these patients are shown in figure 1. For 2 of these patients (7 and 9) the influence of IOV and interaction of IOV with other parameters was negligible (S(IOV)+S(interaction,IOV) < 5%). In both cases, RU had the largest impact (S(RU) was 53 % and 49%, respectively). For the other two patients (2 and 3), uncertainty was mostly driven by IOV (S(IOV)+S(interaction,IOV) was 43% and 63%, respectively). SU played a minor role (for patients 2, 3, 7 and 9, S(SU) was 12%, 14%, 25% and 13%, respectively).

Table 1: Mean values, CI and SI for the D 95%.

of the CTV s

for 10 patients.

Figure 1: DVH variances for the 4 patients with the largest D 95%

uncertainties. SI for D 95%

are shown in the pie

charts.

Conclusion A framework to evaluate the combined impact of SU, RU and IOV is presented. Although not explicitly accounted for in the margin concept, IOV did not lead to a relevant increase in D 95% uncertainty for 8 out of 10 patients. The impact of future RU and SU reductions can be assessed with the SA. PH-0599 Three-dimensional dosimetric verification of spot-scanning proton therapy in a breast-shaped phantom S.E. Jensen 1 , L. Barbosa Valdetaro 1 , M. Fuglsang Jensen 1 , P. Balling 2 , P. Sandegaard Skyt 1 , J. Breede Baltzer Petersen 3 , L. Paul Muren 1 1 Danish Centre for Particle Therapy, Aarhus University Hospital, 8200 Aarhus N, Denmark; 2 Department of