ESTRO 2023 - Abstract Book

S1769

Digital Posters

ESTRO 2023

highly dependent on the tissue density, studies have shown that MBRT optimization must be performed robustly to setup uncertainties. This study investigates the impact of range uncertainties associated with assigning voxel densities from CT images on MBRT plans. Materials and Methods To evaluate the impact of range uncertainties, MBRT plans were generated for 2 patients: a sarcoma of the chest wall and a sarcoma of the leg. MBRT plans were optimized using a column generation-based algorithm on an in-house web-based treatment planning system called Brems. For each patient, a plan was robustly optimized considering both setup and range uncertainties. Robust optimization was performed using the stochastic programming approach, where each scenario was assigned equal weight. Robust setup scenarios consisted of calculating beamlets with +/- isocenter shifts in the 3 cardinal directions by 5 mm. To handle range uncertainties, 2 additional phantoms were generated by applying a +/-3.5% factor on every voxel’s CT number prior to their conversion to mass densities. For each +/-range scenarios, the beamlets would be calculated using their respective downscaled/upscaled phantoms. The impact of range scenarios was assessed by comparing the DVHs in these range scenarios to DVHs obtained from setup scenarios.

Results

For each plan, 3 DVHs are plotted for each relevant structure. The full line DVH represents the nominal scenario (with no setup or range error), the shaded bands represent the setup scenarios, while the DVH in dotted lines represent the range scenarios. In the leg sarcoma case, DVHs of range scenarios were observed to mostly fall within the setup scenario bands for all relevant structures. This indicates that range uncertainties represent a smaller deviation from the nominal scenario than setup scenarios. In the chest wall case, DVHs of range scenarios exceeded setup scenario bands in the ipsilateral lung and the heart. Range uncertainties therefore seem to have a greater impact on MBRT dose distributions in sites of large tissue heterogeneity and low mass density. Conclusion The impact of range uncertainties on MBRT doses was assessed on 2 representative cases. In the mostly homogeneous leg case, range uncertainties were found to be negligible compared to the deviations evaluated in setup scenarios. However, in the chest wall case, DVHs of structures near low density voxels were found to be more sensitive to range uncertainties.