[OA010] A 4D Monte Carlo (MC) dose calculation framework with statistical breathing phase sampling to quantify interplay effects

Purpose The interplay between dose application by complex techniques and respiratory motion of a tumor can potentially lead to undesirable and non-intuitive deviations from the planned dose distribution. We developed a 4D dose calculation framework to precisely simulate dose distributions for moving target volumes. In this study we randomly sample the simulated application of delivered dose fragments on 4D-CT phases. Methods The workflow combines MC dose calculation with linac-log files and dose accumulation based on 4D-CT images. Treatment plan fragments of 0.2s duration are retrieved from linac-log data and calculated on ten 4D-CT phases using MCverify/Hyperion V2.4 (research version of Monaco 3.2). The resulting dose fragments allow the simulation of arbitrary respiratory curves (e.g. changes in breathing frequency and pattern) with a resolution of 0.2 s by assigning every fragment to a distinct 4D-CT phase. Using deformable image registration (plastimatch) the dose fragments are accumulated by AVID, a software system for automated processing and analysis of radiotherapy data. In addition to the patient’s recorded normalized breathing curve, three statistical approaches are implemented: (1) random phase shift of the breathing curve, (2) random phase assignment of every 0.2s dose fragment and (3) random phase assignment of 1MU dose fragments. Results An exemplary 3 Gy, VMAT, SBRT treatment of a 9 cm 3 lung tumor with 1.6 cm crano-caudal movement was analyzed. 128 random treatments were simulated for the three statistical approaches. In all three cases the mean 4D-CT calculated dose (original plan calculated on ten 4D-CT phases) and the mean of the randomly simulated doses agree. The dose deviations for the 128 runs are very similar for the three methods (average dose deviations: σ D 2 % ≈ 0.41 % , σ D 50 % ≈ 0.25 % and σ D 98 % ≈ 0.68 % for the three approaches). Conclusions The described random assignments (2) and (3) have similar statistics as the random start phase approach (1) and are hence promising methods to comprehensively cover treatment plan and technique specific interplay effects without the need for daily breathing curves. The MU-based random sampling approach (3) is in addition independent of the linac-log data. The introduced framework has the potential to comprehensively include breathing motion induced interplay effects in the treatment planning and evaluation process.