ESTRO 2022 - Abstract Book

S421

Abstract book

ESTRO 2022

4D dose calculation (4DDC) is an important tool for evaluating pencil beam scanned (PBS) proton treatments for moving tumors. We investigate the limitations of the conventional phase-sorting (PS) 4DDC based on a single 4DCT, due to the lack of breathing variability and its rather coarse temporal resolution. Materials and Methods Multicycle motions from two 4DMRIs were used to generate six synthetic 4DCT(MRI) 1,2 datasets for three liver and three lung cancer patient CTs, providing irregular motion scenarios and their associated density variations (11/15 breathing cycles for liver/lung cases, respectively). Using superior-inferior (SI) isocenter motions as a surrogate, each breathing cycle was divided into 8 or 10 phases with equal number of phases in in- and exhalation (Fig1; temporal resolution 0.4-1 s). Based on its time of delivery, each pencil beam of a plan was then assigned to one motion phase either from a single reference cycle to simulate PS for different reference 4DCTs (PS single cycle-PSs) or from the full irregular motion pattern (PS multicycle- PSm). 4D dose distributions of both scenarios were compared to dose distributions calculated using the deforming-dose- grid algorithm, which works at a temporal resolution determined by the time between the delivery of pencil beams (typically 4-200 ms) (spot wise multicycle-SWm) 1 . Two-field plans were optimized on the static reference CT of each dataset using a geometric internal target volume (gITV). 4DDCs were assessed for delivery scenarios of single scanning (VS1) and 8- times volumetric rescanning (VS8). The dosimetric impact of the 4DDC scenarios and number of phases per cycle (8/10) was studied. Results Dose distributions (VS1) for example liver and lung cases are depicted in Fig1. Pronounced differences between PSs and PSm show the impact of motion irregularity, as PSs only considers a single reference 4DCT. Fig2 compares the dosimetric indices of the different 4DDC scenarios and the influence of the reference 4DCT selection (boxplots). Depending on the choice of reference 4DCT, PSs8 can underestimate motion effects and therefore misestimate D 5% -D 95% and V 95% by up to - 8.9% and +16.9% compared to SWm results. On average, V 95% is overestimated by 2.2% (PSs8) for interplay effect evaluations (VS1). Moreover, the influence of the reference 4DCT selection on the dosimetric indices (box sizes) are comparable for VS1 and VS8 and liver and lung anatomies. However, PS results for rescanning are more similar to the SWm results (mean difference V 95% SWm-PSs8: -0.2%). Furthermore, dosimetric differences due to either the different temporal resolution for PS and SWm or different numbers of phases (8 or 10) are only marginal.