ESTRO 2025 - Abstract Book

S2767

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2025

Keywords: plan aperture modulation, modulation, complexity

References: [1] D. A. Low et al. Dosimetry tools and techniques for IMRT., Med Phys 38 (3) (2011) 1313–1338. [2] M. Miften et al. Med Phys 45 (2018) e53–e83. [3] M. W. Geurts et al. J Appl Clin Med Phys 23 (9) (2022) e13641. [5] S. Chiavassa et al.Br J Radiol 92 (2019) 20190270. [6] M. Antoine et al. Phys Med 64 (2019) 98–108. [7] E. Kamperis et al. Int J Radiat Oncol Biol Phys 106 182–184. [8] arXiv:2410.02632v2 [physics.med-ph]. https://doi.org/10.48550/arXiv.2410.02632 (PrePrint)

1434

Poster Discussion A lexicographic optimization-based treatment planning system for automated VMAT-TBI plans Giulia Zorzi 1,2 , Angelo F Monti 1 , Cristina De Mattia 1 , Maria G Brambilla 1 , Roberto Pellegrini 3 , Edoardo Salmeri 3 , Virginia M Arienti 4 , Barbara Orlandini Bertarini 4 , Mauro F Palazzi 4 , Paola E Colombo 1 1 Medical Physics, ASST GOM Niguarda, Milan, Italy. 2 Specialization School in Medical Physics, University of Milan, Milan, Italy. 3 Medical Affairs, Elekta AB, Stockholm, Sweden. 4 Radiotherapy, ASST GOM Niguarda, Milan, Italy Purpose/Objective: VMAT-TBI planning is complex and time-consuming. Automatic-planning aims to reduce manual operator time and inter-operator variability without compromising treatment quality. This study evaluates the performance of “mCycle”, the Monaco Research version ( 6.09.00, Elekta) of Elekta One Autoplanning commercial version, for VMAT TBI. Material/Methods: Fifteen patients treated with VMAT-TBI (12Gy/6frs) between March 2023 and July 2024 were retrospectively selected and re-planned using mCycle. Constraints and objectives were sequentially optimized through multi-criteria optimization (MCO) using a priority Wish-List (WL). Two mCycle WLs were tuned on a subset of 3 patients and tested on all patients: WL1 aimed to reproduce manual plans, in which OARs are R/L lungs, and WL2, including additional OARs (lenses, kidneys, spinal cord). The effect of the collimator orientation (0°-90°) was analyzed too. The standard clinical Hyperion Sequencer was compared with the new Pseudo Gradient Descent (PGD) Sequencer. Manual plans (MP) were compared with automatic plans (AP1 generated with WL1, AP2 generated with WL2) in terms of dose volume constraints, plan complexity (MU, Modulation Degree(MD)), and planning time. Based on the Shapiro-Wilk normality test, statistical significance was assessed by performing either the parametric t-test or the non-parametric Wilcoxon signed-rank test (significance level at p<0.05). Results: All automatic plans were clinically acceptable without any adjustment. The effect of collimator orientation (0°-90°) showed no significant differences. PGD outperformed Hyperion in dosimetric terms, with 3-6 hours lower calculation time. Target coverage was comparable across plans, with median V 95% values of 96.9 (MP), 96.0 (AP1; p=0.15), and 95.8 (AP2; p=0.06). Lung doses showed significant sparing in AP1 and AP2, with median lung (Right-Left) mean doses and D 1% of 9.3-9.2Gy and 12.5-12.5Gy for MP, 8.5-8.3Gy and 12-12.1Gy for AP1, and 8.7-8.9Gy and 12 12Gy for AP2 respectively (p<0.01 for all the comparisons). Additional OARs in AP2 were better spared than in MP and met all the planning criteria.Median MD values were 5.5 (MP), 3.9 (AP1; p<0.01), and 4.1 (AP2; p<0.01), while median MU values were 5289 (MP), 4822 (AP1; p<0.01), and 5171 (AP2; p=0.41) (Figure1). Planning time (hh:mm) decreased, with manual times of 3:25-6:48 and automatic times of 1:40-2:18 (p<0.01).