ESTRO 2025 - Abstract Book

S3229

Physics - Intra-fraction motion management and real-time adaptive radiotherapy

ESTRO 2025

References: 1.

Kishan, A.U. et al . JAMA Oncology, 2023. 9 : p. 365-373. Zhang, P. et al . Int J Radiat Oncol Biol Phys, 2024. Mylonas, A. et al. Medical Physics, 2019. 46 (5): p. 2286-2297. Hurkmans, C. et al. Radiother Oncol, 2024. 197 : p. 110345. Willoughby, T. et al. Med Phys, 2012. 39 (4): p. 1728-47.

2. 3. 4. 5. 6. 7. 8.

Ng, J.A. et al. Med Phys, 2014. 41 (11): p. 111712.

Keall, P. et al . Int J Radiat Oncol Biol Phys, 2020. 107 (3): p. 530-538. Sengupta, C. et al . Radiother Oncol, 2024. 190 : p. 110031.

9. S Alnaghy, C Sengupta, K Makhija. 6 DoF Robotic Motion Phantom . https://github.com/Image-X-Institute/6 DoF-Robotic-Motion-Phantom. 10. Worm, E.S. et al . Med Phys, 2023. 50 (6): p. 3289-3298. 11. Lim Joon, D. et al . Sci Rep, 2021. 11 (1): p. 8931.

1459

Digital Poster Speed meets precision: Systematic optimization for the rapid generation of high-quality online-adapted proton therapy plans in a commercial TPS Lena Nenoff 1,2,3 , Annabell Eberhardt 3 , Rebecca Bütof 1,2,3 , Albin Fredriksson 4 , Stefan Menkel 3 , Virginia Gambetta 1,2 , Esther GC Troost 1,2,3 , Christian Richter 1,2,3 , Kristin Stützer 1,2 1 OncoRay – National Center for Radiation Research in Oncology, Faculty of Medicine, Dresden, Germany. 2 Helmholtz-Zentrum Dresden - Rossendorf, Institute of Radiooncology – OncoRay, Dresden, Germany. 3 Department of Radiotherapy and Radiation Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, TUD Dresden University of Technology, Dresden, Germany. 4 RaySearch Laboratories AB, Research Department, Stockholm, Sweden Purpose/Objective: A rapid treatment plan optimization is essential for online-adaptive proton therapy (OAPT). For our envisioned OAPT realization, we systematically investigated the impact of altered optimization settings on both the quality of the optimized dose distribution and the optimization time for non-small cell lung cancer (NSCLC) patients. Material/Methods: From a randomized clinical trial 1 , we selected ten advanced NSCLC patients who underwent 1-3 offline adaptations. An experienced radiation oncologist scored automated segmentation results. Accordingly, we selected deep learning-based organ-at-risk (OAR) contours for the planning and 5-7 repeat CTs (pCT/rCTs), and clinical target volumes (CTVs) rigidly or deformably propagated to those rCTs without offline adaptation. We generated robust PT plans with a clinical objectives/constraints template and patient-individually adjusted values in RayStation23B-R. A medical physicist confirmed proper quality of these reference plans before starting systematic impact analyses for varied optimization settings on the pCT. OAR and CTV quality scores were employed to assess patient-averaged alterations in all clinical goal-related dose-volume-histogram parameters relative to the reference plans. The optimization setting with best time-quality trade-off was applied for OAPT re-optimization on all rCTs. Results: The optimization time depended mainly on the maximum number of iterations, which we already reduced from 200 to 100 in the reference plan, resulting in a median time saving of 2:25min and only minimal compromise on OAR sparing (Fig.1a). Reducing relative spot spacing (from 1 to 0.7) improved quality scores but added a median 2:19min, which was not justified. Reducing the energy layer spacing (from 1 to 0.7) or the number of range uncertainty scenarios (from 3 to 2) had no impact on quality, but the latter saved a median 0:57min. Replacing two maximum