S1917
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
automated AMBRA plans were generated with a robustly-tuned optimization template using VITAT technique [2] for L-WBI clinical patients (CLIN), mimicking optimal TF plans.AMBRA and CLIN plan performances were compared across 83 previously treated patients enrolled from participating institutions. A blinded qualitative assessment was performed at six institutions, involving at least two MPs and two ROs per site. Each center independently evaluated only its own locally treated cases. Plans were rated using a 5-point Likert scale (1 = poor, 5 = optimal) grouped into three categories: clinically unacceptable (1–2), acceptable (3–4), and optimal (5). Results: The study included 166 plans, 25 evaluators, and 694 total plan evaluations. AMBRA plans achieved dosimetric equivalence to CLIN plans in target coverage and OAR sparing, with minor center-specific differences (INST1) linked to local heart-sparing constraints. Mean qualitative raw scores were similar for CLIN and AMBRA plans (Figure 1), 3.81 ± 1.14 vs. 3.68 ± 1.17, though the difference was statistically significant in favour of CLIN (p = 0.0039); once pooled by categories, differences were not significant. Considerable inter-observer variability was observed, with comparable intra-institutional variability. AMBRA plans demonstrated higher inter-observer agreement than manual CLIN plans, as confirmed by Cohen’s kappa analysis (Figure 2). No significant differences emerged between internal and external institutions in either raw or categorical scoring.
Conclusion: Large-scale automatic AMBRA planning achieved clinical evaluations not inferior in term of score categories to manually generated plans across multiple institutions and a large number of evaluators. Higher inter-observer agreement was observed for AMBRA compared with manually generated plans. These results support its potential as a reliable standardized tool for multi-centric large - scale plan optimization. Acknowledgements: current study was supported by a grant from AIRC (IG23150) References: [1] Tudda A et al. Phys Med. 2025 Feb;130:104889. doi: 10.1016/j.ejmp.2024.104889. [2] Esposito PG et al. Phys Med. 2020 Sep;77:160-168. doi: 10.1016/j.ejmp.2020.08.011.. Keywords: breast cancer , automatic plan, plan evaluation Mini-Oral 3120 Validation of Dual-Energy CT-Based Dose Calculation for Photon Radiotherapy of Head and Neck Cancer Luisa Tischendorf 1 , Gary Razinskas 1 , Andrea Wittig- Sauerwein 2 , Anne Richter 2 1 Department of Radiation Oncology, University Hospital Würzburg, Würzburg, Germany. 2 Department, University, Würzburg, Germany Purpose/Objective: Dual-Energy CT (DECT) enhances soft-tissue contrast and tumor delineation in head and neck (H&N) imaging [1]. While its dosimetric potential is established for protons [2], its use in photon radiotherapy remains limited. Conventional treatment planning systems (TPS) rely on single-energy CT (SECT)- based calibration curves, complicating DECT
Made with FlippingBook - Share PDF online