S638
Clinical – Head & neck
ESTRO 2026
Digital Poster 3828 Inter-Beam Setup Uncertainty Robustness Evaluation of IMPT and Proton Arc Therapy for HN Cancer: Independent-Beam versus Universal Robust Optimization Bin Yang, Yeung Sum Wong, Hui Geng, Wai Wang Lam, Wing Ki Claudia Chan, Shu Ting Hung, Ting Chuan Li, Ka Keung Tang, Chin Chak Ho, Kin Yin Cheung, Siu Ki Yu Medical Physics Department, Hong Kong Sanatorium & Hospital, Hong Kong, Hong Kong Purpose/Objective: Intensity-modulated proton therapy (IMPT) with multifield optimization and Proton Arc Therapy (PAT) offer superior dose conformity and OAR sparing for head and neck (HN) cancers. However, concerns remain regarding independent beam positioning errors between fields, which are not fully captured by conventional universal robustness evaluation (U-RE). This study evaluates plan robustness of IMPT with independent-beam robust optimization (IB-RO) versus PAT with universal robust optimization (U-RO) by considering inter-beam setup uncertainties. Material/Methods: 10 HN patients were retrospectively planned in RayStation-2024A using three-field IMPT with IB-RO and two half-arc PAT (360° arc length, 30 beam angles) with U-RO. Both techniques applied 3 mm setup and 3% density uncertainties during robust optimization according to institutional guidelines. To evaluate inter- beam setup uncertainty effects, an in-house script generated independent-beam robustness evaluation (IB-RE) scenarios by randomly combining different setup errors per beam while maintaining consistent inter-beam density uncertainty. Sixty IB-RE scenarios (30 per density direction) were sampled for setup uncertainties of 1-3 mm. Target dose metrics (CTV D95% ≥ 100% and D1% ≤ 110%) and robustness passing rates were compared between techniques under both U-RE and IB-RE methods using Wilcoxon test, with p<0.05 as statistically significant.
convention). These automatically-detected geometric changes were shown to be reliable surrogates for dosimetric impact. The analysis confirmed these changes led to significant dosimetric failures in the non-adapted (CT1_Hybrid) plans (Fig. 2C). Compared to the adapted CT1_Plan, the non-adapted plan had significantly higher "Fail" rates for key goals, including CTV_High V97% (p=0.0018) and both Parotid_L/R V30Gy (p=0.0117 / p=0.0156).
Conclusion: While a non-adaptive strategy leads to known dosimetric failures, the primary clinical bottleneck is the resource-intensive process of identifying which patients require replanning. Our automated workflow, leveraging AI-contouring and scripted analysis, provides a practical and efficient solution. It successfully quantifies geometric changes and provides a reliable, low-effort, and consistent tool for adaptive decision support, enabling clinics to efficiently flag high-priority patients and implement ART more broadly. Keywords: NPC, Decision Support, Adaptive
Results: Under U-RE, IMPT and PAT demonstrated clinically equivalent robustness with worst-case CTV D95% of
Made with FlippingBook - Share PDF online