S1859
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
conventional and adaptive radiotherapy. Material/Methods:
achieved the highest overall composite score (0.871) for the Central lesion, standing out with excellent dose homogeneity (HI ≈ 0.06°) and the lowest MU (1368). For peripheral lesions, the more efficient 2-arc plans were found to be most advantageous: 2Arc_R1-L2 (FL,0.867), 2Arc_M-L2 (BL,0.854), 2Arc_R1-M (BR,0.851), and 2Arc_R1-R2 (FR,0.826). Conclusion: Optimal beam-angle selection in the HyperArc™ technique is fundamentally a function of tumor location. The 3 arc configuration (R1-R2-L1) offered a superior balance of PTV coverage, Normal Brain sparing, and treatment efficiency, justifying its use for central lesions, while 2-arc plans remained the most effective and cost-efficient choice for peripheral lesions. These findings provide practical clinical guidance for arc selection in HyperArc planning. References: 1. Cho B, Kim JI, Lee J, Park JM, Chung JB, Kim W, et al. HyperArc VMAT planning for single and multiple brain metastases stereotactic radiosurgery: a new treatment planning approach. Radiat Oncol. 2018 Jan 29;13(1):16.2. Lin MH, Zhang X, Huang BT, Jiang X. A data-driven approach to optimal beam/arc angle selection for liver stereotactic body radiation therapy treatment planning. Phys Med. 2021 Dec;8(4):618- 26. 3.Yu J, Huang V, Liu D, Ma L, Li H. Comparison of beam angle selection heuristics using the glioblastoma case. Res Gate. 2015 May 20;8(2):123-30. Keywords: Brain Tumors, Weighted Scoring System, HyperArc™ Digital Poster 1848 Addressing target and contour propagation uncertainty using probabilistic treatment planning Ilamparithi Balasubramanian 1 , Luciano Rivetti 1 , Andrej Studen 1,2 , Manju Sharma 3 , Robert Jeraj 1,2 1 Faculty of Mathematics and Physics, University of Ljubljana, Ljubljana, Slovenia. 2 Jo ž ef Stefan Institute, Jo ž ef Stefan Institute, Ljubljana, Slovenia. 3 Department of Radiation Oncology, University of San Francisco, San Francisco CA, USA
A head and neck (HN) dataset comprising 4 patients with oropharyngeal cancer was used in the study. Probabilistic Clinical Target Maps (CTMs) (Fig. 1a) were first generated using a Gaussian-based tumour infiltration model to represent microscopic disease extension beyond the gross tumour volume (2). These CTMs, along with OAR masks, were subsequently propagated across fractions using a deep-learning deformable image registration (DL-DIR) model (7). The DL-DIR propagation produced voxel-wise probability maps capturing inter-fraction anatomical variation.These probability maps were incorporated into the optimisation objective function as probability- weighted voxel values (1,2). Robust optimisation incorporated setup uncertainties across 13 scenarios, including the nominal and 12 error scenarios with ±1 and ±2 mm shifts in all three dimensions (3). Optimisation was executed using OpenTPS (4). All frameworks had the same beam geometries.For dosimetric plan evaluation, the Statistically Sound Dosimetric Selection (SSDS) (Fig. 1b) method was applied across the high-dimensional scenario space (5,6). Multiple realisations within the space were sampled at confidence levels of 0.9, 0.45, and 0.2, and were normalised to have the same target coverage across frameworks. The dose distributions were then compared.
Results: Preliminary analyses show that the probabilistic framework enhances OAR sparing in comparison to classical and adaptive whilst maintaining the same target coverage. Evaluation indicated lower mean dose to parotids, submandibular glands, and constrictor muscle under the probabilistic plan (Fig. 2). The percentage of fractions meeting constraints for each framework [Classical, Adaptive, Probabilistic] are: left sub-mandibular gland [0, 63.6, 27.3], right sub- mandibular gland [100, 100, 100], left parotid [36.7, 72.7, 100], right parotid [36.7, 100, 63.6], constrictor muscles [0, 100, 72.7]. Even with only the worst-case reported, probabilistic planning outperforms classical and approaches adaptive, even by exceeding it for the left parotid.
Purpose/Objective: Classical planning treats uncertainties
deterministically, producing plans that may lose efficacy under real-world variation (1). In contrast, Probabilistic planning provides a better trade-off between target and Organs-At-Risk (OARs). But questions remain about the benefit of probabilistic planning when multiple sources of uncertainty are involved (2). We implemented a probabilistic treatment planning framework that integrates uncertainty modelling to improve OAR sparing without compromising target coverage compared to
Made with FlippingBook - Share PDF online