S1780
Physics - Dose prediction/calculation, optimisation and applications for particle therapy planning
ESTRO 2026
power ratio (SPR) for a common cranial titanium implant improves proton dose calculation accuracy using the pencil beam algorithm in our clinical treatment planning system (TPS). Material/Methods:
An 18 cm diameter cylindrical solid water phantom was used as test geometry. Physical samples and CAD drawings of a CranioFix implant (B. Braun Aesculap AG, Tuttlingen, Germany) were obtained. The phantom was CT scanned with a 1 cm spindle CranioFix on top, using a clinical single-energy CT protocol calibrated for proton therapy via a look-up table (LUT) to SPR, and images were reconstructed at standard and high resolution. To assess the CT-representation influence on calculated dose, a virtual ground-truth geometry was created with the CAD-drawing representing the implant. The different implant representations are shown in Fig. 1. Three single-beam proton pencil beam scanning (PBS) plans were generated in Eclipse TPS (Varian, Siemens Healthineers, USA) and calculated with the pencil beam algorithm for the ground-truth and CT-based geometries. Additional calculations were performed after manually contouring the implant and overriding to the theoretical titanium SPR (3.2) in the CT-based geometries. TOPAS Monte Carlo simulations in the ground-truth geometry served as reference. Dose distributions behind the CranioFix were compared in terms of end-of-range (R80 depth) and dose uniformity. Results: Eclipse consistently underestimated end-of-range behind the CranioFix compared to TOPAS, with the smallest deviation in the ground-truth geometry, followed by CT-based SPR from the LUT, and largest with manual SPR override for both CT resolutions. For one example plan (see depth doses and transversal profiles in Fig. 2), Eclipse R80 differences to TOPAS were 2 mm in ground-truth, and 4.6 mm (LUT SPR) and 6.4 mm (overridden SPR) in standard-resolution CT geometry. TOPAS revealed localized dose drops some centimetres behind the CranioFix, not correctly reproduced by Eclipse in any case (see transversal
References: [1] Babier A, Mahmood R, Zhang B, et al. OpenKBP- Opt: an international and reproducible evaluation of 76 knowledge-based planning pipelines. Phys Med Biol. 2022;67(18):185012. [2] Podell D, English Z, Lacey K, et al. SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis. In: Kim B, Yue Y, Chaudhuri S, Fragkiadaki K, Khan M, Sun Y, eds. International Conference on Representation Learning. Vol 2024. 2024:1862-1874. [3] Babier A, Zhang B, Mahmood R, et al. OpenKBP: The open- access knowledge-based planning grand challenge and dataset. Medical Physics. 2021;48(9):5549-5561. Keywords: 3D Dose Modeling, Standardization, Deep Learning On proton dose accuracy: does manual override of stopping-power ratios for small titanium objects enhance pencil beam algorithm performance? Jens Zimmerman 1,2 , Åsa Carlsson Tedgren 1,2 , Gavin Poludniowski 3,4 1 Radiotherapy physics and engineering, Karolinska University Hospital, Stockholm, Sweden. 2 Department of Oncology-Pathology, Karolinska Institutet, Stockholm, Sweden. 3 Department of Nuclear Medicine and Medical Physics, Karolinska University Hospital, Stockholm, Sweden. 4 Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Stockholm, Sweden Digital Poster Highlight 4512
Purpose/Objective: To evaluate whether manually overriding the stopping-
Made with FlippingBook - Share PDF online