S1764
Physics - Dose prediction/calculation, optimisation and applications for particle therapy planning
ESTRO 2026
References: [1] Neishabouri A, Bauer J, Abdollahi A, Debus J, Mairani A. Real-time adaptive proton therapy: An AI- based spatio-temporal mono-energetic dose calculation model (CC-LSTM). Comput Biol Med. 2025 Apr;188:109777. doi: 10.1016/j.compbiomed.2025.109777. Epub 2025 Feb 12[2] Pastor-Serrano, O., & Perkó, Z. (2022). Millisecond speed deep learning based proton dose calculation with Monte Carlo accuracy. Physics in Medicine & Biology, 67(10), 105006. doi:10.1088/1361- 6560/ac692e[3] Radonic D, Xiao F, Wahl N, Voss L, Neishabouri A, Delopoulos N, Marschner S, Corradini S, Belka C, Dedes G, Kurz C, Landry G. Proton dose calculation with LSTM networks in presence of a magnetic field. Phys Med Biol. 2024 Oct 21;69(21). doi: 10.1088/1361-6560/ad7f1e. Keywords: AI, proton therapy, dose calculation, Performance of single- and dual-energy CT techniques for proton range prediction Isidora Sofia Muñoz Hernandez 1,2 , Torbjörn Näsmark 3 , Christina Vallhagen Dahlgren 2 , Jonas Andersson 3 , Alexandru Dasu 1,2 1 Cancer Precision Medicine, Department of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden. 2 Department of Medical physics, The Skandion Clinic, Uppsala, Sweden. 3 Department of Intervention and Diagnostics, Umeå University, Umeå, Sweden Digital Poster 3972 Purpose/Objective: Proton therapy relies on accurate prediction of proton range in tissues to ensure that the prescribed dose conforms precisely to the target volume while sparing surrounding healthy tissues. The proton stopping power ratio (SPR) derived from computed tomography (CT) images is a key factor in determining proton range. Single-energy CT (SECT) is typically used for this, while dual-energy CT (DECT) acquiring images at two different energy spectra, provides material-specific information for estimation of SPR with a potential to reduce range uncertainties. This study aims to evaluate and compare several CT acquisition technologies for their ability to estimate SPR and water-equivalent thickness (WET), to identify feasible approaches for integration into proton therapy planning. Material/Methods: Phantom measurements were performed using the Sun Nuclear Advanced Electron Density (AED) phantom in both head and body setups. Scans were acquired at three clinical sites using Siemens CT systems implementing different DECT technologies:
twin-scan, twin-beam and dual-source. Scanning protocols were harmonized to ensure comparable CTDI dose levels (20/50 mGy for SECT and 10/40 mGy for DECT, in body/head phantoms). SPRs were estimated with the EPTN consensus method for SECT scans[1], Siemens DirectSPR and a virtual monoenergetic image (VMI)-based method[2] for DECT scans. SPR predictions were compared to calculated values for the phantom materials, and their impact on
proton range was quantified as WET for three simulated anatomical sites: brain, lung, and prostate[3]. Results:
The residual root mean square error (RMSE) of SPR for all tissue types showed best agreement for the EPTN consensus method (1.1%), followed by the VMI-based method (1.8%) and DirectSPR (2.4%). Among scanner technology, dual-source achieved the lowest residual RMSE(1.7%) for DECT conversions, followed closely by twin-scan (1.7%) and twin-beam (2.0%). In contrast, the analysis of WET accuracy showed superior performance for the VMI method (brain 0.2%, lung 0.8%, prostate 0.1%), followed by the EPTN consensus method (brain 0.4%, lung 0.7%, prostate 0.3%) and DirectSPR (brain 0.4%, lung 1.2%, prostate 0.7%). The dual-source scanner provided the lowest deviations (brain 0.2%, lung 0.8%, prostate 0.2%), followed by twin-scan (brain 0.4%, lung 0.7%, prostate 0.4%) and twin-beam (brain 0.5%, lung 1.4%, prostate 0.8%). Conclusion: SECT-based EPTN showed slightly better SPR accuracy for the chosen tissue substitutes, whereas VMI-based DECT improved WET accuracy across sites. These findings suggest SECT is robust for calibration, whereas DECT, particularly the VMI method, has potential to enhance proton range prediction in heterogeneous or complex tissues. References: [1] N. Peters et al., “Consensus guide on CT-based prediction of stopping-power ratio using a Hounsfield look-up table for proton therapy,” Radiotherapy and Oncology, vol. 184, Jul. 2023, doi: 10.1016/j.radonc.2023.109675.[2] T. Näsmark and J. Andersson, “Proton stopping power prediction based on dual-energy CT-generated virtual monoenergetic images,” Med Phys, vol. 48, no. 9, pp. 5232–5243, Sep. 2021, doi: 10.1002/mp.15066.[3] N. Peters et al., “Experimental assessment of inter-centre variation in stopping-power and range prediction in particle therapy,” Radiotherapy and Oncology, vol. 163, pp. 7– 13, Oct. 2021, doi: 10.1016/j.radonc.2021.07.019. Keywords: Proton therapy, SPR prediction, CT calibration
Made with FlippingBook - Share PDF online