S1755
Physics - Dose prediction/calculation, optimisation and applications for particle therapy planning
ESTRO 2026
machine learning (ML) models of Random Forest (RF), Multi-Layer Perceptron (MLP) and Gradient Boosting Regression.
Conclusion: Optimization of MRIgRT planning parameters and a clear understanding of PTV geometric complexity enable a balanced trade-off between plan quality and efficiency. Reducing the number of unnecessary small- area or low-MU segments enhances delivery robustness and workflow efficiency, consistently with the recommendations outlined in AAPM TG-155[1,2]. References: [1]Ezzell GA, Galvin JM, Low D, Palta JR, Rosen I, Sharpe MB, Xia P, Xiao Y, Xing L, Yu CX; IMRT subcommitte; AAPM Radiation Therapy committee. Guidance document on delivery, treatment planning, and clinical implementation of IMRT: report of the IMRT Subcommittee of the AAPM Radiation Therapy Committee. Med Phys. 2003 Aug;30(8):2089-115. doi: 10.1118/1.1591194. PMID: 12945975.[2] Pappas, E., Georg, D., McEwen, M., et al. “AAPM Task Group 155 report: Megavoltage photon beam dosimetry in small fields and non-equilibrium conditions.” Medical Physics, 48(1): e171–e210, 2021. Keywords: Workflow Efficiency,PTV Complexity Analysis,MRIgRT Robustness of Proton Arc Therapy Plans Against Anatomical Changes in Nasopharyngeal Carcinoma Feng Yang radiotherapy, sichuan cancer hospital, chengdu, China. College of Nuclear Technology and Automation Engineering, Chengdu University of Technology, chengdu, China Purpose/Objective: The introduction of Proton Arc Therapy (PAT) represents a significant advancement in the field of radiation oncology. This study aims to evaluate and Digital Poster 3605 compare the robustness of PAT and Intensity Modulated Proton Therapy (IMPT) in managing anatomical variations during radiotherapy for nasopharyngeal carcinoma (NPC), thereby providing
Results: IMRT Efficiency and Accuracy Goal showed a strong inverse relationship with total treatment time, reducing beam-on and gantry/MLC time without compromising target coverage. Conversely, plans with higher gantry and MLC times demonstrated a greater number of low-MU (<7) and small-field (<4 cm ² ) segments, indicating increased modulation complexity. PCA revealed MU per fraction, total segments, and beam-on time as key contributors to plan variance, while the ML models identified the same parameters as primary predictors of efficiency with the Coefficient of Determination R2 equally 0.97. Across all plans, increasing PTV geometric complexity was correlated with longer treatment times and a higher prevalence of MU per fraction, low-MU (<7) and small-field (<4 cm ² ) segments, indicating greater modulation demands and reduced delivery efficiency. Crucially, the PTV complexity was the only variable that created distinct clusters in the PCA plot, as illustrated
in Figure 2, confirming its role as the primary separator of treatment planning parameters’ optimization.Figure2-PCA Scatter plot
Made with FlippingBook - Share PDF online