S1722
Physics - Dose prediction/calculation, optimisation and applications for particle therapy planning
ESTRO 2026
sparing1. However, variations in patient anatomy and tissue density along beam paths may compromise plan robustness. This study aimed to (i) evaluate PAT plan robustness against treatment uncertainties and anatomical variations for head-and-neck cancer (HNC) patients, and (ii) compare it against VMAT and IMPT. Material/Methods: Twenty HNC patients were included, selected for IMPT following the Dutch model-based selection. Clinical VMAT and IMPT plans were available from the plan comparison. Prescription doses (Dpres) were 70 Gy/GyRBE for CTV70.00 and 54.25 Gy/GyRBE for CTV54.25. A 3mm PTV margin was applied for VMAT, while 3mm/3% setup and range robustness settings were used for IMPT and PAT. Dose was prescribed to PTV-D98% ≥ 95%Dpres for VMAT and to the voxel-wise minimum dose across 28 robustness evaluation scenarios: VWmin-D98%,CTV ≥ 94%Dpres, for IMPT and PAT. Polynomial Chaos Expansion (PCE) was used to generate plan-specific voxel-dose models, enabling probabilistic robustness evaluations of 10,000 fractionated treatments per plan. To assess plan robustness, PCE models were built for the nominal plan and six repeat CTs to accumulate dose distributions across all sampled treatments. Probabilistic adequate CTV dose was defined as D99.8%,CTV ≥ 95%Dpres for 90% of treatments, consistent with PTV-based VMAT and robust IMPT criteria2. Probabilistic OAR doses and NTCPs were evaluated for 95% of treatments. No plan adaptations were assumed. Results: Figure 1 shows the probabilistic accumulated D99.8% doses for both CTVs. Table I summarizes additional metrics. CTV dose reductions were greater for PAT than for VMAT or IMPT. For CTV54.25, PAT achieved a lower D99.8% of 49.2 GyRBE (41.6–52.4) on average at 90% probability, compared with the goal (51.5 GyRBE), VMAT (50.7 Gy; 31.7–54.6), and IMPT (50.6 GyRBE; 47.5–54.8). For CTV70.00, D99.8% was 65.1 GyRBE (58.5–67.7) for PAT, below the goal (66.5 GyRBE) and comparable to 65.7 Gy (52.0–67.8) for VMAT but lower than 66.2 GyRBE (62.3–67.8) for IMPT. Patient 1 showed lower robustness for VMAT due to OAR sparing. However, PAT lead to the lowest probabilistic OAR Dmean of 27.7 GyRBE (7.8–50.0) on average compared to 40.9 Gy (20.5–60.1) for VMAT and 33.1 GyRBE (10.6–55.3) for IMPT, leading to lower NTCP values for xerostomia and dysphagia.
References: [1] Langendijk et al. 2021,DOI: 10.14338/IJPT-20- 00089.1 Keywords: deep learning, IMPT, dose mimicking
Digital Poster Highlight 1769
The price of robustness vs OAR sparing in proton arc therapy: A probabilistic robustness evaluation against VMAT and IMPT for Head-and-Neck Cancer Jesus Rojo-Santiago 1,2 , Bas A de Jong 3 , Stefan Both 3 , Erik W Korevaar 3 , Mischa S Hoogeman 1,2 1 Radiotherapy, ErasmusMC Cancer Institute, Rotterdam, Netherlands. 2 Medical Physics & Informatics, HollandPTC, Delft, Netherlands. 3 Radiation Oncology, University Medical Center Groningen, Groningen, Netherlands Purpose/Objective: Compared to IMPT, proton arc therapy (PAT) offers more optimization flexibility as more gantry angles are used, enabling improved dose conformity and OAR
Made with FlippingBook - Share PDF online