S1981
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
and allowing more focus on complex and stereotactic treatment QA. Keywords: in-vivo, dosimetry, iQM
Digital Poster 4162 DVH prediction to assure treatment plan quality and consistency when changing practice Geert Wortel, Casper Carbaat, Georgios Tsekas, Marjan Admiraal, Roel Zinkstok, Tomas Janssen Department of Radiation Oncology, The Netherlands Cancer Institute, Amsterdam, Netherlands Purpose/Objective: Assessing the quality of radiotherapy treatment plans is a complex task, both for individual plans and in the context of practice changes. To support our clinical transition from Pinnacle (Philips) Autoplanning to Monaco (Elekta AB) manual planning and upcoming Autoplanning, we developed a plan QA tool based on a patient anatomy-specific dose prediction. Inter- planner variability is well known and such a transition likely affects different planners differently. Therefore, the analysis addresses not only average plan quality but also variability between planners. Material/Methods: This study was conducted for 70 Gy head and neck SIB plans. The model estimated the achievable mean dose to an organ-at-risk (OAR) based on the patient anatomy by splitting the OAR into different sub volumes according to their proximity to the high-dose PTV, low-dose PTV, or both. The dose in these regions was established using a linear fit to the fractional sub volumes and achieved mean doses in the training data.125 Clinical Pinnacle Auto-plans were available, of which 64 were used to train the model and 61 for evaluation. The model was next applied to 61 manually created Monaco plans. Differences between predicted and achieved dose were analyzed per OAR. Regimes where OAR sparing differed were identified using unpaired sample t-tests. Second, OAR sparing per planner was analyzed and quantified in terms of z- scores. Results: Differences between predicted and achieved dose per OAR and dataset are shown in Figure 1. The mean dose difference for the Pinnacle evaluation set is - 0.7±3.3 Gy.In the manual Monaco plans, suboptimal OAR sparing was observed for patients with a favorable anatomy for the submandibular glands, the oral cavity and the parotid glands, whereas improved OAR sparing is observed for high-risk constrictor muscles (all p<0.05).
The dose differences per OAR per planner shown in Figure 2 reveal inter-planner variation. The most significant variation is the superior gland sparing of planner 1 at the expense of the oral cavity (|z|>2).
Conclusion: A plan QA tool based on patient anatomy-specific DVH prediction can support the evaluation of radiotherapy treatment plan quality and consistency when changing practice. We used the tool to monitor plan quality following a TPS transition, found that not all OARs are optimally spared, and quantified inter-planner variability. Based on these findings, we are revising our class solution template, and will use this tool to detect individual outlier plans and to support future implementation of Monaco Autoplanning. Keywords: plan quality, TPS transition, planner variation
Made with FlippingBook - Share PDF online