S1991
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
Conclusion: the dosimetric model by Upadhyay et al.1 can be reproduced. Our data confirm that the risk of ≥ G3 RN is ≤ 6% if V23 to the total irradiated volume is < 20 cc. Systemic in fl ammation, measured by NLR, was associated with overall survival after linac-based SFRT in patients with brain metastases. References: Upadhyay R, Ayan AS, Jain S, Klamer BG, Perlow HK, Zoller W, Blakaj DM, Beyer S, Grecula J, Arnett A, Thomas E, Chakravarti A, Raval RR, Palmer JD. Dose- Volume Tolerance of the Brain and Predictors of Radiation Necrosis After 3-Fraction Radiosurgery for Brain Metastases: A Large Single-Institutional Analysis. Int J Radiation Oncol Biol Phys, Vol.118, No. 1, pp. 275- 284, 2024 Keywords: brain, stereotactic, radionecrosis Inter-Observer Variability in Dose Accumulation for Lung Reirradiation: A Multicentre Comparison of Clinical and Automated Approaches Pauline Dupuis 1 , Chelmis Muthoni Thiong'o 2 , Anthony Alexis 3 , Charbel Kassis 4 , Thomas Baudier 5,6 , Myriam Ayadi 1 , Marie-Pierre Sunyach 7 , Eliana Vasquez Osorio 2 1 Medical Physics Department, Centre Léon Bérard, Lyon, France. 2 Division of Cancer Sciences, The University of Manchester, Manchester, United Kingdom. 3 Medical Physics Department, Institut de Cancérologie de l'Ouest, Saint Herblain, France. 4 Radiation Oncology Department, Institut de Cancérologie de Seine et Marne, Jossigny, France. 5 CREATIS UMR 5220, INSA - INSERM - CNRS, Lyon, France. 6 Tomoradio Team, Centre Léon Bérard, Lyon, France. 7 Radiation Oncology Department, Centre Léon Bérard, Lyon, France Digital Poster 4299 Purpose/Objective: Accurate assessment of cumulative doses to Organs of interest (OOI) is key for clinical decision-making in reirradiation (reRT). Due to anatomical changes occurring between treatments, deformable image registration (DIR) is often used [1]. Multiple DIR solutions can impact dose accumulation [2], therefore carefully DIR evaluation and uncertainty assessment is crucial [3]. This study compared dose accumulation results from three clinical teams and one automated pipeline to assess inter-observer variability in a lung reRT case. Material/Methods: Anonymised DICOM data from a lung cancer Type 1 reRT case showing major anatomical changes (Figure 1) were shared with four teams: three clinical and one research. All teams converted doses to EQD2Gy using alpha/beta=2Gy for the spinal canal and 3Gy for other
OOI. Clinical teams performed their best achievable DIR with their clinical software (MIM v7.3.5, RayStation v2024A SP1, Velocity v4.1) and conducted thorough evaluation. The research team run an automated RayStation v11BR script performing eight registrations (varying similarity measure, resolution levels and smoothing sigma) for two set-ups: intensity-based and hybrid (using all OOI as controlling structures) DIRs. For each OOI, only DIRs achieving mean distance to agreement (mDTA) ≤ 3mm were recorded. All teams recorded cumulative D0.035cc, D0.1cc, and D5cc within 2-mm Planning organ at Risk Volumes (PRV). Results: All clinical teams performed contour-guided DIR, either hybrid (MIM) or contour-based only (Velocity and RayStation). Figure 2 shows cumulative estimates. Largest manual estimate differences ( Δ ) were within 2Gy for the contralateral main bronchus and spinal canal for all estimates, consistent with the automated results. For the heart, differences varied ( Δ D0.035cc=7.3 Gy; Δ D1cc=1.9 Gy; Δ D5cc=1.3 Gy), mostly aligning with intensity-based DIR results. The oesophagus showed large discrepancies ( Δ D0.035cc=12.1Gy, Δ D1cc=6.4Gy, Δ D5cc=3.5Gy), consistent with the automated results. Manual estimates for the aorta were consistent ( Δ D0.035cc=1.1Gy, Δ D1cc=2.7Gy, Δ D5cc=2.1Gy), however automated results were not recorded due to inconsistent delineation lengths, Figure 1.For the ipsilateral main bronchus, Δ D were within 4Gy. This variation was consistent with hybrid-DIR results for D0.035cc, and intensity-based DIR results for D5cc, while Δ D1cc between manual and automated estimates reached up to 17Gy. Differences were observed for the lobar bronchus ( Δ D0.035cc=3.2Gy, Δ D1cc=2.1Gy, Δ D5cc=3.2Gy) and sublobar bronchus ( Δ D0.035cc=1Gy, Δ D1cc=5.1Gy). For these structures only hybrid DIRs produced plausible results, producing estimates slightly below manual estimates.
Made with FlippingBook - Share PDF online