ESTRO 2025 - Abstract Book

S3077

Physics - Inter-fraction motion management and offline adaptive radiotherapy

ESTRO 2025

Conclusion: The tattoo-based and SGRT setups perform comparably. Outliers are less pronounced in the SGRT setup for translations and pitch rotation, and in the tattoo-based setup for yaw and roll rotation. Unfortunately, clinical and technical problems impede implementation of a tattoo less workflow, requiring improvement of the SGRT system. Simultaneously, other IGRT patient groups need to be evaluated similarly.

Keywords: SGRT, tattooless, thorax

355

Digital Poster Retrospective automated daily synthetic CT-based dose calculations in a French randomized phase II study for prostate cancer

Jessica Prunaretty, Adrien Rouhana, Riou Olivier, Pascal Fenoglietto Radiotherapy, Institut du Cancer de Montpellier, Montpellier, France

Purpose/Objective: This study aimed to calculate the daily delivered dose in RCMI-GI trial using a new automated artificial intelligence based software: Adaptbox (Therapanacea). Material/Methods: The delivered dose was evaluated for 15 prostate cancer patients treated as part of the RCMIGI trial. The patients were included in the experimental arm using reduced PTV margin and a real-time tracking system (Calypso system, Varian). PTV2 was defined as a 3mm expansion around the prostate, while PTV1 was the combination of PTV2 and an anisotropic margin around the seminal vesicles (1 cm cranial, 1 cm lateral, 0.3 cm anterior and 0.5 cm posterior). The study was conducted with a schedule of 80 Gy in 40 fractions to the PTV2 and simultaneously 56 Gy to the PTV1. All treatment were delivered with Volumetric Modulated Arc therapy (VMAT) using 2 arcs and a TrueBeam Stx. Daily cone beam CT (CBCT) was employed for patient positioning and the Calypso system tracked the prostate during the irradiation. Each daily CBCT scan was imported into the Adaptbox software (v2.3.1). A synthetic CT was first generated from the CBCT images using a deep learning model. Organs at risk (OAR) and CTVs (prostate and seminal vesicles) were then delineated using AI and rigid propagation, respectively. Finally, the dose calculation was performed using a collapse cone algorithm. Variations (Δ) were defined as the difference between the dose delivered during each session and the initial planned dose. These variations were evaluated for the PTV coverage (ΔV 95% ) and for several dose metrics for the OARs (ΔV 76Gy , ΔV 70Gy and ΔD 50% for the rectum; ΔV 80Gy , ΔV 70Gy and ΔD 50% for the bladder). Results: A total of 567 fractions in 15 patients were analyzed. The median PTV coverage variation was below 0.6% and 0.4% for the PTV2 and PTV1, respectively (Figure 1). For the rectum, the highest median dose variations were observed for patient n°4 with 6.54%, 7.91% and 8.36% for ΔV 76Gy , ΔV 70Gy and ΔD 50% , respectively. Additionally, outlier results were identified for patient n°5 due to inaccuracies in rectum segmentation. Regarding the bladder, the highest median dose variations were observed for patient n°1 with 14.96% and 20.01% for ΔV 80Gy and ΔV 70Gy , respectively. ΔD 50% exceeded 10% for the patient n°4.