S2130
Physics - Inter-fraction motion management and daily adaptive radiotherapy
ESTRO 2026
treatment dose is translationally shifted to match the daily MRI. During an adapt-to-shape (ATS) workflow, treatment plans are made based on daily contours. Although ATS results in more anatomically accurate treatment plans than ATP, it requires a longer timeslot due to manual contour adjustments and plan re- optimization. At our institute, in-house developed auto-contouring software was recently introduced clinically. The software is based on combined deep learning and deformable image registration (DL-EVO), requiring almost no manual adjustments. Therefore, we investigated a fast ‘ATS-lite’ workflow for high-risk PCa patients using the DL-EVO software. Material/Methods An ATS-lite workflow (Figure 1) was developed using data from 20 high risk PCa patients treated with 20 x 3.1 Gy (focal GTV boost to 74 Gy) using ATP. The first and last two fractions (80 total) were selected for analysis. Daily patient contours were generated using the new auto-contouring framework (ATS-lite contours), and treatment plans were created on the contours without manual adjustments. A 5 mm CTV- PTV margin was applied. Next, a second set of contours was made, the physician-adjusted ‘ground truth’ contours. This resulted in two contour sets: 1. ATS-lite contours (auto-generated from daily MRI), 2. Ground truth contours (physician-adjusted, accurately matching the anatomy of the scan). Both the ATS-lite and the clinically used ATP treatment plans were evaluated on the ground truth contours. Dose-volume histogram (DVH) parameters were calculated for targets and organs-at-risk. The performance was assessed against clinical dose constraints. Additionally, one patient who showed extreme rectum variations was treated using the ATS- lite workflow for 12 fractions to demonstrate feasibility.
Conclusion Without any contour adjustments, the ATS-lite workflow improves target coverage compared to the current ATP workflow for patients with high risk PCa. We currently include more patients into the ATS-lite workflow and investigate feasibility of PTV margin reduction. Digital Poster 1703 Online adaptive radiotherapy for endometrial cancer with a DLAS model: individualized dosimetry using a linear mixed-effects model Min Li, Runhong Lei, Yongping Zheng, Weijuan Jiang, Ping Jiang, Ang Qu, Junjie Wang Department of Radiation Oncology, Peking University Third Hospital, Beijing, China Purpose/Objective: This study assesses the dosimetry of online adaptive radiotherapy (oART) for endometrial cancer, using FBCT guidance and deep learning auto-segmentation (DLAS). Material/Methods: Nine postoperative endometrial cancer patients underwent pelvic radiotherapy with a prescribed dose of 45 Gy in 25 fractions. Weekly oART was performed, and the total workflow time for each session was recorded. For each fraction, three plans were generated: (1) ART plan, which was recontoured and recalculated on the daily anatomy; (2) Primary plan,
Results Compared to ATP, the rectum and bladder were less likely to violate the planning constraints in the ATS-lite workflow, evaluated on ground truth contours (Figure 2b). Coverage of the CTV (Figure 2a), PTV and BV was higher for the ATS-lite treatment plan. The GTV boost did not differ significantly between both plans. The ATS-lite workflow including contouring and treatment delivery took approximately 20 minutes.
Made with FlippingBook - Share PDF online