S2202
Physics - Intra-fraction motion management and real-time adaptive radiotherapy
ESTRO 2026
treatment plans with regional boosts and large target volumes, which may prolong fraction times and increase intrafraction bladder filling. Therefore, assessing the feasibility of oART for this patient group is particularly important. This study describes the clinical implementation of oART for LACC N+ and evaluates its feasibility in clinical practice. Material/Methods: Fifteen LACC N+ patients were treated with oART using the Ethos Therapy system equipped with HyperSight CBCT. The prescribed dose was 25x1.8Gy, with a simultaneous integrated boost to pathological lymph nodes of 2.2 or 2.3Gy. A CTV-to-PTV margin of 5mm was applied, extended to 8mm in the cranial/posterior direction of the low-risk CTV. Due to the extended treatment volume, eight patients required a double- isocenter plan for irradiation and imaging. In total, 375 fractions were analyzed. Information regarding plan selection, manual editing, and post-treatment visual target positioning was collected from treatment logs. Timing data were obtained from DICOM metadata. Bladder volumes were calculated from CBCT delineations. Results: The online adapted plan was selected over the reference plan in all fractions. The median total fraction time was 27.9 min (IQR 23.6-32.3), with 23.5 min (IQR 21.1-26.4) for single-isocenter and 31.5 min (IQR 28.4-35.1) for double-isocenter treatments. Target delineations required manual editing in 14.7% of fractions, with a median duration of 7.5 min. The duration of the different steps is shown in Figure 1. Despite identical drinking instructions, bladder volumes showed considerable variation across patients (Figure 2), with a median initial volume of 122.6 mL (IQR 76.1-207.1) and a median increase of 48.6 mL (IQR 14.4-101.1) during fractions. The target remained positioned within the PTV on the post- treatment CBCT in 80.3% of cases. Loss of coverage was minimal and occurred most frequently at the tip of the uterus, predominantly caused by anatomical shifts due to bladder filling. Conclusion: This study demonstrates the feasibility of oART for LACC N+ patients. Despite extended fraction times and intrafraction bladder filling, the applied margins were sufficient in most fractions.
margins (data-aware epistemic uncertainty)(Figure.2). By N=5-10, the model was clinically informative: MAE=0.17mm, 90% of residuals were<0.3mm, posterior-predictive densities closely matched the empirical distribution (supported by lack of significance: paired Wilcoxon p=0.20, Levene variance test p=0.17 and a Wasserstein distance of 0.04mm). Beyond N ≥ 15, incremental gains were small relative to diminishing actionability late in the course.
Conclusion: A BNN can provide patient-specific, uncertainty- quantified intrafraction PTV margins after ~5-10 fractions, supporting human-in-the-loop margin adaption decisions. Providing confidence in moving to tighter margins for consistently stable movers with narrow-HDIs early in treatment, or additional monitoring with potentially enlarged margins for outliers. This is a practical path away from uniform margins and toward confidence-aware planning in the SGRT/adaptive era. Keywords: Bayesian, Neural Network, H&N Margins Clinical implementation of online adaptive radiotherapy for patients with node-positive locally advanced cervical cancer Harriëtte H.G. Dankers 1 , Jeremy Godart 1,2 , Judith H. Sluijter 1 , Maarten L.P. Dirkx 1 , Miranda E.M.C. Christianen 1 , Jan Willem M. Mens 1 , Mischa S. Hoogeman 1,2 , Remi A. Nout 1 , Henrike Westerveld 1 1 Department of Radiotherapy, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Rotterdam, Netherlands. 2 Department of Medical Physics & Informatics, HollandPTC, Delft, Netherlands Purpose/Objective: Online adaptive external beam radiotherapy (oART) enables daily plan adaptation to account for Digital Poster Highlight 426 anatomical variations. This is particularly relevant in the treatment of locally advanced cervical cancer (LACC), which can be challenging due to variability in uterine shape and position, influenced by bladder and rectal filling as well as tumor regression. While the benefits of oART have been demonstrated for node- negative LACC, its feasibility for node-positive (N+) patients is little explored. Treating N+ patients introduces additional challenges, including complex
Made with FlippingBook - Share PDF online