S2153
Physics - Inter-fraction motion management and daily adaptive radiotherapy
ESTRO 2026
Digital Poster 3308
oncologists review propagated contours and dose distributions. The selected plan is exported to the beam delivery system, and delivered doses are recorded in dose-tracking module for accumulated- dose–based optimization in subsequent fractions.
Patient specific models of bladder deformation using multichannel image registration for library of plan creation Poppy Nikou 1 , Sarah Osman 1,2 , Asma Sarwar 3 , Gemma Eminowicz 3 , Colin Baker 1 , Catharine Clark 1,4 1 Dept of Radiotherapy Physics, University College London Hospital, London, United Kingdom. 2 National RTQA, Radiotherapy Trials Quality Assurance (RTTQA) Group, London, United Kingdom. 3 Dept of Radiation Oncology, University College London Hospital, London, United Kingdom. 4 Dept of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom Purpose/Objective: Anatomical changes during radiotherapy treatment of cervical cancer patients can impact treatment efficacy [1,2]. Large, random anatomical variations during treatment have been reported for this cohort, especially in bladder filling [3,4]. In this work, patient- specific models (PSMs) were developed to capture this change for individual patients. Predictions from these models could be used to create a personalized library of plans, enabling an adaptive treatment workflow. Material/Methods: Fifteen patients treated with photon radiotherapy were included in this study. Each patient had both a full-bladder and an empty-bladder CT scan. Rigid and affine registrations were first applied to align the bony anatomy.Deformable image registration can fail when anatomical changes are large or discontinuous, as found in the bladder region [5]. To better constrain the bladder registration, a multichannel image registration was implemented. Each multichannel image contained the CT and segmentations of the bladder and body.Deformable registration was performed using a B-spline free-form deformation algorithm (NiftyReg [6]). Parameters were tuned to best capture bladder deformation. Registrations were performed using the stationary velocity field approach, resulting in a diffeomorphic transformation. To build the PSMs, the stationary velocity field was assumed to be linear and sampled at intervals of σ = 0.1. At each interval, the PSM transformation was then used to resample the CT image and bladder segmentation. The model performance was assessed by tracking bladder volume during emptying, as no other intermediate ground truth data was available. Results: The optimal parameters for capturing bladder deformation included channel weights of 5:5:3 for CT, bladder, and body respectively. The loss functions used were the local normalized cross-correlation for the CT channel and sum of squared differences for the segmentation channels. A linear energy penalty of 0.1,
Results: The automated workflow in RayStation executed all adaptive steps without user intervention, including deformable registration, contour propagation, and plan generation. The computation time was under seven minutes, and the entire adaptive cycle, including physician review, was finished in 20 minutes. Conclusion: A fully scripted workflow for online adaptive upright proton therapy was established, enabling rapid plan adaptation and minimizing manual workload. Future work will validate the process in clinical pilot cases and integrate automated plan evaluation and dose accumulation for routine clinical use. References: 1. Feldman J, Pryanichnikov A, Achkienasi A, et al. Commissioning of a novel gantry-less proton therapy system. Front. Oncol. 2024; 14:1417393.2. Feldman J, Pryanichnikov A, Shwartz D, et al. A. Study of upright patient positioning reproducibility in image-guided proton therapy for head and neck cancers. Radiother Oncol. 2024; 201:110572.3. Blumenfeld P, Pryanichnikov A, Hillman Y, et al. Prospective clinical trial of upright image-guided proton therapy for locally recurrent head and neck and brain malignancies.Radiother Oncol. 2025; 211:111097.4. Taasti VT, Hazelaar C, Vaassen F, et al. Clinical implementation and validation of an automated adaptive workflow for proton therapy. Phys Imaging Radiat Oncol. 2022; 24:59-64. Keywords: Online adaptive, upright proton therapy
Made with FlippingBook - Share PDF online