S2215
Physics - Intra-fraction motion management and real-time adaptive radiotherapy
ESTRO 2026
[2] Morrison, K., Tree, A., Khoo, V., & Van As, N. J. (2018). The PACE trial: International randomised study of laparoscopic prostatectomy vs. stereotactic body radiotherapy (SBRT) and standard radiotherapy vs. SBRT for early stage organ-confined prostate cancer. Journal of Clinical Oncology , 36 (6_suppl). https://doi.org/10.1200/jco.2018.36.6_suppl.tps153
first 30 patients in the SI (superior-inferior), LR (left- right) and AP (anterior-posterior) directions by acquiring a post-treatment CBCT image in addition to the pre-treatment CBCT used for IGRT. Intrafractional shifts were obtained based on a prostate soft-tissue match. To limit prostate motion during treatment, IGRT was performed approximately every 3-4 minutes. Results For 28 of 30 patients, the CBCT image quality allowed a clear visualization of prostate boundaries. The average time between pre-treatment CBCT acquisition and end of treatment delivery was 3min27, with values ranging between 2min56 and 4min09. The average time between pre- and post-treatment CBCTs was 5min24. Histograms of estimated intrafractional prostate displacements in each direction based on 30 patients and a total of 136 prostate shifts are presented in figure 1, while table 1 provides a statistical description of these results. In the SI and LR directions, all displacements are contained within the 5mm PTV margins, with maximum shifts of ±3mm observed. The largest range of motion is obtained in the AP direction, with few shifts exceeding or being equal to the PTV margin used anteriorly. Out of all 136 shifts recorded in each direction, the prostate was located outside of the PTV only once (+6.0mm AP).
Mini-Oral 1184
Real-Time, Binning/Template-Free 4D-MRI Reconstruction using Continuous Neural
Representations for Online Adaptive Radiotherapy Muheng Li 1,2 , Xinyang Wu 1 , Xia Li 1,3 , Orso Pusterla 4,5 , Sairos Safai 1 , Philippe Claude Cattin 5 , Antony John Lomax 1,2 , Ye Zhang 1 1 Center for Proton Therapy, Paul Scherrer Institute, Villigen, Switzerland. 2 Department of Physics, ETH Zurich, Zurich, Switzerland. 3 Department of Computer Science, ETH Zurich, Zurich, Switzerland. 4 Department of Radiology, University Hospital Basel, Basel, Switzerland. 5 Department of Biomedical Engineering, University of Basel, Basel, Switzerland Purpose/Objective: Four-dimensional MRI (4D-MRI) offers superior soft- tissue contrast for respiratory motion management in radiotherapy, yet clinical implementation remains limited by spatial-temporal resolution trade-offs, prohibitive reconstruction times, and workflow complexity. MR-guided radiotherapy systems demand sub-second volumetric reconstruction for real-time beam tracking [1]. However, conventional binning- based methods underestimate motion and have artifacts due to irregular breathing, while deep learning approaches require pre-acquired template volumes, complicating workflows. Here, we introduce a template-free neural representation framework achieving real-time 4D-MRI reconstruction from continuous free-breathing data without binning. Material/Methods: A dual-network implicit neural representation (INR) architecture (Fig. 1a) has been developed that jointly optimizes anatomical structures and temporal deformations. A Spatial Anatomy Network (SAN) learns 3D canonical volumes (Fig. 2a), while a Temporal Motion Network (TMN) predicts continuous deformation vector fields guided by motions extracted from diaphragm points, tracked on sagittal navigator slices using Co-tracker [2]. Unlike binning and template-dependent methods, our framework learns anatomy and motion simultaneously from 2D slices alone, enabling binning-free reconstruction across arbitrary respiratory states. Data were acquired on a 1.5T MRI (MAGNETOM Aera Siemens) using custom bSSFP [3] with interleaved coronal stacks and sagittal
Conclusion For almost all patients, accurate prostate soft-tissue match was possible with CBCT IGRT without the need for fiducials. Based on the estimated intrafractional prostate shifts, small displacements were observed, with 98% of shifts remaining within ±3mm of initial prostate positioning. By minimizing all uncertainties, isotropic 5mm PTV margins were therefore adequate for prostate SBRT treatments without fiducial markers. References [1] van As, N., et al. (2024). Phase 3 trial of stereotactic body radiotherapy in localized prostate cancer. New England Journal of Medicine , 391 (15), 1413–1425. https://doi.org/10.1056/nejmoa2403365
Made with FlippingBook - Share PDF online