S2242
Physics - Intra-fraction motion management and real-time adaptive radiotherapy
ESTRO 2026
identifying bronchiolar branches that moved coherently with the lesion within a defined range of motion similarity and delineated as a Correlation Object. This object, encompassing the lesion, served as a motion surrogate and was rendered as a Digitally Reconstructed Radiograph (DRR).For each lesion, three Correlation Objects were created, each representing a different degree of motion similarity and varying in size. To assess the impact on the accuracy and robustness of the registration, each Correlation Object was used for retrospective localization across three distinct X-ray image pairs per patient. Tumor visibility was scored with and without surrogates, and subgroup analyses were performed according to lesion size and lobe location. Statistical correlations between visualization failure and tumor characteristics were assessed. Geometric accuracy was evaluated by calculating 3D vector discrepancies between surrogate DRRs and corresponding X-rays. Results: Without surrogates, the lung lesion was visible in 14.3% of X-ray pairs. When applying surrogate reconstructions, visibility increased to 75.4%. Visualization tended to be higher for lesions in the upper and middle lobes than in the lower lobe, while tumor size was not a significant predictor of visualization success.The mean geometric deviation was 3.3 mm (SD 2.4 mm). After excluding 27 data points affected by patient motion, 61.1% of vectors showed errors <4 mm and 75.7% <5 mm. Across all analyses, mean 3D discrepancies clustered around 4.0 ± 3.0 mm, with median values near 3 mm for all surrogate–X-ray combinations. No statistically significant differences were observed between surrogates for any given X-ray or between X-rays for any surrogate, indicating that within a defined motion similarity threshold, correlation objects does not affect geometric targeting. Conclusion: Surrogate-based X-ray targeting using ExacTrac Dynamic significantly enhanced lesion visualization and demonstrated consistent geometric accuracy in markerless lung SBRT. These findings support the feasibility and robustness of correlation object-driven registration. The absence of significant variation across surrogates and X-ray acquisitions underscores the clinical potential of this fully markerless intra- fraction verification approach for lung SBRT workflows. Keywords: Markerless Lung SBRT, stereoscopic x-rays, DIBH
ability of the method to recover three-dimensional information on beam position and dose deposition from full-flux secondary radiation data. Conclusion: This study introduces a novel AI-driven framework for real-time in vivo monitoring in PT. Combining advanced detector technology, Monte Carlo simulations, and deep learning algorithms, the system enables motion-resilient dose delivery. This innovation is critical for achieving truly high-precision adaptive PT, fundamentally enhancing treatment safety and clinical
confidence. References:
[1] M. Durante, R. Orecchia, J.S. Loeffler. Charged- particle therapy in cancer: clinical uses and future perspectives. Nat Rev Clin Oncol. 2017;14:483–495.[2] Y. Liu et al. Frontiers and challenges in silicon-based single-photon avalanche diodes and key readout circuits. Microelectron J. 2024;147:106165.[3] M. Marafini et al. Mondo: a neutron tracker for particle therapy secondary emission characterisation. Phys Med Biol. 2017;62:3299.[4] A. Ferrari et al. FLUKA: a multi-particle transport code. CERN-2005-010, INFN/TC_05/11, SLAC-R-773 (2005).[5] A. Schiavi et al. Fred: a GPU-accelerated fast Monte Carlo code for rapid treatment plan recalculation in ion beam therapy. Phys Med Biol. 2017;62(18):7482–7504. Keywords: Secondary particle, dose verification, AI Evaluation of Correlation-Object-Based X-ray Targeting in Markerless Stereotactic Lung Radiotherapy Marlies Boussaer 1 , Cristina Teixeira 1 , Kajetan Berlinger 2 , Selma Ben Mustapha 1 , Anne-Sophie Bom 1 , Sven Van Laere 1 , Mark De Ridder 1 , Thierry Gevaert 1 1 Department of Radiotherapy, Research Centre for Digital Medicine, UZ Brussel - Vrije Universiteit Brussel, Brussels, Belgium. 2 Brainlab SE, Brainlab, Munich, Germany Purpose/Objective: Accurate positioning and real-time intra-fraction monitoring of moving lung targets remain major challenges in markerless stereotactic body radiotherapy (SBRT), despite advances in image guidance. This study aimed to evaluate the feasibility and geometric accuracy of using anatomical surrogates instead of the lesion itself for patient positioning and intra-fraction X-ray monitoring with ExacTrac Dynamic (Brainlab, Munich, Germany). Material/Methods: Digital Poster 2985 Twenty-one patients were retrospectively analyzed. For each patient, surrogate-based reconstructions derived from 4DCT datasets were generated by
Made with FlippingBook - Share PDF online