S2071
Physics - Image acquisition and processing
ESTRO 2026
Radiother. 2025; 30(3):298-305.3.Emin S et al., Clinical implementation of a commercial synthetic computed tomography solution for radiotherapy treatment of glioblastoma. Phys Imaging Radiat Oncol. 2024; 30:100589. Keywords: MRI-RT, Brain, Synthetic CT, Deep Learning, cGAN
and 95 % confidence intervals (CIs) were calculated for D2%, D50%, D95% and D98% for PTV, and for mean doses to all OARs. Results:
Digital Poster 3083
Feasibility of liquid fiducial markers for target localization in abdominal SBRT: a kick-off study Valeria Faccenda 1,2 , Denis Panizza 1,2 , Carlo Alberto Capodaglio 3 , Ilenia Manno 2,4 , Antonio Rovere 3 , Davide Leni 3 , Rocco Corso 3 , Elena De Ponti 1,2 , Stefano Arcangeli 2,4 , Rita Marina Niespolo 4 1 Medical Physics, IRCCS San Gerardo dei Tintori, Monza, Italy. 2 School of Medicine and Surgery, University of Milan Bicocca, Milan, Italy. 3 Diagnostic and Interventional Radiology, IRCCS San Gerardo dei Tintori, Monza, Italy. 4 Radiation Oncology, IRCCS San Gerardo dei Tintori, Monza, Italy Purpose/Objective: Delivering accurate abdominal stereotactic body radiotherapy (SBRT) using standard linear accelerators is challenging due to target positioning uncertainties caused by poor visibility and gastrointestinal motion artifacts in cone-beam CT (CBCT). Liquid fiducial markers placed in proximity of the target offer a potential option for precise treatment delivery. This study assesses the feasibility of the novel BioXmark (CQ Medical) carbohydrate/iodine-based liquid fiducials for target localization in liver and pancreatic tumors for the first time worldwide. Material/Methods: From November 2024 to January 2025, four patients (2 HCC, 1 liver met, 1 pancreas) underwent BioXmark implantation one week before contrast-enhanced four-dimensional CT (4DCT) simulation. Two to three shots were performed per patient, with one to two marker injections per puncture under contrast- enhanced ultrasound guidance. A low-dose CT scan was acquired post-implantation to verify marker positioning. Markers were used for image registration between the baseline 4DCT and contrast-enhanced 4DCT for target contouring across all respiratory phases and between the reference CT and daily CBCT for treatment delivery. The accuracy of marker-based registration was compared to conventional registration methods (liver parenchyma or abdominal vessels) through mean absolute error (MAE) to assess the benefit of marker-guided matching. Results: Three of four patients had 1–4 visible markers across different imaging modalities. One patient received an
The mean PTV dose differences between sCT and pCT plans were -0.16 %, -0.01 %,0.21% and 0.31 % for D2 %, D50 %, D95% and D98%, respectively. All OAR dose differences were below 3.56 %. Patients exhibited deviations exceeding ±2%, attributed to incomplete bone representation at the skull base and MRI signal loss near the orbits. MRI artifacts in the eye region caused underestimation of dose in the lens and orbit. Overall, the distribution of sCT–pCT dose differences across all cases remained within clinically acceptable limits for the target.
Conclusion: The deep learning–based sCT workflow achieved clinically acceptable dosimetric accuracy, with PTV dose deviations within ±2% and OAR differences below 3.5%, supporting its feasibility for MRI-only brain radiotherapy. Integration of the Siemens RT Pro Edition with syngo.via RT Image Suite ensured geometric reliability and consistent sCT generation. Minor artifacts and bone-mapping errors were identifiable but did not affect overall dosimetric accuracy, confirming the clinical readiness of this commercial solution for MRI-only treatment planning. References: 1. Villegas F et al., Challenges and opportunities in the development and clinical implementation of artificial intelligence based synthetic computed tomography for magnetic resonance only radiotherapy. Radiother Oncol. 2024.2.Babka V et al., Quality evaluation of synthetic CT using a gamma analysis for prostate cancer MR-only radiotherapy. Rep Pract Oncol
Made with FlippingBook - Share PDF online