S2072
Physics - Image acquisition and processing
ESTRO 2026
subjective image quality compared to standard CBCT system. Material/Methods: Quantitative image quality was evaluated using a Catphan 504 phantom to measure HU linearity, uniformity, CNR, spatial resolution, noise texture characteristics, and low-contrast visibility. Images were acquired from four TrueBeam LINACs in our department—two equipped with HS and two with the standard CBCT system—using four default CBCT protocols (two at 100 kV and two at 125 kV). A reference scans were obtained with the departmental CT. All available reconstruction algorithms were assessed, and analyses were conducted with an in- house automated analysis tool based on the Pylinac library developed by Kerns [3]. For subjective evaluation, CBCT images from 15 patients with various anatomical targets were reviewed before and after HS installation. Eight radiographers performed blinded assessments of image quality and registration feasibility. Results: HS improved several image quality parameters compared to the standard CBCT system (Fig 1.). HU linearity showed smaller deviations, particularly for high-density materials, and HU uniformity was more consistent. CNR values were higher with HS, indicating improved contrast detectability, while low-contrast visibility showed minor improvements with one of the 125 kV protocols. Spatial resolution and noise texture characteristics remained similar between techniques, as expected. Subjective assessments revealed no significant differences in perceived image quality or registration performance (Fig 2.).
insufficient marker volume, leading to poor visibility and requiring conventional matching. The mean translational MAE between conventional and marker- based registration for simulation 4DCT was 0.8 mm. Treatment was successfully delivered with a total of 20 CBCT scans. Optimal marker visibility in CBCT was achieved with injection volumes resulting in markers of 5–6 mm in diameter. Smaller markers were prone to non-visualization due to imaging artifacts and patient positioning variations. The mean translational MAE for CBCT was 2 mm, with a maximum variation of 4 mm in the lateral and vertical directions. Conclusion: BioXmark fiducials improve target localization in abdominal SBRT by balancing poor CBCT visibility, facilitating accurate image registration, and enabling margin reduction and potential dose escalation. A learning curve is necessary to optimize marker injection parameters for consistent visibility throughout treatment. Further studies with larger patient cohorts are required to assess clinical outcomes. Keywords: Abdominal SBRT, Liquid fiducial markers Digital Poster 3304 Evaluating Image Quality of Cone-Beam CT Using Hypersight® on the Varian TrueBeam: A Quantitative and Subjective Study Jesse J, K Lohela 1,2 , Santeri Rytky 1,2 , Sakari S. Karhula 1,2 , Juha Nikkinen 1,2 1 Cancer Centre, Oulu University Hospital, The Wellbeing Service County of North Ostrobothnia, Oulu,
Finland. 2 Research Unit of Health Sicence and Technology, University of Oulu, Oulu, Finland
Purpose/Objective: Cone-beam computed tomography (CBCT) plays a crucial role in image-guided radiotherapy and has attracted growing interest for adaptive radiotherapy workflows. However, several limitations compared to conventional computed tomography (CT) have restricted its broader use [1]. Varian recently introduced Hypersight (HS), a novel CBCT imaging technique designed to overcome many of these limitations through hardware upgrades, a larger FOV, faster acquisition, and advanced reconstruction algorithms. Originally developed for O-ring linear accelerators (LINACs), HS has only recently become available for C-arm LINACs. However, previous image quality findings from O-ring systems cannot be directly applied to C-arm configurations due to differences in panel size, acquisition speed, and imaging geometry [2]. Here, we have assessed the suitability and potential of HS for offline adaptive radiotherapy with C-arm LINACs by investigating both quantitative and
Made with FlippingBook - Share PDF online