S2091
Physics - Image acquisition and processing
ESTRO 2026
Digital Poster 4414 Quantitative and qualitative evaluation of HyperSight cone-beam CT imaging for contouring in lung cancer patients Nadine Coorens, Colien Hazelaar, Stephanie Peeters, Daniel Portik, Vicki Taasti, Rik Hansen, Marta Bogowicz, Wouter van Elmpt Department of Radiation Oncology (Maastro), Maastricht University Medical Centre+, Maastricht, Netherlands Purpose/Objective: To quantitatively and qualitatively evaluate the quality of cone-beam computed tomography (CBCT) scans acquired with an advanced CBCT imaging system for contouring of organs-at-risk (OARs) in lung cancer patients. Material/Methods: For seven lung cancer stage II-IV patients, a planning CT (pCT, 50% exhale phase of 4DCT), a conventional CBCT (TrueBeam v2.7) and two CBCT scans on a Halcyon with HyperSight imaging (v4.0, Varian) were acquired in a clinical imaging trial (NCT05524454). TrueBeam and HyperSight CBCTs were acquired the same day, while the pCT was acquired 8-11 days earlier. Acquisition time of the TrueBeam scans was 60 seconds reconstructed using a filtered back-projection algorithm (FDK). Halcyon CBCT scans were acquired with a standard acquisition time of 6 seconds and additionally using the 60 seconds Slow Thorax imaging mode. Scans were reconstructed both with FDK and iterative reconstruction (IR). All scans were acquired in free-breathing.Image quality of the CBCTs was assessed quantitatively and qualitatively. For quantitative assessment, TrueBeam and HyperSight CBCT scans were rigidly registered to the pCT, regions- of-interest (0.1-4.1 cm3) were placed in homogeneous regions of various tissues (muscle, fat, lung, and tumor). Mean CT number and standard deviation (SD) were calculated. Qualitatively, two radiation oncologists scored confidence in contouring OARs using a 5-point Likert scale (1 lowest, 5 highest confidence). Ratings of the observers were averaged and outcomes were correlated with SD in CT number. Results: Quantitative analysis revealed that the variation in CT numbers was lowest for pCT scans, moderate for HyperSight scans reconstructed with IR, and highest for both Halcyon FDK and conventional TrueBeam scans (Figure 1A). Image noise was minimal in the pCT scan and comparable between Halcyon IR and FDK 6s scans. However, noise levels increased in the 60s Halcyon scans and were highest in TrueBeam scans (Figure 1B).Qualitative assessment (Figure 2A) showed the highest observer confidence in contouring OARs on the pCT scan (mean score: 4.4), followed by
Conclusion: DIR geometric performance varied by structure, with inferior regions showing poorer registration results. Further evaluation in a larger cohort is needed to clarify the relationship between treatment interval and DIR performance. Larger mDTA values for autosegmentations than DIRs suggest variation in contouring guidelines. References: 1. Tsang DS, et al. Outcomes After Reirradiation for Recurrent Pediatric Intracranial Ependymoma. Int J Radiat Oncol Biol Phys. 2018 Feb;100(2):507-515. 2. Asklid A, et al. Reirradiation in Paediatric Tumours of the Central Nervous System: Outcome and Side Effects After Implementing National Guidelines. Clin Oncol (R Coll Radiol). 2025;37:103667. 3. Thiong'o CM, et al. Dose Mapping using Image Registration for Reirradiation: A Systematic Review. Int J Radiat Oncol Biol Phys. 2025 Oct 11:S0360-3016(25)06326-6. 4. Weistrand O, Svensson S. The ANACONDA algorithm for deformable image registration in radiotherapy. Med Phys. 2015 Jan;42(1):40-53. Keywords: deformable image registration, reirradiation
Made with FlippingBook - Share PDF online