ESTRO 2025 - Abstract Book

S3011

Physics - Image acquisition and processing

ESTRO 2025

4. Korte, J.C., et al., A radiation therapy platform to enable upright cone beam computed tomography and future upright treatment on existing photon therapy machines. Medical Physics, 2024. doi:10.1002/mp.17523 5. Rit, S., et al. The Reconstruction Toolkit (RTK). in Journal of Physics: Conference Series. 2014. doi:10.1088/1742 6596/489/1/012079

2282

Proffered Paper Evaluation of 2D antiscatter Grid-Based Quantitative CBCT in a Prospective Trial: Analysis of Image Quality Metrics Farhang Bayat 1 , Ryan Sabounchi 1 , Uttam Pyakurel 1 , Junxiao Hu 2 , Roy Bliley 1 , Brian Kavanagh 1 , Ryan Lanning 1 , Tyler Robin 1 , Cem Altunbas 1 1 Radiation Oncology, University of Colorado Anschutz Medical Campus, Aurora, USA. 2 Pediatrics, University of Colorado Anschutz Medical Campus, Aurora, USA Purpose/Objective: Scattered radiation is one of the leading causes of image quality degradation in CBCT. To address this challenge, a novel quantitative CBCT (qCBCT) pipeline with a 2D antiscatter grid was developed, and its performance was evaluated in a prospective trial. Material/Methods: In this single-arm study, 31 participants were enrolled, and each participant received two CBCT scans: one using the Varian TrueBeam's clinical CBCT protocol and one using the qCBCT. Acquisition parameters were identical for both scans. Each clinical CBCT scan was reconstructed twice using standard and advanced (iCBCT) reconstruction options. The qCBCT scans were processed offline using an in-house developed data processing chain and reconstructed using the standard filtered backprojection algorithm. In total, 93 CBCT sets were analyzed, with 4,464 regions of interest placed in soft tissues. Absolute HU error relative to planning CT, artifact amplitude, and contrast to-noise ratio (CNR) were calculated. The effect of anatomical region and patient size on image quality metrics was evaluated. Differences in image quality metrics between CBCT types were assessed using a Linear Mixed Model. Results: Mean HU errors and 95% confidence intervals (CI) for standard clinical CBCT, advanced clinical CBCT, and qCBCT were 43 (CI: 36–50), 33 (CI: 28–39), and 19 (CI: 16–22) HU, respectively. Artifact amplitudes were 168 (CI: 142–198), 117 (CI: 99–138), and 78 (CI: 66–92) HU for the respective CBCT types. Differences in HU errors and artifact amplitudes between clinical CBCT and qCBCT were significant (p < 0.01). Soft tissue CNRs were 6.45 (CI: 5.7–7.3), 8.7 (CI: 7.7–9.9), and 7.26 (CI: 6.4–8.2), respectively. qCBCT showed an improvement in CNR over standard clinical CBCT (p = 0.02). Higher CNRs were observed in advanced clinical CBCT due to the lower noise with iterative reconstruction. HU error and artifact amplitude reduction by qCBCT was more pronounced in lateral separations greater than 40 cm and in the lower neck region. Differences in HU errors and artifact amplitudes between advanced clinical CBCT and qCBCT were not significant in the upper neck region. Conclusion: The 2D antiscatter grid-based qCBCT improves CT number accuracy and reduces artifacts, particularly in anatomical regions with larger lateral separations. CNR was also improved in qCBCT images compared to clinical CBCT images when the reconstruction algorithms of the two were comparable. These results suggest that qCBCT has the potential to enhance tissue visualization and improve CT-based treatment monitoring tasks, such as CBCT-based dose calculations.