방사선의학 웹진

바로가기  >

Abstract
Background: Online adaptive radiotherapy (ART) relies on cone-beam computed tomography (CBCT) imaging for daily treatment adaptation, but CBCT's physical limitations can compromise contour delineation and dose calculation accuracy. The relationship between image quality metrics and clinical ART workflow accuracy remains unclear.

Purpose: To investigate the clinical impact of CBCT image quality on ART workflow accuracy in breast cancer patients by comparing contour agreement and dose calculation accuracy between Halcyon and TrueBeam systems, establishing how quantitative image quality metrics affect contour agreement and dose calculation accuracy within the ART workflow.

Methods: Forty breast cancer patients treated with VMAT were equally divided between Halcyon (n = 20) and TrueBeam (n = 20). Phantom-based image quality assessment evaluated HU uniformity, contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR), and modulation transfer function (MTF) using Catphan 504. Planning CT and first-day CBCT images were aligned using deformable registration. AI auto-segmentation generated contours for six structures, evaluated using Dice Similarity Coefficient (DSC) and other geometric metrics. VMAT plans were recalculated on CBCT images using system-specific HU-to-electron density calibration curves for dose-volume histogram, 3D gamma analysis, and dose difference analysis.

Results: Halcyon demonstrated 1.65-fold higher CNR (22.21 vs. 13.42) and 1.78-fold better HU uniformity (9 vs. 16 HU) compared to TrueBeam, despite comparable spatial resolution (MTF50: 0.36 lp/mm). Contour agreement analysis showed Halcyon achieved higher accuracy (treated breast DSC: 0.92 ± 0.06 vs. 0.87 ± 0.21, p < 0.001) with 3.5-fold lower inter-patient variability. Dosimetric analysis revealed paradoxical patterns: TrueBeam maintained better PTV coverage (D95% difference: 0.013 vs. 0.207 Gy) but showed significantly increased internal dose metrics (Dmean difference: -0.320 vs. 0.046 Gy, p < 0.001). Although both systems achieved > 99.8% gamma passing rates at 3%/3 mm criteria, stringent 1%/2 mm criteria revealed substantial performance differences (Halcyon: 97.66 ± 1.91% vs. TrueBeam: 90.28 ± 6.51%, p < 0.001).

Conclusion: Quantitative CBCT image quality metrics, particularly CNR and HU uniformity, directly affect both contour agreement and dosimetric concordance in breast ART workflows. These findings provide quantitative evidence linking CBCT image quality metrics to both contour agreement and dose calculation accuracy within the ART workflow.

Affiliations

Sunhwa Kim 1, Young Kyu Lee 2, Wonjoong Cheon 2, Yunji Seol 3, Byung-Ock Choi 2, Young Nam Kang 2
1Department of Medical Sciences, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea.
2Department of Radiation Oncology, College of Medicine, Seoul St. Mary's Hospital, The Catholic University of Korea, Seoul, Republic of Korea.
3Department of Proton Center, National Cancer Center of Korea, Goyang-si, Gyeonggi-do, Republic of Korea.