A Systematic Evaluation of Imaging Performance across a Fleet of Cone-Beam CT Scanners
Abstract
Purpose
Cone-beam CT (CBCT) has rapidly expanded across diverse clinical applications, yet fleet-wide benchmarking of imaging performance remains challenging due to wide variation in acquisition hardware, reconstruction pipelines, and vendor-driven quality control (QC) methods. This study systematically evaluates and compares imaging performance across a clinical CBCT fleet using a vendor-agnostic, standardized phantom-based framework under default clinical imaging protocols.
Methods
A Catphan phantom was scanned on five CBCT systems: three interventional radiology (IR) CBCTs (Philips-Azurion, Siemens-AXIOM-Artis, Canon-Alphenix) using default vascular protocols, and two dedicated extremity-CBCTs (CurveBeam-PedCAT, Planmed-Verity) using bone imaging protocols. Default acquisition settings varied widely across the fleet with limited ability to modify imaging attributes, including tube voltage (90–120 kVp), tube current (5–337 mA), FOV (20–48cm), angular sampling (185°–360°), and voxel size (0.2×0.2×0.2 to 0.65×0.65×0.65mm³). System-reported CTDI ranged from 0.57–10 mGy, although some systems did not provide CTDI. Imaging performance was evaluated in terms of gray-scale response for material inserts, CNR, spatial resolution, uniformity, and artifact.
Results
Across the CBCT fleet, gray-scale values spanned a wide range across material inserts, from air (−1028.56 to −658.18) to Teflon (140.60 to 955.34). Gray-scale response patterns were highly correlated across systems (r = 0.96453–0.99938), with the highest similarity between Philips-Azurion and Canon-Alphenix and the lowest between Siemens-AXIOM-Artis and CurveBeam-PedCAT. CNR showed marked fleet-wide variability, highest for Canon-Alphenix, intermediate for Philips-Azurion and Siemens-AXIOM-Artis, and lowest for the extremity-CBCTs. Spatial resolution was higher for extremity-CBCTs (9 lp/cm) than IR- CBCTs (7–8 lp/cm). IR-CBCTs showing lower non-uniformity and cupping/shading artifacts than extremity-CBCTs.
Conclusion
CBCT imaging performance varied across the fleet under clinical protocols despite highly correlated gray-scale response patterns. Extremity-CBCTs demonstrated higher spatial resolution but lower CNR and greater non-uniformity and cupping artifact than IR-CBCTs. Standardized phantom-based QC, coupled with protocol harmonization may enable fleet-wide benchmarking of CBCT imaging performance.