Premium
TH‐D‐L100J‐08: Imaging Performance of a Mobile Cone‐Beam CT C‐Arm for Image‐Guided Interventions
Author(s) -
Daly M,
Siewerdsen J,
Moseley D,
Cho Y,
Ansell S,
Wilson G,
Jaffray D
Publication year - 2007
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.2761698
Subject(s) - imaging phantom , cone beam computed tomography , image quality , image noise , artifact (error) , pixel , flat panel detector , calibration , image resolution , nuclear medicine , detector , iterative reconstruction , optics , artificial intelligence , physics , computer science , medicine , radiology , image (mathematics) , computed tomography , quantum mechanics
Purpose: To characterize the imaging performance of a mobile cone‐beam CT (CBCT) C‐arm for image‐guided interventions. This work reports on 3D image quality of a flat‐panel detector with multiple gain modes (Varian PaxScan 4030CB), radiation dose, and robust methods for geometric calibration and artifact management. Method and Materials: A prototype imaging system based on a mobile C‐arm (Siemens PowerMobil) has been developed to provide flat‐panel CBCT. Three readout modes (fixed‐, dual‐, and dynamic‐gain) were evaluated in CBCT phantom images across a range of dose (0.6–18.8 mGy). An analytic (non‐iterative) geometric calibration method capable of determining all nine degrees of freedom in source‐detector geometry was implemented. Image artifacts associated with x‐ray scatter and lateral truncation were characterized, and methods for artifact management (scatter estimation and projection extrapolation, respectively) were evaluated. Results: CBCT images exhibit soft‐tissue visibility (∼20 HU) and high spatial resolution (∼1 mm) at dose (∼10 mGy) sufficiently low as to permit repeat intraoperative imaging. Dynamic‐gain readout demonstrated the highest level of soft‐tissue and bony‐detail visibility across all doses, whereas fixed‐gain was degraded at high dose due to pixel saturation, and dual‐gain was degraded due to image noise. The C‐arm exhibits large geometric non‐idealities (>15 mm departure from semicircular orbit) due to mechanical flex; however, the geometric calibration restored image quality (e.g., 0.77 mm FWHM) and was reproducible to sub‐pixel precision. Lateral truncation artifacts were effectively minimized via mixed linear‐exponential extrapolation of projections at the detector edges, and x‐ray scatter was managed to a large extent by subtraction of 2D scatter fluence estimates based on the measured detector signal (patient thickness). Conclusion: The prototype C‐arm demonstrates sufficient image quality for guidance at doses low enough for repeat intraoperative imaging. The C‐arm is currently being deployed in patient protocols ranging from brachytheraphy to chest, breast, spine and head‐and‐neck surgery.