Purpose: To quantify inter- and intrascanner conformity in computed tomographic (CT) densitometry of the lungs.
Materials and methods: With six scanners from four manufacturers, a lung densitometry protocol with several variations was applied for performance comparison. Phantoms included water, air, and a humanoid thorax phantom equipped with a dog lung and exchangeable pseudolungs of polyethylene foam.
Results: All scanners produced acceptable CT numbers (Hounsfield units) for water, but some not for air. An incorrect calibration of air density affected all CT numbers at lung densities, but the error was easily correctable. Some systems were more sensitive to object size than others were. Sensitivity of CT numbers to section thickness, reconstruction filter, zoom factor, and table height was small, except for two scanners in relation to section thickness.
Conclusion: After correction for poor air calibration, scanner conformity was acceptable when the reproducibility of lung densitometry in clinical practice was set as a reference.