Fast kVp-switching dual energy contrast-enhanced thorax and cardiac CT: A phantom study on the accuracy of iodine concentration and effective atomic number measurement.

PURPOSE

To assess the effect of vessel diameter and exposure parameters on the estimation accuracy of concentration and effective atomic number (Z ) of iodine (I) in contrast-enhanced thorax and cardiac dual-energy CT using a modern fast kVp-switching CT scanner.

METHODS

A standard semi-anthropomorphic cardiac CT phantom devised to simulate the human chest at three different body habitus i.e., medium-sized, large-sized, and obese, was scanned using a fast kVp-switching Revolution-GSI GE CT scanner. Five cylindrical, 10 mm diameter, vials were filled with solutions prepared by diluting I contrast at five concentrations (2.5, 5, 10, 15, and 20 mg I/ml). To simulate small vessels, pipette tips with a diameter ranging from 5 mm to 0.5 mm were employed. The vials and pipette tips were accommodated within the semi-anthropomorphic phantom. CT acquisitions were performed in the fast kVp-switching dual-energy mode at six different CTDI values. Acquisitions were also performed at 80, 100, 120, and 140 kVp. Images were acquired at 64 × 0.625 mm beam collimation and reconstructed at 2.5 mm using all available reconstruction filter kernels. Virtual monochromatic spectral (VMS) images, iodine concentration (I ), and Z maps were reconstructed. Hounsfield unit as a function of energy (HU ) in VMS and single-kVp (HU ), I and Z were measured at each CTDI . The effect of vessel diameter on I and Z was investigated. Measured HU and Z were compared to theoretically estimated values and I were compared to nominal (I ) values.

RESULTS

In 10 mm diameter vessels, HU values were accurate to 18% for the medium-sized, 22% for the large-sized and 39% for the obese phantoms. I was underestimated by up to 10% for the medium-sized, 26% for the large-sized and 33% for the obese phantom. I error decreased with increasing CTDI from ±0.799 mg/ml at 8.61 mGy to ±0.082 mg/ml at 32.01 mGy. The percentage difference between measured and theoretically estimated Z ranged from -3.9% to -14.5%. In pipette tip vessels, I was found to depend on the kernel employed. At the standard kernel, I , for I = 20 mg/ml, was reduced with vessel diameter from 19.25 ± 0.39 mg/ml, at 10 mm, to 2.52 ± 0.31 mg/ml, at 1 mm. Linear regression between I and I resulted in I /I factors of 0.925 for 5 mm, 0.815 for 4 mm, 0.651 for 3 mm, 0.377 for 2 mm, and 0.129 for 1 mm vessel diameter. Measured Z values were underestimated when vessel diameter was decreased from 5 mm to 1 mm by 27% for the 20 mg I/ml and 21% for the 2.5 mg I/ml.

CONCLUSIONS

HU , I , and Z depend on several parameters such as body size, vessel size, exposure parameters, and reconstruction kernel. The limiting spatial resolution of the CT system results in considerable underestimation of HU , I , and Z in vessels smaller than 5 mm diameter. The underestimation of I uptake may be experimentally corrected, if the diameter of the investigated vessel is measured and the correction factors produced in this study are employed.

© 2017 American Association of Physicists in Medicine.