Background: Accurate estimation of calculus size is an important prediction factor for spontaneous passage. The method used for calculus-size estimation needs to be as precise as possible. Previous studies have shown that the image quality of virtual non-contrast (VNC) images obtained with the new generation of dual-energy CT (DECT) is still inferior compared to native images. However, the difference in assessed image quality compared to native series is smaler with the new than with the older system.
Purpose: The aim of this study is to evaluate the accuracy of measurements of renal calculi in a VNC image series obtained with the new generation of dual-energy CT, compared to actual stone size and single-energy image series in a phantom study.
Material and Methods: For the purpose of this study a ureter phantom was created. The phantom was scanned with DECT (Somatom Definition Flash) at three different radiation-dose settings with single and dual energy, first with water in the ureters then with i.v. contrast medium solution. At each energy level CDTI were set equal for single and dual-energy. VNC series were created. All image series were assessed for stone visibility and size.
Results: All calculi were visible in single- and dual-energy images with water in phantom tubes. With the VNC calculi in the size range 1.42 – 2.02 mm were missed. Results of intraobserver variability in investigated dose levels shows very good agreement between assessments of stone size by both evaluators, with correlation coefficient (CC) variying from 0.97 (95% CI 0.91–0.99) to 1.0 (95% CI 0.99–1.0). Results of correlation tests between the mean of maximal stone size and maximal stone size measured with electronic callipers show good agreement, with a CC variability from 0.93 (95% CI 0.76-0.98) to 0.99 (95% CI 0.96-0.99), a limit of agreement of "1.65 to 1.38 and bias "0.14. With one of the observers, the results of correlation tests in assessments made by the same reader of stone-size in dual-energy and VNC image series for corresponding dose levels show lower ICC for the low dose level, 0.77 (CI 0.36–0.93), with limits of agreement of -2.23–0.57 and bias "0.33.
Conclusions: This study shows that detection of small stones is not reliable despite better image quality with the new DECT and that small stones will be missed with VNC imaging. In larger stones, the inherent measurement error with CT is magnified with VNC imaging.