Radiology. 2012 Jul;264(1):119-25. Epub 2012 May 8.
Mangold S, Thomas C, Fenchel M, Vuust M, Krauss B, Ketelsen D, Tsiflikas I, Claussen CD, Heuschmid M.
Purpose: To retrospectively determine which features of urinary calculi are associated with their detection after virtual elimination of contrast medium at dual-energy computed tomographic (CT) urography by using a novel tin filter. Materials and Methods: The institutional ethics committee approved this retrospective study, with waiver of informed consent. A total of 152 patients were examined with single-energy nonenhanced CT and dual-energy CT urography in the excretory phase (either 140 and 80 kV [n = 44] or 140 and 100 kV [n = 108], with tin filtration at 140 kV). The contrast medium in the renal pelvis and ureters was virtually removed from excretory phase images by using postprocessing software, resulting in virtual nonenhanced (VNE) images. The sensitivity regarding the detection of calculi on VNE images compared with true nonenhanced (TNE) images was determined, and interrater agreement was evaluated by using the Cohen k test. By using logistic regression, the influences of image noise, attenuation, and stone size, as well as attenuation of the contrast medium, on the stone detection rate were assessed. Threshold values with maximal sensitivity and specificity were calculated by means of receiver operating characteristic analyses. Results: Eighty-seven stones were detected on TNE images; 46 calculi were identified on VNE images (sensitivity, 52.9%). Interrater agreement revealed a κ value of 0.95 with TNE images and 0.91 with VNE data. Size (long-axis diameter, P = .005; short-axis diameter, P = .041) and attenuation (P = .0005) of the calculi and image noise (P = .0031) were significantly associated with the detection rate on VNE images. As threshold values, size larger than 2.9 mm, maximum attenuation of the calculi greater than 387 HU, and image noise less than 20 HU were found. Conclusion: After virtual elimination of contrast medium, large (>2.9 mm) and high-attenuation (>387 HU) calculi can be detected with good reliability; smaller and lower attenuation calculi might be erased from images, especially with increased image noise. © RSNA, 2012.