Abstract
PURPOSE: Prospective protocol optimization, determination of image quality and diagnostic performance of virtual non-enhanced images (VNEI) derived from split-bolus dual-energy computed tomography (DECT) urography in patients with urinary stones. METHODS: IRB-approved, prospective study of 100 patients who, after written informed consent, underwent single-energy, non-enhanced CT and split-bolus, contrast-enhanced DECT (30 + 50 mL of contrast media; combined nephro-urographic acquisition). DECT was performed using setting A (80/140 kVp) in the first 20, and setting B (100/140 kVp) in the second 20 patients. Tin filtration was used in all patients. After a pre-analysis of VNEI quality, 60 additional patients were examined using setting B. Two readers qualitatively and quantitatively determined image quality of all weighted-average DECT images regarding urinary tract opacification (n = 100), and all VNEI regarding quality of iodine subtraction and urinary stone detection (n = 80). True nonenhanced (TNEI) images were the standard of reference for statistical analysis (inter-reader variability and diagnostic performance characteristics). RESULTS: The urinary tract was completely opacified in 94% (94/100) of patients. Iodine subtraction was improved (p < 0.01) and image noise of VNEI was lower (p < 0.05) in DECT setting B. On VNEI, 83% (86/104) of urinary stones were correctly identified and 17% (18/104) were missed. Stones missed (2.5 mm, 1-4) were significantly smaller than stones correctly identified (5 mm, 2-27; p < 0.001). Diagnostic accuracy was 98% on a per-renal-unit basis and 96% on a per-patient basis. Inter-reader agreements were excellent (κ = 0.91-1.00; ICC = 0.86-0.99). CONCLUSIONS: Split-bolus DECT urography was technically feasible and quality of VNEI was improved with the 100/140 kVp setting. Detection of urinary stones <4 mm on VNEI was limited.