Diagnostic Accuracy of Split-Bolus Single-Phase Contrast-Enhanced Cone-Beam CT for the Detection of Liver Tumors before Transarterial Chemoembolization.

PURPOSE:To evaluate detectability of hepatocellular carcinoma (HCC) using split-bolus cone-beam CT in intraindividual comparison between cone-beam CT and contrast-enhanced MR imaging. MATERIALS AND METHODS:In a retrospective, single-center study, 28 patients with 85 HCC tumors were treated with transarterial chemoembolization between May 2015 and June 2016. All patients underwent arterial and hepatobiliary phase (HBP) MR imaging within 1 month before transarterial chemoembolization. Cone-beam CT images were acquired using a split-bolus contrast injection with 2 contrast injections and 1 cone-beam CT acquisition. Statistical analyses included Friedman 2-way analysis, Kendall coefficient of concordance, and Wilcoxon test. Tumor detectability was scored using a 5-point system (1 = best; 5 = worst) by 2 independent readers resulting in 170 evaluated tumors. Quantitative analysis included signal-to-noise and contrast-to-noise ratio and contrast measurements. P values < .05 were considered significant. RESULTS:Better tumor detection was provided with split-bolus cone-beam CT (2.91/2.73) and HBP MR imaging (2.93/2.21) compared with arterial MR imaging (3.72/3.05; P < .001) without statistical difference between cone-beam CT and HBP MR imaging in terms of detectability (P = .154) and sensitivity for hypervascularized tumors. More tumors were identified on cone-beam CT (n = 121/170) than on arterial MR imaging (n = 94/170). Average contrast-to-noise ratio values of arterial and HBP MR imaging were higher than for cone-beam CT (7.79, 8.58, 4.43), whereas contrast values were higher for cone-beam CT than for MR imaging (0.11, 0.13, 0.97). CONCLUSIONS:Split-bolus cone-beam CT showed excellent detectability of HCC. Sensitivity is comparable to HBP MR imaging and better than arterial phase MR imaging.