Non-obese patients (38 men, eight women) with 68 HCCs underwent quadruple-phase CT at 16 MDCT (using low tube voltage, 80 kVp; moderately high tube current, 280 mAs) and Gd-EOB-DTPA-enhanced 3 T MRI. Three observers independently and randomly reviewed the CT and MR images on a tumor-by-tumor basis. The diagnostic accuracy of these techniques for detecting HCC was assessed using alternative free-response receiver operating characteristic analysis. Sensitivity and positive predictive values were evaluated. The mean effective doses for the low dose CT were also evaluated.
The areas under the ROC curves were 0.963, 0.959, and 0.941 for low dose CT and 0.981, 0.982, and 0.976 for MRI. Differences in Az of the two techniques for each observer were not statistically significant (P > .05). Differences in sensitivity and positive predictive values between the two techniques for each observer were not also statistically significant: sensitivity (86.8%, 82.4%, 85.3%for CT and 95.6%, 94.1%, 91.2%for MRI) and positive predictive values (92.2%, 90.3%, 89.2%for CT and 92.9%, 92.8%, 92.5%for MRI). Six HCCs (8.8%) in five patients were observed only on hepatobiliary phase of MRI, and all were smaller than 1.5 cm. The mean effective dose for CT was approximately 10.2 mSv.
Quadruple-phase low-dose liver CT (80 kVp, 280 mAs) had relatively good diagnostic performance for detecting HCC in non-obese patients. Because no significant difference was observed between low-dose CT and MRI, the use of low-dose liver CT can be justified based on its reduced radiation effects.