Detection of hepatocellular carcinoma in patients with cirrhosis: added value of coronal reformations from isotropic voxels with 64-MDCT.

OBJECTIVE: The purpose of our study was to prospectively assess the added value of isotropic coronal reformations of the liver when using 64-MDCT for the detection of hepatocellular carcinoma (HCC).

SUBJECTS AND METHODS: Seventy-one consecutive patients (60 men, 11 women; mean age, 65 years) suspected of having HCC underwent 64-MDCT with coronal reformations. A multiphasic CT protocol that included unenhanced, hepatic arterial, portal venous, and equilibrium phases was performed. Three independent, blinded readers interpreted the transverse scan alone, the coronal scan alone, and the combined transverse and coronal scans for the presence of HCC. Sensitivity, positive predictive value, area under the receiver operating characteristic curve (A(z)), and interpretation time were calculated for each reading session.

RESULTS: Seventy-six HCC nodules were confirmed in 48 patients using histopathologic analysis or follow-up with long-term CT, MRI, or both (mean follow-up time, 12 months; range, 12-15 months) as the reference standard. Mean sensitivity, positive predictive value, and A(z) value for HCC detection were, respectively, 84% (191/228 readings), 91% (191/210 readings), and 0.85 for the transverse scan alone; 83% (189/228 readings), 93% (189/203 readings), and 0.86 for the coronal scan alone; and 87% (198/228 readings), 93% (198/213 readings), and 0.87 for combined interpretation of transverse and coronal scans. No comparisons were statistically significant. Forty-eight false-positive interpretations were recorded (19 for the transverse, 14 for the coronal, and 15 for the combined interpretation sets). The reading session in which combined transverse and coronal scans were available for interpretation showed significantly superior reader confidence for HCC detection as well as longer interpretation times (p<0.05 for both comparisons). The average reading time for the combined interpretation of transverse and coronal image sets (mean, 12.1+/-0.8 minutes) was significantly longer than for the transverse image set (7.4+/-1.5 minutes) or the coronal image set (7.1+/-1.3 minutes) (p<0.01).

CONCLUSION: With 64-MDCT, the addition of isotropic coronal reformations to transverse images significantly improved reader confidence for the detection of HCC, with no statistically significant improvement in sensitivity, positive predictive value, or diagnostic accuracy (as determined by the A(z) value). This improvement comes at the cost of a longer interpretation time.