Re-evaluation of detectability of liver metastases by contrast-enhanced CT: added value of hepatic arterial phase imaging.

Abstract

We compared the detectability of liver metastases on 2- and 3-phase images using robust statistical methods. Nine radiologists evaluated unenhanced CT plus portal venous phase (2-phase) images and 2-phase plus hepatic arterial phase (HAP) (3-phase) images. Our study included 54 patients with primary malignant tumors who underwent 3-phase hepatic dynamic CT more than twice to screen for liver metastases; 24 had 1-6 liver metastases measuring 4-27 mm in diameter (median 13 mm). The other 30 had no metastases. Nine board-certified radiologists participated in our observer performance study. They specified the location of the metastatic lesions and rated the probability of metastatic nodules on 2-phase images acquired with and without HAP imaging. We used jackknife alternative free-response receiver operating characteristic (JAFROC) analysis to compare their performances. For all radiologists the area under the curve without and with HAP imaging was 0.86 and 0.88, respectively; the difference was significant (p = 0.04). For metastases smaller than 10 mm the averaged lesion localization fraction without and with HAP imaging was 0.17 and 0.26 (p < 0.01). Adding HAP to 2-phase imaging improved the detectability of liver metastases, especially of lesions smaller than 10 mm.