A Comparative Study of CT and MRI for Evaluating Hepatic Malignant and Focal Nodules Based on ROC Curves

Abstract

The clinical use of magnetic resonance imaging (MRI) and multiphase enhanced computed tomography (CT) with the contrast media (Gd-EOB-DTPA) for detecting hepatic malignant and focal nodules prior to operation was examined based on the receiver operating characteristic (ROC) curve. This study included 70 patients with malignant and focal liver nodules who underwent MRI and multiphase CT scans before operation. Both scans for each patient were conducted within 1 month. For MRI, the T (2)-weighted image (single shot fast spin echo) and two-dimensional (2-D) and 3-D T (1)-gradient magnetic signals were obtained for all patients before administering the contrast media. The 2-D and 3-D T (1)-gradient magnetic signals were obtained in the same location after delivering the contrast media. For the CT scans, images of artery phase, portal phase, and delayed phase were obtained at a thickness of 5 mm or less after administering contrast similar to MRI. An ROC curve was used (paired-samples T test, P < 0.05) to evaluate the images. When the analysis was based on the ROC curve, MRI showed high values (P < 0.05) for area under curve (AUC), sensitivity, and specificity in terms of detection rates of small lesions (less than 2 cm and more than 2 cm) compared to multidetector computed tomography (MDCT) (for a parts per thousand currency sign2 cm, MRI: 0.928, 70, 93%, CT: 0.775, 30, 90%; for a parts per thousand yen2 cm, MRI: 0.744, 80%, 84%; CT: 0.692, 40%, 84%). Gd-EOB-DTPA contrast media-enhanced MRI scanner for detecting malignant and focal liver nodules before operation showed the higher detection rate of lesion and classification of lesion as either benign or malignant than multiphase enhanced MDCT when the ROC curve was used for analysis. Based on these results, we believe that analysis based on the ROC curve will provide guidelines for evaluating malignant and focal hepatic lesions prior to operation.