Hepatic tumors: quantitative tissue characterization with MR imaging.

Publisher Website
Google Scholar
Abstract
To determine which quantitative methods of image analysis are most suitable for the differential diagnosis of benign and malignant hepatic lesions, the authors analyzed magnetic resonance images obtained at 0.6 T in 42 patients with proved hepatic cavernous hemangioma and 63 patients with vaious hepatic malignancies. The lesion-liver signal-intensity ratio for images obtained with a repetition time of 2,350 msec and echo time of 180 msec was most helpful in distinguishing hemangiomas from cancer (area under the receiving operator characteristic curve [ROC] = 0.99 .+-. 0.005). All 38 lesions with lesion-liver signal-intensity ratios greater than 3.5 were hemangiomas, whereas all 57 hepatic tumors with a ratio less than 2.5 were malignant neoplasms. A specificity of 93% and sensitivity of 89% can be achieved with use of quantitative signal-intensity data only. The authors conclude that in conjunction with heavily T2-weighted pulse sequences, signal-intensity ratios are an important adjunct to morphologic analysis in the differential diagnosis of hepatic neoplasms.