Abstract

Background/Aims
To investigate the diagnostic performance of diffusion-weighted imaging (DWI) for hepatic neuroendocrine tumors (NET) compared with combined DWI and contrastenhanced magnetic resonance imaging (MRI) .
Methods
Fifteen patients with hepatic NET (n=128) underwent enhanced MRI and DWI with multiple-b values. We analyzed three different sets: Precontrast set; DWI set (added DWI); combined set (added enhanced image). Two reviewers rated possibility of NET using a 5-point scale for each image set. Their diagnostic performance was compared using Jackknife alternative free-response ROC (JAFROC).
Results
Diagnostic performance was better on the combined set (figure of merit [FOM]=0.852, 0.761) than the precontrast set (FOM=0.427, 0.572, P<0.05) and the DWI set (FOM=0.682, 0.620, P<0.05). However, DWI improved performance compared with precontrast set without statistical difference. In small NETs (<1 cm), all sets showed low sensitivity (10.7-65.9%) with high specificity (95.4-100%). Interobserver agreement was moderate in all image sets (k=0.521 to 0.589).
Conclusions
Combined DWI and enhanced MRI were more useful for detecting NET. Although statistically insignficant, there was a trend in improved diagnostic performance with DWI.

• Keywords: Diffusion-weighted imaging;Hepatic neuroendocrine tumor;Contrast-enhanced MRI;Comparison study