Aim: To determine the diagnostic performance of unenhanced magnetic resonance imaging (MRI) with diffusion-weighted imaging (DWI; NonMRI) for the detection of primary small (≤20 mm) pancreatic solid tumours and prediction of pancreatic ductal adenocarcinoma (PDAC) in comparison with pancreatic computed tomography (CT; PanCT) and pancreatic MRI with magnetic resonance cholangiopancreatography (PanMRI).
Methods and materials: The institutional review board approved this retrospective study and waived the requirement for informed consent. A total of 126 patients who underwent PanCT and PanMRI, including 94 small (≤20 mm) pancreatic tumours (51 PDACs, 34 neuroendocrine tumours [NETs], nine solid pseudopapillary tumours [SPTs]), and 32 patients with a normal pancreas, comprised the study population. Two observers assessed three sets of images: PanCT, PanMRI and NonMRI (T1- and T2-weighted images and DWI). Receiver operating characteristic curve analysis and diagnostic accuracy using the area under the receiver operating characteristic curve (Az) were used for statistical analysis.
Results: On NonMRI and PanMRI, all of tumours except one NET were detected, but eight tumours (six NETs, one PDAC, one SPT) were not detected on PanCT (p<0.01). For prediction of PDAC, the Az value of the NonMRI (0.884 for observer 1; 0.930 for observer 2) was comparable with PanCT (0.922; 0.924; p>0.05), and inferior to PanMRI (0.930; 0.977; p<0.05), but all of 51 PDACs were considered as probable or definite PDAC on NonMRI by both observers.
Conclusion: NonMRI showed better performance than PanCT, and competitive performance to PanMRI for the detection of primary small solid pancreatic tumours, and showed reasonable sensitivity for prediction of PDACs compared with PanCT and PanMRI.
Copyright © 2017 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.