Diffusion-weighted MR Imaging for the Differentiation of Malignant from Benign Strictures in the Periampullary Region

Objectives: To investigate the value of DWI for differentiating malignant from benign strictures in the periampullary region.

Methods: We retrospectively analysed data from 78 patients who had undergone magnetic resonance cholangiopanreatography (MRCP) and diffusion-weighted imaging (DWI), in whom biliary strictures in the periampullary region were suspected. Twenty-two malignant and 56 benign lesions were included. One radiologist compared the signal intensity of malignant and benign periampullary lesions on DWI using b = 500 and 800 s/mm(2). The signal intensity of bile was also compared, and an optimal b value was determined for periampullary lesions. Two other radiologists reviewed MRCP alone and combined DWI and MRCP for the possibility of malignant periampullary lesions. Diagnostic accuracy was calculated for each reviewer by receiver operating characteristic (ROC) curve analysis.

Results: Malignant periampullary lesions more frequently appeared hyperintense than benign lesions on DWI using the two b values (P < 0.001). Bile more frequently appeared hyperintense on DWI using b = 500 s/mm(2) (87.2 %) than b = 800 s/mm(2) (24.4 %). Therefore, b = 800 s/mm(2) was determined as the preferred sequence. Diagnostic accuracy for malignant periampullary lesions improved for both reviewers after adding DWI; from 0.714 to 0.924 (P = 0.006, for reviewer 1) and from 0.714 to 0.919 (P = 0.007, reviewer 2).

Conclusions: Combined DWI with MRCP can improve the diagnostic accuracy for differentiating malignant from benign strictures in the periampullary region.

Key Points: • Diffusion-weighted magnetic resonance imaging provides yet more information about hepatobiliary structures. • Diffusion-weighted imaging (DWI) has now been applied to the biliary tree. • Most periampullary carcinomas appear hyperintense on high b value DWI. • DWI can help differentiate between malignant and benign periampullary lesions.