Non-gaussian models of 3-Tesla diffusion-weighted MRI for the differentiation of pancreatic ductal adenocarcinomas from neuroendocrine tumors and solid pseudopapillary neoplasms

•The DWI model yielded AUCs of 0.87 and 0.86 for distinguishing PDACs from PNETs in primary and validation groups.•This DWI model achieved AUCs of 0.916 and 0.891 for differentiating PDACs from SPNs in primary and validation groups.•The error rates were reduced when radiologists were assisted by the DWI models.