PET-MR Imaging and MR Texture Analysis in the Diagnosis of Pancreatic Cysts: A Prospective Preliminary Study.

Abstract

RATIONALE AND OBJECTIVES\nOur aim was to evaluate the capability of textural and metabolic parameters measured at pretreatment 18F-fluorodeoxyglucose Positron emission tomography (PET)-MR in differentiating malignant from benign pancreatic cystic lesions.\n\n\nMATERIALS AND METHOD\nForty consecutive patients were prospectively enrolled in this study. They underwent simultaneous PET-MR for the diagnosis of pancreatic cysts. Thirty texture parameters were extracted from manually contoured axial T2-weighted imaging with fat suppression (T2FS) and apparent diffusion coefficient images, respectively. Maximal and mean standardized uptake values (SUVmax and SUVmean, respectively) of pancreatic cysts were measured at PET-MR imaging. The Mann-Whitney test was used to compare both textural and metabolic parameters between benign and malignant group.\n\n\nRESULTS\nFDG uptake was significantly higher in patients with malignant pancreatic cysts (SUVmaxp\u202f=\u202f0.002, SUVmeanp < 0.001). Malignant cysts showed significantly lower standard deviation for spatial scaling factor at 3-6mm on T2FS images and lower skewness for spatial scaling factor at 2-4mm on apparent diffusion coefficient images (p < 0.01). SUVmean had the highest Area under the curve of 0.892 on receiver-operating characteristic analysis with a sensitivity, specificity, and accuracy of 88.9%, 87.1%, and 87.6%, respectively. When metabolic and textural features were combined into a single diagnostic model, the AUC increased to 0.961, with a sensitivity, specificity, and accuracy of 88.9%, 96.8%, and 95.0%, respectively.\n\n\nCONCLUSION\nOur study implied that PET-MR showed no obvious advantages over traditional PET-related imaging in differentiating malignant from benign pancreatic cystic lesions. Diagnostic model based on the combination of metabolic and textural parameters showed satisfactory performance.