Differential diagnosis of pancreatic serous oligocystic adenoma and mucinous cystic neoplasm with spectral CT imaging: initial results.

ABSTRACT

AIM: To investigate the imaging characteristics of pancreatic serous oligocystic adenoma (SOA) and mucinous cystic neoplasms (MCNs) using spectral computed tomography (CT) and to evaluate whether quantitative information derived from spectral imaging can improve the differential diagnosis of these diseases. MATERIALS AND METHODS: From February 2010 to June 2013, 44 patients (24 SOAs and 20 MCNs) who underwent spectral CT imaging were included in the study. Conventional characteristics and quantitative parameters were compared between the two disease groups. Logistic regression was used for multiparametric analysis. The receiver-operating characteristic curve was used to evaluate the diagnostic performance of single parameter and multiparametric analysis. Two radiologists diagnosed the diseases blinded and independently, without and with the information of the statistical analysis. RESULTS: Tumour location, contour, size, and monochromatic CT values at 40 keV to 70 keV, iodine concentration, and effective atomic number (effective-Z) in the late arterial phase were the independent factors correlated with category. Multiparametric analysis with logistic regression showed that tumour size, location, and contour were the most effective variations, and obtained an area under the ROC curve (AUC) of 0.934. With the knowledge of statistical analysis, the accuracy of the first reader increased from 70.5% to 86.4%, and the accuracy of the second reader increased from 81.8% to 90.9%. CONCLUSIONS: Although CT spectral imaging provided additional information and multiparametric analysis obtained better performance than single-parameter analysis in differentiating MCNs from SOAs, multiparametric analysis with the combination of quantitative parameters derived from CT spectral imaging did not improve the diagnostic performance. Tumour size, location, and contour played an important role in differentiating MCNs from SOAs.