Sarica Ozgur, Kaymaz Goksen, Ucar Nese
From the Department of Radiology, Taksim Education and Research Hospital, Istanbul, Turkey.
J Comput Assist Tomogr. 2014 Nov-Dec;38(6):985-91. doi: 10.1097/RCT.0000000000000132.
This study aims to evaluate the strength of magnetic resonance (MR) lesion descriptors for malignancy and to determine the effectiveness of a scoring system that combines BI-RADS parameters with additional criteria.
Three hundred thirty-nine histopathologically proven lesions that had undergone MR imaging were analyzed retrospectively. Based on the Fischer scoring system, an optimal cutoff value was calculated for combined parameters. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated for all lesions with MR BI-RADS classification without using additional parameters. Finally, the results of the scoring system and MR BI-RADS classification were compared.
The optimal cutoff value according to the total score was calculated as 5. The sensitivity and the specificity of BI-RADS classification were calculated to be 94.20% and 56.12%, respectively. The scoring system using a cutoff value of 5 resulted in a little loss of sensitivity (89.86%) but resulted in a reasonable increase in the specificity (88.49%).
Additional parameters can improve the specificity of MR imaging. T2-weighted signal features, adjacent vessel sign, unilateral-focal edema, and hook sign were considered as effective parameters.