Use of US BIRADS for the Differentiation of Male Breast Masses and Interobserver Agreement Assessment

Abstract

Purpose To investigate the efficiency of Breast Imaging Reporting and Data System (BIRADS) proposed by the American College of Radiology for male breast lesions and to evaluate assess inter-observer agreement. Methods Six breast radiologists were divided into three groups to retrospectively analyze 90 male breast nodules by BIRADS classifications. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and accuracy were calculated for comparative analysis. Interobserver agreement was assessed. Results The group II showed the higher specificity (75.9%) than other two groups while the group III showed the best AUC, sensitivity and accuracy, which were 0.904, 97.2%, 62.9% and 76.7%, respectively. Besides, the whole of three groups expressed excellent ICC on 2(0.934) category and good on BIRADS 3 (0.677) category. The ICC of 4a (0.379) was expressed as poor level. The categories of 4b (0.483) ,4c (0.521) and5(0.491) was expressed as fair level. Only group I had significant difference with the group III in diagnostic effectiveness of the same systems among three groups. Conclusions Male sex did not have a large impact on the diagnosis of BIRADS and use of US BIRADS lexicon shows excellent or good levels of concordance for mass characterizations. But a lower concordance respect to BIRADS classifications in BIRADS 4 categories.