We wanted to determine if the analysis and categorization of breast lesions with using BI-RADS for US would be useful.

We retrospectively reviewed 1,024 cases (976 persons), in which US-guided core needle biopsy had been performed. The breast lesions were described and categorized according to the BI-RADS for US. Each US characteristic was analyzed in order to determine its association with a benign versus malignant tissue diagnosis. The diagnostic sensitivity, specificity, positive predictive value and negative predictive value, and the accuracy of breast US were examined.

Logistic regression analyses of the US features in terms of their ability to distinguish between benign and malignant breast lesions showed that all eight US features (shape, orientation, margin, echogenicity, posterior acoustic features, surrounding tissue, calcification and special cases) were statistically significant. Moreover, the shape, margin, microcalcification and enlarged axillary lymph nodes were found to be the most useful diagnostic features. In addition, a clinically palpable mass increased the diagnostic accuracy up to 92.7%. The positive predictive value of a category 3 lesion was 0.8%, and those of category 4 and 5 lesions were 30.4% and 99.4%, respectively. The diagnostic sensitivity, specificity, positive predictive value, negative predictive value and accuracy of breast US were 99.0%, 57.2%, 49.5%, 99.3% and 69.6%, respectively.

Using BI-RADS for US in breast can be successful for characterizing and differentiating both the malignant and benign lesions of the breast.