The aim of our study was to determine the positive predictive value (PPV) and to evaluate the usefulness of ultrasound BI-RADS, as compared with the histologic results of breast lesions that were categorized and classified by the ultrasound BI-RADS lexicon.

Between January and December 2004, the ultrasound features of 471 breast lesions in 368 patients were analyzed and categorized with using ultrasound BI-RADS. All of the lesions were compared with the histological results. We categorized category 2 and 3 lesions as benign lesions, and category 4 and 5 lesions as malignant lesions. We then calculated the sensitivity, specificity, positive predictive value, negative predictive value and accuracy.

The breast lesions were histologically diagnosed as 298 cases of category 2, 21 cases of category 3, 108 cases of category 4 and 44 cases of category 5. The categorical malignancy rate was 1% (3/298) in category 2 lesions, 4.7% (1/21) in category 3 lesions, 8.3% (9/108) in category 4 lesions and 90.9% (40/44) in category 5 lesions. Within category 4, the malignancy rate for category 4a lesions was 5.4% (5/92), the malignancy rate for category 4b lesions was 1.3% (1/8) and the malignancy rate for category 4c lesions was 50% (4/8). The sensitivity, specificity, positive predictive value and negative predictive value were 90.6%, 75.1%, 31.6% and 98.4%, respectively. The statistically valid ultrasound features that were correlated with malignancy were an ill defined margin, a microlobulated mass, an irregular mass, inhomogenic echogenicity, an echogenic halo and an older patient age.

Ultrasound BI-RADS was useful in differentiating benign from malignant breast lesions.