PURPOSE: to analyze which characteristics proposed by the BIRADS lexicon for ultrasound have the greatest impact on distinguishing between benign and malignant lesions. METHODS: ultrasonography features from the third edition of the BIRADS were studied in 384 nodes submitted to percutaneous biopsy from February 2003 to December 2006, at the Medical School of Botucatu. For the ultrasonography, the equipment Logic 5 with a 7.5-12 MHz multifrequential linear transducer was used. The ultrasonography analysis of the node considered the features proposed by the BIRADS lexicon for ultrasound. The data were submitted to statistical analysis by the logistic regression model. RESULTS: the benign lesions represented 42.4% and the malignant, 57.6%. The logistic regression analysis found an odds ratio (OR) for cancer of 7.69 times when the surrounding tissue was altered, of 6.25 times when there were microcalcifications in the lesions interior, of 1.95 when the acoustic effect is shadowing, of 25.0 times when there was the echogenic halo, and of 7.14 times when the orientation was non-parallel. CONCLUSIONS: among the features studied, the lesion limit, represented by the presence or not of the halogenic halo, is the most important differentiator of the benign from the malignant masses.
Breast neoplasms; Biopsy; Logistic models; Breast diseases; Radiology information systems
