A breast sonography classification proposal - Revista Brasileira de Ginecologia e Obstetrícia

PURPOSE: the technological improvements in image quality have increased the importance of ultrasound as an imaging method in the study of breast pathologies. The need for a standardized method for lesion characterization, description and reporting in image analysis motivated the development of a breast sonographic report classification system. METHODS: the classification grouped the breast sonographic images in five classes: I – normal; II – benign; III – indeterminate, IV – suspect, and V – highly suspect. The used morphologic ultrasound features were shape, border, contour, echogenicity, echotexture, sound transmission, orientation, and secondary signals. The gold standard test, in the study of 450 lesions, considered sonographic follow-up of the lesions for a period from 6 to 24 months and the histopathology of surgical cases. RESULTS: breast sonographic classification for the diagnosis of breast cancer showed a sensitivity of 90.2% (CI: 82.8-94.9%), a specificity of 96.2% (CI: 94.0-97.6%), a positive predictive value of 84.1% (CI: 76.0-89.9%), and a negative predictive value of 97.8% (CI: 95.9-98.9%), obtaining an accuracy of 95.1%. CONCLUSIONS: the adoption of a sonographic classification system results in the standardization and optimization of the reports. It also aids the comparison with clinical findings, histopathological tests and breast images, avoiding unnecessary procedures and therefore leading to more adequate therapeutical management.