Comparison of retraction phenomenon and BI-RADS-US descriptors in differentiating benign and malignant breast masses using an automated breast volume scanner

• We compare the diagnostic performance of retraction phenomenon with BI-RADS-US descriptors using ABVS.
• Retraction phenomenon has high diagnostic performance with the highest specificity and odds ratio.
• Microlobulated margins have high diagnostic performance with the highest sensitivity and area under the ROC curve.
• We suggest retraction phenomenon be written to the BI-RADS-US lexicon in the future.
• Microlobulated margins require additional attention in ABVS image interpretation.

ObjectiveTo compare the diagnostic values of retraction phenomenon in the coronal planes and descriptors in the Breast Imaging Reporting and Data System-Ultrasound (BI-RADS-US) lexicon in differentiating benign and malignant breast masses using an automated breast volume scanner (ABVS).Materials and methodsTwo hundred and eight female patients with 237 pathologically proven breast masses (120 benign and 117 malignant) were included in this study. ABVS was performed for each mass after preoperative localization by conventional ultrasonography (US). Multivariate logistic regression analysis was performed to assess independent variables for malignancy prediction. Diagnostic performance was evaluated through the receiver operating characteristic (ROC) curve analysis.ResultsRetraction phenomenon (odds ratio [OR]: 76.70; 95% confidence interval [CI]: 12.55, 468.70; P < 0.001) was the strongest independent predictor for malignant masses, followed by microlobulated margins (OR: 55.87; 95% CI: 12.56, 248.44; P < 0.001), angular margins (OR: 36.44; 95% CI: 4.55, 292.06; P = 0.001), calcifications (OR: 5.53; 95% CI: 1.34, 22.88; P = 0.018,) and patient age (OR: 1.10; 95% CI: 1.03, 1.17; P = 0.004). Mass shape, orientation, echo pattern, indistinct margins, spiculated margins, and mass size were not significantly associated with breast malignancy. Area under the ROC curve (Az) for microlobulated margins and retraction phenomenon was higher than that for other significant independent predictors. Az, sensitivity, and specificity were 0.877 (95% CI: 0.829, 0.926) and 0.838 (95% CI: 0.783, 0.892), 82.9% and 70.1%, and 92.5% and 98.3%, respectively, for microlobulated margins and retraction phenomenon.ConclusionsRetraction phenomenon and microlobulated margins have high diagnostic values in the differentiation of benign and malignant breast masses using an ABVS.