A Comparative Study of the Diagnostic Performance of Evaluating Breast Masses for Breast Surgeons versus S-DetectTM(Samsung Medison Co., Ltd, Seoul, Korea)

Abstract

Purpose: Ultrasonography is widely used for examining breast mass. We used the Breast Imaging-Reporting and Data System (BI-RADS) to characterize breast lesions found on ultrasonography. Among various ultrasound techniques, we used S-DetectTM (Samsung Medison Co., Ltd, Seoul, Korea), which supports the morphological analysis of breast masses found according to BI-RADS. In addition, we compared the breast surgeons’ categorization of breast masses with that by S-DetectTM. Methods: Breast surgeons evaluated the breast masses found using ultrasonography between April 2016 and December 2016. A total of 139 masses, which were categorized as BI-RADS 3 or 4, from 112 patients were reevaluated by S-DetectTM before performing vacuum-assisted resection or surgical excision. Results: Of the 139 masses, 118 were benign tumors and 21 were malignant tumors. With regard to the diagnostic performance, the sensitivity of categorization was 95% for breast surgeons, but the sensitivity was relatively lower for S-detectTM (85%). However, the specificity and accuracy of S-detectTM were 70.6% and 74.1%, respectively, which were higher than those values obtained from breast surgeons (18.5% and 30.9%, respectively). Conclusion: S-detectTM can be used by breast surgeons as a diagnostic aid when evaluating and diagnosing breast masses found on ultrasonography.

Keywords

• breast
• ultrasonography
• computer-assisted diagnosis
• artificial intelligence