대한유방검진의학회

Review Article

Full Text(PDF)

Automated breast ultrasound (ABUS) was developed to overcome the operator dependence and limited reproducibility of conventional handheld ultrasound while preserving its diagnostic strengths. Standardized image acquisition by technologists enables consistent coverage of the whole breast and more efficient interpretation by radiologists. In addition, multiplnar reconstructions—particularly the coronal view—provide a wide field of view that improves lesion detection and characterization, making ABUS a useful supplemental screening tool to mammography. However, the large volume of image data may prolong reading time and increase inter-reader variability. To address these challenges, artificial intelligence-based computer-aided diagnosis (AI-CAD) systems have introduced into ABUS workflow. By rapidly analyzing large datasets and highlighting suspicious lesions, AI-CAD has demonstrated improvements in workflow efficiency and diagnostic consistency. Prior studies have reported reduced reading time and improved diagnostic performance, with a particularly greater benefit among less-experienced readers. This article reviews current evidence regarding the clinical utility of AI-CAD for ABUS, focusing on clinical advantages, key considerations, and potential pitfalls.
Key Words: Breast; Automated breast ultrasound; Artificial intelligence; Computer-aided diagnosis