Artificial intelligence in breast ultrasound: a systematic review of research advances

Objective Through bibliometric visualization analysis, this study aims to summarize research progress in artificial intelligence (AI)-integrated ultrasound technology for breast cancer, reveal research hotspots, development trends, and international collaboration patterns, thereby providing references for clinical diagnosis and therapeutic decision-making. Methods Based on the Web of Science Core Collection (SCI-Expanded), we retrieved relevant literature from 2004-2025 (1,876 articles finally included). VOSviewer (v1.6.20), CiteSpace (v6.3.1 Basic), and Microsoft Excel 2019 were employed for visual analysis of publication volume, national/institutional collaboration, author networks, keywords, and co-citation relationships. Results Annual publications have shown a progressive increase since 2024. The United States (485 articles, 15,394 total citations) demonstrated the highest academic influence. Core researchers included Moon Woo Kyung (38 articles), while Seoul National University Hospital (47 articles) emerged as a key collaborative institution. Keyword clustering identified “deep learning”, “breast ultrasound”, and “machine learning” as research hotspots, with burst detection analysis revealing “deep learning” as the most prominent emerging theme (post-2020 surge). Radiology ranked as the most cited journal (4,258 citations), with foundational works by Berg WA (2008) and Al-Dhabyani W (2020) constituting the highest-impact literature. Conclusion AI-ultrasound integration is suggested to have potential for enhancing diagnostic accuracy in breast cancer, although global research still exhibits regional disparities. Future efforts should strengthen international collaboration, optimize deep learning-based imaging analysis, leverage big data for treatment optimization and prognosis prediction, while addressing technical challenges including data quality assurance and algorithm sharing mechanisms.