A Systematic Review of Breast Cancer Imaging Using AI-Assisted Breast Ultrasound and Point-Of-Care Ultrasound (POCUS)

Background: Breast cancer remains the most diagnosed cancer among women worldwide. Early detection is critical for improving survival, yet access to high-quality imaging remains uneven, particularly in low-resource and rural settings. Ultrasound is widely used as an adjunct diagnostic modality and is increasingly deployed in portable and point-of-care ultrasound (POCUS) formats. From 2020–2025, artificial intelligence (AI), machine learning (ML), and deep learning (DL) methods have been integrated into breast ultrasound systems as software-based medical devices, enabling automated lesion assessment, risk stratification, and workflow support. Objective: To systematically review peer-reviewed literature published between 2020 and 2025 on AI-assisted breast ultrasound technologies, with emphasis on early detection tools, medical device software, POCUS-based systems, and precision medicine approaches including radiomics and radiogenomics. Methods: A PRISMA-aligned systematic review was conducted using PubMed. Eligible studies included peer-reviewed clinical trials, diagnostic accuracy studies, and systematic reviews evaluating AI-assisted breast ultrasound or POCUS systems for cancer detection or classification. Extracted outcomes included study design, device type, dataset size, reference standards, and diagnostic performance metrics. Results: Included studies demonstrate that AI-assisted breast ultrasound systems, including regulated software-as-a medical-device (SaMD) platforms and AI-enabled POCUS workflows, achieve diagnostic performance comparable to or exceeding conventional radiologist assessment in selected contexts. That said, the existing studies are limited in number. Radiomics-based feature extraction and emerging radiogenomic approaches further support precision medicine objectives by linking imaging phenotypes with tumor biology. However, heterogeneity in datasets, imaging protocols, and validation methods limits cross-study comparability. Conclusion: Between 2020 and 2025, AI-assisted breast ultrasound evolved from experimental CAD tools into clinically evaluated medical device software, including applications in POCUS and low-resource environments. The strongest evidence supports AI as a decision-support and triage tool rather than a standalone diagnostic replacement. Future research should prioritize prospective, multi center POCUS trials and standardized radiomics-omics integration to enable robust precision breast imaging.