A Review of Artificial Intelligence Models for Detecting Breast Arterial Calcification on Mammograms and Their Clinical Implications.

Breast arterial calcification (BAC), traditionally regarded as an incidental mammogram finding, is now recognized as a clinically significant marker associated with cardiovascular disease (CVD), particularly in women. Despite this association, BAC remains underreported in clinical practice due to the lack of standardized screening protocols and the manual burden of identification. With CVD being the leading cause of mortality among women worldwide and traditional risk calculators often underestimating female cardiovascular risk, the potential of BAC as a surrogate biomarker is increasingly being explored. Mammography, already a widely used screening tool for breast cancer, offers an opportunity to identify BAC and thereby enable dual-purpose screening for both breast and cardiovascular health. Recent advancements in artificial intelligence (AI) and machine learning (ML), particularly in deep learning models such as convolutional neural networks (CNNs), have shown tremendous promise in detecting and quantifying BAC with high accuracy. Models such as DU-Net, difference-of-Gaussian generative adversarial network (DoG-GAN), and Simple Context U-Net (SCU-Net) utilize U-Net architectures optimized for segmentation and demonstrate performance metrics that rival or surpass human experts. Other approaches, including hybrid models, transfer learning, and ensemble methods, have also achieved strong diagnostic metrics, improving the reliability and scalability of BAC detection. This review consolidates findings from recent studies and technical innovations, evaluating various ML algorithms and their applications in automating BAC identification. In doing so, it highlights the potential of AI to address the long-standing challenge of underreporting and inconsistent quantification of BAC. The clinical implications of AI-enhanced BAC detection are significant. Accurate, automated identification of BAC can improve cardiovascular risk stratification, especially in women whose disease may otherwise go unnoticed by traditional tools. Moreover, at a population level, integrating BAC detection into routine mammogram workflows could yield substantial public health benefits, enabling earlier interventions and reducing overall healthcare costs. By consolidating current models and emphasizing the need for standardized reporting, this review aims to support the integration of AI-based BAC detection into routine clinical practice, thereby enhancing both diagnostic accuracy and preventive care for cardiovascular disease.