Recent advances in the field of artificial intelligence (AI) have the potential to revolutionize breast cancer (BC) screening. They could offer capability to enhance diagnostic accuracy and reduce the workload of radiologists. The global burden of BC highlights the need for innovative diagnostic tools to improve early detection rates. AI tools such as iCAD (ProFound AI v3.0, USA) have shown promise in assisting the interpretation of mammography images by identifying suspicious lesions and reducing human error. Although traditional screening methods have proven to be effective, there is still a substantial gap in the accuracy and efficiency of early detection of malignancies, especially in dense breast tissue.

Contrast-enhanced mammography (CEM), a relatively new imaging technique, has become a valuable tool in the detection and characterization of breast lesions. By combining standard mammography with the administration of an iodinated contrast agent, CEM improves the visibility of suspicious lesions by highlighting areas of increased vascularity. This dual-energy approach improves the detection of malignant tumors, especially in cases where dense breast tissue may obscure findings on standard mammograms. The integration of CEM with AI tools such as iCAD represents a significant advance in BC diagnostics, providing radiologists with a powerful combination for greater accuracy and efficiency.

The aim of this study was to retrospectively evaluate the accuracy of iCAD in the analysis of suspicious lesions categorized as BIRADS 0 and referred for CEM. By focusing on this subset of patients, the clinical utility of AI in differentiating benign from malignant lesions and its potential integration into routine diagnostic workflows will be determined.