AI-Enabled POCUS for Breast Cancer Risk Stratification

This original research article evaluates the diagnostic performance of an AI-enabled point-of-care ultrasound (POCUS) system, referred to as Breast AI, in predicting malignancy among women with palpable breast abnormalities. Conducted at Groote Schuur Hospital, the prospective cohort study involved women aged 25 years and older who presented with suspicious breast lesions. The study aimed to assess the effectiveness of Breast AI in providing real-time malignancy risk scores compared to histopathological results. The findings indicate that Breast AI achieved a sensitivity of 67.2%, specificity of 79.4%, and a positive predictive value of 70.3% at a 51% threshold. The area under the curve (AUC) was calculated at 0.76, suggesting moderate discriminatory performance. The study also developed a three-tiered risk model for malignancy prediction. Overall, the research supports the integration of AI into POCUS to enhance breast cancer detection and clinical decision-making, particularly in resource-limited settings.