AI-Based Risk Models May Change the Process of Breast Cancer Screening
A recent study published in Nature Medicine examines the role of AI in improving screening programs by advancing early detection of breast cancer while reducing over-screening.
Breast cancer is currently the most diagnosed cancer and the leading cause of cancer-related deaths among women worldwide, with 281,550 new cases of invasive breast cancer among women in the United States in 2021 that led to 43,600 deaths.
Although mammography is typically used for screening and early detection of breast cancer, there are limitations, especially for women with dense breast tissue, and the service is not always accessible, especially in limited-resource settings, due to the high cost of the equipment.
Ultrasounds also play an important role in breast cancer diagnosis but still pose challenges for radiologists in evaluating the images to determine if the findings are benign, or if a short-term follow-up is needed.
Over a decade ago, computer-aided diagnosis, known as CAD systems, were proposed to assist radiologists in interpreting breast exams and recent advances in deep learning have facilitated the development of AI systems for automating diagnosis of breast cancer from U.S. images.
Using the developments, a team of researchers from New York previously developed an AI system using New York University’s breast ultrasound dataset that consisted of 5,442,907 images within 288,767 breast exams in the U.S. from 143,203 patients who were examined between 2012 and 2019 at the NYU Langone Health system.
Researchers found that the AI system maintained a high diagnostic accuracy among all age groups, mammographic breast density and U.S. device manufacturers.
Alexa can be reached at [email protected]