AI Tool Predicts Breast Cancer Risk Years in Advance Using Mammogram Data

An artificial intelligence (AI) tool is showing promise in predicting a woman's risk of developing breast cancer years before it might otherwise be detected. Developed by researchers at NYU Langone Health, the AI model analyzes years of a woman's 3D mammograms to track changes in breast tissue, a departure from traditional methods that often focus only on the latest scan.

This AI technology has the potential to identify women at increased risk who might otherwise go unnoticed, according to breast imaging specialists. While conventional risk assessments consider factors like family history and genetic mutations, AI can uncover additional clues within the breast tissue itself. "A mammogram can tell us more than whether cancer is visible today — it can also provide information about a woman's risk of developing cancer in the years ahead," said Dr. Connie Lehman, a breast imaging specialist and professor of radiology at Harvard Medical School, who was not involved in the study.

IMAGE1:

The AI model was trained on more than 313,000 mammograms from over 161,000 women. The researchers believe this approach could eventually help tailor breast cancer screening to individual risk levels. Women identified as high-risk might discuss additional screenings, such as MRI, with their healthcare providers, while those at average risk could continue with routine mammograms.

Furthermore, the AI could analyze previously acquired mammograms, meaning the risk assessment would not necessitate additional imaging exams or radiation exposure for the patient. "A higher score can prompt a conversation with your health care provider about additional screening and risk-reduction options," Lehman advised. "A lower score does not eliminate risk or replace regular mammography."

IMAGE2:

Experts caution that an AI-generated risk score is not a cancer diagnosis and cannot definitively predict if a woman will develop the disease. The NYU model is currently a research tool and has not yet received FDA authorization for clinical use. The study also highlighted the need for further testing, as it utilized data from NYU Langone hospitals and equipment from a single manufacturer. The research team plans to validate the model with data from diverse medical centers and different 3D mammography systems.