#United States #New York #Breast Cancer #NYU Langone #AI Imaging

AI-Assisted Imaging Tool for Breast Cancer

In a groundbreaking study, researchers at NYU Langone Health and its Perlmutter Cancer Center have developed an advanced artificial intelligence (AI) model designed to enhance breast cancer screening. This innovative tool, known as NYU-DRP, utilizes data from women’s annual 3D mammograms over several years to predict the five-year risk of developing breast cancer with remarkable accuracy. Unlike traditional methods that typically analyze only the most recent mammogram, NYU-DRP offers a more comprehensive assessment by considering how breast tissue has changed over time.

The research, published online in the American Journal of Roentgenology, highlights the model’s superior predictive capabilities. According to the findings, NYU-DRP successfully identified women at higher risk for breast cancer with a 72% accuracy rate, notably outperforming other methods including single 3D mammograms (70%) and AI-assisted 2D mammograms (68%).

The Development of NYU-DRP

NYU-DRP was crafted using a colossal dataset comprising 313,531 yearly 3D mammograms from 161,165 women who were not diagnosed with breast cancer during screenings conducted between 2016 and 2020. The model's superior efficacy was further validated through comparisons with the Tyrer-Cuzick risk assessment, a widely recognized tool that estimates a woman’s lifetime risk for breast cancer. NYU-DRP surpassed Tyrer-Cuzick, successfully predicting higher risk in 67% of cases compared to 56% for Tyrer-Cuzick.

While Tyrer-Cuzick relies on personal and familial medical histories—such as age, genetic predispositions, and breast density—NYU-DRP’s reliance on mammogram data allows it to capture fluctuations in the anatomy and structure of breast tissue that may indicate increased risk.

Key Findings

Among the study's key revelations was the inadequacy of breast density alone as a risk predictor. Contrary to established beliefs, the model classified 37.6% of women with extremely dense breasts as average risk, although only 0.7% were actually diagnosed with breast cancer after five years. Conversely, 15.5% of women with less dense breast tissue were classified as high risk, with a higher actual incidence of cancer at 2.5%. These observations emphasize the nuanced understanding of breast cancer risk beyond traditional metrics like breast density.

Study co-investigator Dr. Laura Heacock asserts that by harnessing repeated 3D mammography data, healthcare providers can better tailor screening strategies based on actual risk levels. This approach avoids unnecessary tests for low-risk individuals while ensuring that high-risk patients receive the monitoring they need.

Future Directions and Implications

Looking ahead, the research team plans to explore the longitudinal impacts of the NYU-DRP model on women’s breast health over time. They intend to expand their data collection to include information from other academic health institutions and a variety of 3D mammography technologies, enhancing the model’s robustness and adaptability.

As breast cancer persists as one of the most common cancers among women in the U.S.—with approximately 382,640 expected new diagnoses in 2026—the development of predictive tools like NYU-DRP could fundamentally change how screening is conducted. Currently, it is estimated that 1 in 8 women in the U.S. will receive a breast cancer diagnosis, making early detection paramount; with advancements in early assessment methodologies such as this, survival rates continue to improve, with over 4 million breast cancer survivors presently in the country.

Annual mammography screening is recommended to commence at age 40, with over 43 million mammograms performed in 2025 alone. The implications of AI in this domain could revolutionize the medical approach and lead to better outcomes for countless women.

Funding for this progressive study was generously provided by several institutions, including the National Science Foundation and the National Institutes of Health, ensuring that advancements in breast cancer screening continue to grow from research to real-world application.