Published in Radiology, paper shows general radiologists achieve cancer detection rates comparable to breast imaging specialists with integration of DeepHealth’s AI-powered workflow
A new study published in Radiology shows that DeepHealth’s AI-powered breast cancer screening workflow is associated with improved cancer detection among general radiologists, making their performance comparable to that of breast imaging specialists using the same workflow.1 The adjusted cancer detection rate among general radiologists increased by 32.7%, from 3.76 to 4.99 cancers per 1,000 examinations, reaching a level comparable to that observed among fellowship-trained breast imaging specialists using the same workflow.
With general radiologists currently interpreting approximately 70% of U.S. screening mammograms, the findings suggest that AI-powered workflows could help radiologists detect more cancers, support more consistent screening for patients, and enable health systems to extend access to breast imaging specialist expertise across more sites.
The research builds on the ASSURE study published in Nature Health, the largest real-world analysis of AI-powered breast cancer screening in the United States that demonstrated a 21.6% higher cancer detection rate enabled by DeepHealth’s novel AI-powered breast cancer detection workflow. The findings supported the clinical effectiveness and benefit of DeepHealth’s AI technology to deliver equitable results across several racial, ethnic and breast density patient groups.2 Drawing from the same large-scale, real-world implementation, the new Radiology study extends those findings by showing how the workflow helped general radiologists achieve cancer detection rates comparable to those of fellowship-trained breast imaging specialists.
The study evaluated more than 577,000 screening mammograms interpreted by 95 radiologists across 109 imaging sites in the United States. The workflow combined DeepHealth’s computer-aided detection and diagnosis software, now branded Mammo Dx,3 with an AI-supported Safeguard Review, which uses AI to route suspicious screening exams that are not initially recalled by the interpreting radiologist and route them for additional expert review.
Among general radiologists, the adjusted recall rate increased from 9.06% to 10.40%, meaning that more patients were called back for additional evaluation. At the same time, the positive predictive value of recalls increased from 3.38% to 3.89%. Together, the findings show that although more patients were recalled for additional evaluation, a greater proportion of those recalls resulted in a cancer diagnosis.
“The goal of routine breast screening is to find cancers as early as possible, when patients have the best opportunity for timely, effective treatment,” said Christoph Lee, M.D., breast imaging specialist and study author. “What is compelling about this study is that it shows how AI-powered solutions can support general radiologists in detecting more cancers across a large, real-world screening population. This workflow offers a practical way to expand access to specialist-comparable cancer detection while keeping the interpreting radiologist’s judgment at the center of every decision.”
“With limited breast imaging specialists around the country, AI-driven workflows can help extend specialized expertise across screening sites nationwide and support general radiologists in detecting more cancers,” said Niccolò Stefani, M.D., Business and Product Leader, Clinical AI, DeepHealth.“By enabling more patients to benefit from high-quality breast screening, regardless of where they receive care, this approach has the potential to support earlier diagnosis and, ultimately, better patient outcomes.”
- McCabe et al. Closing the Performance Gap between Generalists and Breast Imaging Specialists Using a Nationally Deployed AI Workflow for Screening Mammography. Radiology. July 21, 2026.
- Landmark Nature Health Study Demonstrates the Effectiveness of DeepHealth’s Novel AI-Powered Breast Cancer Detection Workflow. DeepHealth. November 17, 2025.
- The FDA-cleared software previously known as ProFound Pro is now marketed as Mammo Dx including priors. Mammo Dx retains the FDA-cleared capabilities of ProFound Pro and serves as the foundation for ongoing innovation, with additional features and enhancements being introduced over time.
