RadNet, Inc., the largest provider of outpatient diagnostic imaging services in the United States, has announced groundbreaking results from the most extensive real-world analysis of AI-driven breast cancer screening ever conducted in the country. This study, published in Nature Health, showcases the effectiveness of DeepHealth’s AI technology in enhancing cancer detection rates while ensuring equitable outcomes for various patient demographics.
The AI-Supported Safeguard Review Evaluation (ASSURE) study analyzed data from over 579,000 mammograms collected from 109 community-based imaging centers across California, Delaware, Maryland, and New York. By comparing traditional 3D mammography to an AI-enhanced protocol that includes DeepHealth’s FDA-cleared computer-aided detection and diagnosis (CADe/x) software and a specialized Safeguard Review workflow, the study highlights significant improvements in detection rates.
Dr. Howard Berger, President and CEO of RadNet, emphasized the unique aspects of this research. He noted its unprecedented scale, diverse patient population, and the real-world implications of its findings. The study revealed a remarkable 21.6% increase in the cancer detection rate using the AI-driven protocol compared to standard 3D mammography, all while adhering to recall rates established by the American College of Radiology. Additionally, the positive predictive value saw an increase of 15%.
Importantly, the methodology employed in the ASSURE study highlighted the availability of this AI-enabled workflow for all patients at no additional cost during the study, eliminating selection bias. This approach ensures that women across all backgrounds, particularly those from historically underserved communities, gain access to higher-quality care. Notably, the findings are especially significant for Black women, who historically face a 40% higher mortality rate from breast cancer in the U.S. The study identified a 22.7% improvement in cancer detection for women with dense breasts—those who often encounter diagnostic challenges and increased cancer risks.
Dr. Gregory Sorensen, co-author of the study and Chief Science Officer at RadNet, reiterated the importance of conducting research in community settings where most women receive their mammograms. The early detection of breast cancer can provide women with a broader range of treatment options, ultimately improving health outcomes.
The Enhanced Breast Cancer Detection (EBCD™) program, utilizing this AI technology, was launched across RadNet-affiliated centers in 2023. It is designed to assist in detecting difficult-to-find lesions and to enhance the overall effectiveness of breast cancer screenings.
RadNet and DeepHealth’s commitment to innovation is evident, as they continue to leverage AI to improve radiology services, aiming for earlier and more accurate disease detection. This progress not only holds promise for enhancing patient care but also offers hope for reducing breast cancer disparities among various populations.
In a broader context, RadNet’s extensive operations, which include over 400 outpatient imaging centers, underscore its pivotal role in the U.S. diagnostic imaging landscape, emphasizing the need for equitable healthcare solutions for all communities. With the integration of AI technologies, RadNet aims to push the boundaries of what is possible in breast cancer screening, ultimately striving to save lives through timely and effective detection.