A prospective study finds AI increases cancer detection by 13.8% in single-reader settings without raising recall rates, offering a new solution for countries relying on single-reader mammography screening.
Lunit, a provider of AI-powered solutions for cancer diagnostics and therapeutics, announced the publication of a prospective study in Nature Communications validating the real-world effectiveness of AI-powered mammography screening in South Korea’s national breast cancer screening program.
Conducted as a large-scale, multicenter prospective study in a single-reading setting, the research underscores how AI-assisted mammography interpretation significantly improves breast cancer detection rates without increasing recall rates.
The study was led by professor Yun-Woo Chang, MD, PhD, from Soonchunhyang University Seoul Hospital, in collaboration with both breast and general radiologists from six academic hospitals in South Korea. The study analyzed 24,543 women aged 40 and above who underwent routine biennial mammography (2D full-field digital mammogram) between February 2021 and December 2022 as part of Korea’s national breast cancer screening program.
The study compared the performance of breast radiologists interpreting screening mammograms with and without Lunit’s AI-based computer-aided detection, Lunit INSIGHT MMG. The results revealed that AI-assisted breast radiologists detected 13.8% more screen-detected breast cancers than those relying solely on conventional interpretation methods, or without AI assistance.
The cancer detection rate increased from 5.01 (per 1,000) to 5.70 with AI assistance, while recall rates remained statistically unchanged, ensuring improved clinical effectiveness and minimizing unnecessary additional recalls. Moreover, AI assistance led to a significant improvement in detecting small-sized tumors and node-negative cancers, key indicators of improved early detection.
Additionally, the study evaluated Lunit INSIGHT MMG’s impact on general radiologists who do not specialize in breast imaging through a simulated retrospective study. The results demonstrated that AI assistance led to an even greater improvement in cancer detection rates for general radiologists, with an increase of 26.4% (3.87 to 4.89 per 1,000), underscoring AI’s potential to support radiologists with varying levels of expertise.
“This prospective study offers compelling real-world evidence that AI can improve early cancer detection while maintaining effectiveness and reducing unnecessary patient anxiety caused by false positives in a single-reading setting,” says Chang, the lead and corresponding author of the study, in a release.
A Solution for Single and Double-Reading Screening Systems
The publication further builds upon Lunit INSIGHT MMG’s track record in improving breast cancer detection. In Sweden, Capio St. Göran Hospital conducted a prospective clinical trial to assess Lunit AI’s effectiveness in breast cancer screening when replacing one reader within a double-reading setting. Following positive trial results, Capio St. Göran Hospital became the world’s first to replace one of two radiologists in its double-reading workflow with AI.
“This latest prospective study in South Korea further solidifies Lunit INSIGHT MMG’s potential as an indispensable tool in breast cancer screening,” says Brandon Suh, CEO of Lunit, in a release. “Whether in a single-reader or double-reader setting, AI can serve as a powerful force multiplier, assisting radiologists in detecting cancer earlier and more accurately.”
Shaping the Future of National Breast Cancer Screening Programs
With mounting evidence supporting AI’s role in breast cancer screening. Lunit INSIGHT MMG has already been deployed in national breast screening programs in Australia, Sweden, Iceland, Singapore, Saudi Arabia, and Qatar, helping healthcare systems enhance efficiency and optimize diagnostic accuracy.
The study’s findings are expected to influence global breast cancer screening protocols, particularly in countries where single-reading mammography interpretation is widely practiced.
By demonstrating AI’s ability to enhance radiologists’ accuracy without increasing recall rates, the study provides critical data supporting AI’s role in optimizing cancer detection while reducing the workload burden on radiologists. This is particularly relevant as many countries face a shortage of specialized breast imaging professionals, making AI an important tool to improve efficiency and maintain high diagnostic standards.
“As AI adoption accelerates, we remain committed to driving innovation and making AI-powered cancer screening the new standard of care,” says Suh in a release. “By working closely with healthcare providers worldwide, we aim to ensure that AI benefits as many patients as possible.”
Photo caption: Lunit’s AI-powered mammography analysis solution, Lunit INSIGHT MMG
Photo credit: Lunit
