Lunit, a company specializing in AI-powered solutions for cancer diagnostics, has announced the publication of a study in Nature Communications demonstrating the effectiveness of its AI mammography analysis solution, Lunit INSIGHT MMG, in a real-world screening setting. The study, described as the world’s first large-scale, multicenter prospective study in a single-reading mammography environment, was conducted within South Korea’s national breast cancer screening program.
Study findings
The study, led by Professor Yun-Woo Chang from Soonchunhyang University Seoul Hospital and involving breast and general radiologists from six academic hospitals in South Korea, analyzed data from 24,543 women aged 40 and above who underwent routine biennial mammography screenings between February 2021 and December 2022. The research compared the performance of breast radiologists interpreting mammograms with and without the assistance of Lunit INSIGHT MMG.
Key findings of the study include:
- Increased Cancer Detection Rate (CDR): AI-assisted radiologists detected 13.8% more screen-detected breast cancers than those relying solely on traditional interpretation methods. The CDR increased from 5.01 to 5.70 (per 1,000 screenings) with AI assistance.
- Unchanged Recall Rates (RRs): The use of AI did not lead to a statistically significant increase in recall rates, indicating improved clinical effectiveness without an increase in unnecessary follow-up procedures.
- Improved detection of early-stage cancers: AI assistance resulted in a significant improvement in the detection of small-sized tumors and node-negative cancers, suggesting enhanced early detection capabilities.
- Benefits for general radiologists: A simulated retrospective study component showed that AI assistance led to an even greater improvement in CDRs for general radiologists (a 26.4% increase), highlighting AI’s potential to support radiologists with varying levels of experience.
The results suggest that AI can serve as a valuable tool in single-reader mammography settings, common in many countries, by enhancing radiologists’ accuracy without increasing the number of false positives. This is particularly significant given the global shortage of specialized breast imaging professionals. Lunit INSIGHT MMG has already been implemented in national breast screening programs in several countries, including Australia, Sweden, Iceland, Singapore, Saudi Arabia, and Qatar. The study builds on previous research, including a trial at Capio St. Göran Hospital in Sweden, where AI replaced one radiologist in a double-reading workflow.
Featured image credit: Lunit
