News
Article
AI Improves Breast Cancer Screening and Detection Rates in Real-World Study
Author(s):
Key Takeaways
- AI in breast cancer screening improved detection rates without increasing false positives, enhancing accuracy and efficiency compared to standard methods.
- The study involved over 460,000 women in Germany, comparing AI-supported double reading with standard double reading in mammography screening.
Inclusion of AI improved detection and reduced instances of false-positive test results.
A real-world study found that use of artificial intelligence (AI) in breast cancer (BC) screening improved detection rate without a higher rate of false positives, according to study findings published in Nature Medicine.1
Woman getting a mammogram | Image Credit: © IYearDesign - stock.adobe.com
Mammographs and early screening has greatly reduced instances of BC and BC-related mortalities; however, there is a need for increased sensitivity and specificity to minimize false-positive results. Multiple retrospective studies show that AI can have comparable and sometimes superior accuracy to radiologists.1
In the observational, multicenter, real-world, noninferiority, implementation PRAIM study (DRKS00027322), investigators compared the performance of AI-supported double reading to standard double reading among over 460,000 women who underwent organized mammography screening at 12 sites in Germany from July 2021 to February 2023.2
The screenings were collected from the German mammography screening program, which is aimed at early detection of BC and screens over 3 million women between the ages of 50 to 75 annually. Four 2-dimensional mammograms are taken from each woman, which are read independently by 2 radiologists. If one finds something suspicious, a consensus meeting is held with at least 2 initial readers and a lead radiologist. If the concern is confirmed, the woman is called back for further tests such as ultrasound, tomosynthesis, magnification views, contrast mammography, or MRI.1
Of the screenings, 260,739 samples were screened with AI and 119 by radiologists. According to the data, radiologists in the AI screening group achieved a detection rate of 17.6% (95% confidence interval: +5.7%, +30.8%) higher than the rate achieved in the control group. The recall rate was 37.4 per 1000 and 38.3 per 1000 in the AI and control groups, respectively (percentage difference: −2.5% (−6.5%, +1.7%)).1
Additionally, the AI group saw a positive predictive value (PPV) of recall of 17.9% compared with 14.9% in the control group. The PPV of biopsy was 64.5% in the AI group versus 59.2% in the control group.1
“Our initial aim was to demonstrate that AI-based evaluations are equivalent to human assessments,” explained Alexander Katalinic, MD, professor, principal investigator and Director of the Institute of Social Medicine and Epidemiology at the University of Luebeck and UKSH, Campus Luebeck, in a press release. “However, the findings exceeded our expectations: AI significantly improves [BC] detection rates.”3
The findings of the study highlight a pivotal advancement in BC screening, showcasing the potential of AI to enhance detection rates while maintaining low false-positive rates. By demonstrating superior accuracy and efficiency compared to standard methods, AI-supported screening represents a transformative step toward improving early breast cancer detection and outcomes on a large scale. These results reinforce the value of integrating innovative technologies into established screening programs to elevate the standard of care and save more lives.
REFERENCES
