AI-Supported Mammography Increases Early Detection and Reduces Aggressive Breast Cancers: Study
AI-supported screening detected 81% of cancers at screening and reduced interval cancers by 12%, lowering aggressive cancer rates without increasing false positives, researchers found.
- Published in The Lancet, the MASAI randomized trial led by Kristina Lång, PhD, Lund University, involved more than 100,000 Swedish women and was the first randomized AI mammography study.
- ScreenPoint Medical's AI was trained on more than 200,000 mammography scans from 10 countries and used a 1–10 triage scoring system to assign low-risk scans to single reading and high-risk to double reading by radiologists.
- Follow-Up showed 1.55 interval cancers per 1,000 women in the AI group, a 12% reduction from 1.76 per 1,000 in control, with sensitivity at 80.5% and specificity at 98.5%.
- Lång expects rollout across south-west Sweden within months, with AI easing radiologists' workloads and shortening waiting times, potentially cutting waiting lists.
- Authors warned the trial has generalizability limits, including one mammography vendor and experienced radiologists in Sweden, and said countries need to assess local impact and cost-effectiveness; experts estimate comparable trials may take about five years.
36 Articles
36 Articles
AI-assisted mammograms cut risk of developing aggressive breast cancer
Interval cancers are aggressive tumours that grow during the interval after someone has been screened for cancer and before they are screened again, and AI seems to be able to identify them at an early stage
More Evidence Backs AI-Supported Mammography
(MedPage Today) -- Mammography screening supported by artificial intelligence (AI) showed consistently favorable outcomes compared with standard double reading by radiologists, including a noninferior interval cancer rate, according to results...
A Swedish clinical trial with more than 100,000 women shows that AI in mammographic screening detects more relevant cancers and reduces interval tumors by 12% without increasing false positives Read
Coverage Details
Bias Distribution
- 48% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium





















