AI Detects 20% More Breast Cancers Than Human Radiologists Alone in Landmark Swedish Study

A landmark randomized controlled trial conducted across four hospitals in Sweden has shown that artificial intelligence can detect significantly more breast cancers than the standard practice of having two radiologists independently review each mammogram. The study, published in The Lancet Oncology, involved over 80,000 women aged 40-74 and represents the largest prospective trial of AI in breast cancer screening to date.

In the trial, women were randomly assigned to either AI-supported screening or standard double reading. In the AI-supported group, an AI system first analyzed each mammogram and assigned a risk score. Low-risk mammograms were read by a single radiologist, while higher-risk ones received the traditional double reading by two radiologists. This approach detected 20% more cancers compared to the standard care group — 244 cancers versus 203 in similarly sized cohorts.

“The study, published in The Lancet Oncology, involved over 80,000 women aged 40-74 and represents the largest prospective trial of AI in breast cancer screening to date.”

Crucially, the increased detection did not come at the cost of more false positives. The recall rate — the percentage of women called back for additional testing — was virtually identical between the two groups (2.2% for AI-supported versus 2.0% for standard care). This means the AI was genuinely finding real cancers that would have been missed, not just flagging more suspicious-looking but ultimately benign findings.

The workload implications were equally significant. AI-supported screening reduced the total reading workload for radiologists by 44%, as the AI effectively triaged the straightforward cases, allowing radiologists to focus their expertise on the more complex mammograms. Given the global shortage of breast imaging radiologists, this efficiency gain could help expand screening access to underserved populations.

The study's lead author, Dr. Kristina Lång of Lund University, emphasized that the AI was designed to assist radiologists, not replace them. The system served as an intelligent first reader, flagging cases that warranted closer human attention while providing reassurance on clearly normal scans.

Multiple countries are now considering incorporating AI into their national breast cancer screening programs based on these results. The UK National Health Service has launched pilot programs at several screening centers, and similar trials are underway in Germany, the Netherlands, and Australia.