An artificial intelligence (AI) analysis of mammograms detected more cancers than two breast radiologists working together, according to a new study, without increasing false positives and almost halving the radiologists’ workload.
Interim findings from the first randomized study investigating the use of AI in a national breast cancer screening program, published in the journal The Lancet Oncology, suggested AI-supported screening detected 20 percent more cancers compared with the routine double reading of mammograms by two breast radiologists.
European guidelines recommend double reading of screening mammograms to ensure high sensitivity. The U.S. does not have the same standard but like many other countries is experiencing a shortage of breast radiologists.
Breast cancer is the most common cancer among women in the U.S., accounting for about 30 percent of all new cancer cases in women each year.
The study showed that AI use in detecting breast cancer is safe, and it supports the potential of the technology to improve screening efficacy and reduce the workload of radiologists. The AI reduced the mammogram reading workload by 44 percent.
“Taken together, the evidence suggests that use of AI could potentially benefit mammography screening by reducing the screen reading workload and the number of interval cancers, but randomized trials are needed to assess the efficacy of AI-supported screening,” the study concluded.
Previous studies have looked at using AI to diagnose breast cancer in mammograms, but they were retrospective, meaning the screenings had already been examined by radiologists.
The study’s co-author said it’s too early to start implementing AI in hospitals right now.
“These promising interim safety results should be used to inform new trials and program-based evaluations to address the pronounced radiologist shortage in many countries. But they are not enough on their own to confirm that AI is ready to be implemented in mammography screening,” Kristina Lång, an associate professor of radiology diagnostics at Lund University in Sweden, said in a statement.
The study examined mammograms of more than 80,000 women across Sweden between April 2021 and July 2022. Half of the women had a mammogram read by an AI program before it was examined by a radiologist, and the other half had mammograms read by two radiologists without the use of AI.
Among the participants screened with AI, 244 cancers were detected, compared with 203 cancers in the non-AI group. The false positive rate was the same in both groups. Overall, AI-supported screening resulted in a cancer detection rate of 6 in 1,000 screened women compared with 5 in 1,000 for standard double reading without AI, the equivalent of detecting one additional cancer for every 1,000 women screened.
The final results will look at whether AI can reduce the number of interval cancers — cases detected between screenings — and whether the use of AI in screening is justified. But those findings are not expected for several years.
“The greatest potential of AI right now is that it could allow radiologists to be less burdened by the excessive amount of reading,” Lång said. “While our AI-supported screening system requires at least one radiologist in charge of detection, it could potentially do away with the need for double reading of the majority of mammograms easing the pressure on workloads and enabling radiologists to focus on more advanced diagnostics while shortening waiting times for patients.”
