AI-supported mammography more effective than standard screening, new large-scale study contends
Artificial intelligence-supported mammography is more effective than standard breast cancer screening, according to results from a new large-scale study.
In particular, the technology can help to identify more cancer cases during screening, reducing the rate of diagnosis by 12% in subsequent years. That’s compared to the two-radiologist, double-reading approach that is typical in Europe, researchers detailed in The Lancet.
The findings come from Sweden's Mammography Screening with Artificial Intelligence, or MASAI, a randomized clinical trial incorporating over 105,000 women. Those involved believe results from this first-of-its-kind investigation provide more fuel for implementing AI in regular clinical practice.
“Widely rolling out AI-supported mammography in breast cancer screening programs could help reduce workload pressures amongst radiologists, as well as helping to detect more cancers at an early stage, including those with aggressive subtypes,” radiologist and lead author Kristina Lång, MD, PhD, with Lund University, Sweden, said in a statement Jan. 29, cautioning that implementation must be executed “cautiously, using tested AI tools and with continuous monitoring in place to ensure we have good data.”
The study randomly assigned women to either receive standard breast cancer screening with double reading or artificial-intelligence-backed mammography (using technology from Screenpoint Medical). AI was used for detection support and to triage exams to single or double reading scenarios. Women were assigned to each study group between 2021 and 2022 across four sites in Sweden, with participants at a median age of 54. Scientists trained and tested the AI system with over 200,000 exams from multiple institutions across 10-plus countries.
During the two years of follow-up, there were 1.55 interval cancers per 1,000 women in the AI group versus 1.76 in the control arm. This represents a roughly 12% reduction in interval cancer diagnoses when using AI. Additionally, there were 16% fewer invasive, 21% fewer large, and 27% fewer aggressive subtype cancers in the AI group compared to the control arm. In the AI group, about 81% of cancer cases were detected at screening versus 74% in the control group, representing a 9% increase. Both groups had a similar rate of false positives at 1.5% for AI-backed screening compared to 1.4% without.
The study had several limitations, including being conducted in one country, its focus on one type of mammography device, and one AI system. Also, radiologists involved in the trial were moderately to highly experienced, further limiting generalizability of the findings.
“Further analyses of subsequent screening rounds and cost-effectiveness will clarify the long-term balance of benefits and harms and could provide a strong rationale for implementing AI in population-based mammography screening programs, particularly in the context of workforce shortages,” the authors concluded.
Marty Stempniak has covered healthcare since 2012, with his byline appearing in the American Hospital Association's member magazine, Modern Healthcare and McKnight's. Prior to that, he wrote about village government and local business for his hometown newspaper in Oak Park, Illinois. He won a Peter Lisagor and Gold EXCEL awards in 2017 for his coverage of the opioid epidemic.