Royal Surrey research explores how AI could support radiologists in breast cancer screening | News

Researchers at Royal Surrey have helped lead major research showing that artificial intelligence (AI) could support radiologists in detecting breast cancer in routine screening mammograms, helping to ease pressure on NHS services.

The AIMs study analysed screening data from more than 125,000 women, making it the largest NHS study to date on AI in breast cancer screening. Findings were published this week in two companion papers in scientific journal, Nature Cancer.

Royal Surrey staff contributed to both studies through the work of the Trust’s Scientific Computing team, led by Prof Mark Halling-Brown. One of the papers was led by Lucy Warren, who carried out the research during her time as AI Research Lead at the trust.

In the UK, mammograms taken through the NHS breast screening programme are assessed by two specialist readers. The AIMs study investigated the impact of one human reader being replaced with an AI tool. There were two papers describing the results of this research. The first looked at the standalone performance of the AI. It found that the AI detected more cancers overall than one human reader, including invasive cancers.

The second paper was Royal Surrey’s arbitration study. In NHS breast screening, when two human readers disagree, arbitration is performed by a panel to come to a final decision. In this study radiologists and radiographers from partner trusts, St Georges and Imperial, arbitrated human and AI decisions. The study found that after arbitration, one human reader working with AI achieved a similar performance to two human readers, while cutting down workload.

Many members of staff from Scientific Computing at Royal Surrey worked on the study - providing web portals and viewing software allowing for readers (radiologists and radiographers) to view images as well as results from the AI readers. The software enabled the teams to record their results, and performed statistical analysis of the results.

Lucy Warren, who led the arbitration study during her time at Royal Surrey, said:
“Breast screening programmes rely on highly skilled specialists, but there is increasing pressure on the workforce. It was encouraging to find that a combination of human expertise and AI achieved a similar level of performance to two human readers.

“The AIMs study was a success because of the collaborative work between multi-disciplinary teams from multiple trusts and institutions.”

Hutan Ashrafian from the Institute of Global Health Innovation (IGHI) at Imperial College London and an author on both papers said: “This is the closest AI has ever come to helping reduce breast cancer deaths within the NHS, so the potential for the NHS to take this forward is significant.”

The AIMs study used Royal Surrey’s OPTIMAM database, funded by Cancer Research UK, which holds almost 7 million mammogram images and associated clinical data collected through the UK national breast screening programme. This allowed researchers to carry out a detailed retrospective analysis to test AI developed by Google in a real-world setting.

The research was carried out in partnership with Imperial College London, Imperial College Healthcare NHS Trust, St George’s University Hospitals NHS Foundation Trust, and Google Research.