AI spots breast cancer too small for doctors to see in landmark UK study

Yvonne Cook's tumour was so small it was missed by medics, but AI identified a potential problem.

As a result of the abnormality detected by the technology, she was invited for another test where the cancer was detected.

Her cancer was found as part of a clinical trial which found the use of AI in breast cancer screening could increase detection of disease by 10.4 per cent.

Read More: Strictly's Dr Punam Krishan, 42, reveals breast cancer diagnosis

Read More: Stop age discrimination over breast cancer checks, urges former minister

It comes as a leading health expert - Lord Darzi, who authored an influential report into the state of the NHS - said AI has the potential to transform how the NHS prevents, detects and treats diseases like cancer.

Ms Cook, from Aberdeen, said she felt "lucky that AI was used and lucky that it caught something so small at exactly the right time".

She believes that without AI, her cancer would not have been found until her next routine mammogram appointment some three years later, or if the tumour had grown so large she could have felt it.

The new trial, published in the journal Nature Cancer, found that as well as increasing cancer detection rates, AI could also reduce the time to notify affected women from two weeks to three days.

It was also found to reduce the number of women unnecessarily recalled for further tests, including unnecessary biopsies.

And it can cut the workload of healthcare staff by as much as 31 per cent, found the researchers form the University of Aberdeen, NHS Grampian and Kheiron Medical Technologies, now part of DeepHealth.

The AI tool, known as Mia, was used to support healthcare workers in the routine breast screening of 10,889 women in NHS Grampian.

Of these 106 were diagnosed with with routine screening and an additional 11 cancers were diagnosed with the support of AI, seven of which were deemed to be "invasive".

At present woman in the UK aged 50 to 70 are invited for breast cancer screening every three years.

Their scans, known as mammograms, are assessed by two radiologists.

But because some cancers are hard to detect, a number are missed.

And a number are said to be "unnecessarily" recalled.

The University of Aberdeen said using AI as a second scan reader - substituting one human reader, and as an extra reader serving as a safeguard - resulted in the best combination of workload savings and increased early cancer detection without recalling more women for additional tests.

"Not only did we find optimal ways to detect breast cancer, quicker and more accurately, we also found ways to reduce the number of women having to return for unnecessary tests," said Dr Clarisse de Vries, lecturer in data science at the University of Glasgow, lead author and former research fellow at the University of Aberdeen.

She pointed out that the UK National Screening Committee does not currently recommend the use of AI in breast cancer screening, adding: "Our work adds high-quality evidence to the scientific literature in support of AI."

Professor Gerald Lip, lead for artificial intelligence in clinical practice at the University of Aberdeen, added: "The bottom line here is - without AI, doctors would not have caught these cancers as early."

Ms Cook, who is in her 60s, attended a routine mammogram appointment in May 2023.

After her first test she received a letter calling her back for additional imaging.

"I had a scan and the consultant confirmed that the AI diagnosis was correct, that there was a small, Grade 2 tumour there, too small to be detected by the human eye," she said.

"Overwhelmingly I just felt incredibly lucky that I was part of the research programme and that it had been picked up at this early stage."

Ms Cook added: "Had the AI not picked up the small tumour when it did, then either it would have been discovered at my next routine mammogram three years later, or I would have picked it up when it had grown to a stage that I was able to feel it," she said.

It comes as experts from Imperial College London, Google, the universities of Cambridge and Surrey, NHS Trusts at Cambridge University Hospitals, Imperial College Healthcare, the Royal Marsden, the Royal Surrey and St George's University Hospitals examined data on more than 175,000 women, with the study split into three parts.

The first examined data on 116,000 women who took part in screening services in 2015/16 and found AI could increase the cancer detection rate, identify more invasive cancers, reduced false positives and reduced the time taken to read a scan.

The second part of the study looked at 9,266 cases at two screening services in London.

The average time for AI to complete a read was 17.7 mins compared to two days for the first medic to assess a scan, they found.

Researchers also looked at the use of AI in arbitration - when medics do not agree on a diagnosis and a third person is called in to make a decision - on scans involving 50,000 women.

They found AI performed comparably to humans and reduced the overall screening workload.

Dr Hutan Ashrafian, from the Institute of Global Health Innovation at Imperial College London, 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."

Author Lord Darzi, director of IGHI who wrote an influential report on the NHS in 2024, added: "AI has the potential to transform how the NHS prevents, detects and treats diseases like cancer.

"These findings highlight how AI can support clinicians to identify more cancers earlier, reduce errors and deliver higher quality care to patients."

Simon Vincent, chief scientific officer at Breast Cancer Now, said: "This research underlines the huge potential of AI to support radiologists in breast cancer screening."