AI improves cancer detection in major UK breast screening study

A large UK evaluation finds artificial intelligence (AI) could boost breast cancer detection rates while easing pressure on screening services.

AI could significantly improve cancer detection in routine breast screening while reducing pressure on overstretched radiology services, according to a large UK study published today in Nature Cancer.

Researchers analysing the use of an AI-assisted system within the National Health Service (NHS) found that integrating the technology into the screening process increased breast cancer detection by more than 10% and reduced clinical workload by over 30%.

The findings add new evidence to ongoing discussions about the role of AI in national cancer screening programmes.

The evaluation involved more than 10,000 mammograms and examined how AI could support radiologists in identifying potential cancers during routine breast imaging.

Large-scale evaluation of AI in NHS breast cancer screening

The research was carried out by scientists and clinicians from the University of Aberdeen, NHS Grampian and health technology company Kheiron Medical Technologies, now part of DeepHealth Inc.

The work formed part of the GEMINI project, which investigated how AI tools could operate in real-world screening settings.

In total, the team analysed mammograms from 10,889 women who attended routine breast screening appointments in the NHS Grampian region.

The study examined how an AI system called Mia could be integrated into existing clinical workflows used to identify abnormalities in mammography images.

Across the UK, women aged 50 to 70 are invited for mammography every three years through the national breast screening programme. This results in more than two million examinations annually, creating a substantial demand for specialist radiologists to review each image.

Under current practice, two human readers independently examine each mammogram to minimise the chance that a cancer is missed. Despite this double-reading approach, an estimated one in five cancers may still go undetected, partly because some tumours are difficult to detect early on in imaging.

The new study evaluated whether AI could support clinicians in detecting cancers that might otherwise be overlooked while also improving efficiency in the screening process.

AI supports radiologists in analysing mammograms

The AI system used in the study performs tasks similar to those carried out by trained radiologists. After analysing mammography images, it highlights areas that may contain suspicious tissue changes, allowing clinicians to examine those regions more closely.

Researchers tested 17 different workflow scenarios in which AI was introduced at various stages of the screening process. These included configurations in which the technology served as an additional reviewer or replaced one of the two human readers in certain situations.

The results showed that the most effective model involved AI acting as a second reader, effectively substituting one human reader while also functioning as a safety net to flag potential abnormalities.

In this configuration, the system improved cancer detection by 10.4% without increasing the number of women recalled unnecessarily for further testing.

Faster results and fewer unnecessary recalls

The researchers also found that AI-assisted workflows could shorten the time required to notify women of screening results. Under standard procedures, patients typically wait around two weeks to receive results from their mammogram.

Using the AI-supported approach, the team estimated that this waiting period could be reduced to approximately three days.

Earlier notification may allow faster follow-up investigations and treatment for women diagnosed with cancer. According to the researchers, this is particularly important for high-grade or aggressive tumours, where early intervention significantly improves treatment outcomes.

Another potential benefit involves reducing unnecessary recalls for additional testing.

In traditional screening pathways, many women are asked to return for further examinations, such as additional imaging or biopsies, even though the majority ultimately do not have cancer.

Current data suggest that only about one in five women recalled after screening receives a cancer diagnosis. By improving image analysis and triage, AI tools could help clinicians limit these unnecessary recalls. This would reduce patient anxiety while also conserving healthcare resources.

Addressing workforce pressures in radiology

Radiology services across the UK face mounting pressure due to increasing demand for imaging and a shortage of specialist clinicians. The authors of the study argue that AI could help mitigate these challenges.

Breast screening programmes require thousands of mammograms to be reviewed every week. Integrating AI into the workflow could allow radiologists to focus more attention on complex or ambiguous cases rather than reviewing large numbers of normal scans.

In the study, the optimal AI configuration was estimated to reduce radiologists’ reading workload by more than 30%. Healthcare researchers note that this type of efficiency improvement may be particularly valuable as populations age and screening programmes expand.

Evidence gap in national cancer screening policy

Despite growing interest in AI for medical imaging, the UK National Screening Committee has not yet recommended the routine use of AI within the NHS breast screening programme.

Previous assessments concluded that available evidence was insufficient to determine whether the technology improves outcomes across large populations.

The authors of the new study argue that their findings help address some of these evidence gaps. By evaluating AI in a prospective, real-world screening environment rather than relying solely on retrospective data, the research offers insight into how such systems might operate in everyday clinical practice.

However, the researchers emphasise that further studies are still required to assess potential benefits and risks before fully implementing on a large scale.

Professor Mike Lewis, NIHR Scientific Director for Innovation, added: “By generating high-quality evidence on the safe and effective use of AI in breast cancer screening, the team has shown its potential to improve detection, reduce unnecessary stress for patients, and ease pressure on the NHS workforce.

“The NIHR is proud to have funded this work, helping to ensure that cutting-edge technologies are tested rigorously and can be translated into real-world benefits for patients. This is exactly the kind of innovation we want to see delivering tangible improvements across the health system.”

Next phase: Nationwide trials of AI in cancer screening

The findings also support the upcoming EDITH trial, a larger UK research programme designed to examine the role of AI in breast cancer screening across multiple NHS sites.

The Scottish component of the trial will involve collaboration between the University of Aberdeen, NHS Grampian and the University of Glasgow.

The study aims to evaluate AI tools across diverse screening settings to determine whether similar improvements in cancer detection and workflow efficiency can be replicated at scale.

As healthcare systems increasingly explore digital technologies to manage growing demand, the results suggest that carefully integrated AI tools could become an important support for clinicians involved in early cancer diagnosis.

If future trials confirm these findings, AI may play a growing role in helping screening programmes detect cancers earlier while maintaining efficiency within strained healthcare services.