A major UK evaluation has found that artificial intelligence (AI) has the potential to increase breast cancer detection rates while reducing pressure on screening services.
AI has the potential to significantly improve cancer detection during routine breast screening, while reducing pressure on overstretched radiology services, according to a large UK study published today in Nature Cancer.
Researchers analyzing the use of AI-assisted systems within the National Health Service (NHS) found that integrating the technology into screening processes increased breast cancer detection by more than 10% and reduced clinical workload by more than 30%.
The findings add new evidence to the ongoing debate about the role of AI in national cancer screening programs.
The evaluation included more than 10,000 mammograms and looked at how AI can help radiologists identify potential cancers during routine breast imaging.
Large-scale evaluation of AI in NHS breast cancer screening
The study 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.
This research formed part of the GEMINI project, which investigates how AI tools can operate in real-world screening environments.
In total, the research team analyzed the mammograms of 10,889 women who attended routine breast cancer screening in the NHS Grampian region.
This study looked at 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 to have a mammogram every three years through the national breast cancer screening programme. As a result, more than 2 million examinations are performed annually, creating a huge demand for specialized radiologists to examine each image.
In current practice, two readers examine each mammogram independently to minimize the chance that cancer is missed. Despite this dual-read approach, some tumors are difficult to detect early with imaging, and it is estimated that one in five cancers remains undetected.
A new study assessed whether AI can help clinicians detect cancers that might otherwise be missed, while also improving the efficiency of the screening process.
AI helps radiologists analyze mammograms
The AI system used in this study performs tasks similar to those performed by trained radiologists. After analyzing mammography images, areas that may contain suspicious tissue changes are highlighted, allowing clinicians to examine them more closely.
Researchers tested 17 different workflow scenarios that introduced AI at various stages of the review process. These included configurations where the technology acted as an additional reviewer or replaced one of the two human readers in certain situations.
The results showed that the most effective models had AI acting as a second leader, effectively replacing a single human leader, while also acting as a safety net to warn of potential anomalies.
This configuration increased cancer detection rates by 10.4% without increasing the number of women unnecessarily recalled by the system for further testing.
Faster results and fewer unnecessary recalls
Researchers also found that AI-assisted workflows could reduce the time needed to notify women of their screening results. Under standard procedures, it usually takes about two weeks for patients to receive their mammogram results.
Using an AI-powered approach, the team estimated that this waiting period could be reduced to about three days.
Early notification may mean faster follow-up and treatment for women diagnosed with cancer. According to the researchers, this is especially important in cases of high-grade or aggressive tumors, where early intervention can significantly improve treatment outcomes.
Another potential benefit includes reducing unnecessary recalls for additional testing.
In traditional screening pathways, the majority of women ultimately do not have cancer, but are asked to return for further tests, such as additional imaging or biopsies.
Current data suggest that only about 1 in 5 women who are recalled after screening receive a cancer diagnosis. AI tools could help clinicians limit these unnecessary recalls by improving image analysis and triage. This reduces patient anxiety and at the same time saves medical resources.
Addressing workforce pressures in radiology
Radiology services across the UK are facing increasing pressure due to increased demand for imaging and a shortage of specialist clinicians. The study authors claim that AI could help alleviate these challenges.
Breast screening programs require thousands of mammograms to be performed each week. By integrating AI into the workflow, radiologists may be able to focus more attention on complex or ambiguous cases rather than reviewing large numbers of routine scans.
This study estimated that an optimal AI configuration could reduce radiologists’ interpretation workload by more than 30%. Medical researchers say this kind of efficiency improvement could be especially valuable as the population ages and screening programs expand.
Evidence gaps in national cancer screening policies
Despite the growing interest in AI in medical imaging, the UK National Examination Board has not yet recommended the routine use of AI within NHS breast screening programmes.
A previous evaluation concluded that there was insufficient evidence available to determine whether this technology improves outcomes in large populations.
The authors of the new study claim their findings help address some of these evidence gaps. By evaluating the AI in a prospective, real-world screening setting, rather than relying solely on retrospective data, this study provides insight into how such systems would perform in daily clinical practice.
However, the researchers stress that further research is still needed to assess the potential benefits and risks before it can be fully implemented 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 demonstrated its potential to improve detection, reduce unnecessary stress for patients and reduce pressure on NHS staff.”
“NIHR is proud to fund this research and help ensure cutting-edge technology is rigorously tested and translates into real-world benefits for patients. This is exactly the kind of innovation we want to see that delivers tangible improvements across the health system.”
Next step: National trial of AI in cancer screening
The findings also support the upcoming EDITH trial, a large UK research program designed to investigate the role of AI in breast cancer screening across multiple NHS sites.
The Scottish part of the trial includes a collaboration between the University of Aberdeen, NHS Grampian and the University of Glasgow.
This study aims to evaluate AI tools in different screening settings and determine whether similar improvements in cancer detection and workflow efficiency can be replicated at scale.
As health systems increasingly explore digital technologies to meet growing demands, our results suggest that carefully integrated AI tools could provide important support for clinicians involved in early cancer diagnosis.
If future trials confirm these findings, AI could play an increasing role in helping screening programs detect cancer earlier, while maintaining the efficiency of strained health services.
Source link
