How AI can improve breast cancer detection in the UK

Your browser does not support the audio element.

Breast cancer affects one in every eight women in the UK. In this fight, early detection is crucial to giving people the best chance of overcoming the disease. New research from Google, Imperial College London and the UK's National Health Service (NHS), published as a pair of studies in Nature Cancer today, marks a turning point in screening technology and reveals how AI can strengthen early detection efforts.

Our experimental research AI system identified 25% of the "interval cancers" that were previously missed - the cases that typically slip through traditional screenings and only surface after symptoms appear, when they become more challenging to treat. But this research goes beyond the accuracy of the scans. It offers a first-of-its kind, large-scale look at how radiologists react when AI challenges or confirms their diagnosis in a clinical setting.

Confronting a growing challenge

In the UK's NHS, the frontline of breast cancer screening relies on a rigorous "double-reading" process: Two specialists must agree on every mammogram, with an arbitration panel deciding any disputes. It is a vital safety net, but one that's stretched to its limit. Each specialist must review roughly 5,000 scans annually, with just four hours of dedicated time per week, all amidst a global shortage of radiologists. We set out to determine how AI could help to tackle this challenge.

Validating accuracy at scale

The first step was comparing the accuracy of AI-based mammography interpretation to that of expert radiologists. We tested this by using AI to review the mammograms of 125,000 women, and the results were definitive: The AI-based screening detected 25% of the total interval cancers (cancers detected between scans) previously missed. AI also identified more invasive cancers and more cancers overall than the expert radiologists, and identified fewer false positives for women having their first-time scan.

Giving radiologists more time for patient care

Having proven AI's standalone accuracy, we turned to a practical question: Can this technology effectively give radiologists more time for patient care? Our second study looked at the scans of over 50,000 women and showed that AI is capable of reducing screening workloads by an estimated 40% when used as the "second reader" in the workflow. This work revealed that AI-assisted workflows could enable healthcare professionals to address the nationwide screening backlog, giving them more time to focus on complex cases all while upholding the rigorous clinical benchmarks of the traditional double-reading standard.

Building trust through human-AI collaboration

While AI-assisted screening works in theory, the true test of its value lies in how medical professionals respond to AI-driven diagnoses in practice. To bridge this gap, we looked at the entire clinical pathway, paying careful attention to "arbitration" by medical professionals - the final step where specialists resolve diagnostic disagreements. While arbitration successfully filters out false positives, we observed a critical tension during our simulated review: Arbitration panel specialists occasionally overruled AI-detected cancers that would have otherwise gone undetected.

These findings highlight the need for continued research on human-AI interaction to build specialist trust in AI's ability to catch subtle, early-stage cancers.

We also evaluated the challenges of practical clinical AI integration into clinical workflow through an observational feasibility study across 12 NHS screening sites in London, processing over 9,000 cases in real-time without using AI results to impact patient care. A key lesson was that AI isn't a "plug-and-play" solution. It requires careful and continuous calibration to the unique heartbeat of each hospital and adaptation to shifting workflows, evolving equipment and diverse patient populations.

Bringing it all together

These findings build on our previous work that found an earlier version of this AI-based screening system could detect cancers in a single reader setting and lead to shorter diagnostic waiting periods for women. Collectively, these studies show how AI can strengthen early detection efforts, paving the way for more women to be diagnosed and treated sooner, with the ultimate goal of saving lives.

• Health