Using artificial intelligence to improve early detection of breast cancer

The challenge:

BreastScreen Australia (BSA), the population-based mammography screening program plays a critical role in detecting breast cancer early, when treatment is most effective and the chances of survival are greatest. Every two years, more than 2 million women have screening mammograms, with each image independently reviewed by at least two specialist radiologists (doctors specialising in medical imaging).  

This double-reading approach helps improve cancer detection, but it is becoming increasingly difficult to sustain. Australia faces a shortage of breast imaging radiologists, creating pressure on the screening system and increasing the risk of delays in women receiving their results. Delays can mean women wait longer for diagnosis and treatment, potentially reducing the benefits of early detection. 

As demand for breast screening continues to grow, new approaches are needed to ensure women continue to receive accurate and timely diagnoses that give them the best possible chance of surviving breast cancer. 

Project description:

Recent advances in artificial intelligence (AI) have shown that AI tools can detect signs of breast cancer with an accuracy comparable to experienced radiologists. While this technology has enormous potential, there are still important questions about how AI can be safely and effectively used alongside clinicians to improve outcomes for women. 

Professor Sarah Lewis from the University of Sydney is leading an NBCF-funded study to investigate how AI can be best support breast cancer screening and early detection. The team will develop and test an AI tool that can work in partnership with radiologists in several ways:  as a second reader, a pre-screening tool or as a triage system that helps prioritise potentially suspicious cases.  

The researchers will also assess whether the AI performs fairly across different groups of women in Australia and investigate how radiologists and AI can work together to achieve the highest levels of accuracy without compromising clinical expertise. 

By identifying the safest and most effective way to integrate AI into breast screening, the team hopes to improve the speed and accuracy of breast cancer detection, helping more women have their breast cancers detected at earliest possible stage. 

Potential impact:

Professor Sarah Lewis’ NBCF-funded study could provide the evidence needed to safely and effectively integrate AI tools into BreastScreen Australia, helping more women receive an accurate diagnosis as early as possible.  

By combining the strengths of AI with the expertise of radiologists, breast cancers could be identified earlier, more accurately and a with fewer delays. Detecting cancer at an earlier stage gives women the best chance of successful treatment and survival. 

The research could also help radiologists prioritise women with suspicious findings, reducing delays in diagnosis when timely treatment matters most. At the same time, it may improve the sustainability of Australia’s breast screening program by reducing pressure on the specialist radiology workforce. 

A number of countries are now developing guidelines and trialling AI for use to support radiologists in screening, but there are a number of unknowns about how humans can interact with AI safely. If successful, this project could position Australia as a global leader in AI-enabled breast screening while delivering a tangible benefit to women in Australia through faster, more accurate detection of breast cancer. 

Grant code: IIRS-22-087
Active years: 2022-2026
Scientific project title: Effective Artificial Intelligence for breast cancer screening: a second reader, pre-screening and triage model