Capacity, Confidence, Care - Using Artificial Intelligence and Machine Learning to support Breast Screening

"The diagnosis of cancer and other diseases often uses imaging tests (CT, MRI, x-ray, ultrasound) as part of the pathway of investigation. In the NHS, the number of imaging tests which need to be performed, outstrips the ability of the healthcare services to provide and/or interpret these tests. This can lead to delays in diagnosis and delays to subsequent treatment.

Artificial Intelligence (AI) could be part of the solution. It involves using complex computer-based tools to look at data. In the context of healthcare this data can either be the pictures obtained when a patient has an imaging test, or information about how those tests are arranged and how services are run at a wider administrative and operational level. Using AI tools to understand this data has enormous potential, and could allow us to improve the care that patients receive, reducing delays.

Although some of these tools have been approved by international regulators, they are often developed using small sets of data. Many of these tools have not yet been tested at the scale often encountered in the NHS. In addition, the use of them needs a clear set of rules for the NHS, to ensure that they are used correctly and patients can be certain their healthcare data is secure.

Within the East Midlands, for the last five years, eight hospital Trusts have been working closely together to create and use a shared imaging test system. This is called the EMRAD consortium and is an NHS group, working directly with commercial software suppliers. EMRAD has already developed ways of sharing, being responsible for patient data, and testing new tools. These processes have themselves been confirmed as working well, as part of an independent audit.

We will use these existing models and strong governance arrangements of the consortium and the specific expertise of our partners, to test an AI tool developed specifically for breast cancer screening. This will help radiologists and breast units deliver a more accurate, faster breast screening service, reducing unnecessary recalls and biopsies, saving the NHS significant costs, and cutting anxiety for patients too. We will also test the use of AI tools to streamline operational aspects of the service, making sure clinics and screenings are scheduled in the best way.

By the time we have completed the project, we aim to have developed new models for the safe, trusted use of AI across the NHS."