Supporting breast screening services post COVID-19: How can an AI solution help?

 

 

Last weekend (1 June 2020), our breast screening AI solution, Mia, was showcased by the UK Government’s BEIS Department (Department of Business, Energy and Industrial Strategy). 

This case study highlights the potential for AI to support breast screening services which are currently on pause because of #Covid-19, putting women's lives at risk. We have heard that there could be delays of up to 12 months before normal screening resumes - we believe Mia can help when it does.

COVID-19 has ushered in an unprecedented time for technology adoption in the NHS. The focus now is on front line healthcare services but there will be a second wave of problems to deal with as normal services are resumed. For example, breast screening services are effectively on hold (1).

Every day, week and month that passes with screening services on hold, invitations not being sent and women not attending increases the risk of having their breast cancer detected later. This is a risk because statistics show that survival rates for women with screen detected breast cancer are significantly better than without (2,3). 

It is therefore extremely important that screening services resume as quickly as possible and we believe that Mia may be able to help radiologists process the backlog of cases. In fact, the UK Government recently awarded us a grant to help determine the best use of Mia to increase the automation of breast screening services while maintaining patient safety and the standard of care. It’s going to accelerate our efforts at this unique moment in time. But we think what sets Mia apart is the clinical rigour that has gone into the solution to ensure it’s safe and effective for patients and users. 

We’ve been developing Mia for several years, and we are already running a multi site, multi hardware trial, using real world screening images. We think it’s the best way to get the evidence that shows Mia can work on any woman in the UK and that Mia can help readers in breast screening services get through the COVID-19 backlog. 

One of the things that makes Mia really different is that it can make a critical decision - whether a woman should be recalled for further testing or not. We focused on this use case because we heard from radiologists and breast screening programmes that the escalating shortage of qualified readers is the biggest challenge they face. 

Importantly, Mia has been developed to support the work of a radiologist, not beat them or replace them. Because at Kheiron we believe that AI+humans gives the best outcomes for patients so there is always a human reading mammograms alongside Mia.

References: 

  1. CRUK blog
  2. https://acsjournals.onlinelibrary.wiley.com/doi/full/10.1002/cncr.32859
  3. http://www.ncin.org.uk/publications/data_briefings/improved_survival_for_screen_detected_breast_cancer