EU pledges €35M for artificial intelligence research to assist early cancer diagnosis

09 Jul 2019 | News

European Commission calls on researchers to overcome the problems of pooling patient scans, to provide data around which to develop image recognition systems, and speed up cancer diagnoses  

The European Commission has pledged €35 million to promote the development of image recognition systems for diagnosing the most common forms of cancer. The aim is to speed up diagnosis, leading to faster referrals, reducing anxiety for patients and leading to better outcomes.

The competition, which comes under the Horizon 2020 research programme, has a deadline of November 13.

Image recognition systems have now reached the stage where it is acknowledged they can be faster and more accurate than humans in interpreting scans from microscopes, magnetic resonance imaging and positron emission tomography machines.

At the same time, there is a shortage of experts able to interpret scans. For example, the clinical radiology U.K. workforce report, published by the Royal College of Radiologists in April, shows the number of radiologists is “depressingly short of demand,” with staff shortages resulting in delayed cancer diagnoses and inadequate emergency diagnostics.

The shortage of skills is set against an increasing demand for diagnostic imaging. There is also an increase in the complexity and diversity of images, and growth in the number of images to be reviewed. A total of 32 million X-rays, CT and MRI scans were carried out in England alone in 2017 - 2018.

With faster diagnosis, patients would be referred for treatment sooner, avoiding late intervention that is often expensive and ultimately unsuccessful.

The fuel for these efforts is anonymised patient scans, but creating large enough pools of images is difficult. Patient data repositories in Europe are relatively small, both because of the average hospital size, and because regulations make it difficult to merge data from multiple facilities.

Applicants for EU funding will need to contribute to data stores, by building a repository of “high quality, interoperable, anonymised or pseudo-anonymised data sets of annotated cases”, the commission says.





Never miss an update from Science|Business:   Newsletter sign-up