Using artificial intelligence to diagnose brain tumours

Project title: Moving Artificial Intelligence of Functional Imaging for children’s tumours into a multi-centre environment through a Clinical Decision Support System

Funded by The Little Princess Trust and administered by CCLG
Lead investigator: Prof Andrew Peet, University of Birmingham
Award: £99,959.67
Awarded August 2019

When a child is suspected of having a tumour, the doctor who sees them invariably asks for a scan to be carried out at the hospital. Modern scans are impressive in the way that they can clearly see the inside of the body and detect whether a lump is present and exactly where it is.

However, the scans typically cannot tell whether the lump is definitely a tumour or provide more information about what type of tumour it is. That usually requires an operation to remove part of the lump.

Professor Peet’s team have shown that new scans called functional imaging can give answers to some of these important questions. Whilst functional imaging is becoming widely available, these scans are often very difficult to interpret and so currently are not used much. The team has developed a computer app that helps the doctors to interpret the functional images and have shown in one hospital that this helps doctors to distinguish between cancer and non-cancerous lumps in children.

One of the important features of the app is that it contains a library of children’s tumours for the functional imaging to compare against. This project aims to increase the number of cases in the library using a national database that the team have developed over the last 15 years and use computerised artificial intelligence to help the doctors interpret the functional imaging.

The team will also extend the app so that it can tell the difference between different types of tumour and say how likely the tumour is to respond to treatment. Access to the app will then be given to 10 major children’s cancer hospitals to find out how much it helps the doctors diagnose tumours.