Lead investigator: Prof Andrew Peet, University of Birmingham
Funded by The Little Princess Trust and administered by CCLG
Funded July 2017
Award: £99,027.00
Brain tumours are the most common solid tumours in childhood and are responsible for the majority of cancer related deaths.
Diagnosis usually starts with recognition of symptoms commonly by parents and leads to a scan which confirms the presence of a lump in the brain. The best scanning method is Magnetic Resonance Imaging, usually known as MRI. These scans provide exquisite structural detail of the brain and can easily detect any lump present but often we cannot be certain that the lump is a tumour or tell what type of tumour it is.
Over the past 15 years we have been developing a set of advanced MRI scans called functional imaging which use an MRI scanner to give a detailed picture of the make-up of any lumps seen on the MRI. We have collected these images from around the UK and stored them on the largest database of its type in the world, the CCLG Functional Imaging Database.
The various Functional Imaging scans give information on changes in chemicals in the tumour, the blood flow in it and whether it is starting to invade nearby brain, the hallmarks of cancer. Each of these properties can be used to help diagnose brain tumours but they have not yet really been combined together to give an overall picture. This is important since the best diagnostic accuracy is likely to come from combining the techniques and can provide new diagnostic insights. There have been major advances over the past few years in diagnosing tumours by their inherent genetic make-up, however, how a tumour behaves is also governed by how it interacts with the brain and functional imaging gives a unique opportunity to probe these properties whilst the tumour is still within the child.
Whilst functional imaging can give a detailed picture of a brain tumour, its properties and its interaction with the brain, we have shown that the vast amount of information that is provided by these images is best analysed using specialised computer programmes, often called machine learning. We will combine functional imaging with machine learning to provide the best possible non-invasive diagnostic schemes for children’s brain tumours. This strategy can provide a more timely diagnosis and give a more accurate diagnosis.