Synthetic Contrast MRI for Sustainable and Equal Healthcare

Project ID: 2531ad1519

(You will need this ID for your application)

Project Summary:

Contrast based MRI is widely used for Radiological diagnosis of patients with brain tumours and Multiple Sclerosis (MS). Contrast-enhanced MRI shows regions of blood brain barrier disruption and is associated with disease. Patients require injection of a contrast agent known as gadolinium which has several issues: 1) contra-indicated in patients with severe renal impairment, pregnant patients, patients with allergic reactions to it. 2) MRI takes longer and injection can be uncomfortable to the patient. 3) The long-term effects of repeated use is unknown. 4) There is an associated cost to the NHS. 5) Gadolinium is toxic to the environment.

The above concerns could be addressed using advanced deep learning techniques to synthesize contrast enhancement from non-enhancing images.

Patients having an MRI will have multiple imaging modalities within one session. Clinically these will constitute of images that probe the structure of the brain (e.g. pre- and post-contrast T1, T2 and T2 FLAIR). More advanced quantitative brain imaging techniques can be used to probe the underlying biology, e.g. diffusion MRI probes the brain microstructure. Datasets with advanced imaging techniques exist within both brain tumours and MS.

The end goal of this research will be to provide a clinically validated tool that can replace or minimise the use of gadolinium in patients to provide a more sustainable and equal healthcare system. The following will be done during the project: i) Use radiomics and other image analysis techniques on advanced brain tumour and MS datasets to probe the biological composition of contrast enhanced regions vs non-enhaced regions. ii) Compare results from the two diseases to understand any similarity/differences in biological compositions of enhanced regions. iii) Develop new deep learning based AI tools that incorporate the most significant imaging parameters for creation of synthetic contrast enhanced images without use of contrast agents.