Deep unsupervised clustering of combined multi-modal neuroimages and clinical data to improve subtyping of population-wide mental and behavioural disorders.
Approved Research ID: 79105
Approval date: July 6th 2022
The aim of this project is to group subjects in the UK Biobank neuroimaging cohort according to characteristics in the images taken of the brain of each subject. Several different types of brain images were taken of each patient. Using computer algorithms we want to leverage this richness of information to separate subjects into groups according to how their brains look in each imaging modality.
After we have separated subjects into different groups we will see what characteristics subjects in a group share and how they are different to subjects in other groups. We will focus both on characteristics from their brain images (e.g. people in a group might have larger brain structures) and on clinically relevant data (e.g. people in a group might have all been diagnosed with Alzeihmer's disease). Finally, genetic analysis will be carried out to see if in each group genetics plays a role.
Most mental and behavioural disorders typically share intersecting symptoms making it challenging to perform a precise categorization. It is important to do distinguish diseases and look at each disease as possibly having different symptoms in different patients. Therefore by looking at groups of subjects that share similar characteristics we can help clinicians better identify patients that might share similar disease progression. This can only be done nowadays because large amounts of data are required for these types of grouping strategies and these were not available until initiatives like the UK Biobank were carried out.
This project is part of a PhD thesis expected to last 3 years (36 months). The first part of the project will last about 2 years and will focus on grouping patients together that share similar characteristics according to neuroimaging data. Once these groups have been determined the second part of the project will involve consulting with clinicians on how to use the information extracted to characterise each group.
The prevalence of mental and behavioural disorders is only increasing. For example, the two most prevalent neurological disorders, Alzheimer's and Parkinson's disease already affect over a million people just in the UK. As popoluation's age the prevalence of such diseases is only going to increase. Mental and behavioural disorders are not only terrible for the patient but can have an impact in the life of relatives and friends. Therefore being able to diagnose them early is crucial to help all involved better understand the situation.