Enriched Image Based Phenotypes for Health Exploration and Discovery
Approved Research ID: 44584
Approval date: January 22nd 2019
Chronic metabolic, cardiovascular, and liver disease are amongst the greatest public health burdens, and there is a paucity of effective treatments. The UK Biobank imaging data provides a new window into the organs involved in these conditions (e.g. heart, liver, pancreas, spleen), as well as systemic changes. The aim of this project is to develop new machine learning methods to better understand the relationship between organ form (as visible on MRI or DEXA) and function. We propose to apply recent advances in artificial intelligence, techniques which have already transformed other fields, to better understand the changes that occur within specific organs, and how these relate to disease. These methods could be used to identify people at risk of disease, to identify genetic or lifestyle factors, or to inform the development of new treatments.
We will investigate the mechanisms and pathways by which predisposing genetic and environmental factors alter metabolic health and phenotype and therefore health-span. The process involved in maintaining these parameters comprises a series of interlinked molecular and physiological networks in the area of stress responses, including oxidative and inflammatory stress, genetic/epigenetic variations, carbohydrate/lipid challenges and psychological/cultural context. By creating highly enriched multifactorial phenotypes, we hope to identify the key components in this network and through this to develop novel approaches for the reversal of metabolic dysfunction and the de-acceleration of ageing to maintain optimal health throughout adult life.
We will also develop methods to extract features from brain MRI sequences (e.g. segmenting the temporalis muscle) and explore how these are related to skeletal muscle as well as metabolic, genetic, and environmental factors in the context of aging.