Deep Learning of Cerebral Cortex Microstructure (360G-Wellcome-215901_Z_19_Z)

My research aims to bridge the gap between in vivo neuroimaging and underlying brain microstructure. MRI can inform us of shape or signal intensity changes in neurological and psychiatric diseases but has limited interpretability with respect to underlying neurobiological disturbances. Currently such specific measurement of neurobiology is only accessible through ex vivo histological techniques. I wish to further our understanding of how in vivo MRI reflects microstructure through three interrelated projects. 1) I will map the distribution of microstructural properties, including cytoarchitecture, myeloarchitecture and receptor-architecture to whole-brain reference atlases, thus representing these features in the space of brain scans. 2) I will create deep learning frameworks for analysing cerebral cortical structure. 3) I will use computational tools to link in vivo MRI cortical structure to its underlying microstructure in health and disease. I will use high-resolution and multi-modal structural MRI data to link interregional and interlaminar MRI differences to their underlying microstructural source. I will apply deep learning to a large cohort of patients with a form of surgically remediable epilepsy, focal cortical dysplasia, which is characterised by abnormal microstructure. This work aims to solve the clinical challenge of identifying subtle epileptogenic structural abnormalities from a patient’s presurgical MRI scan.