Background: The brain is inherently an electrically conductive organ, but standard imaging modalities are not well equipped to measure its electrical properties (EPs). This is particularly relevant for studying neurodevelopment – pre-clinical studies show that control of neuronal polarisation and hence conductivity is a crucial step in the emergence of interconnected circuits across the brain and subsequently networks responsible for cognition and behaviour. We have extracted brain EPs from neonatal MRI (Fig. 1 left) and found that average brain conductivity correlates with age (Fig. 1 middle), prematurity (Fig. 1 right) and social and cognitive outcomes at 18 months.
Aim: To develop novel early biomarkers of atypical brain development by mapping MRI-based electrical properties.
Workplan: We will explore developing brain EPs using the existing dataset of ~1000 fetal and neonatal MRI with genetics and neurodevelopmental outcomes from the Developing Human Connectome Project.
Rotation project: Create a spatio-temporal atlas of the developing brain and obtain average EP maps across neonatal period. Assess changes of EP due to prematurity at a cohort level.
Year 1: Develop a convolutional-neural-network (CNN) based scheme for simultaneous denoising and calculation of high-resolution EP maps for individual babies.
Year 2: Extend methods to fetal data to explore changes in brain EPs over a large sweep of development.
Year 3: Discover links between local EP in the brain and prematurity, outcomes and genetic markers.
Skills: The project requires computational background and provides training in machine learning methods and MRI data acquisition/analysis.