Spotting signs of damage in the brain: connectivity-aware white matter hyperintensity segmentation from routine MRI using deep learning

This project aims to revolutionise the way we assess damage in the brain and understand its impact on brain functions from routine MRI scans. The damage of particular importance is the damage to white matter, brain's communication pathways. Damaged white matter is known as white matter hyperintensity because it appears brighter than healthy white matter in routine MRI scans. Current approaches. Current approaches to assess this, known as white matter hyperintensity segmentation, identify spatial locations and extent of the damage. However, this information offers limited value for quantifying the nature of functional impairment. Particularly, they do not inform on the extent to which individual communication pathways supporting specific brain functions have been disrupted. Currently, accessing this information on brain connectivity would require collection of additional specialised MRI scans unavailable routinely. The key idea behind this project is that we can gain access to this information by learning from existing large-scale MRI studies for which specialised MRI scans are available to infer brain connectivity. This project will develop this novel idea of connectivity-aware white matter hyperintensity segmentation, evaluate its performance on quantifying brain functions, and ultimately produce a well-tested and documented software tool.