Deep learning of cerebral cortex microstructure
University College London
Brain scanning techniques offer powerful ways of examining the living brain non-invasively and diagnosing neurological and psychiatric illness. But brain changes associated with such conditions are often too complex for the human eye. Artificial intelligence can identify such hidden patterns, but we also need to understand their underlying, microscopic causes.
To link brain imaging with microscopic changes, I will work on three related projects. Using newly-developed microscopic-level imaging technology, I will create maps from donor brains with detailed information on neurons and connections that make up the brain. I will adapt deep learning tools, the basic instruments behind the present-day advances in artificial intelligence, to analyse these microscopic-level brain maps. I will also use brain maps and deep learning to analyse high resolution MRI scans. In doing so, I will push the boundaries of what we can understand from brain scans.
My findings will be extended to identify abnormalities causing epilepsy.