Mapping the structural connectivity of the human brain is one of the main goals of modern neuroscience. Diffusion MRI tractography is the only method that allows researchers probing white matter connections in-vivo, making it a unique tool for research and clinical investigations.
However, it remains difficult to scale high quality tractography dissections or tract segmentations to large datasets in a reliable and un-supervised manner. While a new generation of deep learning (DL) tractography methods have been recently introduced to make better use of anatomical knowledge to guide dissections, time-consuming manual dissections remain the reference benchmark in terms of anatomical accuracy and reproducibility.
Aim of this study is to develop a new framework for tractography dissections using novel DL tractography methods that can be anatomically accurate, reliable, and fully deployable to large neuroimaging projects.
In this PhD, the student will familiarise first with existing DL approaches while also learning the fundamental of diffusion tractography and neuroanatomy. During this period the student will compare DL tractography methods with conventional approaches. Next, the student will focus on improving DL tractography by combining novel fibre specific information and new tractography strategies with prior anatomical knowledge of individual tracts. New DL networks will be also trained for connections not commonly available with other methods or unique to specific studies. During the final part of the PhD, the student will test the new tractography framework on both large external datasets (e.g.HCP, UK-Biobank) and datasets with selected pathologies (e.g. Parkinson, Huntington diseases) available within the Neuroimaging Department.
Advanced programming experience is required.