Most clinical diffusion MRI studies rely on the statistical comparison of a group of patients against a group of healthy controls to make inference about disease. This stymies the potential power of microstructural MRI in the clinic, i.e., to identify microstructural abnormalities in a single patient. We present a framework to address this problem on a case-by-case basis, extending the reach of microstructural imaging to rare cases, where group comparisons are otherwise impossible. Our framework operates on the manifold of white matter pathways and uses autoencoders to learn normative microstructural features, and discriminate patients from controls in a paediatric population.
This abstract and the presentation materials are available to members only; a login is required.