Linking MRI and microscopy for multi-scale neuroscience: Mechanisms, diagnostics and anatomy

MRI has tremendous potential to provide diagnostic and mechanistic insights into brain health and disease in living subjects, but is limited by its poorly defined relationship to histology. My group has pioneered techniques for combining MRI and histology that will enable us to define these relationships to provide more biologically interpretable MRI measures. We will construct models relating MRI and histology both from the bottom up (using microscopy to predict MRI signals) and top down (predicting histopathology from MRI). The top-down approach will use machine-learning methods to predict histological stains from in-vivo MRI, providing “virtual neuropathology” for improved diagnosis in living patients. The bottom-up approach will use electron and optical microscopy for hyper-realistic predictions of the MRI signal with the goal of identifying novel MRI signatures relating to mechanisms of neural health and disease. My group is poised to leverage our unique expertise within a comprehensive research program spanning scales (microscopic to macroscopic), species (rodent to humans) and expertise (physics to neuroscience). We will deploy these methods with neuroscience collaborators for: (i) virtual neuropathology in ALS; (ii) mechanisms of experience-induced plasticity; and (iii) high-resolution neuroanatomy. A primary output will be the “Oxford Digital Brain Bank”, a freely available data repository.