Cellular deconvolution of the brain with topological diffusion mapping MRI

Motivation: MRI indirectly measures neuropathology by producing imaging phenotypes, which are non-specific macroscale proxies for the molecular and cellular processes underlying neuropathology. Can MRI more specifically measure cellular changes associated with neuropathology? Goal(s): Develop an image analysis approach able to non-invasively measure and detect cell populations in the brain. Approach: We performed diffusion MRI (dMRI) in rat models of neurologic illness and using topological data analysis (TDA), were able to train machine learning models to deconvolve cellular populations (e.g. microglia, neurons) in the brain. Results: TDA of dMRI data is feasible and allows for non-invasive cellular deconvolution of the brain. Impact: Topological diffusion mapping MRI (tdmMRI) is an innovative spatially-informed image analysis technique providing enhanced imaging sensitivity and specificity to neurobiology and further, is able to non-invasively deconvolve cellular populations in the brain.