Dr Julien Cohen-Adad: In vivo histology with MRI

Recent advances in MRI have led to the development of techniques that provide quantitative measures of tissue microstructure. Axon diameter and density can be inferred from q-space diffusion protocols, while myelin density can be inferred from quantitative magnetization transfer (qMT), myelin water fraction or macromolecular tissue volume (MTV). Surprisingly, these techniques have mostly been studied independently despite their complementarity: together they can provide comprehensive characterization of the fibre microstructure, including quantifying the myelin thickness or the so-called g-ratio. Moreover, most validation work has been done using tissue staining approaches (e.g., luxol fast blue), without looking at the axon microstructure itself. In this talk we will briefly describe some of the techniques for in vivo histology with MRI, how to circumvent acquisition challenges related to susceptibility artifacts in the spinal cord (e.g., high order real time shimming), how to process those data using template-based analysis (github.com/neuropoly/spinalcordtoolbox), and finally how to validate those techniques using electron microscopy and automatic segmentation of axon/myelin using deep learning (github.com/neuropoly/axondeepseg).