Diffusion as a probe of tissue microstructure

Diffusion MRI (dMRI) has become one of the most popular contrast mechanisms that MRI offers, with applications spanning a wide range of research fields and clinical needs. The main reason for the popularity of dMRI is that it enables rapid extraction of unique tissue information that cannot be obtained by any other MRI methods or in-vivo imaging modalities. The popularity and usage of dMRI is even more striking considering the number of artifacts, pitfalls, and conceptual issues of this method. Regarding the latter, the most dramatic is that dMRI does not directly measure diffusion, but rather the molecules' displacement. Throughout the evolution of dMRI, the acquisition and analysis framework originally developed by Stejskal and Tanner to study the bulk diffusion of liquids has been used under one significant assumption: that the motion of molecules should be Brownian, meaning that diffusion should be free and extend in a Gaussian manner in time and space. When that assumption holds, the original Stejskal–Tanner framework can measure the physical properties of diffusion, and hence the method can be called diffusion MRI. However, in biological tissues, this main assumption of Stejskal and Tanner does not hold. The motion of water molecules in biological tissues, and specifically in the brain, is not Brownian and does not expand in a Gaussian manner, either in space or in time. This specific misalignment between theory and reality causes significant modeling and interpretation issues, yet has not affected the burgeoning use of dMRI in neuroscience. Thirty years of research on dMRI in the brain have highlighted these pitfalls and provided several insights and solutions to some extent. This chapter will review the current state-of-the-art dMRI modeling in the brain, the innovative applications in neuroscience, and future prospects with emphasis of using dMRI as a microstructural probe.