Diffusion MRI is widely used for the clinical examination of a variety of diseases of the
nervous system. However, clinical MRI scanners are mostly capable of low magnetic field
gradients. The availability of the Connectome MRI scanners with ultra-high gradient strengths
up to 300 mT/m enables the resolution of a wider range of diffusion coefficients and tissue
structures. In addition, biological tissues are heterogeneous on many scales and the
complexity of tissue microstructure may not be accurately captured by simplistic models
based on pre-existing assumptions. For example, an axon may not be considered as a straight tube as in most of the diffusion literature. Here, we study the intrinsic diffusion physics in thousands of realistic axon models and show a diversity of diffusion behavior. We outline a framework for analyzing diffusion MRI to obtain a model-free Full Diffusion Tensor Distribution (FDTD). We then apply this method and use a machine learning method to identify features in FDTD to visualize and characterize tissue heterogeneity in diffuse gliomas.