A microstructure fingerprinting approach is proposed in which an optimal single-fascicle configuration is selected from a pre-computed collection of Monte Carlo diffusion signals, or fingerprints, uniquely relating the diffusion MRI signals to the underlying axon properties. The simulated 3D geometries feature randomly-placed cylinders with diameter heterogeneity. The approach is validated on a public dataset of ex vivo cat spinal cord and exhibits indices of axon density and of the mean and standard deviation of the axon diameter distribution in agreement with histological measurements. The method is shown to outperform AMICO, which relies on approximate analytical expressions for the signal, and a simpler model using Monte Carlo simulations in a homogeneous packing of identical cylinders.
This abstract and the presentation materials are available to members only; a login is required.