Abstract #0077

Inferring Micron-scale Tissue Structure using Extreme Value Theory for Cylindrically-restricted Diffusion

Leigh A. Johnston^1,2, David Wright², Rick H. H. M. Philipsen³, Scott C. Kolbe², James A. Bourne⁴, Iven M. Y. Mareels¹, Gary F. Egan²

¹Electrical & Electronic Engineering & NICTA VRL, University of Melbourne, Parkville, VIC, Australia; ²Howard Florey Institute, Florey Neuroscience Institutes, Parkville, VIC, Australia; ³Technical University of Eindhoven, Netherlands; ⁴Australian Regenerative Medicine Institute, Monash University, Australia

The non-exponential signal decay observed in q-space diffusion acqusitions is derived for restricted diffusion in cylinders, using a probabilistic approach based on extreme value theory and expectation over continuum distributions for geometry-dependent apparent diffusion coefficients. Simulation and experimental results demonstrate the accuracy of the resultant non-exponential signal decay and the ability to infer axon densities without need for pulse duration or diffusion time approximations.