Title: 

Sparse Laplace Inversion for Multidimensional Diffusion MRI

Abstract:

Multidimensional Diffusion MRI signals arise theoretically as the Laplace transform of a high dimensional probability density. Extracting this density from the signal is a horribly ill-conditioned problem and for it to be feasible at all, one typically has to assume that the density only consists of a few point masses. The signal model is then a sum of exponential functions of pure decay. In this talk, I will present an algorithm for inversion of such a signal model based on a connection to PSD Hankel matrices.

Speaker: Erik Troedsson