Effective medium theory of a diffusion-weighted signal.

Living tissues and other heterogeneous media generally consist of structural units with different diffusion coefficients and NMR properties. These blocks, such as cells or clusters of cells, can be much smaller than the imaging voxel, and are often comparable with the diffusion length. We have developed a general approach to quantify the medium heterogeneity when it is much finer than the sample size or the imaging resolution. The approach is based on the treatment of the medium statistically in terms of the correlation functions of the local parameters. The diffusion-weighted signal is explicity found for the case in which the local diffusivity varies in space, in the lowest order in the diffusivity variance. We demonstrate how the correlation length and the variance of the local diffusivity contribute to the time-dependent diffusion coefficient and the time-dependent kurtosis. Our results are corroborated by Monte Carlo simulations of diffusion in a two-dimensional heterogeneous medium.