A Model for Some Subcortical DTI Planar and Linear Anisotropy

Abstract

Linear anisotropy, planar anisotropy and isotropy [1] are used as metrics for different kinds of diffusion in diffusion imaging. While linear anisotropy is reported to correlate to coherent neural fiber structures, the cause for planar anisotropy remains ambiguous. We hypothesize that overlapping linear structures and partial-volume averaging generate the planar anisotropy. We identify a subcortical region containing both linear and planar anisotropy in a human volumetric diffusion tensor image (DTI), propose a model of the anatomy and of the imaging process, and calculate simulated diffusion images of the anatomical model that qualitatively agree with the human DTI. Regions of planar anisotropy are common immediately beneath the cortex. Choosing one such region as representative, we model the anatomy with isotropic regions and linearly anisotropic structures. From the possibly overlapping model structures we simulate the diffusion imaging process, generating a series of diffusion weighted images (DWIs) of the anatomical structures. We then fit the DTI from these simulated DWIs and visualize it. The visualization agrees qualitatively with the visualization of a subcortical human DTI.