Using diffusion anisotropy to characterize neuronal morphology in gray matter: the orientation distribution of axons and dendrites in the NeuroMorpho.org database

Abstract

Read online

Accurate mathematical modeling is integral to the ability to interpret diffusion magneticresonance (MR) imaging data in terms of cellular structure in brain gray matter (GM). Inprevious work, we derived expressions to facilitate the determination of the orientationdistribution of axonal and dendritic processes from diffusion MR data. Here we utilize neuronreconstructions available in the NeuroMorpho database (www.neuromorpho.org) to assess thevalidity of the model we proposed by comparing morphological properties of the neurons topredictions based on diffusion MR simulations using the reconstructed neuron models. Initially,the method for directly determining neurite orientation distributions is shown to not depend onthe line length used to quantify cylindrical elements. Further variability in neuron morphology ischaracterized relative to neuron type, species, and laboratory of origin. Subsequently, diffusionMR signals are simulated based on human neocortical neuron reconstructions. This reveals a biasin which diffusion MR data predict neuron orientation distributions to have artificially lowanisotropy. This bias is shown to arise from shortcomings (already at relatively low diffusionweighting) in the Gaussian approximation of diffusion, in the presence of restrictive barriers, anddata analysis methods involving higher moments of the cumulant expansion are shown to becapable of reducing the magnitude of the observed bias.

Keywords

• Cerebral Cortex
• Diffusion
• MRI
• simulation
• neuron morphology
• Cytoarchitecture