Group Statistics of DTI Fiber Bundles Using Spatial Functions of Tensor Measures

Abstract

We present a framework for hypothesis testing of differences between groups of DTI fiber tracts. An anatomical, tract-oriented coordinate system provides a basis for estimating the distribution of diffusion properties. The parametrization of sampled, smooth functions is normalized across a population using DTI atlas building. Functional data analysis, an extension of multivariate statistics to continuous functions is applied to the problem of hypothesis testing and discrimination. B-spline models of fractional anisotropy (FA) and Frobenius norm measures are analyzed jointly. Plots of the discrimination direction provide a clinical interpretation of the group differences. The methodology is tested on a pediatric study of subjects aged one and two years.

Registration, atlas estimation and variability analysis of white matter fiber bundles modeled as currents

“…This is the reason why the variability analysis of the white matter often relies on diffusion images. This includes images of fractional anisotropy (Smith et al, 2006;Goodlett et al, 2008;Faria et al, 2010;Delmaire et al, 2009) or images of tensors (Alexander et al, 2001;Jones et al, 2002;Cao et al, 2005;Zhang et al, 2007;Yeo et al, 2009), for instance. The problem with these approaches is that there is no guarantee that the variability of the measures of diffusivity reflects the variability of the underlying neural pathways.…”

Tract-Based Probability Densities of Diffusivity Measures in DT-MRI

“…Using voxel or integral curve (as in tractography) representations of fiber tracts, previous research has proposed tract-based analysis of diffusion-derived measures in order to increase the specificity and robustness of related statistics [1,2,3]. A typical approach in these earlier studies is to find a representative skeleton (or curve) first and then project the diffusivity measures of individual subjects on this representative.…”

“…Parametrization of tracts based on arc-length distance has already been shown to be useful [5], and tract arc length by itself has been proposed as a metric [6]. As in TBSS, previous studies using tractography have also been able to detect differences between groups of subjects by analyzing diffusivity measures projected onto an arc-length parametrized representative tract [2,3].…”

“…Tract-based approaches to analysis of diffusion brain data sets are attracting interest because they can geometrically localize quantities while reducing sensitivity of associated statistics to computational errors and noise (e.g., [1,2,3]). In this context, we propose using probability densities in characterizing diffusion along diffusion-tensor MRI (DTI) fiber tracts generated from diffusion-weighted imaging data sets.…”

A Statistical Model of White Matter Fiber Bundles Based on Currents

“…These structures are of great importance as they may contribute to map brain connections between functional areas, or to understand effects of neurological pathologies (like Alzheimer's disease) onto the brain white matter. Most recent approaches are based on the nonlinear registration of fractional anisotropy (FA) maps [5,6] or tensor images [7,8]. The deformation resulting from this image registration is then applied to fibers.…”