This work presents a method that permits the characterization, quantification, and 3D visualization of white matter structural information contained within diffusion tensor MR imaging (DTMRI) data. In this method, regions within the brain are defined as possessing linear, planar, or spherical diffusion. Visualization of this diffusion metric data is realized by generating streamtube and streamsurface models to represent regions of linear and planar diffusion. Quantification of differences in diffusion anisotropy between different regions of interest (ROIs) is then achieved by analyzing 2D barycentric histograms created from the complete distribution of diffusion metric values measured in each region. In four healthy volunteers, there was only a small degree of asymmetry (epsilon) in the number of linear, planar, or spherical diffusion voxels between the right and left hemispheres (epsilon similar or equal to +/- 2%). However, in a patient with a metastatic brain lesion there was marked asymmetry in both linear (epsilon similar or equal to -10%) and planar (epsilon similar or equal to 5%) diffusion between comparable ipsilateral and contralateral regions, with a significant reduction in the number of linear diffusion voxels and an increase in the number of planar diffusion voxels in the tumor-bearing hemisphere. These results demonstrate the potential of this approach to characterize brain structure in both healthy and diseased subjects. (C) 2003 Wiley-Liss, Inc.