Biexponential diffusion tensor analysis of human brain diffusion data

Magn Reson Med. 2004 Feb;51(2):321-30. doi: 10.1002/mrm.10685.

Abstract

Several studies have shown that in tissues over an extended range of b-factors, the signal decay deviates significantly from the basic monoexponential model. The true nature of this departure has to date not been identified. For the current study, line scan diffusion images of brain suitable for biexponential diffusion tensor analysis were acquired in normal subjects on a clinical MR system. For each of six noncollinear directions, 32 images with b-factors ranging from 5 to 5000 s/mm2 were collected. Biexponential fits yielded parameter maps for a fast and a slow diffusion component. A subset of the diffusion data, consisting of the images obtained at the conventional range of b-factors between 5 and 972 s/mm2, was used for monoexponential diffusion tensor analysis. Fractional anisotropy (FA) of the fast-diffusion component and the monoexponential fit exhibited no significant difference. FA of the slow-diffusion biexponential component was significantly higher, particularly in areas of lower fiber density. The principal diffusion directions for the two biexponential components and the monoexponential solution were largely the same and in agreement with known fiber tracts. The second and third diffusion eigenvector directions also appeared to be aligned, but they exhibited significant deviations in localized areas.

Publication types

  • Research Support, Non-U.S. Gov't
  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Adult
  • Brain / anatomy & histology*
  • Brain / physiology*
  • Brain Mapping / methods*
  • Chi-Square Distribution
  • Diffusion Magnetic Resonance Imaging*
  • Humans
  • Image Processing, Computer-Assisted
  • Male