A Sheet Probability Index from Diffusion Tensor Imaging

In proceedings of Computational Diffusion MRI, Springer, 2017
 

Abstract

A sheet probability index (SPI) has recently been derived from high angular resolution diffusion MRI to quantify the hypothesis that white matter tracts are organized in parallel sheets of interwoven paths. In this work, we derive the DTI-SPI, a variant of the SPI that can be computed from the widely available, simple, and fast diffusion tensor imaging, by considering the normal component of the Lie bracket of the major and medium eigenvector fields. We observe that, despite the fact that DTI does not allow us to infer crossing fiber orientations, the DTI-SPI has a meaningful interpretation in terms of sheet structure if the largest pair of eigenvectors spans the same plane as the two dominant fibers. We report empirical results that support this assumption. We also show a direct comparison to the previously proposed SPI on data from the human connectome project, and demonstrate that major features in maps of our DTI-SPI remain recognizable in standard clinical DTI data.

Images

Bibtex

@INPROCEEDINGS{ankele:cdmri17,
     author = {Ankele, Michael and Schultz, Thomas},
      title = {A Sheet Probability Index from Diffusion Tensor Imaging},
  booktitle = {Computational Diffusion MRI},
       year = {2017},
  publisher = {Springer},
   abstract = {A sheet probability index (SPI) has recently been derived from high angular resolution diffusion MRI
               to quantify the hypothesis that white matter tracts are organized in parallel sheets of interwoven
               paths. In this work, we derive the DTI-SPI, a variant of the SPI that can be computed from the
               widely available, simple, and fast diffusion tensor imaging, by considering the normal component of
               the Lie bracket of the major and medium eigenvector fields. We observe that, despite the fact that
               DTI does not allow us to infer crossing fiber orientations, the DTI-SPI has a meaningful
               interpretation in terms of sheet structure if the largest pair of eigenvectors spans the same plane
               as the two dominant fibers. We report empirical results that support this assumption. We also show a
               direct comparison to the previously proposed SPI on data from the human connectome project, and
               demonstrate that major features in maps of our DTI-SPI remain recognizable in standard clinical DTI
               data.}
}