Single-shell free-water elimination

1

Dear,

I am working on dMRI metrics after free-water elimination using single-shell data (e.g., b=0 and b=1000), as proposed in the paper: Free water elimination and mapping from diffusion MRI - PubMed . I am wondering is this method implemented in MRtrix3?

Thanks!

2

Hi @sunpan,

We don’t have that specific model implemented. The 3.1.0 release will have both better diffusion kurtosis fitting and a crude IVIM implementation, but nobody has invested the time in a FWE DTI fit, maybe in part due to it being fundamentally ill-posed.

Don’t know if it’s been proposed here previously, but a dirty solution does come to mind:

  1. Do a two-tissue CSD using WM-like and CSF-like response functions
  2. Perform a spherical convolution of the CSF ODF with the CSF-like response function to get the estimated contribution to the DWI signal from free water
  3. Subtract this from the empirical DWI data
  4. Perform a diffusion tensor fit

No idea how effective it might be, and no guarantee or even particular advocacy for its use, but it perhaps at least demonstrates the relationship between “free water elimination” and the way we approach spherical deconvolution.

Rob