Hello, every MRtrix3 expert,

It provides me lots of convenience using MRtrix3 to calculate egienvalues. But I am confused that in some voxels, the value of some elements in tensor or eigenvalue is negative. It seems that these negative eigenvalues or elements in tensor are unphysical. Would you like to give some suggestions about how to deal with these negative values or if adding postivie constraint when reconstructing tensor in MRtrix3.

Thanks very much,

Zhaoqing

You are right, these are indeed unphysical. We are in the process of adding exactly this constraint to dwi2tensor, so you will have to hang in there just a little bit longer.

For the time being, I suggest that you make sure that you use dwidenoise and mrdegibbs to reduce the presence of physically implausible signals as much as possible. This should already fix the problem in most voxels.

Thanks much for your suggestions. I am sorry I used raw data to reconstruct tensor. I will follow your suggestions and try again.