Continuing the discussion from Vector orientation confusion between MRTrix and FSL in rat data as a new topic since the underlying issues are distinct:
Quick recap: directions are flipped along the left-right axis in
mrview compared to
With a recent installation (more recent than April 15 2016, when this fix was released), there should be no mismatch. So there are two possibilities that I can think of:
your version of MRtrix3 is outdated. Check with:
$ dwi2tensor -version == dwi2tensor 3.0_RC1-191-g100da720 == ...
Then to really verify what date this corresponds to, take the SHA1 checksum, which in the example above is the last bit of the version string, minus the initial ‘g’:
100da720. Then from your MRtrix3 installation:
$ git log 100da720 commit 100da72015f1fa9541b94a1f28d71172029b46f5 (master) Merge: be66a20a 63664af0 Author: Robert Smith <firstname.lastname@example.org> Date: Tue Jul 25 10:39:42 2017 +1000 Merge pull request #1069 from MRtrix3/properties_get_step_size Tractography::Properties::get_step_size() function
From which you can read the commit date.
You are using a buggy version of
fslview. If I remember correctly,
fslviewversion 4.0 was affected by exactly the same bug that we fixed back in April 2016… Previous versions of
fslviewwere not affected, and I’m assuming more recent versions will work fine too.
In both cases, you can ‘fix’ this by changing the strides in the NIfTI image to LAS (the opposite to the NIfTI standard coordinate system, but what used to be the Analyze standard). You can do this with:
$ mrconvert dwi.nii -stride -1,2,3,4 dwi_LAS.nii
and then recompute the tensor and vectors.
Of course, the better ‘fix’ would be to upgrade either your MRtrix3 installation or
fslview, whichever requires it…