I’ve been using Ubuntu 16.04 on Bash on Windows (but this problem has occurred within regular linux/ubuntu as well).
After running DTIPrep (through their own tool, not through Slicer module), cannot convert .nrrd to .nii.gz due to “unsupported source pixel type” / “unsigned_short” (as has been discussed)
I know that this error can be fixed with the -allowLossyConversion option, but another problem is that this slicer conversion output has wrong header info (flipped on anterior/posterior axis). In the past, I’ve fixed this by following the FSL instructions here, with the example code below:
fslorient -deleteorient filename.nii.gz
[which is output from DWIConvert]
fslorient -setsformcode 1 filename.nii.gz
However, this fixed header file causes problems for registrations of this diffusion data to T1 weighted data.
I’m wondering:
-
can this “unsupported source pixel type” be avoided in the DTIPrep output by running DTIPrep as a Slicer module? (in other words, does DTIPrep as a Slicer module create output that is already “short”?)
-
Is there any other option other than -allowLossyConversion to move from unsigned_short to short?
-
or, is there a better way to fix the header of the output after Lossy conversion from nrrd to nifti?
Thanks so much!
Micah