TBSS registration have orientation problem but files are all in standard orientation

Summary of what happened:

Here is a FA image I tried to align to the standard FA


Here is the final results:

Command used (and if a helper script was used, a link to the helper script or the command generated):

First used dcm2niix to do the conversion:
dcm2niix -a y -d n -g y -z y -o
Then process through a standard fdt pipeline;
Copy all FA images to a TBSS folder, then run
tbss_1_preproc
tbss_2_reg -T
tbss_3_postreg -T

People suggested to do a reorientation for the FA image, but I tried that and didn’t work. My images appear to be in the right orientation and fslreorient2std didn’t do anything.
FSL TBSS question… Warping? - Neuro Questions - Neurostars

Version:

fsl 5.0.3

Hi @zzhou, first of all, your brain extraction is quite poor - there are a large number of non-brain voxels in the inferior+anterior region. I’d suggest improving the brain extraction before running dtifit. Second of all, I would double check that you have re-oriented your data correctly - what is the output of fslhd on the image that you input to the tbss scripts?

Thanks, poor brain extraction is indeed the issue. I adjusted the bet threshold and got a good registration.