Hello FreeSurfer team,

I am preprocessing the IXI dataset and noticed that my T2 results look incorrect, even when I run the same pipeline as for T1 with only the input changed to the T2 file. I would appreciate guidance on what I’m doing wrong or how I should process T2 properly.

Environment

- OS: Ubuntu 22.04.5 LTS

- FreeSurfer: 7.4.1

- SUBJECTS_DIR: /mnt/e/freesurfer_subjects

- FSF_OUTPUT_FORMAT: nii.gz (default unless noted)

What I ran

- For a single subject, I tried:

input="/mnt/e/T2_all/12subject_12.nii.gz"

sid="subject_12T2"

outdir="/mnt/e/output"

1) recon-all stage 1

recon-all -parallel -i "$input" -autorecon1 -subjid "$sid"

2) Apply Talairach transform

mri_convert "$SUBJECTS_DIR/$sid/mri/brainmask.mgz"

--apply_transform "$SUBJECTS_DIR/$sid/mri/transforms/talairach.xfm"

-o "$SUBJECTS_DIR/$sid/mri/brainmask_align.mgz"

3) Convert to NIfTI

mri_convert "$SUBJECTS_DIR/$sid/mri/brainmask_align.mgz"

"$outdir/${sid}_brainmask_align.nii.gz"

Issue

- The T2 result looks abnormal (while T1 looks OK). I assumed I could reuse the exact same steps for T2, just swapping the input file.

- Additionally, when I follow the steps from this guide:

https://github.com/junyuchen245/Preprocessed_IXI_Dataset/blob/main/PreprocessingMRI.md

my output NIfTI “info” (dimensions, sform/qform, etc.) is slightly different from the author’s preprocessed files (e.g., they have 160×192 while mine often become 256×256×256 after conformation). I’d like to understand why.

Questions

1) Is it correct that FreeSurfer’s recon-all is primarily designed for T1, and that T2 should NOT be run through the same standalone pipeline?

 

2)Then, how can I preprocess T2 using the same code as T1?  

Is there another method within FreeSurfer to handle T2 properly?

3) “Could you explain why some of the ‘info’ fields differ between the author’s preprocessed files and my outputs, even though I followed the same code?”

Thank you very much for your time and guidance!

Best regards,

[JHL]