Skip to content

run_T2_preprocessing.sh — Core T2 Preprocessing

Target Audience

Neuroimaging researchers using T2-FLAIR images.

Location: BRC_structural_pipeline/scripts/run_T2_preprocessing.sh
Authors: Ali-Reza Mohammadi-Nejad, Stamatios N Sotiropoulos
Last updated: 02 Oct 2018

Purpose

Preprocesses the T2-weighted (FLAIR) image by co-registering it to the T1, applying the T1 brain mask, computing the T2→MNI registration, applying de-facing, and performing bias field correction using the T1-derived bias field. For RegType 2 and 3, it also invokes BIANCA for white matter lesion detection.

Processing Steps

1. Co-registration to T1

The T2 is registered to the T1 space. The approach differs by --regtype:

# Crop + BET the T2
bet T2_tmp T2_tmp_brain -R -m -f 0.50
# Register T2 brain to T1 brain (6 DOF, corratio cost)
flirt -in T2_tmp_brain -ref T1_brain -dof 6 -cost corratio \
      -omat T2_orig_ud_to_T2.mat

# Initial T2→T1_orig_ud alignment
flirt -in T2_orig_ud -ref T1_orig_ud -dof 6 -omat T2_tmp.mat
# Refine using T1 brain mask as weighting
flirt -in T2_orig_ud -ref T1_brain -refweight T1_brain_mask \
      -nosearch -init T2_tmp2.mat -dof 6 -omat T2_orig_ud_to_T2.mat

The T1 brain mask is used directly as the T2 brain mask.

2. T2→MNI Linear Transform

convert_xfm -omat T2_orig_ud_to_MNI_linear.mat \
    -concat T1_to_MNI_linear.mat T2_orig_ud_to_T2.mat

Combines the T2→T1 affine with the T1→MNI affine to get the T2→MNI transform without re-running registration.

3. Apply Linear Warp to MNI

applywarp --rel -i T2_orig -r MNI152_T1_1mm \
    -o T2_to_MNI_linear \
    --premat=T2_orig_ud_to_MNI_linear.mat --interp=spline

4. Non-Linear Warp to MNI (RegType 2 or 3)

applywarp --rel --interp=spline \
    -i T2_unbiased -r MNI152_T1_1mm \
    -w T1_to_MNI_nonlin_field \
    -o T2_to_MNI

The T1 non-linear warp field is reused — no separate T2 registration is run.

5. De-Facing (optional)

The MNI facemask is warped to T2 native space using the concatenated T2→MNI affine and then inverted. Facial voxels are zeroed in both the T2_orig and T2.

6. Bias Field Correction

fslmaths T2.nii.gz -div T1_brain_bias.nii.gz T2_unbiased.nii.gz
fslmaths T2_brain.nii.gz -div T1_brain_bias.nii.gz T2_unbiased_brain.nii.gz

The T1-derived bias field (from FSL FAST) is applied to the T2.

7. BIANCA Lesion Detection (RegType 2 or 3)

Calls run_T2_bianca.sh for white matter lesion detection.

Parameters

Parameter Type Description
--workingdir Path T2 temporary working directory
--t2input File path Path to T2_orig.nii.gz
--tempt1folder Path T1 temp folder (co-registration reference)
--fastfolder Path T1 FAST folder (provides bias field)
--regtempt1folder Path T1 registration temp folder (T1→MNI transforms)
--regtempt2folder Path T2 registration temp folder (output)
--dodefacing yes/no Apply de-facing to T2
--regtype 1/2/3 Registration strategy (must match T1)
--docrop yes/no Apply FOV cropping (RegType 1 only)
--biancatempfolder Path BIANCA temporary directory
--t2lesionpath Path Pre-existing lesion mask to propagate to T1 space
--logfile Path Log file path

Key Outputs

File Description
T2.nii.gz T2 aligned to T1 native space
T2_brain.nii.gz Skull-stripped T2 (T1 brain mask applied)
T2_brain_mask.nii.gz Brain mask (copied from T1)
T2_unbiased.nii.gz Bias-corrected T2
T2_unbiased_brain.nii.gz Bias-corrected skull-stripped T2
T2_orig_defaced.nii.gz De-faced T2 original (if defacing enabled)
reg/T2_to_MNI_linear.mat T2 native → MNI affine
reg/T2_to_MNI_linear.nii.gz Bias-corrected T2 in MNI (linear)
reg/T2_to_MNI.nii.gz Bias-corrected T2 in MNI (non-linear, RegType 2/3)