run_T1_preprocessing.sh — Core T1 Preprocessing¶

Target Audience

Neuroimaging researchers and pipeline developers.

Location: BRC_structural_pipeline/scripts/run_T1_preprocessing.sh
Authors: Ali-Reza Mohammadi-Nejad, Stamatios N Sotiropoulos
Last updated: 02 Oct 2018

Purpose¶

Performs all core T1 image preprocessing. This is the most computationally intensive script in the pipeline. It combines brain extraction, registration to standard space, bias correction, tissue segmentation, subcortical segmentation, and brain volume estimation into a single coordinated workflow.

Processing Steps¶

1. Copy and Reorient¶

fslreorient2std T1_orig_ud T1_orig_ud

The raw input is copied to the working directory as T1_orig_ud and reoriented to standard (RAS) orientation.

2. Field-of-View Cropping (optional)¶

robustfov -i T1_orig_ud
fslmaths T1_orig_ud -roi 0 -1 0 -1 $head_top 170 0 1 T1_tmp

Automatically crops the superior FOV to remove neck. Skipped when --docrop=no.

3. Initial Brain Extraction¶

bet T1_tmp T1_tmp_brain -R -m -f <fBET>

Recursive BET with fractional intensity threshold fBET (default 0.5). Produces an initial brain mask used in subsequent steps.

4. FOV Reduction to Head Region¶

standard_space_roi T1_tmp_brain T1_tmp2 -maskNONE -ssref MNI152_T1_1mm_brain -altinput T1_orig_ud -d

Reduces the FOV to the head region using a standard space reference. Output is renamed to T1.

5. Linear Registration to MNI152¶

flirt -interp spline -dof 12 -in T1 -ref MNI152_T1_1mm -omat T1_to_MNI_linear.mat

12-DOF affine registration to MNI152 1mm. The resulting .mat file is used to initialise non-linear registration.

6. Non-Linear Registration (RegType 2 or 3)¶

RegType 2 — FNIRTRegType 3 — ANTs SyN

fnirt --in=T1 \
      --ref=MNI152_T1_1mm \
      --aff=T1_to_MNI_linear.mat \
      --config=bb_fnirt.cnf \
      --refmask=MNI152_T1_1mm_brain_mask_dil_GD7 \
      --cout=T1_to_MNI_nonlin_coeff \
      --fout=T1_to_MNI_nonlin_field

antsRegistrationSyN.sh -d 3 \
    -f MNI152_T1_1mm.nii.gz \
    -m T1.nii.gz \
    -o SynOut \
    -n 48 -j 1

ANTs output is converted to FSL format using c3d_affine_tool and convertwarp.

7. Brain Extraction Using Inverse Warp¶

The MNI brain mask is warped back to native T1 space using the inverse of the non-linear warp (RegType 2/3) or the linear transform inverse (RegType 1):

applywarp --rel --interp=trilinear \
    --in=MNI152_T1_1mm_brain_mask \
    --ref=T1 \
    -w T1_to_MNI_nonlin_coeff_inv \
    -o T1_brain_mask
fslmaths T1 -mul T1_brain_mask T1_brain

8. De-Facing (optional)¶

# Warp MNI facemask to native T1 space
convert_xfm -omat grot.mat -concat MNI_to_MNI_BigFoV_facemask.mat T1_to_MNI_linear.mat
convert_xfm -omat grot.mat -inverse grot.mat
flirt -in MNI152_T1_1mm_BigFoV_facemask -ref T1 -out grot -applyxfm -init grot.mat
# Zero-out face voxels
fslmaths grot -binv -mul T1 T1

A QC metric is computed: number of voxels where the facemask overlaps the brain mask.

9. Bias Field Estimation & Correction¶

fast -b -o FAST/T1_brain T1_brain
fslmaths T1 -div FAST/T1_brain_bias T1_unbiased
fslmaths T1_brain -div FAST/T1_brain_bias T1_unbiased_brain

FSL FAST estimates the bias field (-b flag). The field is divided out to produce bias-corrected volumes.

10. Tissue Segmentation (optional)¶

fslmaths FAST/T1_brain_pve_0 -thr 0.5 -bin FAST/T1_brain_CSF_mask
fslmaths FAST/T1_brain_pve_1 -thr 0.5 -bin FAST/T1_brain_GM_mask
fslmaths FAST/T1_brain_pve_2 -thr 0.5 -bin FAST/T1_brain_WM_mask

Thresholds the FAST partial volume estimates at 0.5 to produce binary tissue masks. Skipped when --dotissueseg=no.

11. Apply Registration to Bias-Corrected Volumes¶

Linear and (if applicable) non-linear warps are applied to T1_unbiased and T1_unbiased_brain to produce images in MNI space.

12. Subcortical Segmentation (optional)¶

run_first_all -i T1_unbiased_brain -b -o FIRST/T1_first

FSL FIRST segments 15 subcortical structures. Skipped when --dosubseg=no.

13. SIENAX Brain Volume Estimation (RegType 2 or 3 only)¶

Calls run_T1_sienax.sh to produce normalised brain volume estimates.

Parameters¶

Parameter	Type	Description
`--workingdir`	Path	T1 temporary working directory
`--t1input`	File path	Path to `T1_orig.nii.gz`
`--dosubseg`	yes/no	Enable FSL FIRST subcortical segmentation
`--dotissueseg`	yes/no	Enable FAST binary tissue masks
`--docrop`	yes/no	Enable FOV cropping
`--dodefacing`	yes/no	Enable de-facing
`--fastfolder`	Path	FSL FAST output directory
`--firstfolder`	Path	FSL FIRST output directory
`--sienaxtempfolder`	Path	SIENAX temporary directory
`--regtempt1folder`	Path	Registration outputs directory
`--regtype`	1/2/3	Registration strategy
`--fbet`	Float 0–1	BET fractional intensity threshold (default 0.5)
`--logfile`	Path	Log file path

Key Outputs¶

File	Description
`T1.nii.gz`	Cropped, reoriented T1
`T1_brain.nii.gz`	Skull-stripped T1
`T1_brain_mask.nii.gz`	Binary brain mask
`T1_unbiased.nii.gz`	Bias-corrected whole-head T1
`T1_unbiased_brain.nii.gz`	Bias-corrected skull-stripped T1
`T1_orig_defaced.nii.gz`	De-faced original T1 (if defacing enabled)
`reg/T1_to_MNI_linear.mat`	T1 native → MNI affine matrix
`reg/T1_to_MNI_nonlin_field.nii.gz`	Non-linear warp field T1 → MNI
`reg/T1_to_MNI_nonlin_coeff_inv.nii.gz`	Inverse warp: MNI → T1
`FAST/T1_brain_bias.nii.gz`	Estimated bias field
`FAST/T1_brain_pve_{0,1,2}.nii.gz`	CSF / GM / WM partial volume estimates
`FAST/T1_brain_{CSF,GM,WM}_mask.nii.gz`	Binary tissue masks