FSL Lesion Normalization¶
Goal: Warp a native space T1 weighted image and lesion mask into standard (MNI) space using FSL.
Requirements: You need to have FSL installed and working. Place the three scripts described below (lesion_norm_fsl.sh, fsl_anat_alt.sh, and optiBET.sh) in your path. Some familiarity with Unix command line processing is assumed. Once everything is in your path, you will run lesion_norm_fsl.sh as illustrated below.
Images can be zipped or not (i.e., T1w.nii and T1w.nii.gz are OK).
A structural (anatomical) T1 weighted image.
A lesion mask with values of 1 in the lesion and 0 elsewhere.
The 3 Scripts¶
lesion_norm_fsl.sh assumes we are in the directory with the images we need. It expects 2 arguments:
a T1 anatomical image
a lesion mask (1=lesion; 0=non-lesion)
lesion_norm_fsl.sh sub-001.nii.gz sub-001_LesionSmooth.nii.gz
lesion_norm_fsl.sh calls the two scripts described below (fsl_anat_alt.sh and optiBET.sh)
fsl_anat_alt.sh is an altered version of fsl_anat that does several things:
It uncomments the lesion mask handling calls that fsl_anat was not using.
It calls optiBET.sh to improve the quality of BET, especially for lesioned brains.
It assumes that because optiBET is so much better, we can use the skull-stripped brain for linear registration.
optiBET.sh is a simplified version of the original optiBET. It calls bin/bash instead of bin/sh, and does the skull stripping described below using the FSL options.
Why Skull Strip with optiBET?¶
Registration of the participant brain to the standard brain is generally better accomplished if both brains are skull stripped (but see SPM12). Skull stripping is difficult, especially for individuals with large lesions.
The optiBET.sh script is impressive, but you should always look at the skull stripping results and edit them if they are bad. Once you have a good skull strip (all the brain, and nothing but the brain), you are ready to go to the next step in your processing journey.