Help with Cropped Images in FSL Preprocessing

Software Support
, , ,
1

Summary of what happened:

I am currently working on preprocessing images that will be used in a model using FSL. However, I noticed that some parts of the fmri_in_mni image appear slightly cropped. After reviewing the intermediate images, I found that the issue starts from the flirt_fmri_to_t1 image, where a small part of the image seems to be cut off. Could you please advise on how to handle this situation? Any guidance would be greatly appreciated.

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

import os
import subprocess
import nibabel as nib
import time

# Environment setup
os.environ['FSLDIR'] = '/path/to/fsl'
os.environ['PATH'] += os.pathsep + os.path.join(os.environ['FSLDIR'], 'bin')

# File paths (redacted)
fmri_file = '/path/to/fmri_data.nii.gz'
t1_file = '/path/to/t1_image.nii.gz'
mni_template = os.path.join(os.environ['FSLDIR'], 'data', 'standard', 'MNI152_T1_1mm_brain.nii.gz')
output_dir = '/path/to/output_directory'
os.makedirs(output_dir, exist_ok=True)

def run_command(command):
    """Utility function to run shell commands"""
    print(f"Running: {command}")
    subprocess.run(command, shell=True, check=True)

# 1. Load the resting-state fMRI file
print("Loading fMRI data...")
fmri_img = nib.load(fmri_file)

# 2. BET: Perform brain extraction on fMRI data
print("Starting BET brain extraction for fMRI...")
bet_fmri_output = os.path.join(output_dir, "fmri_brain")
bet_command = f"bet {fmri_file} {bet_fmri_output} -F -f 0.5 -m"
run_command(bet_command)

# 2.1 Extract the first volume from the fMRI data
print("Extracting the first volume from fMRI data...")
fmri_img = nib.load(bet_fmri_output + ".nii.gz")
first_volume_data = fmri_img.get_fdata()[..., 0]  # Extract the first volume
first_volume_img = nib.Nifti1Image(first_volume_data, fmri_img.affine, fmri_img.header)
first_volume_path = os.path.join(output_dir, "fmri_brain_first_volume.nii.gz")
nib.save(first_volume_img, first_volume_path)

# 3. BET: Perform brain extraction on the T1-weighted image
print("Starting BET brain extraction for T1w...")
bet_t1_output = os.path.join(output_dir, "t1_brain")
bet_t1_command = f"bet {t1_file} {bet_t1_output} -f 0.5 -m"
run_command(bet_t1_command)

# 4. FLIRT: Perform linear registration of the brain-extracted fMRI data to the brain-extracted T1w image
print("Starting FLIRT registration (fMRI -> T1w)...")
flirt_fmri_to_t1_output = os.path.join(output_dir, "flirt_fmri_to_t1")
flirt_fmri_to_t1_mat_output = os.path.join(output_dir, "fmri_to_t1.mat")
flirt_fmri_to_t1_command = f"flirt -in {first_volume_path} -ref {bet_t1_output} -out {flirt_fmri_to_t1_output} -omat {flirt_fmri_to_t1_mat_output} -dof 6"
run_command(flirt_fmri_to_t1_command)

# 5. FLIRT: Perform linear registration of the brain-extracted T1w image to the MNI152 template
print("Starting FLIRT registration (T1w -> MNI)...")
flirt_t1_to_mni_output = os.path.join(output_dir, "flirt_t1_to_mni")
flirt_t1_to_mni_mat_output = os.path.join(output_dir, "t1_to_mni.mat")
flirt_t1_to_mni_command = f"flirt -in {bet_t1_output} -ref {mni_template} -out {flirt_t1_to_mni_output} -omat {flirt_t1_to_mni_mat_output} -dof 12"
run_command(flirt_t1_to_mni_command)

# 6. FNIRT: Perform non-linear registration of the T1w image to the MNI152 template
print("Starting FNIRT registration (T1w -> MNI)...")
fnirt_output = os.path.join(output_dir, "t1_to_mni_warp")
fnirt_command = f"fnirt --in={bet_t1_output} --ref={mni_template} --aff={flirt_t1_to_mni_mat_output} --cout={fnirt_output}"
run_command(fnirt_command)

# 7. applywarp: Apply the warp from FNIRT to transform the fMRI data into MNI152 template space
print("Applying warp to fMRI data to MNI space...")
apply_warp_output = os.path.join(output_dir, "fmri_in_mni")
apply_warp_command = f"applywarp --in={flirt_fmri_to_t1_output} --ref={mni_template} --warp={fnirt_output} --out={apply_warp_output}"
run_command(apply_warp_command)

print("fMRI preprocessing pipeline completed successfully.")

Screenshots / relevant information:

2

Hi @e.lee, I don’t really see anything wrong with the screenshot you have posted. However, it looks like the field-of-view of your images is quite large (lots of empty space around the brain). So you may see some improvement if you reduce the FOV of both your functional and structural data before any other steps. You can do this using the robustfov command, e.g.:

robustfov -i fmri_data.nii.gz -r fmri_data_cropped.nii.gz
1 Like