Clinical Applications of Quantitative MRI in Healthy Adolescents - Project Pipeline

Pipeline & Research Notebook - Neuroscience MSc Research Project in "Clinical Applications of Quantitative MRI in Healthy Adolescents"

This Source Code Form is subject to the terms of the Mozilla Public License, v. 2.0. If a copy of the MPL was not distributed with this file, You can obtain one at http://mozilla.org/MPL/2.0/.

Step 1 - Creating the HD-BET brain masks to be used in the QUIT MPM pipeline

• First, create and run fsl_roi.sh as documented in “Step 1” in the notebook.

• Create an SGE job file that receives the index file and and index, and run it as following:

  qsub -t 1:[N] brain_mask_creation.job
  

  where N is the number of rows in the brain_mask_creation.index file.

• Then, run the QUIT MPM pipeline while providing the correct brain mask for each acquistion.

• Validate your steps and results.

Step 2 - Copying all the original MTsat files to the relevant working directory

• After the QUIT MPM pipeline finishes running, copy all the resulting MTsat MPM files to the mt_delta_template_work_dir.
• Follow step number 2 in the notebook and validate your steps and results.

Step 3 - Running a manual quality control (QC) step

• Create an "excluded" directory under each age_group specific directory.
• Move the images that are not suitable for template creation into the /excluded directory.
• Validate your steps and results.

Step 4 - Creating the MTsat templates per age group using ANTs

• Copy the ANTs antsMultivariateTemplateConstruction2.sh script into each one of the age_group directories under mt_delta_template_work_dir
• Edit it to your liking according to your wanted SGE runtime configuration (e.g. QSUBOPTS)
• Run, for example:

./antsMultivariateTemplateConstruction2.sh -b 0 -c 1 -d 3 -i 4 -g 0.2 -k 1 -n 0 -r 1 -o ${PWD}/mt_delta_template_14_and_above_output_ ${PWD}/sub*.nii.gz

• Validate your steps and results.

Step 5 - registering the remaining excluded MTsat images into the resulting age-appropriate MTsat template using ANTs

• For each age group, create and run the register_mt_to_{age_group_name}_template.sh file.
• Follow step number 5 in the notebook and validate your steps and results

Step 6 - copying the resulting transformation files for each MTsat image in the age-appropriate MTsat template to the relevant MPM folder

• Two transformation files will be copied - the SyN deformation file (*1Warp.nii.gz file) and the affine transformation file (0GenericAffine.mat).
• The MTsat images transformed into the MTsat template space are also copied.
• Follow step number 6 in the notebook and validate your steps and results.

Step 7 - Creating the index file to transform all the MPM images (R1, R2*, PD) from their original space into the age-appropriate MTsat template space

• To create the index files, follow step number 7 in the notebook and validate your steps and results.

Step 8 - now, we want to transform the R1, R2* and PD images from their original space into the age-appropriate MTsat template space

• Create an SGE job file that receives the index file and and index, and run it as following:

  qsub -t 1:[N] mpm_to_mt_template.job  
  

  where N is the number of rows in the mpm_to_mt_template.index file

• Follow step number 8 in the notebook and validate your steps and results.

Step 9 - Creating the index file to transform all the MPM images (R1, R2*, PD) from the age-appropriate MTsat template space into the MNI template space

• The MNI template chosen is the tpl-MNIPediatricAsym cohort 5 1mm template.
• Follow step number 9 in the notebook and validate your steps and results.

Step 10 - now, we want to transform the R1, R2* and PD images from the age-appropriate MTsat template space into the MNI template space

• Create an SGE job file that receives the index file and and index, and run it as following:

  qsub -t 1:[N] mpm_in_mt_to_mni.job
  

• where N is the number of rows in the mpm_in_mt_to_mni.index file.

• Follow step number 10 in the notebook and validate your steps and results.

Step 11 - merging all MTsat images in the MNI space for a 2nd round of quality control (following registration), before any statistical analysis takes place

• Create an "fsl merge" command to concat all (before any quality control) MTsat MPM files into a single 4d nii.gz file.
• Run the resulting fsl merge command and manually perform a quality control step on the resulting 4d file. Follow a similar pattern for the 4d files of the remaining modalities as well.

Step 12 - merging all MTsat, R1, R2*, PD images that passed QC in order to perform statistical analysis using fsl randomise on each 4d file

• Create the “fsl merge” command and run it on each modality, to create the final 4d files with the images that passed the QC step.
• Validate your steps and results.

Step 13 - Statistical analysis:

FSL randomise:

• Create design and matrix files (age_corr_sorted.mat and age_corr.con, assuming each 4d file contain acquisitions from the same participants and sessions in the same temporal order).
• A mask containing only white matter and gray matter was created on top of the existing WM/GM masks for the provided MNI template and theshloded.
• Run permutation tests with age as a single covariate.

randomise_parallel -i 4d_all_mt_sorted.nii.gz -o mt_all_age_corr_sorted_qc2 -d age_corr_sorted.mat -t age_corr.con -m wm_gm_mask.nii.gz -n 10000 -T -D
    
randomise_parallel -i 4d_all_r1_sorted.nii.gz -o r1_all_age_corr_sorted_qc2 -d age_corr_sorted.mat -t age_corr.con -m wm_gm_mask.nii.gz -n 10000 -T -D
    
randomise_parallel -i 4d_all_r2s_sorted.nii.gz -o r2s_all_age_corr_sorted_qc2 -d age_corr_sorted.mat -t age_corr.con -m wm_gm_mask.nii.gz -n 10000 -T -D

• Threshold the resulting images with values above 0.95 for statistically significancant clusters.

ROI analysis:

• Create a binary mask for each region. For instance, in the Harvard-Oxford Subcortical Atlas, the left pallidum has a value of "7". For example:

  fslmaths HarvardOxford-sub-maxprob-thr0-1mm.nii.gz -thr 7 -uthr 7 -bin l_pallidum.nii.gz  
  

• For each permutation of (modality, ROI) run fslstats in order output the mean value. For example:

  fslstats -t 4d_all_mt_sorted.nii.gz -k l_pallidum.nii.gz -m
  

• Finally, analyse the relationship between the mean MTsat, R1 and R2* values in each ROI and age by using the output of fslstats. Our statistical analysis was performed using Graphpad Prism.