ASL-Variability-MRI

Authors: Luis C Sanmiguel, Soetkin Beun, Liza Vanhauwaert, Pim Pullens, Patricia Clement.

Introduction

Pipeline Code to:

1. Sort DICOM images with DICOM sort.
2. Convert DICOM images to .nii.
3. [Human]:
   • Use Freesurfer to create a 3D map of the brain.
   • Use FSL to register ASL and M0 images with the Freesurfer output.
4. [QASPER]:
   • Use FSL to register ASL and M0 images.
5. Compare perfusion values for Human and QASPER in different sessions using Python.

Requirements

1. FSL - FMRIB Software Library and FreeSurfer.
2. DICOM Sort.
3. Software to convert DICOM to NIfTI (suggested).
4. Python/Jupyter Notebook.
5. Suggested: use Neurodesk.

Experiment Setup

Human Pipeline

Deface step:

mri_deface T1_mprage_SX.nii   talairach_mixed_with_skull.gca   face.gca   t1_mprage_defaced_SX.nii

recon-all: Creates a 3D map of the brain, identifying where gray and white matter are located (using the T1_mprage as input).

recon-all -s subject1_0X -i T1_mprage_sX.nii -all

mri_convert Changes file formats so different software programs can read the brain maps.

mri_convert subject1_0X/mri/brain.mgz subject1_0X/mri/brain.nii.gz

flirt Performs an initial "rough" alignment to get the ASL and T1 images into the same neighborhood.

flirt -in subject1_0X/M0X.nii -ref subject1_0X/mri/brain.nii.gz -omat subject1_0X/init1.mat -dof 6 -cost mutualinfo

bbregister Fine-tunes the alignment by matching the ASL image specifically to the boundary between gray and white matter.

bbregister --s subject1_0X --mov subject1_0X/M0X.nii --reg subject1_0X/initX.mat --o subject1_0X/cM0X.nii.gz --t2

lta_convert Translates the alignment "instructions" into a format FreeSurfer understands.

lta_convert --inreg subject1_0X/initX.mat --src subject1_0X/M0X.nii --trg subject1_0X/mri/T1.mgz --outlta subject1_0X/fmriX_to_T1.lta

mri_vol2vol Moves the ASL data into the T1 anatomical space using those alignment instructions

mri_vol2vol --mov subject1_0X/ASLX.nii --targ subject1_0X/mri/T1.mgz --lta subject1_0X/fmriX_to_T1.lta  --o subject1_0X/cASLX.nii.gz --interp trilinear

mri_convert Exports the final tissue segmentation (the labels for GM/WM) into a standard format for data extraction.

mri_convert subject1_0X/mri/aseg.mgz subject1_0X/asegX.nii.gz

QASPER Pipeline

flirt (Step 1): Compares the "calibration" image from the X session to the first session to calculate exactly how the subject's “head” moved (the "map" of the shift).

flirt -in ScanX/M0X.nii -ref Scan1/M01.nii -omat ScanX/scanX_to_scan1.mat -dof 6 -out ScanX/cM0X.nii.gz

flirt (Step 2): Uses that "map" to physically move the actual perfusion (ASL) data from the X session so it overlaps perfectly with the first session.

flirt -in ScanX/ASLX.nii -ref Scan1/M01.nii -applyxfm -init ScanX/scanX_to_scan1.mat -out ScanX/cASLX.nii.gz

Pipeline Summary