Optic Nerve dMRI Registration¹

Non-linear movement/distortion correction in human optic nerve diffusion imaging.

---

Installation

with docker

• run docker build -t on_reg .

without docker

• install Python 2.7 and its libraries
  • numpy
  • scipy
  • nibabel >= 2.0
  • Priority dictionary (priodict.py, http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/117228)
  • ANTs 2.2.0

---

Usage:

run Docker image

• run_on_reg_docker.sh will run bash in the on_reg container as a current user. It also mount a current directory to /data inside the container.

Preparation

Registration

• example data file names for following commands:
  • fMRI: dwi.nii.gz

1. Create an initial ROI file (dwi_on_init.nii.gz) containing the followings:

   • posterior end voxel of right optic nerve: one voxel of value 1
   • posterior end voxel of left optic nerve: one voxel of value 2
   • anterior end region (right before eyeball) of optic nerve: 4-5 voxels of value 4 at each of left and right optic nerve on a coronal slice
   • Note: we recommend users set these points on both the mean b0 image and the mean high-b-value image.

2. Create an exclusion ROI file (dwi_exclusion.nii.gz) containing masks of recti muscle near optic nerve.

3. Run the following command

on_reg.py \
    -i dwi_on_init.nii.gz \
    -e dwi_exclusion.nii.gz \
    dwi.nii.gz

4. Check the initial centerlines for all volumes in the output file of step 3 (dwi_on_centerline.nii.gz and its dilated image dwi_on_centerline_dilated.nii.gz). If some centerline is incorrect, modify exclusion ROI of 2 and re-run 3.

5. Run the following command (the same as step 3 but it reads the created initial centerline generated at step 4)

on_reg.py \
    -i dwi_on_init.nii.gz \
    -e dwi_exclusion.nii.gz \
    dwi.nii.gz

6. Check the resulting optic nerve segmentation (dwi_on_model.nii.gz and its projection image dwi_on_model_1st_bin.nii.gz). e.g., fslview dwi dwi_on_model -l Red -b 0.8,1.2 -t 0.3 dwi_on_model_1st_bin -l Red -t 0.3. If necessary, modify the exclusion ROI file and re-run step 5.

7. Check the resistered image (dwi_nonlin_reg.nii.gz)

Optic nerve center voxel generation

• Create an optic nerve center ROI file (on_center.nii.gz) from the model file (dwi_on_model) at step 5.

on_create_center_from_model.py \
    -o on_center.nii.gz \
    -f on_center_volumes.nii.gz \
    -r dwi.nii.gz \
    dwi_on_model

Reference:

[1] Kim et al, Incorporating non-linear alignment and multi-compartmental modeling for improved human optic nerve diffusion imaging. Neuroimage, 2019, 196:102-113. https://www.ncbi.nlm.nih.gov/pubmed/30930313