Optic Radiation Tutorial


Jump to: navigation, search

This tutorial will take a user from the raw data to a graph showing FA values along the optic radiation. It assumes that data are collected at CNI and downloaded as NIFTI.


[edit] Processing Raw Data

The processing begins with a

  • 3D NIFTI file containing a T1 anatomical volume, and
  • 4D NIFTI file containing multiple volumes of the diffusion measurements in the bvecs directions.
  • You should know the assignment between the 4D NIFTI data and the bvals/bvecs of your diffusion scan. These are usually tab-delimited text files that can be read with matlab's dlmread() function.

We refer to these data sets as T1 and DWI.

The steps will be to

  1. Create a directory structure
  2. Run the dti pre-processing function dtiInit
  3. Explore the data with mrDiffusion to set the ROIs
  4. Run ConTrack to estimate the tracts

[edit] Directories

Suppose the home directory for your data set is <project> and the subject is <subject>.

[edit] Anatomicals

Origiinal anatomical, direct from the scanner, should be in <project>/<subject>/t1. The file is typically called rawt1.nii.gz.

If your subject already has an anatomy, which is typically true for lab members, rather than a file you could create a link to the anatomical in the lab's main anatomical directory. In our lab, anatomicals are stored at /biac2/wandell2/data/anatomy.

Each subject should have a high-resolution 3D anatomical T1-weighted image to be used as a reference anatomy. See Anatomical-Processing for how we collect these data. The anatomy is used to generate a white/gray matter segmentation and surfaces of the brain. It can also be used to establish regions of interest.

[edit] Diffusion

The original (raw) diffusion volumes are usually stored in a file called


or sometimes dwi-40-trilin-rt.nii.gz

The names of the bvec and bvals files are typically

 > <project>/<subject>/raw
 > <project>/<subject>/raw/dti.nii.gz
 > <project>/<subject>/raw/dti.bvecs
 > <project>/<subject>/raw/dti.bvals
 > <project>/<subject>/t1.nii.gz

[edit] Align the T1 to Anterior/Posterior commissure (AC/PC)

mrAnatAverageAcPcNifti: Set Ac, Pc and Mid Sagital points.

When aligning the DTI data to the anatomical t1 it is helpful if the brain you're aligning to is in a "standard" space (i.e., ac-pc). This brain orientation also allows us to more easily align between subjects.

In the directory containing rawt1.nii.gz, use the Matlab function mrAnatAverageAcPcNifti to Align the Base Anatomy to the Anterior and Posterior commissurae.


mrAnatAverageAcPcNifti will ask you to select an Anatomy file and save as. Our naming convention is:

 > <project>/<subject>/t1.nii.gz

After calling the mrAnatAverageAcPcNifti function, the GUI window at right will appear. To align to the standard Ac/Pc space follow these steps:

  1. Place cursor on the anterior commissure > Click "Set Ac" (try using the axial view (bottom left window))
  2. Place cursor on the posterior commissure > Click "Set Pc" (try using the axial view (bottom left window))
  3. Place cursor on the mid-sagital slice (choose a high point on the sagital slice) > Click "Set MidSag". (try using the sagital view (top left window))
  4. Click "Finish"

[edit] Diffusion preprocessing

At this point you have everything you need to run the initialization function dtiInit.m.

To run in Matlab type:

 >> dtiInit

This will open two prompts: (1) Select your raw DWI NIFTI file and (2) Select your T1 anatomy. The script assumes you have a bvecs and bvals file in the folder. It will help you find them.

This function takes a significant amount (> 1 hr, typically) to execute.

The function aligns the DWI data to your T1 anatomical and produces a file that was historically named dt6.mat. We are likely to change this name to dwi<N>.mat before too long.

Once the function has run your directory structure will look as follows:

 dti<#directions>trilin/  fibers/  raw/  t1.nii.gz
 bin/  dt6.mat  t1pdd.png
 b0.nii.gz	  faStd.nii.gz	pddDispersion.nii.gz  vectorRGB.nii.gz	wmProb.nii.gz
 brainMask.nii.gz  mdStd.nii.gz	tensors.nii.gz	      wmMask.nii.gz
 dwi_acpcXform.mat	     dwi.bvals
 dwi_aligned_trilin.bvals   dwi.bvecs
 dwi_aligned_trilin.bvecs   dwi_ecXform.mat
 dwi_aligned_trilin.nii.gz  dwi.nii.gz

Where <N> = the number of diffusion directions you acquired in your scan.

The dti<N>trilin folder contains the resulting dt<N>.mat file, an overlay showing the T1 anatomical and the principal diffusion directions (t1.pdd.png), and a bin/ sub-directory with various computed files needed for visualization or tractography computations.

[edit] Explore the data using mrDiffusion

dtiFiberUI- the mrDiffusion GUI

At this point all the preprocessing of the DTI data is completed. Data are aligned to a standard base Anatomy and the tensors have been estimated. At this point check the data. The mrDiffusion page contains a review of the basic operations.

First, cd to the folder containing the dt6.mat:

 >> cd <project>/<subject>/dt<N>trilin/

Second, in Matlab type:

 >> mrDifffusion

and load the dt6.mat file from the File menu.

[edit] Optic Radiation (OR) ROIs

This will be done using ITKGray and the anatomical image. We might want to do some initial tractography from the chiasm to find the LGN if it's difficult to locate on the T1. Finding the calcarine should not be hard for the user on the t1.

[edit] Steps to trace the OR from V1 to LGN

  1. Load Freesurfer optic radiation roi
  2. Create a 8-mm spherical ROI in the center of the freesurfer OR ROI
  3. Trace using standard tractography
  4. Create two 8-mm spherical ROIs (Left and right LGN) where the fibers created in (2) bend forward in the thalamus
  5. Load the right and left freesurfer Calcarine ROIs
  6. Use contTrack to trace from the two LGN ROIs to the ipsilateral Calcarine ROIs
  7. Score the fibers
  8. Manually delete the excessive fibers

[edit] ConTrack: Generating OR fibers

We will assume that the user has conTrack set-up and ready to run - we can point them to the ConTrack page for help with that.

[edit] Tracking Parameters

Dt6 Roi1 Roi2 NumPaths = 100000 ...

[edit] Pathway Samples

[edit] Scoring Samples

[edit] Fiber Statistics

... add FA, MD, RD, AD to the .pdb structure...

[edit] QUENCH Visualizing and editing pathways

[edit] Diffusion Properties Along the OR

... single subject and groups...

Personal tools