A white matter fiber connects the output of one neuron to one or more target neurons. There are well known bundles or tracts of fibers that run through the brain and spinal cord and connect distant brain regions. Here are a few examples of major tracts:

• Corticospinal tract. A good review is linked here.
• Corpus callosum. Its divisions are discussed here and here.
• Superior longitudinal fasciculus.It is discussed here here.
• Inferior longitudinal fasciculus.

A review of tracts using DTI can be found here.

In this lab we will investigate relatively new methods in MRI to visualize the integrity and direction of white matter tracts.

An important part of the lab will be knowing your way around the white-matter and being able to identity some white-matter tracts, particularly when trying to identity locations of crossing fibers. Let's learn about some tracts now.

1. Start up FSLView and open the MNI152_T1_1mm_brain.nii.gz standard brain.

2. Click on Tools ⇒ Toolbars ⇒ Atlas tools

3. Toggle your view layout to have bigger slices by clicking the following button below.

You want your view to look this below.

4. Click on the Structures button in the Atlas information at the bottom.

5. Then select the JHU White-Matter Tractography Atlas and select the check boxes for Locate selected structure and Preview selected structure's probability map.

Move this window to the side so you can see all three slices in FSLView while being able to change the selected tract in this atlas.

6. Identify each of the following structures one by one. Note we are only going through the left hemisphere tracts here to conserve time. For the following four, make sure to take a screenshot.

• Corticospinal tract L: nerves within the corticospinal tract are involved in movement of muscles of the body.
• Cingulum (cingulate gyrus) L: white matter fibers that project from the cingulate gyrus to the entorhinal cortex.
• Forceps minor: connects the lateral and medial surfaces of the frontal lobes and crosses the midline via the genu of the corpus callosum.
• Inferior fronto-occipital fasciculus L: passes backward from the frontal lobe to the occipital and temporal lobes. Also note the proximity of this tract to the forceps minor and to the corticospinal tract (this might be important later when discussing crossing fibers)!

Feel free to explore some of the other fiber tracts on your own. Do you notice any other tracts that might either cross or come very close to each other?

7. To get an idea of this possibility (of adjacent or crossing fibers), load several of the white matter tracts you just went through at the same time. You can do so by running the following command below in the terminal. (remember, you can just copy and paste it)

cd ~/Desktop/class/input/dti/atlas
 
fslview ${FSLDIR}/data/standard/MNI152_T1_2mm_brain.nii.gz \
  cingulum_L.nii.gz -l 'Green' -t 0.7 -b 0,50 \
  corticospinal_tract_L.nii.gz -l 'Blue' -t 0.7 -b 0,50 \
  forceps_minor.nii.gz -l 'Red' -t 0.5 -b 0,50 \
  inferior_fronto-occipital_fasciculus_L.nii.gz -l 'Green' -t 0.8 -b 0,50 

We have just opened four tracts primarily in the left hemisphere:

• Corticospinal tract in blue indicating it mainly goes between superior and inferior
• Cingulum in green indicating it mainly goes between anterior and posterior
• Forceps minor in red indicating it mainly goes between lateral and medial
• Inferior fronto-occipital fasciculus in green like the cingulum

Feel free to adjust the transparency of each of the tracts via the slider at the bottom or to hide/show each of the tracts by double-clicking the eye icon next to the name of the tract. Note that the values for each tract represent the probability that tract was found for a subject in that dataset. Here, we are displaying tracts where a value was found for any of the subjects.

In this section we will compute FA and DT maps using the FSL Diffusion Toolkit program; FDT diffusion.

Open up a new terminal window and enter in the following:

  mkdir -p ~/Desktop/class/output/lab05
  cp ~/Desktop/class/input/dti/33470_complete/33470_LAS_eddy.nii.gz ~/Desktop/class/output/lab05
  cp ~/Desktop/class/input/dti/33470_complete/bvecs ~/Desktop/class/output/lab05
  cp ~/Desktop/class/input/dti/33470_complete/bvals ~/Desktop/class/output/lab05
  cd ~/Desktop/class/output/lab05

In the same terminal window you used above:

• Type fsl & in the terminal.
• Because you launched fsl within a dti subject folder, this will simplify navigation when selecting the input files.

As this exercise will focus upon white matter tracts, note that FSLView has special options for displaying DTI data and special atlases for exploring fiber tracks. We will go through these in further detail below.

In this section, we will continue to use the JHU White-Matter Tractography Atlas in FSLView as a reference to guide our exploration of the diffusion data. We will also look at another atlas: the JHU ICBM DTI-81 White Matter Labels. To load the atlases:

1. Start FSLView
2. Open a standard brain (File ⇒ Open Standard), I suggest the MNI152_T1_2mm_brain.nii.gz.
3. Reveal the atlas tools: Tools ⇒ Toolbars ⇒ Atlas Tools
4. Once the atlas toolbar appears click the Atlases button to select the appropriate atlas.
5. Select the JHU White-Matter Tractography Atlas and JHU ICBM DTI-81 White Matter Labels
6. (optional) In the same window, de-select the two Harvard-Oxford atlases.
7. Click Ok to exit this window.

Now in the atlas toolbar, you should see your two new white matter atlases. If you move your cursor around on the brain, information in this toolbar will update with the probability that the selected voxel is a given white matter label or tract.

To also visualize the probability distributions for a specific tract (as in the previous walkthrough of white matter tracts):

1. Select the Structures button in the atlas toolbar
2. Select the JHU White-Matter Tractography Atlas in the dropdown
3. Select the check boxes for Locate selected structure and Preview selected structure's probability map
4. Now you can select a particular tract in the list and it will display on your standard brain. Feel free to move this structure list to the side so you can view the whole brain.

Applying strong gradients such as those used in diffusion imaging creates compensatory magnetic fields that oppose the gradients. These are called 'eddy currents' in analogy to the swirls that you can observe in moving water when it encounters an obstruction. These swirls oppose the direction of the stream.

The problem in MRI is that spatial encoding is exquisitely dependent upon the strength of the magnetic field gradient. The spatial location of an hydrogen nucleus (proton) is encoded by applying a magnetic field of a precise strength to a location in space. If the magnetic field strength is augmented or diminished by eddy currents, there will be a change in the perceived location of the proton. Thus, the image will be sheared or otherwise distorted.

There are algorithms that attempt to compensate for eddy current distortions, which are spatially related to the gradient direction. We will first need to remove eddy current distortion from our diffusion data before proceeding. This can be done by choosing FDT diffusion on the FSL menu, and then choosing eddy current correction on the drop down menu. However, due to the time taken for this computation, this has been computed beforehand so you do not need to do it now.

You will need a brain mask for this exercise. A brain mask is a binary image that contains ones wherever there is brain and zeros everywhere else (e.g., skull, scalp, outside of brain). Such a mask can then be used to remove from computation any region that is not brain. Brain masks can be created with FSL's Brain Extraction Tool (BET). Be sure to specify your eddy corrected images as input.

The 33470_LAS_eddy.nii.gz is a T2-weighted image (with no diffusion gradients applied). T2 images provide some challenges to skull stripping, because the skull and scalp are not well defined in T2 contrast.

Also, because we need a brain mask, you need to check the option (under Advanced Options) to output a binary brain mask image, and a brain-extracted image. Skull stripping should only take a few seconds.

Before proceeding, make sure you have a good brain mask and a good skull stripped brain. Quality is important. Make a note of their names, as this will need to be provided in the next step.

Choose the FDT diffusion program from the FSL menu, and the DTIFIT reconstruction from the FDT menu.

Now (note the numbered list below corresponds to the numbers in the image below):

1. Check the Specify input files manually box. This program will run automatically if certain naming conventions are used. However, as we named our image volumes unconventionally, we need to check that box.
2. Be sure to specify the eddy-current corrected diffusion data 33470_LAS_edd.nii.gz (not the skull-stripped data you just computed) as input.
3. Specify your brain mask from the skull stripping step
4. FSL will automatically set the output file
5. The gradient directions are provided in the bvecs file
6. The b vals are the bvals files

Press Go - this runs within a couple of minutes - so be patient and don't hit Go repeatedly.

You will need FSLView for this step. You will need to open a new FSLView window because these data are in a different space than the standard brain used to visualize the white-matter atlas tracts.

1. Examine your FA (fractional anisotropy) image.

• Click File ⇒ Open
• Open the dti_FA.nii.gz file

2. Examine the DTI images for the 3 orthogonal axes that describe the tensor at each voxel.

• Click File ⇒ Add
• Add the dti_V1.nii.gz file followed by the dti_V2.nii.gz and dti_V3.nii.gz files.

You can appreciate this by plotting the color coded principal diffusion axes.

1. First, let's hide dti_V2 and dti_V3 by unchecking the checkbox next to eye icon next to their names. (you can also double-click on the filename to achieve the same thing)
2. Highlight dti_V1 and click on the on the information button (i character on the blue circle) at the bottom of the overlay settings in the FSLView window.

This brings up a box of options. Choose DTI display options and then select Lines (RGB).

In this section we will use TrackVis program to visualize white matter tracts.

1. Open TrackVis by clicking on the icon in your dock.

2. Select File ⇒ Open Track or Scene … and open the tensor_streamlines.trk file in the ~/Desktop/class/input/dti/dipy/dipy directory.

3. You will also want to have an anatomical image to help guide your track selection. Select File ⇒ Load Image and open tensor_fa.nii.gz.

TrackVis has tons of options. We will explicitly cover a few below. For a fuller description of the options you can refer to the TrackVis website here http://www.trackvis.org/. The website offers some brief tutorial movies that can be viewed in your browser that illustrate the use of the program. These are very helpful in quickly learning the interface. You can find a list of these movies here.

The main value of TrackVis is to isolate particular fiber tracks by use of TrackGroups, Slices, Regions of Interests, Balls/Spheres, etc. You can toggle different TrackGroups on and off, for example

• if you right click on Track 1 in the upper-right Objects panel, you can hide this track view, or delete it.
  • NOTE: On a Mac you can “right click” by holding down the control key when while you click
• In the lower-right Property pane, you can double-click on most properties of the selected TrackGroup and change them. For example, you can change the way the current slice looks, which axis it is drawn along, how dense the fibers are drawn, etc.

Play with the controls to get a sense of how you can navigate through space, rotate the brain, adjust which fiber tracts are displayed, etc.

Let's create a ball-shaped region of interest (ROI) to select some tracts. This will help us identify only the tracks that run through areas we're interested in (our ROI)

1. Select TrackGroup ⇒ New Track Group from Sphere

A small ball will show up at the cross-hairs of your 3 orthogonal views. You can make this ball bigger or small by manipulating its properties.

2. Let's make the ball a little bigger.

• Click on the ROI tab in the lower-right Property window.
• Double-click on the 2 next to Radius and set this value to 3

You can drag the ball around within your three orthogonal slice windows along the bottom of the screen.

3. If your three orthogonal slices are not moving as you move your sphere around, click on the Sync Slice to ROI center button. (this button is in the Image window. It is the button that looks like a window-pane with a ball in the middle of it)

4 You can add a second ROI sphere by simply repeating steps 1-3.

Find Some Tracts

Use these methods, or others you discover, to isolate the following tracts.

• minor forceps
• major forceps
• corticospinal tract
• cingulum