User Tools

Site Tools


biac:courses:sample_fsl_2ndlevel

FEAT Level 2 and MRIcron

Higher level FEAT (e.g., level 2) is used to combine one subject's data across runs. The 1st level analyses you run are conducted in native space; however, at 2nd level, everything is done in standard space.

The data you will use are 5 runs of a blocked-design motor squeezing task. They are available at /afs/acpub/project/neurobio381/datasets/Class.01/Examples/FSL_sample_NEW. We've already done the first-level analyses for you. You are going to do the second-level analyses.

To copy the 1st level feat directories into your folder, you can use the following command (but remember to change the destination directory). First, you'll need to open the Terminal (it's in the “Utilities” folder and is distinct from the “Xterm” window that comes up when you open MATLAB).

cp -r /afs/acpub/project/neurobio381/datasets/Class.01/Examples/FSL_sample_NEW/*.feat /afs/acpub/project/neurobio381/classwork/Fall_2008/Students/David/FSL/. 


  • Take a minute or two to look through the output of the files you copied over. You can do this by opening the report.html file within each *.feat directory
  • The first run is exactly what you've analyzed before, except for all of these, prestats and stats were run together (i.e., not separately like you did for the 1st level lab).
  • Within the Poststats part of the outputs, you'll notice that the other runs look very similar with respect to the activation pattern.
  • Looking at the Prestats part of the outputs, you'll see that subject didn't really move a whole a lot (anything over 1mm is bad!)
  • In the Registration part of the outputs, you'll see that all of the registrations look OK. If these were messed up, your 2nd level analysis would crash.

Higher level FEAT

Now we're ready to start FSL and proceed to the analysis.

  • To make your life easier, you'll first want to navigate to your FSL folder where you copied all of the *feat directories.
  • Once you're in your folder, type Feat_gui & into the command terminal. This will bring up the FEAT GUI without going through the main FSL GUI.
  • Before you do anything else, you need to change the pull-down menu in the upper left of the FEAT GUI from “First-Level Analysis” to “Higher-Level Analysis”
  • The only tabs that appear “active” now should be the “Misc”, “Data”, “Stats”, and “Post-Stats” tabs.

Misc tab

  1. The “Progress Watcher” and “Balloon Help” should be checked by default.
  2. There's also an option to “Clean up first-level standard space images”. This is off by default, but you should turn it on to save diskspace.

Data Tab

  1. The uppermost button within the Data tab should read Inputs are Lower Level FEAT directories
  2. The number of analyses for should be 5, for all of the odd numbered runs (1,3,5,7,9}.
  3. Click Select FEAT directories, you should already be in your FSL directory and click into the run# folders. Once you click into all the run# directories in the Select input data GUI you are done and click ok. Do this for all the runs that you are combining, 1,3,5,7,9.
  4. Once you have loaded all of the FEAT directories that you need click ok
  5. Under select output directory type /afs/acpub/project/neurobio381/classwork/Fall_2008/Students/YOURFOLDERNAME/FSL/level2 this should create a folder in your FSL folder titled level2.gfeat when we are all done.
  6. Use lower level copes should appear and 1 should be selected (yellow). If we had more contrasts at the first level, this number would be larger.

Stats Tab

  1. In the Stats tab, Mixed effects: FLAME1 should be changed to FIXED because this is a single subject and we're not generalizing to the population.
  2. Click Full Model setup and you should have a column named EV1 and it should have all 1's in the column
  3. Click Done

Post-Stats Tab

  1. You can leave all the defaults checked. You don't have to change anything in this tab.
  2. We will play around with various threshold and displaying the results later in MRIcron

Once everything is set up, click Go and your 2nd level analysis should be begin. It will take about 10 minutes to complete.

MRIcron

This program allows you to create cool images that can be used in figures in your paper/presentation. It's really easy to use, and we'll walk through an application of it using the analysis from above. If you want to play with it at home, you can download it from here.

  1. This program can be found in the “Applications” folder. Double-click on it to open it.
  2. Once it opens, the default background image is the Colin brain, but this version is really high resolution and great for 3D rendering (the file name is ch2better.nii.gz).
    • The Colin brain is based off of a single subject named Colin who was scanned multiple times to achieve a really nice structural scan.
    • In some applications (e.g., subcortical activation), it is probably better to show results on an average anatomical from your subject population.
    • For the purposes of this class, you can show activation on the Colin brain or your subject's anatomical scan. It's only a matter of selecting the background image. The Colin brain will look a lot nicer.
  3. Now we need to overlay an activation map on the background image.
    • If you wanted to do this on your subject's anatomical, you need to click File, and then Open. From here, you'll need to track down your subject's anatomical that has been normalized into standard space. This will be in the level2.gfeat directory and the name of the file will be bg_image.nii.gz. EX: /afs/acpub/project/neurobio381/classwork/Fall_2008/Students/David/FSL/level2.gfeat/bg_image.nii.gz
  4. To overlay a statistical map onto your background image, select Overlay, then Add, and then find the thresh_zstat1.nii.gz image inside the cope1.feat directory within your level2.gfeat directory. EX: /afs/acpub/project/neurobio381/classwork/Fall_2008/Students/David/FSL/level2.gfeat/cope1.feat/thresh_zstat1.nii.gz
  5. The thresh_zstat1.nii.gz image is thresholded based on the parameters you set in post-stats. Sometimes you won't have any activation that exceeds your threshold but you'll still want to see what the unthresholded map looks like. You can find this unthresholded map inside the stats directory under the cope1.feat directory (e.g., level2.gfeat/cope1.feat/stats). The file you want is called zstat1.nii.gz
  6. Now you have an activation map overlayed onto your background image. Take a moment to navigate around the activation.
  7. You can alter the thresholds by changing the min and the max settings in the middle of the window. Try setting the one the left to 5 and the one the right to 10. This will set the lower bound for the z-scores it will display to 5, so it will obscure a lot of the activation that was barely above the 2.3 threshold we set in FEAT.
  8. Now let's take a look at it in 3D. To do this, click on Window and then Render. You can play around with the cutout and the orientation of the brain.

For more information about MRIcron and how to use it, please consult the online guide and tutorials http://www.sph.sc.edu/comd/rorden/mricron/main.html

biac/courses/sample_fsl_2ndlevel.txt · Last modified: 2014/08/04 16:03 (external edit)