| 
  • If you are citizen of an European Union member nation, you may not use this service unless you are at least 16 years old.

  • You already know Dokkio is an AI-powered assistant to organize & manage your digital files & messages. Very soon, Dokkio will support Outlook as well as One Drive. Check it out today!

View
 

Analysis 1: Simple t-test

Page history last edited by PBworks 17 years ago

Analysis 1: Simple t-test

(1) To do this analysis, you'll need to separate your scans into Fix, RealWord, and FalseFont groups. An easy way to do this is to use a Paradigm File. Download the Paradigm File for the IR1 task here, and save it into your tutorial directory.

 

If you're analyzing data for which a Paradigm File has already been defined, you can just re-use the same Paradigm File. But when you're setting up a new fMRI experiment with a new paradigm, you'll need to define a new Paradigm File. See Creating a Paradigm File for further information on how to do this.

 

(2) Open up a new MEDx session, and then open up a New Folder.

 

(3) Load the Mask image that you created in step 12 of Spatial Normalization (the one named SnMask.hdr.

 

(4) Also load the 84 spatially normalized, spatially smoothed, temporally filtered, motion-corrected image volumes which you created in the Spatial Normalization step. You should now be looking at a Group page showing the 84 spatially normalized scans. The name of the Group page should be New Group.

 

(5) Select Toolbox --> Functional --> Define Paradigm.

 

(6) In the Paradigm Editor window, click on the Apply Paradigm button.

 

(7) In the Apply Paradigm dialog box, in the Filter entry box, type in the tutorial directory, followed by *.pdgm, e.g.

 

/export/home/myHomeDir/tutorial/*.pdgm

 

and then click on the Filter button. This will cause the IR1 Paradigm File to be listed in the Files: pane. Select this file, then click on the Ok button. This will cause the IR1 paradigm information to be applied to the 84 scans.

 

(8) Select Page --> Page Manager..., and in the Page Manager window click on the Groups radiobutton. This will cause only Group pages to be listed. (This is a useful trick to know.) Note that in addition to the original New Group group page, you now have four new groups: Fix, RealWord, FalseFont, and All. Select Fix, and then click on the GoTo button. Note that this takes you to a new page in the MEDx folder named Fix, and that it lists the subset of 28 images that correspond to the Fix task.

 

(9) OK, let's do the t-test. Select Toolbox --> Functional --> Group Statistics....

 

(10) In the Group Statistics dialog box, click on the Between Groups tab.

 

(11) Set Test Group to RealWord, and Control Group to FalseFont. Use the Select... buttons if you wish, or just type into the entry boxes.

 

(12) Make sure that the Parametric radiobutton is selected, and that Operation is set to Unpaired t-test. Then click on the OK button. This performs the t-test. A new Group page will be created, containing the output of the t-test; I will touch upon the most important three. The first of these eight images is the t-test map itself. The third is the Mean Diff image, which is the average of the RealWord scans minus the average of the FalseFont scans (the numerator of the t-test). And the sixth is the Z map. Click on the button for the Z map to view it.

 

(13) Note that there's a lot of junk outside of the brain. Let's mask this Z map. Click on the Image Calculator button (has an icon that looks like a calculator on it).

 

(14) In the Image Calculator, click on the Select... button and select Current Image. Note that CurrentImage appears in the white box. (If you mess up, just click on the AC (all clear) button and start over.) Then click on the X button, for the multiplication operation. Then click on the Select... button again, but this time select Page Manager.

 

(15) In the Page Manager, select the Mask that you loaded in step 3. The formula in the white box should now look like this:

 

 

CurrentImage * "PAGE:SnMask.img"

 

 

(16) Set Output Image to Current Image. Make sure that you are currently looking at the Z map (i.e., that it is indeed the CurrentImage).

 

(17) In the Image Calculator, click on the OK button. This causes the formula in the white box to be evaluated. You'll note that most of the voxels outside the brain will be set to zero.

 

(18) Let's look at the range of values in this Z map. Select Display --> Display Range. Note the Min and Max values displayed for Group/Volume. This shows the minimum and maximum values. Sometimes this information isn't updated; to make sure that it's updated, click on the Compute button. In general, the minimum and maximum in Z maps should range between about -5 or -8 and +5 or +8 (ballpark figures). They shouldn't both be zero, and in general they shouldn't go into the thousands. (Caveat: for historical reasons, the Holy Script generates Z maps that are scaled up by a factor of a thousand, because it is stored in 16-bit integer format rather than in floating point format. Whatever!)

 

(19) Select Image --> Save Image As... and save the newly masked Z map into the tutorial directory in AVW format as a file named Z.simple.RealWord-FalseFont.hdr. This filename indicates the type of map ("Z"), that it's from the simple t-test (simple), and that it's from the contrast RealWord Minus FalseFont (RealWord-FalseFont). Try to embed pertinent information into your directory and filenames.

 

(20) Then exit MEDx.


 

If you're up for it, you can also try Analysis 2: Multiple Regression. Or, just proceed to Cluster Detection.

 

Consider at least printing out and going through this stand-alone MEDx tutorial. This particular tutorial covers many useful MEDx features, not only the simple t-test, and is highly recommended. The only caveat is that it uses an obsolete (I wouldn't even say that it was ever really kosher) statistical method implemented in MEDx, called Critical Threshold, found under Toolbox --> Functional --> Final Significance --> Critical Threshold. Don't use this method. Use Cluster Detection (also under Final Significance) or False Discovery Rate (not implemented in MEDx, but implemented outside of MEDx in my own C code, and also implemented in SPM).

 

For some information on creating a paradigm file, see Creating a Paradigm File.

 

Return to tutorial main page

Comments (0)

You don't have permission to comment on this page.