For questions, comments, etc. please contact Aaron Olson (olsonat2@facstaff.wisc.edu; 608 265 4765) or Hillary Schaefer (hsschaefer@students.wisc.edu).
AFNI is a FREE set of C programs for processing, analyzing, and displaying functional MRI (FMRI) data - a technique for mapping human brain activity. The best source for AFNI information is:
http://varda.biophysics.mcw.edu/~cox/index.html#AFNIstuff
It is maintained by AFNI's creator Robert W. Cox, Ph.D., and has many resources including:
If you're working on a computer at the University of Wisconsin Madison's Lab for Affective Nuroscience (LAN) that has AFNI already installed you can try using the directions below (from workshop packet):
I highly recommend reading and working through Dr. Cox's 40-page AFNI manual if you are a beginner (maybe even if you're not). You may be able to follow the instructions below and be successful with out doing so but...
NOTE: Your system must have AFNI properly installed and "set-up" (i.e. paths etc. in .cshrc file, etc.) I'm also making the assumption you are working on a computer on "our" (i.e. LAN's) network.
Login. At the prompt, change directory to /exp/training/AFNI_Oct_5_2000 by typing "cd /exp/training/AFNI_Oct_5_2000" (without quotes and followed by pressing the "enter" or "return" key -- this applies to all typing instructions from here on). Next execute AFNI by typing "afni". It will take a few seconds to start. During this time, text will be "dumped" out to the screen, and a Graphical User Interface (GUI) I'll call the "control-window" will "pop-up" (i.e. a gray box with yellow text and "button" labels) E.G.:
START OF SCREEN DUMP:
lan150% cd /exp/training/AFNI_Oct_5_2000
/exp/training/AFNI_Oct_5_2000
lan150% afni
MCW AFNI: Analysis of Functional NeuroImages, by R.W. Cox (rwcox@mcw.edu)
version 2.25g: Copyright Medical College of Wisconsin: 10 February 2000
Development supported by MCW funds and by NIH grants MH51358 & NS34798.
Clinical uses are not recommended, and have not been evaluated by the FDA.
Thanks go to E.A. DeYoe for useful feedback.
Initializing: X11.. Widgets...... Input files:
session # 1 = /da/exp/training/AFNI_Oct_5_2000/ 10 anatomical datasets, 0 functional datasets
dataset count = 10
Time series = 1 files read
NLfit & NLerr= Found 16 models
Plugins = 33 libraries read
-orient = RAI
END OF SCREEN DUMP!
Note that the GUI is NOT shown here. Also typing "afni -h" or "afni -help" may help you if you're having problems or want to know other features etc. These additions (called "options" or "flags") to the "afni" command prevent AFNI from starting. Instead a bunch of "help" text is dumped to the screen.
Ok, finally play time. Go to the GUI with your mouse! To better understand what a button does you can "click" (i.e. press and release left mouse button) on the "BHelp" button then click on the button you want help with. You may also be able to just place the "pointer" over a button and "hit" the "F1" key. To make the help box "go away" click inside it.
Now click (without using the Bhelp feature) on the "Image" button to the right of the word "Axial" in the left-center portion of the GUI. Another "image-window" should pop-up with an axial brain image (probably will look "blurry") inside it. You should also see text and buttons along both the right and bottom sides of the box. (The window may have to be resized -- click and "hold" (down mouse button) on corner of box and "drag" until you're sure you can see everything) (It may also need to be moved. To do this, click and hold on the "title-bar" (i.e. where it says "[A] AFNI ...") near the top of the window and drag the window to its desired location. Play with buttons and click on the green "bar" along bottom -- note effects. When you've had enough or want to start over, inside the image-window click the "Done" button. This will close only the image window, but AFNI will remain running.
Now click on the "Switch Anatomy" button. A list of items should appear. Click on "17_neu..." so it becomes highlighted and then click on the "Set" button below the list. Now click on the Axial "Image" button again. You may notice that the image is not the same as before. This is because your "Switch Anatomy" selection caused the 3D dataset that you are viewing to change. Now click on the "Graph" button to the right of the highlighted Axial "Image" button. This should cause a "graph-window" to pop-up with 9 boxes in it. You may have to move all your windows around a little, so you can see them. Next click somewhere on the brain in your image-window. You should see the green cross-hairs (Xhairs) and box move to where you clicked. You should also note that the graphs in the graph-window changed. That's because those 9 boxes are 9 different graphs that correspond to the 9 voxels inside the Xhairs and green box in your image-window. The Y-axis of all graphs represent voxel intensity (i.e. brightness) and the X-axis represents time. This data set is the average of the 8, 17 second imaging cycles from the experiment. Each graph has 17 points (one for each second along the x-axis).
Next we want to see if any brain voxels show evidence of "activation" due to the stimulus. One method to do this involves choosing an "ideal" waveform of what we speculate the activation should look like and computing the cross-correlation coefficient between it and all the voxels in the dataset. To do this, click on the "FIM" button in the lower right hand corner of the graph-window. This should bring up a list of options. Click on "Pick Ideal." Another window appears that should have a "...ideal.1D..." choice. Click to highlight it and then click the "Set" button. You should see a red "hill-like" waveform appear in the center graph of the graph-window - this is the "ideal hemodynamic response" for the given stimulus. Now click on the "FIM" button again and choose the "compute FIM" option. Another window will pop-up. Choose the "fico" option. After a moment you will probably see areas of the brain "light up." These are areas that "correlated" to the ideal response in some way.
Now click on "Switch Anatomy" again and choose the "T1high..." dataset. A much "nicer" brain image will replace the blurry one and it will be "overlaid" with the functional calculation you just made. Next, click on the "Define Function" button in the control-window. Another window will become attached to the right of the control-window. By "sliding" the gray piece on the green bar using your mouse you can control the threshold controlling the functional overlay (i.e. it will change the number of voxels that appear active). The decimal number below the green bar represents the confidence interval for the correlation coefficient statistic (?? I think ??).
Ok, that's all for me. Play around all you want though. When you want to be completely finished, or start completely over, click the "done" button in the control-window twice!
Good Luck!! Please feel free to contact me with any questions, constructive criticism, comments, etc.
Aaron Olson Email: atolson2@facstaff.wisc.edu phone: 608 265 4765