fMRI Research Page

This is a resource page for information and software needed to analyze fMRI data. Beware that some information is obsolete because it describes how we used the clinical MRI at Royal University Hospital to get our data. The MRI we used was retired in March of 2009. However the information here may be of use to others who want to try and use a clinical MRI for research purposes.

SIEMENS IDEA resources


Our computers run Linux - here's a short Linux tutorial.


How to use afni on thalassa, wilder or triceratops remotely.

If you are using a windows machine you will need to download two programs and install them on your computer: Tightvnc and PuTTY. Then

  1. Use PuTTY to ssh to the into the machine of your choice (thalassa, wilder or triceratops) -- the first time you connect it will ask you a question - answer yes.
  2. Start the vncserver - making note of your X desktop number (i.e. thalassa.usask.ca:1) by typing
    # vncserver
  3. On your local machine start the tightvnc client and connect to the X desktop, the password should be the same as your logon.
  4. You should now have a window displaying the linux X desktop in which you can run afni.
  5. When finished close the tightvnc window and kill the vncserver in your putty client by typing:
    # vncserver -kill :1
    (where :1 is whatever desktop you first started with.)
If you are using a linux machine follow these steps:
  1. ssh to the remote machine you wish to work on
  2. Start the vncserver - making note of your X desktop number (i.e. thalassa.usask.ca:1)
    # vncserver
  3. On your local machine start the vncviewer and connect to the remote machine:
    # vncviewer thalassa.usask.ca:1
    (where :1 is whatever desktop you first started with.)
  4. You should now have a window displaying the remote linux X desktop in which you can run afni.
  5. When finished close the vncviewer window and kill the vncserver on the remote computer in your ssh session:
    # vncserver -kill :1
    (where :1 is whatever desktop you first started with.)
Old vnc instructions, for reference (please follow more secure procedures outlined above for remote access now):
  1. telnet into the machine of your choice (Jennifer Hadley will have to set things up so that your computer has access through the firewall. Also you will have to have vncviewer running on your computer. You may download an executable copy of the vncviewer for free from http://www.realvnc.com.)
  2. Start vncserver by typing vncserver at the command prompt. Make note of your port name e.g. thalassa.usask.ca:1 (especially the last number).
  3. Use the vncviewer on your local computer to connect to the port (e.g. thalassa.usask.ca:1). A window will pop up from where afni, or any linux program, can be run.
  4. When you are finished, logout of the vnc session, kill the vncviewer and, very important to prevent confusion with other users, kill the vncserver in your telnet session by typing vncserver -kill :1 (if your port number was 1).

Analysis Methods

To get started, click on the first link below, on fMRI data aquisition, to find out how to access your fMRI data. Next, read the step-by-step guide to get an idea of what's required to complete a typical analysis. The step-by-step guide outlines an analysis approach that uses the locally written BOLDfold software in combination with AFNI. But there are many ways to analyze your fMRI data. The documentation on the various fMRI software packages will guide you should you need to or want to depart from the BOLDfold/AFNI approach.

  1. fMRI data acquisition and storage.
  2. Step-by-step guide for a typical fMRI project analysis using BOLDfold and AFNI.
  3. BOLDfold versus deconvolution fMRI experimental designs. Written by Greg Kraushaar, summer 2002. The AFNI program RSFgen is useful for the design of randomized experiments. The analysis of randomized experiments requires the use of AFNI's 3dDeconvolve. An example analysis is given here (note: the missing links may be filled in someday!).
  4. More up-to-date AFNI deconvolve analysis (AKA linear systems IRF parameter model).
  5. Analyzing fMRI Data using Logical AND, Logical OR and Exclusive OR.
  6. Jeff Bird's unedited notes from the summer of 2006 relating to his work with Vasily Vakorin's FDA BOLDfold software to extract parameters characterizing the BOLD response.
  7. Software documentation:
In the summer of 2001, summer students Ben Norris and Greg Kraushaar and compared several fMRI analysis packages. Their comparison work may help you pick an analysis method.

Here is a useful online calculator for p-values, etc.


Resources
  1. Here is a pdf version of my MRI course textbook (offered as BioE 898.3). It is an evolving document.
  2. A compressed tarball of my BOLDfold software and miscelleanous MRI image analysis software.