MEDx
For historical reasons, the CSL uses MEDx for most of its pre-processing.
Official Collection of MEDx Tutorials
Tutorial: First-Level Within-Subject fMRI Processing and Analysis, using MEDx interactively
NCSI 525 Data Analysis Lab Handout (may be useful for broad overview of fMRI pre-processing and first-level analysis)
MEDx regression with motion parameters as effects of no interest
Adding a New fMRI Pre-Processing Protocol
Analysis of Beijing Data for HK R01 -- September 2006
Create Tables of Cluster Overlap With Brodmann Area and Anatomic Area
Threshold a 16-bit Image To Make A Mask
If you have MEDx trouble, just send email to MEDx customer support. Before doing so, however, check with Guinevere or Vlad to make sure that your MEDx subscription is still active. The MEDx subscription covers not only periodic upgrades to MEDx, but also (probably more importantly) customer support. If you paid for customer support, use customer support!!
or
VolView
Kitware's VolView program can be used to make nice 3-D volume renderings of brain activations, with the brain rendered either opaque or transparent. It can also make "spinning brain" movies!
How to Register VolView on app1
Other Software Packages
Meta-Analyses
ALE Meta-Analysis using BrainMap's Search&View Program
ALE Meta-Analysis using CSL software
Making a Fresh Installation of CSL ALE Software
Miscellanea
Obtain data via DICOM transfer
Creating Overlays Using False Discovery Rate
Scrambling an Image Using Fourier Method
Web-Based Method for Distributing Files
Learn Some Computer Programming!!
As Robert Savoy has said, "Functional MRI is difficult." One of the things that makes it difficult is the complexity of the data. Because of this, it would help if at least one person in the lab knows a little computer programming. Well, why can't that person be you? Consider taking the new course entitled Introduction to Computational Software, offered by the biostatistics department during the Fall semester. Maybe you can at least audit it.
Georgetown UIS offers computer training to university employees, free of charge.
Online Titles (consider especially the courses Using the Shell, Using the Shell in Linux, and Shell Programming)
Self-paced Computer-Based Training (courses on CD-ROM)
Scripting
You might want to learn how to script, to avoid doing repetitive tasks. For example, rather than do the same exact processing for each of 40 patients, it might be better to script it all. This will help reduce the tedium, as well as reduce user errors.
For MEDx, use ImageScript. ImageScript is Tcl/Tk with extra MEDx-specific commands added in, so in order to script in ImageScript well, you'll really need to learn Tcl/Tk (see next paragraph). Here is a tutorial introduction to ImageScript. Here is a tutorial on advanced ImageScript scripting in MEDx. See also Chapter 29 of the MEDx User's Guide.
For AFNI or FSL, script using some Unix-based shell. C-shell should be fine (tutorial 1, tutorial 2), although some people argue against it (con 1, con 2). Other options might be the bash shell or the korn shell, or Perl, Python, or Tcl/Tk (tutorial, online documentation). An intriguing possibility which I unfortunately never got around to fully exploring is Fiswidgets (paper 1, paper 2, paper 3); perhaps if you have the time you could try out this automation system.
For SPM, you'll need to learn something about batch jobs in SPM (see this page too), which I don't know much about! From what I've heard, batch jobs are easier to set up in SPM5 than they were in earlier versions of SPM.
Note that automating the analysis can be a double-edged sword, whatever system you choose to go with. On the one hand, automating the analysis can reduce tedium and user errors, and increase the speed in which data gets analyzed. On the other hand, the analysis can then become a black box for users who didn't write the script, and when something goes wrong they might not know what to do! This is related to the old software design problem of choosing between user-friendliness and power -- a rough example of this tradeoff is SPSS (user-friendliness; though the SPSS people might protest that their software is powerful as well) vs. SAS (power; though the SAS people might protest that their software is user-friendly as well). I don't know of an easy solution. It seems too draconian to require each user to choose between doing the analysis manually or writing their own scripts.