Getting Started with fMRI

back to the main page


glossary of technical terms (unix and imaging) and links to other relevant technical information


search the CNL site


printable forms of use to CNL researchers


image analysis software info and links


Instructions and Maintenance information for CNL machines


an introduction to using unix


Interested in fMRI but not Associated with an fMRI Lab?

Using the CNL Website

Overview of the fMRI experiment

Safety Training

Interested in fMRI but not Associated with an fMRI Lab?

If you are a graduate student or faculty member and you think you'd be interested in running your own fMRI experiment, you need to be willing to make a huge time and energy commitment. Here are some reasonable steps to follow if you think you are interested.

1) Review the work of the following principal investigators:

Web pages:

Lee Ryan, Pelagie Beeson, Elena Plante, and Ted Trouard


Lee Ryan, Pelagie Beeson, Elena Plante, and Ted Trouard.

From the information you glean there, decide who you want to approach about a project. Presumably, your proposed project will be closely related to projects of the researcher you choose to talk to. That investigator can help you decide how best to proceed.

If you are a graduate student, and just want to learn what is involved running an fMRI project consider approaching one of the four researchers above about volunteering to help in their lab working on an MRI or fMRI project.

Using the CNL Website

fMRI (functional magnetic resonance imaging) experiments have to be designed and tested behaviorally before you put someone in the magnet (i.e., MRI machine). After you put someone in the magnet, you have a bunch of images to move around, reconstruct and analyze. You'll need people to help you get started with this process, but the CNL website can be a reference when you want to learn more or you are stuck on terminology or "how to" issues.

Glossary: The most important page on the CNL website is the glossary. You'll find an alphabetized list of definitions, with links to other pages and even some standard references for software (since you'll need to reference the software when you write papers). If you want to look up a term, try the glossary first. If you don't find a term that you are looking for, please email me (

Search: If the glossary fails you or you can't figure out how we referenced something, try the search.

Links If you want background information, MRI and fMRI theory, image processing theory, neuroanatomy etc., you should consult the Links page.

Imaging Analysis Archives: If you think a topic has been discussed on the Imaging Analysis listserv, then try the Imaging Analysis archive search. This will sometimes address technical issues, but usually addresses meetings, machine status etc.

Overview of the fMRI experiment

Experimental Design

If you have a background in designing good behavioral experiments, then you are well on your way to designing good fMRI experiments. fMRI experiments are very expensive to run, so avoid using them as fishing expeditions. If you run an experiment first as a behavioral experiment, and you get clear results AND there is a good reason to see what lights up in the brain, then talk to one of the CNL researchers listed above about setting up a pilot project. Be prepared for a lot of issues you didn't think of at this point.

Data Collection

Subjects: Subjects must meet certain criteria to enter the scanner (they can't have metal in their heads, they shouldn't be claustrophobic, they have to be able to lie still while you scan, they can't be pregnant etc.). The CNL has standard questionnaires about relevant issues. See the forms page for the metal screening form, materials you can provide to subjects and instructions on setting up equipment and subjects at the magnet.

The Experiment: How will you present your experimental stimuli? Timing is important, so we usually automate presentation, and you must decide between block design and event related design. The length of each functional scan is typically between 2 and 15 minutes, with an image volume gathered every 2 seconds. You may wish to run 2 or more functional scans on each subject (e.g., scan them doing a task before and after some experimental manipulation). You probably don't want to make anyone lie there for more than 1.5 to 2 hours total. Typically, we create visual and/or auditory stimuli (see the DMDX page) that run on a computer and are presented to the subject via headphones and/or goggles. Subjects can provide feedback by pressing buttons on mice (we have two mice, so they can hold one in each hand). It is difficult to have subjects talk, because they might move their heads if they talk (a bad thing) and the magnet is very noisy....but you can have subjects think about something, and then try to tell you about it later. For some specialized studies, eye tracking hardware and software are available.

Neuroanatomy and the Magnet Setup: You'll have to know something about neuroanatomy. The details of how you set the scan parameters depend on what area of the brain you want to image well and the structure of your experiment (unfortunately, there are always tradeoffs. Some brain areas are easier to image than others, and some brain areas are better imaged in a particular orientation). Issues you'll need to worry about at the scanner are, at least, plane of section, TR, slice thickness, skip, and number of repetitions. Typically, for each subject you will collect some number of functional scans, and two anatomical scans (a T1 "2D" image and a 3D image).

Data Analysis

Save the Data: You will want to transfer your data from the console to mrisun (a linux workstation at UMC). There you can either burn the data to CD or, if you have permission from the CNL, you can have access through mrisun's firewall so you can transfer data to another machine. Instructions for transfering data are here. Instructions for burning data to a CD on mrisun are here. Keep in mind that your data is very valuable and copies will NOT be kept indefinitely on the console. If your data gets mangled in the transfer, doesn't get transferred correctly or the CD is messed up, it is gone forever. Our standard protocol has always been to copy data from one machine to another rather than moving it, to burn 2 cds of the data and keep them in separate places, and to look at the data as soon as possible, to make sure there are no problems. The console machine is sgi and mrisun is linux. These are both distributions of unix. If you are unfamiliar with unix, consult the online tutorial.

Process the Data: There are a number of image processing and analysis programs for looking at your data. You will need to choose one. A wide variety of programs are used by CNL researchers. SPM and Afni are two of the most common. Tutorials on SPM block-design analysis and on Afni block design analysis are available online. As you begin to do analysis, you may want to know something about images. The image properties page may be very helpful.