Introduction

Over the past 20 years, magnetic resonance imaging (MRI) has come to play a dominant role in the field of cognitive neuroscience. However, appropriately designing, analyzing, and interpreting MRI data remains a challenge. In part this is because cognitive neuroscience is inherently multi-disciplinary, and MRI research brings together ideas from a number of fields including (but not limited to!) physics, medical imaging, frequentist statistics, Bayesian inference, cellular and systems neuroscience, and cognitive psychology. This can all be overwhelming, especially for someone new to the field. The goal of this guide is to provide an introduction to MRI for cognitive neuroscientists.

This guide is without question a work in progress that will grow (and hopefully improve) over time. Corrections and additions will appear here from time to time; I will do my best to prevent page names (and thus links) from changing.

Data analysis of any sort is complicated; there is often not an obvious “correct” way of doing something. Rather, it’s a question of appreciating the advantages and drawbacks of various approaches, and using this knowledge to guide both analysis and interpretation. That being said, especially when starting off in MRI data analysis, the large number of options can seem overwhelming. With this in mind, what I’ve tried to do is make clear what a reasonable approach is to provide a good starting point, and perhaps include a few pointers for reasons why one might do something differently.

Why SPM?

Although general principles of MRI analysis hold for any software, practical examples will be given primarily within SPM (http://www.fil.ion.ucl.ac.uk/spm/). This is because:

  • SPM is freely available and open source, promoting replication and transparency (e.g., if you want to know how a particular function is implemented, you can look at the source code).
  • SPM is implemented in Matlab, which facilitates scripting and editing.
  • SPM is widely used (and what I use the most on a day-to-day basis).

Of course, there is no perfect MRI analysis package; they are just tools. The most important thing to focus on is mastering the principles involved so you can make informed decisions about each analysis you do.

Other resources

Mailing list etiquette

There are two reasons to pay attention to mailing list etiquette. One is that it’s a thoughtful and considerate way of interacting with a community. The second is that you are more likely to get a response to your question if you observe a few key principles (most of which just reflect common sense, and become apparent once you have subscribed to a list for a while):

  • Don’t email the list as a first step. At minimum, read software manuals or help text, and search the web to see if you can find clues to help you track down what’s going on.
  • Search the list archives. Many times you are not the first person to have a question, and there’s a chance someone else has already provided an answer. If you find relevant posts that aren’t quite what you need, these still might be helpful (“I’m looking for information on X. I’ve found Y and Z, but they still don’t tell me about _______”).
  • Include enough information so that people understand what the problem or question is. “My results look strange” isn’t particularly helpful. “Even after thresholding at voxelwise p < .05 (FWE corrected), I have significant activation in the eyeballs, as seen in the attached image” is much more helpful.
  • Stick to one question per email. If you write a long, thoughtful email about your entire study that ends with several questions, it discourages people from both reading (because it’s so long) and replying (because answering all of the questions takes quite a bit of time). Keep your description detailed yet succinct, and stick to one main question as often as you can.
  • If someone replies to your question, and you want clarification, in most cases it’s better to reply to the list, rather than the person directly (“off list”). Chances are if you wanted clarification, at least one other person reading your question now or in the future will find themselves in the same position.

Feedback

Corrections, suggestions, and encouragement are definitely welcome! If you prefer not to be acknowledged, please let me know.