This course will show you how to analyze an fMRI dataset from start to finish. We will begin by downloading a sample dataset and inspecting the anatomical and functional images for each subject. We will then preprocess the data, which removes noise and enhances the signal in the images. Once the images have been preprocessed, we will create a model representing what we think the BOLD signal, a measure of neural activity, should look like in our images. During model fitting we compare this model with the signal in different areas of the image. This model fit is a measure of the strength of the signal under different conditions - for example, we can take the difference of the signal between conditions A and B of the experiment to see which condition leads to a larger BOLD response.
Once a model has been created for each subject and the BOLD response has been estimated for each condition, we can do any kind of group analysis we like: Paired t-tests, between-group t-tests, interactions, and so on. The goal of this course is to calculate a simple within-subjects contrast between two conditions, and test whether it is significant across subjects. You will also learn how to create figures showing whole-brain analyses similar to what you see published in the neuroimaging journals, and how to do a region of interest (ROI) analysis.
This course is designed to build your confidence in working with fMRI data, increase your fluency with the basic terms of fMRI analysis, and help you make educated choices during each step. Some chapters have exercises for practicing what you’ve learned and to prepare you for the next chapter. Once you have mastered the fundamentals of this course, you will be able to apply them to other datasets of your choosing.
We will not be covering MRI physics in depth. For a review of that topic, I recommend chapters 1-5 of the book Functional Magnetic Resonance Imaging, by Huettel, Song, & McCarthy (3rd Edition). Also see Allen Elster’s excellent MRI Questions website for useful illustrations of MRI concepts.