This course covers the design, acquisition, and analysis of Functional Magnetic Resonance Imaging (fMRI) data. A book related to the class can be found here: https://leanpub.com/principlesoffmri

Loading...

From the course by Johns Hopkins University

Principles of fMRI 1

259 ratings

Johns Hopkins University

259 ratings

This course covers the design, acquisition, and analysis of Functional Magnetic Resonance Imaging (fMRI) data. A book related to the class can be found here: https://leanpub.com/principlesoffmri

From the lesson

Week 4

The description goes here

- Martin Lindquist, PhD, MScProfessor, Biostatistics

Bloomberg School of Public Health | Johns Hopkins University - Tor WagerPhD

Department of Psychology and Neuroscience, The Institute of Cognitive Science | University of Colorado at Boulder

Welcome back to principles of fMRI.

Â In this module we're going to introduce group analysis.

Â We're first going to give a background and then we'll talk about

Â moving from a single GLM analysis to the group analysis setting.

Â In a multi-level analysis fMRI experiments are often repeated for

Â several runs in the same session, several sessions on the same subject, and

Â several subject, or a number of subjects now drawn from a population.

Â And just to remind us, we've seen this before, but

Â this is the hierarchal structure of the state.

Â So now what you see is we have boxals within images within runs.

Â Run one to K.

Â Runs within sessions potentially, this could be a time one, time two,

Â pre, post-drug intervention.

Â And then sessions within subjects.

Â So for group analysis we run multiple subjects,

Â started with maybe a handful of subjects, but now, it's common to run tens, or

Â even hundreds, or sometimes thousands of subjects nested within groups.

Â Group might be a patient versus control, for example.

Â So the data has this hierarchical structure.

Â And in a multi-level model we, as we said before,

Â do a first level analysis which deals with individual subjects.

Â So we run a model on each subject.

Â And then a second level model deals with groups of subjects, and that could involve

Â patients versus controls, it could involve individual differences and relationships

Â between brain activity and personable characteristics like age or performance.

Â So here's a depiction of the group, and

Â within a group there are subjects, each subject has their own time series.

Â For every voxel, and we're still dealing with inferences

Â that are performed in the mass user variant setting.

Â So we're stealing with one voxel at a time.

Â So here's an overview of the GLM analysis process again, we talked before about

Â design specification or model building in several lectures, we talked about

Â estimation of the design at the first level, that's for every voxel for

Â a subject, and we talked about, defining contrast, which are effects of interest

Â that you care about, and identifying contrast images for each person.

Â Those contrast images for each person, and taken and

Â combined with the contrast images from other subjects, into a group analysis, and

Â that's what allows us to make inferences about which

Â areas are activated in the population, and this is our framework.

Â So we're going to work backwards from a group result to a individual subject

Â result, and this is also rear view.

Â So here there are some true signal mixed with noise,

Â which yields a mixed map of signal plus noise.

Â And if we do a statistical test in the group, then we end up getting a test

Â statistic at each voxel, that could be a T value for example.

Â And we threshold that, correcting for multiple comparisons.

Â And we get some results, which we then interpret.

Â So working backwards again, from the results at one voxel in a group analysis

Â the most basic kind of group result is contrast values between task and

Â control, where each dot here that you see is a score from one subject, and

Â I'm interested in whether those scores, those contrast values,

Â are significantly different different from zero in the group.

Â So there's that group analysis again.

Â Each of those scores then in the group analysis, each individual subject score

Â is resulting from a contrast in a general linear model analysis.

Â So this is an example with two regressers within that individual subject

Â which yield beta 1 and beta 2, two regression perimeter estimates.

Â Those are multiplied by a contrast of interest, this case the difference let's

Â say between the two, one minus one, which yields a contrast value for each subject.

Â So, what we're going to see next is a movie that takes us all the way through

Â the analysis process, from individual subject, design and

Â contrast at one voxel, repeated over a group of subjects, and then

Â the process of doing the group analysis using the very simple one sample T test.

Â So here we go, we're looking at one voxel, now this is a partial fit for trial type

Â A and B within one subject to one voxel, there's the fit all trial types together.

Â Now in the bottom left, we're going to see this difference between trial type one and

Â trial type two, in terms of the human endemic response,

Â the greater the difference is, then the greater the contrast value.

Â Now that's going to be translated into dots, over on the right panel.

Â So we see subjects being repeated.

Â Now we're repeating the analysis several times.

Â The first analysis is where we have all the subjects are actually identical,

Â and the only differences across the subjects

Â are in the actual fMRI random noise, the noise realization.

Â So that's the black line, and that's only one source of air.

Â Then we're going to repeat the analysis and that's the middle row,

Â column of gray dots.

Â And with that, we have two sources of noise.

Â We actually have the design varying as well.

Â So we've randomized the ordering of the events.

Â And the design is randomized.

Â So now we have a little bit greater spread in those points.

Â So we have a little bit greater noise variance.

Â And then in the third case,

Â we've also added the idea that there are real differences across individuals.

Â So now there are individual differences as well, and

Â that adds an additional noise component.

Â And so the spread in the points is greater.

Â The greater the spread, the more that hurts us in terms of statistics, so

Â that's the noise that we have to overcome.

Â But that's a realistic situation.

Â So the realistic idea is really what's happening on the right side,

Â where we have variation due to random noise in the scanner,

Â variation that's related to the design matrix itself, and

Â what I've induced with the design, and then variation that's related to

Â the individual differences in true levels of activation across people.

Â So that's the overview of how the analysis works at one voxel, from soup to nuts.

Â And we'll go back over this now in more detail, and

Â explain it more completely in the next modules.

Â This analysis maps onto a repeated measures analysis

Â from traditional statistics in which we have one or more within person predictors.

Â Now we have just one simple within subject contrast.

Â So that's a generalized linear model, and we'll deal first with this sort of summary

Â statistic case, which has the one simple T test and we'll deal later

Â with models that count more completely for correlated errors and time series models.

Â That's the end of this module.

Â In the next couple of modules, we'll talk more about group analysis.

Â Coursera provides universal access to the worldâ€™s best education, partnering with top universities and organizations to offer courses online.