Principles of FMRI 1

Principles of fMRI 1
by Johns Hopkins University & University of Colorado Boulder
Timeline
START
Week 1
1. Quiz, Quiz 1, Jun 8

WEEK 1
Week 2

WEEK 2
Week 3

WEEK 3
Week 4

WEEK 4
END 07/06
Following weeks
Next Step
WEEK 1
Module 1: Introduction and Ground Rules

It'll take about 9 min. After you're done, continue on and try finishing the week ahead of schedule.
Start
Instructor's Note
Welcome to Principles of fMRI 1! You’re joining thousands of learners currently enrolled in the
course. I'm excited to have you in the class and look forward to your contributions to the learning
community.
To begin, I recommend taking a few minutes to explore the course site. Review the material we’ll
cover each week, and preview the assignments you’ll need to complete to pass the course.
Click Discussions to see forums where you can discuss the course material with fellow students
taking the class.
If you have questions about course content, please post them in the forums to get help from others
in the course community. For technical problems with the Coursera platform, visit the Learner Help
Center.
Good luck as you get started, and I hope you enjoy the course!
More
 Week 1 information
Week 1
Estimated Time: 2h 8m
Week 1
Videos
1h 22m left
Readings
20 min left
Week 1: Week 1
REQUIRED GRADE DUE
Jun 8
Quiz 12:29 PM IST
Quiz 1
26 min
Week 2
Week 2
Videos
1h 39m left
Week 2: Week 2
REQUIRED GRADE DUE
Jun 15
Quiz 12:29 PM IST
Quiz 2
30 min
Week 3
Week 3
Videos
1h 32m left
Week 3: Week 3
REQUIRED GRADE DUE
Jun 22
Quiz 12:29 PM IST
Quiz 3
22 min
Week 4
Week 4
Videos
1h 16m left
Week 4: Week 4
REQUIRED GRADE DUE
Jun 29
Quiz 12:29 PM IST
Quiz 4
24 min
Do you want to receive emails from Johns Hopkins University?
Yes
Overview
Week 1
Week 2
Week 3
Week 4
Grades
Notes
Discussion Forums
Messages
1
Course Info
----------------------------------------------------------------------------
Module 1 video
------------------
Syllabus
Instructors
Martin Lindquist, PhD, MSc and Tor Wager, PhD
Course Description
Functional Magnetic Resonance Imaging (fMRI) is the most widely used technique for investigating
the living, functioning human brain as people perform tasks and experience mental states. It is a
convergence point for multidisciplinary work from many disciplines. Psychologists, statisticians,
physicists, computer scientists, neuroscientists, medical researchers, behavioral scientists,
engineers, public health researchers, biologists, and others are coming together to advance our
understanding of the human mind and brain. In this course we provide an introduction to fMRI,
including physics and acquisition, experimental design, analysis, and inferences about brain and
mind. We discuss both theoretical considerations and practical aspects of conducting fMRI studies,
with an emphasis on statistical analysis. The course is appropriate for scholars from all disciplines
who want to learn more about fMRI, and for researchers engaged in fMRI studies.This is the first
part of a two-part course.
Schedule
Week 1
 Overview
 Acquisition and Reconstruction
Week 2
 Physiology, Signal, and Noise
 Experimental Design
 Preprocessing
Week 3
 Statistical Analysis: The General Linear Model
 Inference and Group Analysis
Week 4
 Inference and Group Analysis (Continued)
 Multiple Comparisons
Quizzes
 There will be four quizzes (one per week).
 Each time that you attempt it, we'll record a score based on your
performance.
 Your effective score will be the highest score of all the allowed attempts
made.
 You must earn a score of at least 80% to pass a quiz.
 Each will be worth 25% of your final grade.

Certificates
 To pass the course, you must pass all four quizzes with a score of 80% or
better.
 If you have signed up for a course certificate and verified you work, you will
receive a certificate when you pass the course.
Complete
-----------------------------
Principles of fMRI Book
Principles of fMRI Book - Click Here To Purchase

Complete
-------------------------
Module2 video
Help Us Translate
You may navigate through the transcript using tab. To save a note for a section of text press CTRL +
S. To expand your selection you may use CTRL + arrow key. You may contract your selection using
shift + CTRL + arrow key. For screen readers that are incompatible with using arrow keys for
shortcuts, you can replace them with the H J K L keys. Some screen readers may require using
CTRL in conjunction with the alt key
Play video starting at 7 seconds and follow transcript0:07
Hi.
In this module, we'll talk about the analysis of fMRI data.
We'll talk a little bit about some nomenclature and
we'll also talk about the goals of fMRI data analysis.
So functional magnetic resonance imaging or fMRI is a non-invasive technique for
studying brain activity.
By non-invasive, we mean that there's no known side effects for
taking frequent fMRI scans.

So during the course of an fMRI experiment,
a series of brain images are acquired while the subject performs a set of scans.
Then changes in the measure signal between the individual images are used to make
inference regarding task-related activations in the brain.
So as a simple example, let's say that you're doing a finger tapping experiment,
where you do finger tapping for 20 seconds, followed by rest for 20 seconds,
followed by finger tapping for 20 seconds.
Under this setting, you might measure the brain multiple times while
you're tapping and while you are at rest.
And then you look for differences in the measured signal between activation and
rest states.
Play video starting at 1 minute 5 seconds and follow transcript1:05

So fMRI is again, functional in the sense that is measured continuously over time,
so you measure the same brain volume multiple times across time.
Each of these brain volumes consists of roughly 100,000 different voxels which
are cubic volumes that span the three-dimensional space of the brain.
So consider that you have your brain and you put it inside sort of a bounding box.
And then you take this box and you split it up into 100,000 equally sized boxes.
Each of these are the fundamental unit that we work with in fMRI.
They're called voxels or volume elements.
Each voxel has a number associated with it, but also a spatial location.
So basically, in this little cartoon,
we take away a voxel from a certain area of the brain and we look at its value.
In this cartoon, it's 39.
So again, each voxel corresponds to a spatial location and
has a number associated with it that represents its intensity.
Play video starting at 2 minutes 2 seconds and follow transcript2:02

During the course of an experiment, several hundred of these types of brain
volumes are acquired, one roughly every two seconds or so.
So basically what we have is we make 100,000 measurements over the brain at
one time point.
Then two seconds later we do it again, etc., etc., for
a couple of hundred time points.
So another way of looking at this,
is we can extract the information from a single voxel.
And as I said, a single voxel represents a spatial location.
So if we take the same voxel across time, we're actually studying what's going on
and how the intensity is changing across that voxel in that spatial location.
So by doing this, we can extract the time series of these intensities and
study to see whether or not there's something in that time series that's
related to the task that we performed.
So in my little example, I was saying we were doing finger tapping, resting,
finger tapping, resting.
Then we might look for a voxel where the activation is going
up while we're finger tapping and going down while we're resting.
Such as in this little cartoon where we see this sort of boxcar activation.
So one of the interesting things is that,
this shows you that fMRI data analysis is fundamentally a time series problem.
Because the data from every voxel is a time series, in this case.
However, it's sort of a time series problem on steroids.
Because what we have is, every voxel of the brain has its own time series and
there's about 100,000 different voxels.
So basically, we're dealing with about 100,000 different time series that we're
studying and looking for at a task-related behavior.
So what is this signal that we get in this time series mean?
Well the most common approach towards fMRI I used is what's called
the Blood Oxygenation Level Dependent or BOLD contrast.
BOLD fMRI measures the ratio of oxygenated to deoxygenated hemoglobin in the blood.
It's important to note that BOLD fMRI doesn't measure neuronal activation
directly.
Instead what it does, is it measures the metabolic demands or
the oxygen consumption of active neurons.
Where neurons are active, they need access to oxygen to replenish their energy.
And it's this oxygen consumption that we can see, so which is a side effect of
the neuronal activation that we're actually interested in studying.

So basically, the way the signal changes in reaction to some
task is described as something called the hemodynamic response function or the HRF.
This represents changes in the fMRI signal that's triggered by neuronal activation.
So let's say that I do something like clap my hands very quickly.
Then neurons in my motor cortex start firing and this leads to an increase
access to oxygenation in that region of the brain.
So basically what we'll see, is we'll see a signal changing in a manner
that corresponding to this cartoon image here.
We'll see a rise in oxygenation levels that
peaks after five to six seconds which goes down.
And after about 10 to 15 seconds, it goes below baseline and
then returns to its baseline form after 25 seconds or so.

So clearly, fMRI data analysis is a massive data problem.
Each brain volume that we're studying consists of roughly 100,000
different voxel measurements.
Each experiment might consist of a 100 of brain volumes.
And each experiment might be repeated for multiple subjects, maybe say 10, 20, 30 or
40 subjects.
In order to facilitate population inference.
At the end of the day,
the total amount of data that needs to be analyzed is staggering.

The statistical analysis of fMRI data is challenging.
It's a massive data problem.
And so it's a sort of like one of these modern big data problems
that's facing Statistics today.
Also, the signal of interest is relatively weak and
the data exhibits a complicated temporal and spatial noise structure.
So throughout this class, we're going to start trying to discuss these different
things and try to understand how we can analyze fMRI data using statistics.

Throughout, we're going to keep coming back to this data processing pipeline
which illustrates the different steps of an fMRI data experiment.
And these consist of the Data Acquisition and Reconstruction step.

There's also a fair amount of preprocess into the data undergoes.
And here, we'll talk about things like Slice-time Correction,
Motion Correction, Co-registration, Normalization and Smoothing.
And then finally, we'll talk about Data Analysis and
there's going to be a couple of key goals that we'll talk about shortly.
Throughout, experiment of the design plays an important role.
So these are all different types of things that we're going to
be covering in future modules.
>> There are three main goals of fMRI data analysis.
The first one is Localization.
The process of determining which specific regions of the brain are active during
a specific task or in relation to a specific psychological event or behavior.

And there are varieties of localization.
And all of these together constitute what we call the Brain Mapping approach.
So the basic framework.
Let's start with the assumption that we have some signal on the brain and
some areas that are truly activated.
Those are shown here in blue.

Those signal areas, when we conduct scans on real people,
are mixed together with noise which is smooth in time and space.
So you see here the red noise pattern.
Where bigger circles mean higher, more positive noise.

What we observe is a mix of the signal and the noise together.

When we do a statistical test, we usually do it at each voxel.
It's called the mass univariate analysis.
And we'll deal with that more later.
And we essentially will conduct a hypothesis test at each voxel.
Because we've made many comparisons, we have to correct for
those multiple comparisons.
And so we end up observing,
usually a small fraction of the areas that are truly active.
And hopefully, in this diagram, all the areas that we've picked out as

results show some true signal.
That is there's some blue in there.
But as you can see, there's one or two areas here that are false positives, so
there's no signal at all.

This brain mapping approach applies whether one is

looking at results from a task.
This might be a task minus control comparison in which each point here in
one voxel constitutes a set of scores from individual subjects.
And we're doing an analysis across those subjects.
That also applies to a Brain-Behavior correlation.
Let's say I have a number of subjects again and
I'm correlating brain activity with some behavioral measure.
It could be their age.
It could be their performance on a cognitive test.

And finally, it applies to what's now called Information-based mapping.
Which is the idea of taking a local area of the brain and
then figuring out how much predictive accuracy that brain area has for
a particular task, condition, or contrast or behavior.

And we'll do which called the searchlight in this approach.
And move a window around the brain and
make a map at each voxel of how much local information there is.
In all three of these cases,

we're essentially doing the same kind of brain mapping procedure.
And the same principles apply.

A second thing we can look at is connectivity.
fMRI in particular, gives us the ability to measure or
assess how brain visions are functionally related to one another.
In connectivity here,
usually means correlations across measured values, across time.

As opposed to functional connectivity in animal studies,
where they're studying direct neuro connections from point to point.

And there are many varieties of connectivity that we can assess.
Here are the three popular ones.
The first one were called Functional connectivity which
relates to correlations across time.
There are varieties of this.
One is seed-based connectivity, in which we take
a region of the brain and we ask what else in the brain correlates with that region.

A second kind of connectivity is called Effective connectivity and
this subsumes many families of statistical models.
Including Path analysis and mediation models,
Granger causality models and Dynamic causal modeling, to name a few.
We'll learn about each of these things later.
Most of them in part two in the next course.
And finally, there are models of Multivariate connectivity.

They fall into two types of categories.
One is the data reduction type, which includes techniques like principle
components analysis, independent components analysis, versions of those.
And the second is Graphical models,
which refers to the process of constructing a visualization and
it announces framework consisting of nodes and edges among a number of regions.
And then deriving properties of that.
That describe and potentially have some inferential power for explaining behavior.
Play video starting at 11 minutes 1 second and follow transcript11:01

Finally, the third type is for Prediction.
We can use a person's brain activity to predict their perceptions, behavior,
health status and a number of different kinds of outcomes.
So on the left here you see an example from our work.
Where we've developed the Classifier Pattern to predict how much pain is
somebody is feeling in response to a given stimulus.
And we can apply that pattern to new brain images coming in
to make a prediction about how much pain they're feeling.
And we can validate that prediction across individuals, across studies and so forth.

There are many emerging applications of prediction that cover
really the whole space of different possible things we might care about in
terms of health outcomes.
So this is really an exciting area in terms of translation and clinical science.
So some of the emerging applications are to Alzheimer's disease, to depression, to
chronic pain and anxiety, to neurological disorders like Parkinson's disease,
to behavioral issues like substance use and
abuse and to basic emotions and other areas of research.

So that's the end of this module.
What we're trying to do is give you an introduction to fMRI data analysis and
what some of the major goals are.
--------------------------------------------------------------------------------------------------------
Module 4.1 video
Help Us Translate
Welcome back to Principles of FMRI.
In this module we're going to talk about psychological inference and
what kinds of questions we can and can't ask with FMRI.

So what can neuroimaging tell us?
Well, the news media is full of examples like this one from The New York Times.
In which, Martin Lindstrom scanned some people with FMRI, and
claimed that he could measure whether they liked or even loved their iPhones.
Here's what he says, most striking of all was the flurry of activation
in the insular cortex of the brain, which is associated with feelings of love and
compassion.
The subjects' brains responded to the sounds of their phone as they would
respond to the presence or proximity of a girlfriend, boyfriend or family member.
In short,
the subjects didn't demonstrate the classic brain-based signs of addiction.
Instead, they loved their iPhones.
Question is, is this something that we can say from the brain?

This isn't just in the media.
There are many new potential applications emerging.
Including applications in law, in clinical science, and others.
So here, this quote is from a judge, Justice John Kennedy, says, scans can
reveal if a person is in pain or mentally competent to stand trial, but not guilt.
We'll ask later on in the course, if that's true and under what circumstances.

When we think about making psychological inferences like these,

what we're really doing is we're trying to understand this question.
Here's a brain map, it's a typical brain map.
What does this map mean?

What kinds of psychological functions is this associated with?
Well, I've circled in red some key areas that should help you decide.
There's the anterior singula, the anterior insula, the medial thalamus.
S2, which is somatosensory cortex.
S1, which is also somatosensory cortex.

All these areas are activated by pain, so maybe it's pain.
We can look at Neurosynth.org Which is a database that now has over 10,000 studies,
each of those blue dots is a point from one study,
and we can ask, studies that activate this pattern,
things like this, what is it that they tend to talk about in their papers?

The top hits for this pattern: Noxious, heat, somatosensory, painful, sensation,
stimulation, muscle, temperature.
So maybe now you're even more confident that it's pain.

I would be, but in fact, you'd be wrong.
This map comes from a study of romantic rejection.
It was published a few years ago in PNAS.

So, that's the pattern of reverse inference.
That's the process of reverse inference.
And that's one idea about what can go wrong.

Let's go back now and think about when we do neuroimaging studies,
what kinds of questions can we actually ask?

And this starting point for
this is really a seminal paper by Steve Kosslyn many years ago, called this.
If neuroimaging is the answer, what is the question?
[LAUGH] And he basically makes the argument that neuroimaging analyses should
be designed to test theories, and so we have to think, what kinds of theories
are actually amenable to testing with neuroimaging studies?
So a typical result is a statistical map, where the colors indicate reliable,
non zero effects.
Statistical maps are used for hypothesis test, at each of the number of voxals, or
regions of the brain.
So effectively,
the question that we can ask is, is this brain voxal responding to my task?
How about this one, or this one, or this region?

That's the question we can ask, in a very strict statistical sense.
And this is the brain mapping approach.
So, the answer that it provides is a hypothesis test on whether a task brain or
brain outcome relationships, whenever I'm testing, is not zero.
There are many choices about how to do my test.
On the left here, I can do tests on each voxel on the brain.
Across the entire map.
I can define some regions of interest.
I've drawn circles around those, so
I can test each fox hole in a set of regions of interest.
I can test each fox hole in one region, or I can just average over one or
more single regions and test that.

And these test differ a lot in certain ways, they differ in the degree
of multiple comparisons correction required, we'll learn about that later.

So, on the left, you have many, many tests so you need very stringent evidence.
On the right, there's really only one test, and so, in one region, so
you need less evidence to conclude, that, that the result is significant.

And you can test activation, you can test correlation maps,
you can test maps of predictive activity, or connectivity, or other things.
But fundamentally, the hypothesis testing framework is the same.
We're asking whether there's a nonzero effect.

So what's the answer?
Given the model assumption, this brain area shows some effect.
But that doesn't necessarily provide any information about how big that effect is,
about how meaningful it is, or about whether it really explains or
predicts a behavior.

And so this slide, Kosslyn to write some of years ago,
this, the demonstration that a particular pattern of brain activity
accompanies the performance of particular tasks isn't really very interesting.

We can only interpret those data in the context of theories.
So this answer is only a good answer,
if your theory requires that the brain data have something to say about it.
Or brain data can say something about it.
So how do we use imaging to address psychological theories?
Well, the key is that result, whether the brain activates or
not, has to constrain your psychological theory.
Has to tell you something that you didn't already know.

And to learn something about the mind,
we need to demonstrate a concept called reverse inference.
I'll explain what that is pretty soon.
The first thing to note, though,
is that brain mapping doesn't provide reverse inference.
It provides something very different.
Which is called forward inference.
So, forward inference is.
Given that I've induced a psychological state,
maybe it's pain, then we observe some brain activity.
And forward inference concerns the probability that observing that
brain activity, whether it's a region or a pattern or
whatever, given the fact that I've induced the state.
I'm in pain. The reverse inference is in some sense
that compliment, can we infer a psychological state given a pattern of
brain activity or activity in any particular region or marker?

These two things are not the same thing, and
they can be very different, qualitatively and quantitatively.
And in fact, there's a name for Fallacious reverse inference for
doing a forward inference mapping study, what's active when I'm in pain,
and then making the conclusion that those regions must imply that I'm in pain.

And that logical fallacy is called the affirming the consequent, and
here how it goes, going back from brain to another simple example.
Dogs like ice cream.
Mary likes ice cream.
Therefore, Mary is a dog.
When you frame it that way, you can see it doesn't make sense, right?
It's not true.
So the fallogical form here is if P, then Q.
Q, and therefore P.
So that's the fallacy.

We'll call a false reverse inference.

So in fact, I can relate forward and
reverse inference quantitatively using Bayes' Rule.
So a Bayes' Rule relates the forward and reverse inference, and what it says,
is that the reverse inference that I care about is
the probability of psychology given the brain, and
that is related to the base rate, the probability of the psychological state.
It's related to the probability of brain activity,
the base rate of brain responses of that type and it's related to the evidence.

So here's a work-through example.

Does punishing wrongdoers feel good?

So the idea of this study here,
just to pick out one among many as an example, is that punishing wrongdoers
feels good because it activates pleasure centers in the brain.
So what do they do?
They gave people the opportunity to punish somebody who was a wrongdoer.
And what they saw when people were doing that, is activation in this brain area
right here, which is part of the caudate nucleus.

So the forward inference is, if you're punishing wrongdoers,
we activate the caudate.
The reverse inference is, if you activate the caudate, you must be enjoying.

So, here's the logical premises of the logical fallacy we talked about earlier,
applied to this case.
So, it would go like this.
One, if I'm enjoying, that's if P.
then my gut is activated Q.
And this can be quantified in terms of sensitivity.
What's the probability of observing colonic activity, given that I'm enjoying.
Something.

Number two, my caudate is activated.
That's Q. And number three, therefore,
am I enjoying?
[LAUGH] That's the fallacy.

And it's only actually true
if enjoyment is the only thing that activates the caudate nucleus.
And this could be quantified in terms of specificity.
Which is 1 minus the probability of getting caudate activity,
if I'm not enjoying, I could be doing anything else.

So let's do a little application of Bayes rule to this particular example scenario.
So for a reverse inference, what we're interested in is another quantity,
called the positive predictive value.
In this case, the positive predicted value that
I'm enjoying something given that I've activated the caudate.

So let's work this through.
So let's say, enjoyment activates the ventral caudate, and
the sensitivity's quite high, so the probability of observing that caudate
activity given I've enjoyed, is .9, 90% So, that's a high sensitivity.
And let's now look at the specificity.
So, if I'm not enjoying things, things seldom activate the caudate.
Let's say probability of observing the caudate without enjoyment is .4.
That's 1 minus the specificity.

And let's say, the base rate of enjoying it is 10%, .1,
and that's a prior enjoyment.

Then I can calculate the positive predictive value using Bayes Rule.
And you can guess, what do you think it is?
Given these number, how likely is it that people are really enjoying something if
they get activation of decani.

And here, the answer would be 0.2, not a very compelling number, so
this demonstrates how you can get very high sensitivity, 90%,
.9, and very low positive predictive value.
Especially if the base rate is low.

So the reverse inference take-home is the following.
Reverse inference is the practice of treating the brain as a marker
for something.
It could be the caudate, it could be any other brain pattern.
When people make reverse inferences,
they assume there's high positive predictive value for that marker.
This positive predictive value can be calculated but
requires the assessment of multiple tasks.
And other states that are potentially confusable with the outcome or
task of interest.
This is a process that is often played out across many studies and
many conditions, not just one.

And for brain activity in a particular region like the caudate to have high
positive predictive value, it has to respond consistently to the task
with high sensitivity, and it must respond only to that task.
Which is with high specificity.
The problem is that, that's rarely the case.
--------------------------------------------------------
Module 4.2 video
Help Us Translate
Now let's look at an example going back to the 1950s.
And this example is going to illustrate how difficult it is to
make valid reverse inferences, even if you have a lot of control and
you can actually manipulate the brain directly.
So this is a picture of Jose Delgado who put himself in the ring with
this brave bull, who's charging him at the moment,
and he has implanted an electrode into the bull's brain.
And he says, I discovered the aggression center of the bull's brain.
To prove it,
I'm going to do this demonstration with the bull charging at me.
And here is with the radio antenna you see there.
Now he's pushing the button, and the bull skids to a halt, and
in his words becomes as placid as Ferdinand.

So did he discover the aggression center?
Well, let's look at where that electrode was in the brain.
This is the brain of a cow, different cow, not the same one,
and there is the caudate nucleus, which is where that electrode was placed.
And this electrode, as we just saw, could be part of the pleasure system, but
mainly it's also heavily involved in the initiation of movement.
And what happened is, when he pressed that button,
the bull was forced to turn continuously to the right.

So you or I might be also very confused, and we might stop being aggressive,
if suddenly something made us turn to the right.
[LAUGH] Right?
But it doesn't have to be the aggression center of the brain.

So this and many other examples demonstrate how difficult it is to really
make reliable reverse inferences about the brain representations
that underlay any particular task, emotional or psychological state.

Is it impossible to make reverse inferences?

Well, Max Coldheart in one challenge, in 2006,
in a nice series of challenge and reply papers, posed this question.
Perhaps functional neuroimaging hasn't taught us anything about the brain so
far [LAUGH].
So, that's one extreme view and I'll let you go through those replies yourself and
see what you think.
But I think, that reverse inference can be done, it's just not very easy.
So, here are two strategies, and
one strategy is to leverage what we know about neuroscience.
So, if you have strong prior knowledge about what does or
doesn't activate a particular brain area,
then activation of that area becomes an interesting answer.
I'll show you an example of that.
And the second strategy is what we'll call quantitative reverse inference.
Which is, we have a brain marker, maybe it's activation of a region like
the caudate, and we assess that activity across our candidate set of tasks.
We can quantify a sensitivity, specificity, positive predictive value.
And this might require testing across many task contexts and study populations.
So it's not the job of one study but the job of many studies together.

Both of these are complimentary.
So they can and should be used together.

So here's an example of the first approach.
And this goes back to Kosslyn's work again, which is still one of my favorite
examples from the field of using brain imaging to test the psychological theory.
So this goes back to a debate between the Kosslyn and Pylyshyn pollution which
started back around 1973, where Pylyshyn claimed that mental imagery,
when you form a mental picture of something in your mind, is propositional.
It's a set of logical rules.
So the feeling of seeing mental images is really an epiphenomenon.

For example, if you imagine a map where the treasure chest is to the west
of the palm tree, then you don't actually see that map in your head,
or at least not in an important way.
There's a set of logical rules.
Like is two left of [LAUGH] and operations like that.
Kosslyn said no, mental imagery is probably depictive.
It's analog, it uses the machinery for perception.
And this debate went on for decades.
And it was finally resolved,
in many people's view, by brain imaging evidence, and it goes like this.
It hinges on this brain area here which is
primary visual cortex along the corcoran fissure.

And this graph here some you a plot from earlier work from Roger.
That shows that the visual cortex, V1, is mapped
where different points on the retina map to different points on the, the cortex.
And what they've done here, is they've shown this stimulus here,
like this ring that you see on the left, and
on the right is the actual primary visual cortex from the monkey where
you can see the pattern of 2-deoxyglucose uptake metabolic changes.

So there's a very strong mapping.
And in particular the fovea in the center of the field is mapped to the back of
the brain and
the peripheral field is mapped more anteriorly to the front of the brain.

So there's a strong prior expectation about what does and
doesn't activate visually cortex.
What Kosslyn did is a series of experiments and
one of them, he asked people to imagine a small letter A versus a big letter A.
And what he saw, was that when they imagined the small A,
there was activity concentrated near the area of V1, and
when they imagined a big A, it spread more anteriorly [LAUGH].
Just like you'd expect it to if it's depictive and analog.
Very hard to explain that with a propositional representation structure.
So that's one kind of answer.
The second example is an example from meta-analysis.
In meta-analysis, as we'll see later,
is the analysis results across many different studies.
In neuroimaging, it's an analysis usually across many types of studies of different
kinds, and that can help quantify our ability to make reverse inferences.

So this graph here, shows you results,
a meta-analysis from six different kinds of studies.
And what you see here is, the task categories are on the bottom,
there's at least 40 studies of each type or so.
And they include studies of emotion, inhibition, long term memory, pain,
switching attention, working memory.
Then what we were able to do is take those seven types of study and
ask if one activates, a secondary somatic sensory cortex, that's two,
what's the probability that they're in pain [LAUGH] or doing a pain task,
rather than any of these other categories?
And what we found here in this case, the probability of being in pain given F2,
which is the positive predictive value, is quite high, almost 90%.
So that's a type of reverse inference and
it illustrates how it can be made among a defined set of alternative candidates.

So now let's think a little bit as we wrap up this lecture
about what brain mapping is good for and what it can and can't be used for.

It can be used to make inferences on the presence of activity,
that can either help us test a theory, as we've shown here, or
it can help us to categorize the pattern of responses to a task for
theory building and for exploratory analysis and development.

It can limit the false positive rates to a specified level,
where we get only the stuff we really believe is true.

And we can leverage hypothesis testing to provide evidence on a variety of theories,
if those theories care about whether there's activity or not.
So we can test, is brain area r involved in task x.

What brain imaging, or brain mapping, isn't really good for
is providing reverse inference, not directly at least.
To do that we need clever experimental designs,
specialized analyses, many studies and conditions.

Secondly, it's not very good for establishing meaningful effects or
effects that are significantly large, or meaningfully large.
So it's really terrible for estimating effect sizes and
predictive accuracy, and we'll talk about this more in later modules.
And finally, there's a number of assumptions that underlie getting
the map that I showed you, or any map.
And brain mapping isn't very good for testing those assumptions.
You have to do it yourself.
Play video starting at 8 minutes 1 second and follow transcript8:01

And finally, brain mapping isn't very good for comparing evidence for
different kinds of theories.
It's a hypothesis test of one theory, but just because a model or
a test shows some effect, it doesn't mean that that's the right model or
the best model or that it's even close.
So other techniques are also needed to actually compare models and to develop,
essentially a picture of what the best models are that explain brain function.
That's the end of this module on psychological inference.
-------------------------------------------------------------
Module 5 Video
Help Us Translate
Hi, in the next couple of modules we'll be talking about how we can acquire and
reconstruct MRI images.
But in order to do that we first have to gain a basic understanding about
MR physics.

So, magnetic resonance imaging, or MRIs,
uses what's called an MR scanner in order to acquire these images, so
an MR scanner consists of an electromagnet with a very strong magnetic field.
Usually this varies between 1.5 and 7 Tesla.
So what does that mean?
Well, to put this in context,
consider that the Earth's magnetic field is 0.00005 Tesla.
So in this case,
a 3 Tesla magnet is roughly 60,000 times stronger than the Earth's magnetic field.
So it's a very strong electromagnet with a strong electromagnetic pull.

So what does MRI measure?
Well MRI is an extremely versatile imaging modality, and
it can be used to study both brain structure and brain function.
And both structural and functional MRI images are acquired using the same magnet.

And different types of brain images can be generated to emphasize contrast,
which is related to different tissue characteristics in the brain.
So all magnetic resonance image techniques rely on the same core set of principles.
And these are the ones that we're going to try to cover now.
So to understand these core principles, we have to begin by studying
a single atomic nuclei and illustrate its impact on the generated MR signal.

So in particular we focus on hydrogen atoms consisting of a single proton.
So these are hydrogen one atoms.

So protons can be viewed as positively charged spheres which are always spinning.
They're spinning around sort of like spinning tops.
Like this.
They give rise to a net magnetic moment along the axis of the spins.

And here we illustrate the direction of the spin.

We can't measure the magnetization of a single hydrogen proton using MR,
instead what we do is we measure the net magnetization of all nuclei within
a volume, so the combination, the ensemble of all the hydrogen atoms within a volume.

So the net magnetization,
which we're going to call M, can be viewed as a vector with two components.
We have a longitudinal component which is parallel to the magnetic field.
And we have a transverse component which is perpendicular to the field.

And so here we see this illustrated, that we have the net magnetization going in
this direction, and we have a longitudinal direction and a transverse direction.

And the transverse direction is in the xy plane.
And the longitudinal is in the z direction.

So in the absence of an external magnetic field,
the nuclear magnetic moments of all these hydrogen atoms are randomly oriented.
So if one hydrogen atom is pointing in this direction,
you have another one pointing in the opposite direction.
So they're cancelling each other out.
For these reasons there's no net magnetization.
However, this changes when they're placed into a strong magnetic field.
This causes the nuclei to align with the field, and
creates a net longitudinal magnetization in the direction of the field.
So here we see all the nuclei are pointing in the same direction, and
this gives rise to a net magnetization in the Z direction.

So the nuclei, while being oriented in this direction, they're also processing
about the field with an angular frequency, which is determined by something called
the Larmor frequency, but at random phase with respect to each other.
So, you can think of these hydrogen atoms as spinning tops, and they're rotating
sort of in the same direction, but at random phase with respect to each other.

So what happens now is that one uses
what's called the radio frequency pulse to align the phase of these nuclei,
so to make them kind of spin in phase with each other, and then tip them over.
And this causes the longitudinal magnetization to decrease and
establishes a new transverse magnetization.
So now all the nucleis are pointing in
the orthoginal direction to the longitudinal direction.
And this causes the net magnetisation to align the transverse plane.

So an analog to this is, I like to think of these hydrogen atoms as these little
compass needles that are always pointing north, and
so north here is the direction of the magnetic field,
and then what we're actually doing here with this RF pulse is, we're taking
the compass needle and we're forcing them to point in the east direction.
So we're excerting a force on it that's forcing the point in the east direction.
But what happens when I remove my finger,
is the compass needle will strive to return to the north direction,
and that's exactly what happens when we remove the RF pulse.
So when remove the RF pulse, the system will seek to return to equilibrium and
this transverse magnetization starts to disappear, and
the longitudinal magnetization grows back to its original size.
And this is a process called longitudinal relaxation.
So basically again,
we have the hydrogen atoms pointing this direction, we remove the RF pulse,
and they start going back into their original longitudinal direction again.

And during this process a signal is created that can be
measured using a receiver coil.

And this is the basic signal that we use in MRI.

So longitudinal relaxation is the restoration of the net magnetization
along the longitudinal direction as the spins return to their parallel state.
This is measured by an exponential growth, described by a time constant called T1.

However there's also something called transverse relaxation,
when this is the loss of net magnetization in the transverse plane
due to the loss of phase coherence.
So remember what we are doing, all these nuclei were out of phase with each other
and the RF pulse put them into phase before it tipped them over.
But once you remove the RF pulse, these nuclei go out of phase with each other.
And this causes a loss of net magnetization in the transverse direction.
And this is also described by an exponential decay
described by a time constant T2.

So here's an illustration of the longitudinal relaxation time, so
basically this is the sort of measure by the time it takes to go from pointing in
this direction in that magnetization, to going back to the original z direction.
And so this is described by this time constant T1 that we see here.
Which is the time it takes to return to 63% of the original net magnetization.

The interesting thing is that different tissue types in the brain have different
T1 values, so the time it takes to relax is different if you're in white matter,
grey matter or in cerebral spinal fluid CSF.

So CSF the relaxation time is much longer,
five times longer than in white matter for example.
And this is going to become important when we start trying to make out images.

The transverse relaxation time, this is due to the lack of,
is the decay of magnetization due to interaction between nuclei, and
this is described by a time constant T2.
And this is due to this lack of phase coherence, and we see this illustrated.
And T2 is how long it takes for
us to go down from 100% down to 37%.
By altering how often we excite the nuclei, which is measured by something
called the TR, and how soon after excitation we begin data collection,
this is something called TE, we can control which characteristic of the image
is emphasized, whether it's T1, or T2, or whatnot.
And so the key to this measure signals approximately equal to the following
equation is M knot, which is the core net magnetization,
times a term that depends on TR and T1.
And remember, TR is something that we can control.
And, a second term that depends on TE and T2.
And again, TE is something that we can control.
And, in contrast, T1 and T2 are tissue properties.
So, the goal of MRI is to construct an image, or
a matrix of numbers that correspond to spatial locations.
And so, the image depicts the spatial distribution of some property of
the nuclei within the sample.

And so this can be the density of nuclei, so how many protons are in
a area of the brain, or the relaxation time of the tissues in which they reside.

So here we see an illustration of this is, is that by altering values of TR, and
TE, we can kind of focus on different characteristics of the tissue.
So, for example if we choose a long TR and a short TE,
the exponential terms here are going to be approximately equal to one, and
the signal's going to be approximately proportional to M knot, so
were going to get the proton density image that we see here.
If we instead have a long TR and
a long TE, well then we have what's called a T2 weighted image.
So this second exponential terms kind of dominates,
and the first exponential term doesn't play a role.
So then what we´re measuring m knot, weighted by the T2 value.
Similarly if we have a short TE, and
a short TE, we get what´s called a T1 weighted image.
And so, as you look at these images, they're quite different from each other
because they're focusing on different tissue characteristics.
So if we want to kind of have a, depending on what type of information we want to
convey with these images, we might choose one of these different image contrasts.
There's one other contrast that we need for fMRI, that's something called T2 star.
An T2 star is the combined effect of T2 and
local inhomogeneities in the magnetic field.

So what happens is that T2 is the dephasing of the protons from each other.
However, it turns out that if you have certain particles in the nearby,
this can cause this dephasing to go quicker.
And this is something that is measured using this T2*.
And so, the MR scanner can be programmed to eliminate the effects of these
inhomogeneities, or alternatively it can emphasize them.
And the later procedure forms the basis of what we call BOLD fMRI.
And it's going to become important for this class.

So, in short, images can be produced that are sensitive primarily to T1,
T2, and T2*.
And because T1 and T2 vary with tissue type,
they're able to represent boundaries between CSF, Gray, and white matter.
And because T2* is sensitive to flow and
oxygenation, it can be used to image brain function.
And that's going to become important for functional MRI.

Okay, so that's the end of this module.
This was just kind of a basic introduction to MR physics and
how we can use an MR scanner to generate a signal.
And so in the next module,
we'll talk about how we can use this signal to construct an image, okay?
See you then, bye.
-----------------------------------------------------------------------------------------------------------------------------
Module 6 video
Help Us Translate
Hi, in this module we'll talk about how we can
take the signal that was acquired using the MR scanner and form an image.

So just to recap, the subject is placed into the MR scanner and
this causes the nuclei of hydrogen 1 atoms to align with the magnetic field.

Now the nuclei precess about the field at similar frequencies but
at random phase with respect to each other, and this causes a net
longitudinal magnetization in the direction of the field.

So within a slice of the brain,
a radio frequency pulse is used to align the phase and then tip over the nuclei.
This causes the longitudinal magnetization to decrease and
establishes a new transversal magnetization.
And then once we return the RF pulse,
the system seeks to return to equilibrium.
So again, everything is pointing in the longitudinal direction.
Then we remove the RF pulse and it seeks to go back to the transverse direction.
So this causes the transverse magnetization to disappear
in a process we call transversal relaxation.
And the longitudinal magnetization grows back to its original size
in something we call longitudinal relaxation.
So longitudinal relaxation is an exponential growth
which is described by time constant T1.
And in contrast,
transverse relaxation is an exponential decay described by a time constant T2.
And so during this process of returning to the equilibrium,
a signal is created that can be measured using a receiver coil.
And so now,
in this module, we want to talk about how we can use that signal to create an image.

So to start with, most structural MRI and fMRI scans involve the construction
of a three dimensional brain volume from a set of two dimensional slices.
So we acquire a set of two dimensional slices and we put them together,
we sort of glue them together,
to make a three-dimensional brain volume as illustrated here.
So for now we're just going to concentrate on how do we acquire a single slice?

So let's imagine that this brain slice is split into a number of
equally sized volume elements, or voxels.
So again, voxels are sort of the three-dimensional analog of pixels.
And so voxels, we have equally sized boxes that we split up.
And so in this little cartoon image, we have a little purple brain here and
it's split up into 16 separate voxels.
So what we want to do is we want to get a measurement of, let's just say for
the sake of argument, the number of protons within each of these voxels.
So we're going to call this measurement the rho(x,y) where (x,y)
represents the location of the voxel.
So basically if we were to kind of measure how many protons we had
in each of these voxels, we could make an image of that information and
we'd get a grayscale image of the object.
So this is a very low spatial resolution because we only have 16 different voxels,
but what we see here is that the darker voxels illustrate that we have more
hydrogen atoms, while the lighter ones illustrate that we have few,
and the white ones that we have very little hydrogen atoms.
So this sort of gives a representation of the object that we're interested in.
Now to get a better representation we would have to split the slice into
more voxels, and that comes with a cost which we'll illustrate in a few slides.

So the measured signal that we have, so
what we want is we want a measurement over each of these voxels.
But alas, the measured signal that we have combines information from the whole brain.
So basically the signal that we're measuring is basically the combination
of all the hydrogen atoms over the whole slice.
So we sort of get a measure of the total number of hydrogen atoms.
Unfortunately, this doesn't give us the information we need to figure out
how many hydrogen atoms are in each of the individual voxels.
So we need to make more measurements to get this information.
And also we need to make different types of measurements.
And so here is one of the clever things about MR scanners,
is here we use a second magnetic field which is called magnetic field gradient.
And using these magnetic field gradients we can sort of sequentially control
the spatial inhomogeneities of the magnetic field, and so
we can change the magnetic field across the brain.
And so this allows us to make a new measurement which is now a weighted
integral of the hydrogen concentration across the brain.
So here you see that now we have for two constants, kx and
ky, we can measure S(kx, ky), which is this row of x,y
weighted by this exponential term which depends on x and y, and kx and ky.
Where, again, kx and ky is controlled by S.
So basically what we can do is we can alter values of kx and ky.
And we can get new measurements of row xy until we have enough for
which we can solve this inverse problem and get a reconstruction of the image,
get the individual rho x y's back.

Now some of you might be familiar with this equation,
this is an example of the Fourier transform.
And so what we can see here is that the measurements that we make, S(kx,
ky), are actually the Fourier transform of the image that we want to reconstruct.
So this is a very useful kind of relationship between the measurements
made by the scanner and the image that we want to view and work with.
Because of this, we can say that the measurements required in the frequency
domain, and in the MR lingo this is usually called k-space.

So by making measurements for multiple values of kx and
ky we can ultimately gain enough information to solve the inverse problem,
and reconstruct rho(x,y).
And by doing this, once we have enough measurements of S(kx,ky),
we can use the inverse Fourier transform and reconstruct rho(x,y).
And that's the image that we want to get.

I've shown you the measurements are integrals, but
in practice the data measurements are made discretely over a finite region.
So instead of using the continuous Fourier
transform we use discrete Fourier transforms.
And so the number of k-space measurements we make
will ultimately influence the spatial resolution of the image.
So we need enough measurements to solve the inverse problem.
So for example, if I want to reconstruct a 4x4 image with 16 different voxels,
I have 16 unknowns.
So I have rho(x,y) in 16 different locations.
So ultimately I have to make 16 k-space measurements so
I have enough information to estimate those guys.
So I need 16 equations because I have 16 unknowns.
If I had made a lower resolution image, only a 2x2 image with 4 voxels,
I would only have 4 unknowns.
So in this case I'd only have to make four different k-space measurements
in order to reconstruct that.
So basically there's a tradeoff between the number of k-space
measurements I need to make and the spatial resolution.
So if I wanted to make a 64x64 image,
I would have to make 4,096 k-space measurements.
And this takes a little bit longer time, and so
there's a temporal cost in doing this.

There's lots of different ways of acquiring data in k-space.
Here are two examples.
One is a echo planar imaging,
which basically samples k-space in sort of a Cartesian grid.
And the other is a spiral which starts from the center and measures outward, so
it changes the values of kx and ky in these manners.
So those are just two popular ways of acquiring the data in k-space.
Another thing that we need to know is that the measured k-space data is
complex values.
So the measurements are complex numbers, they have a real and imaginary part.
And hence, because the k-space data is complex valued,
the measurement at each voxel is also going to be complex value.

So what we typically do here is instead of working with this complex value data,
we work with the magnitude images, or
we just take the square of the real part plus the square of the imaginary part.
We add them up and take the square root, and this is what's called the magnitude

of the complex number, and so we work with magnitude images.
So we take the magnitude of rho(x,y) at each spatial location.

So here's sort of the cartoon showing what happens.
So basically in k-space, we change values of kx and ky and
we make a number of different measurements here using a kind of an EPI trajectory.
And then we make enough measurements of k-space and then we apply the inverse
Fourier transform, and then we get this pretty picture of the brain.
So this is sort of a cool illustration just showing that the measurements
are very indirect.
If you just look at the measurements by itself,
it looks like some salt on a black table here.

But once we do the inverse Fourier transform and
we move into the image space, we get this beautiful brain image.
So this is sort of just illustrating how
we can use the signal from an MR scanner and reconstruct an image of the brain.

So this is the end of the module.
And so here we've talked about how we can take the signal that's acquired by the MR
scanner and we can use it to create an image of the brain that we can use
to study brain function and brain structure.
Okay, I'll see you next time.
Bye.
---------------------------------------------------------------------------------------------
Module 7 Video
Help Us Translate
So in the last module we talked about how we could take this
signal that was acquired from an MR scanner and create an image.
And what we talked about was that the image was not actually created
in image space, but rather in something called k-space.
So in this module, I want to talk a little more about K-space
to gain a little bit more understanding of that concept.
So here's the little cartoon I used last time.
So, data is acquired in the K-space, and
so here I show it being acquired in a grid like fashion.
And so, once you've acquired that data, you apply the inverse for
your transform, and you get this beautiful image in image space.

So it's important to note that there's not a one to one relationship
between image space and K-space.
So we don't have, like there's a single measurement in k-space
that gives you all the information about a single voxel of the brain.
But rather what happens is that,
all the points in k-space contain a little information about every voxel.
So by removing a k-space point,
we lose information about, over every voxel of the brain.

So each individual point in image space depends on all the points contained
in k-space.
So to illustrate the meaning of each k-space based points,
I like to think about this in one dimension.
So let's say that you make three sinusoidal curves as follows.
These are three sinusoidal curves with different frequencies, and
let's say that we take the linear combination of them.
So we take the top sinusoidal curve, and we multiply that by 0.5.
We take the second one and we multiply it by two, then we take the third one and
we multiply it by one, and then we add these curves up, and
then we get the following curve.
So this is a linear combination of three sinusoids.
So if I asked you what three sinusoidal functions went into making this curve,
by looking at it it might be hard to tell.
However, by taking the Fourier transform of this time series
we get the following information.
We get three different spikes in the frequency domain, and so basically,
the frequency domain, the x axis here is frequency,
which is one over the periodicity.
So if we have a sinusoid with a a long period,
we're going to get a spike in the low frequency portions.
So this first spike to the absolute left here, the low one,
which has the magnitude of .5, represents this curve with a height periodicity,
the top one here, and the .5 represents its relative contribution to the signal.

The second peak we have represents the middle curve and
its amplitude is two and the last one, which is the most at high
frequency is the one that oscillates the most, and that has a peak of one.
So by looking at the Fourier transform of this time series we were not only able to
reconstruct the periodicity of the three functions that went into it, but
also the relative contribution to the times series.
And so this is sort of in case base, we do this in two dimensions.
So let's now look at this in two dimensions and
say, well what are the contributions of the each of these k-space points?
Let's say that we have a blank k-space, so it's zero everywhere, and we have our
single measurements here, which is sort of to the Northwest of the origin.
So what happens when you put a value one in this case, and
you take the Fourier transform?
Well, it turns out that if I take the Fourier transform and
transform this into image space, you get a sinusoid here, but in two dimensions.
So, a two dimensional sinusoid, kind of a wave here, going here, and you're seeing
it's going in the direction from which the point, it travels from the origin.
And also the periodicity of this wave depends on how far away
from the center k-space you are.
So this is sort of, in the low frequency parts of k-space, so
it has a high periodicity.
So if you move in the same direction, but towards the high frequency portions,
then we would expect lower periodicity, and so indeed we do that.
So the wave starts oscillating more frequently because we're in the higher
frequency portions, but it's going in the same direction because we're moving in
the same direction from the origin.
Let's say instead we moved
to the Northeast instead of the Northwest from the origin of k-space.
In this case we're still in the same high frequency parts of k-space,
so we would accept the same periodicity, but we're moving in a different direction.
So if we reconstructed this point we would get a wave with the same periodicity,
the same frequency, but
now moving in the Northeast direction instead of the Northwest direction.
So basically, what each point in k-space gives us,
it gives us one these waves, and basically, the value of the point, k-space
point tells us the relative contribution of that wave in reconstructing this image.

Now this is really almost hard to believe because basically when we have
k-space here, what I'm saying here is that the k-space measurements that we see
to the left here are simply weights of these different waves, and we
take the linear combination based on these waves and we get the image to the right.
And so it's hard to believe that this image is just made up of different waves,
but we can show that this is true if we take a single k-space point and
just manipulate it by doubling it.
So let's say that we take this case spaced point, which is sort of to the Northeast
of the center, and I double its value, and now I reconstruct the image again.
Basically what you see is, you see that the way of going in that direction
becomes overvalued, and now we get this kind of grid like artifact over the brain.
So, now we're just overvaluing the wave going in a certain direction,
and this is giving rise to this artifact.
So this sort of illustrates that this image is
a finely kind of balanced combination of these waves, and
if we overvalue one of them, it kind of ruins the whole image.
If we go in the opposite direction, we ge the same type of grid
like pattern, but moving now in the Northwest direction.
So in this case, so the k-space contains information about the entire brain in this
case but now if we're interested in the relative contribution of the high
versus the low frequency parts of k-space, we can do the following example.
Let's split k-space up into nine equally sized boxes, and we take the center box,
and we reconstruct the image using that data, and then we take the outer
eight boxes, just removing the center, and we reconstruct the data using this.
Because the Fourier transform is a linear operation,
the sum of those two should add up to the original image.
So now by doing this little thought experiment, we can see what the relative
contribution of the center of k-space is, versus the outskirts of k-space.
So if we reconstruct the image using the center of k-space,
we get something that looks like this.
It looks very much like the original image, but a little bit blurrier and
you'll see that the detail is not as fine as it was in the original image,
but we've retained most of the information of the brain.
And that's just using one-ninth of the k-space measurement, so
about 11% of the data.
So if we look at what information is conveyed by the additional 88%,
or 89%, we can make that reconstruction.
Here you'll see that we're only getting detail, we're seeing the boundary between
the ventricles in the brain and between the skull and the brain.
So basically, these high frequency parts are the ones

that are oscillating very quickly so those are giving us a lot of the fine detail,
and while the low frequency parts are the things that are changing very slowly and

that's giving us most of the contrast here.

So in general, if we use this as an illustration,
we can see that the low spatial frequency of k-space
represent parts of the object that are changed in a spatially slow manner.
This is contrast.
In contrast, high spatial frequencies represent
small structures whose size in on the order of the voxel size.
So these are usually tissue boundaries and things like that.
So if you want to make out fine spatial resolution,
and you want to make out the difference between grey and
white matter, you need a lot of these high spatial frequencies.
If you're just interested in contrast, you primarily need the center of k-space.
So again, the farther out, the more we sample in k-space,
the more detail we get, and this goes back to spatial resolution.
So if we want to acquire a 32 by 32 image,
we need to sample about 1024 points in k-space.
If we do that, we're primarily using the slowly varying waves and
we're getting this sort of, not much detail.
We're getting very little detail about the brain.
We're just kind of getting a blurry version of the brain.
If instead, we sample a 64 k-space in a 64 by 64 grid,
we have to make 4096 different k-space measurements, and
by doing this, we're now starting to incorporate some of the more high
frequency parts of k-space, and this is giving us more spatial detail.
So now we can start making out differences in the brain.
If we go even higher to a 128 by 128 image,
now we need to sample at 16 roughly 1,6000 points in k-space, but
by doing this we're now getting a lot of high frequency parts of k-space and
I was making wherever to make out a lot of detail.
Like for example, the boundaries between grey and white matter, and
between csf and what not.
And so, what we have here is a very high spatial resolution image of the brain,
which gives us a lot of information and a lot of detail.
Now if we have our druthers, we would like these high resolution images.
But the point is in order to go from the 32 by 32 image to the 128 by 128 image,
we had to make 16 times as many measurements.
And so there's a cost in doing this, because we have to make a lot more
measurements of the brain, so it takes a lot longer for us to do this.
So, in general, there's sort of a trade off between spacial and
temporal resolution here that we want to have adequate

spacial resolution in order to make out what's going on in the brain, but we also
want to acquire the images in a fairly rapid manner because we're going to need
these when we do functional imaging, but we'll come back to this in coming modules.
Okay, so this is the end of this module.
Here we've talked about k-space, and
we talked about the information contact in k-space, and so this sort of ends this set
of three modules where we talked about image acquisition and reconstruction.
And now we're going to move on and talk about some other topics.
Okay, I'll see you in the next module, bye.
-----------------------------------------------------------------------------------------------------------------------------------
Quiz
Quiz 1
TOTAL POINTS 13
1.Question 1
Which field is not relevant to neuroimaging?
None of these options
Engineering
Neuroanatomy
Psychiatry
1 point
2.Question 2
Which technique has the best temporal resolution for fast events?
MEG
PET
ASL fMRI
BOLD fMRI
1 point
3.Question 3
Which of the following are common goals in the statistical analysis of fMRI data?
Making predictions about psychological or disease states
Localizing brain areas activated by a task
All of these options
Determining networks corresponding to brain function
1 point
4.Question 4
Which techniques require the injection of radioactive tracers?
MEG
PET
BOLD-fMRI
PET and BOLD-fMRI
1 point
5.Question 5
Which of the following puts the sequence of processing stages in the correct order?
Reconstruction – Normalization – Connectivity – Localizing brain activity
Acquistion – Reconstruction – Prediction – Smoothing
Acquisition – Slice-timing correction – Connectivity - Prediction
Reconstruction – Motion Correction – Normalization – Localizing brain activity
1 point
6.Question 6
Fill in the blank: The following image is an example of _______ slice.
an axial
a saggital
a coronal
1 point
7.Question 7
The slice thickness determines, in part, the
Voxel size
Field of view
Matrix size
1 point
8.Question 8
Valid reverse inference requires:
High positive predictive value
High sensitivity
Low base rate of brain activation across all tasks
1 point
9.Question 9
Forward inference is related to:
The probability of a psychological state given brain activity
The probability of brain activity given a psychological state
The probability of brain activity given a psychological state and the strength (effect size) of
task-related brain responses
The strength (effect size) of task-related brain responses
1 point
10.Question 10
Transverse Relaxation is the loss of net magnetization in the transverse plane due to loss of phase
coherence and is described by the following constant.
T2
TE
T2*
T1
1 point
11.Question 11
Fill in the blank. Anatomical MR scans are typically __ weighted images.
TE
T2*
TR
T1
1 point
12.Question 12
What analysis technique is used to transform between K-space and image space?
Fourier transform
Partial K-space sampling
Time-frequency analysis
1 point
13.Question 13
One bad value (outlier/artifact) in K-space will appear as _________ in image space:
Random Noise
Ghosting/image wrap-around
Overall image brightness
Bands at a particular angle and spatial frequency
1 point
I, Dr.SELVARAJ DAMODARAN, understand that submitting work that isn’t my

own may result in permanent failure of this course or deactivation of my
Coursera account. Learn more about Coursera’s Honor Code
---------------------------------
Module3 video
Help Us Translate
Hi, in this module we're going to be talking about fMRI data structures.
So this is going to be important, when we get into the data analysis and
the statistical analysis of fMRI data.
We have to try to understand how the fMRI data is structured and
know a little bit about the nomenclature.
So standard fMRI experiments gives rise to massive amounts of data.
We talked about that in the last module, and
most standard fMRI studies consists of both structural and functional data.
And here we'll discuss the structure of the data and
some general terminology that's associated with it.
We're also going to provide a brief overview of
some of the characteristics of the data.

So when deciding an fMRI experiment, you have to balance the need for
adequate spatial resolution with adequate temporal resolution.
So the temporal resolution determines our ability to separate brain events in time.
In fMRI, the temporal resolution is determined by how quickly each
individual image is acquired.
And this is determined by something called the TR,
we'll return to this as we move along.
The spatial resolution determines our ability to distinguish changes in an image
across different spatial locations.

So for example, we we have a structural image, these are often called T1 images,
and again, we'll return to this in a later module.
These have high spatial resolution, but low temporal resolution.
Actually, no temporal resolution because they're just a static image.
But since they have such high spatial resolution,
they can be used to distinguish between different tissue types, and so
we see this cartoon image here that you can separate between gray and
white matter, and you can make out anatomical boundaries and whatnot.

Functional images, which are also called T2* weighted images, and
we'll talk about that again as we move along, have lower spacial resolution, so
they're much blurrier than their structural counterparts.
However, we can measure many of them, and so they have higher temporal resolution,
and they can therefore be used to relate changes in signal
to an experimental manipulation.
So in this cartoon here, we have a bunch of images that are acquired while you're
performing Condition A, and a bunch that are performed by condition B.
So, in the last module we gave an example where you finger tapping, which
might be condition A, and then you're resting, which could be condition B.
So they have measurements under both conditions, and then we might want to look
at the differences in signal between the two different conditions, and
that's something that we can do with functional images.

Another piece of terminology that's important is different
types of slices of the brain.
So we measure three-dimensional brain volume but
we often study slices of the brain.
So we often talk about things like coronal slices,
which is a slice in this direction.

Sagittal slices, which is a slice in this direction and
axial slice which is a slice in this direction.

So these are examples of coronal, sagittal, and axial slices.

FMRI or MRI images are typically acquired in axial slices one at at time.
So we go and do different slices in the x, y plane like this.

They can be performed either sequentially, so one slice at a time, or interleaved,
where we skip and slice and then come back and do it again.
We'll talk about this later on.

Together these slices are kind of glued together and made, to make up a three
dimensional brain volume which is what we ultimately analyze.

There's a bunch of terminology that we need to describe the acquisition.
The first thing that's important to know is what's called the field of view.
The field of view is the extent of the brain that's inside of the image.
So for example, the field of view might be around 20 centimeters.
In this example, I have it at 192 millimeters, so it's 19.2 centimeters.

So that tells us what's the extent in each direction of the brain volume.

The brain volume, again, is made up of multiple slices and
each slice has a certain thickness.
So we also have to say, what's the slice thickness?
Here, in example, we might choose a three millimeter thick slice.

Each slice is then split up into voxels, and for example,
in this cartoon, the voxel, the actual slice is split up
into 4096 different voxels in a 64 by 64 grid.
So if we have 64 voxels and our field of view is 192 millimeter, then
each voxel is going to be three millimeter in both the xy and y directions.
So in this case, if we have a matrix size of 64, a field of view of 192,
and a slice thickness of 3 millimeters, our
eventual voxel size will be 3 millimeters by 3 millimeters by 3 millimeters.

And so that's going to be the size of the unit of measurement that we're going to be
interested in fMR.
We can change that, for example, we can make the slice thickness thinner or
thicker, or we can make the matrix size higher, and
thus make the voxels smaller or bigger in the xy directions.

So an experiment studies many different subjects.
So typically enough from where I would we have experiments where
we look at multiple subjects that perform the same type of task.
And in each of these subjects might be scanned during multiple sessions.
Each of these sessions consists of several runs,
we might repeat the task several times on each subject while they're in the scanner.
And each run is going to consist of a series of brain volumes.
Now each of these brain volumes, again, is made up of multiple axial slices and
each of these slices contains many, many voxels.
And, again, each voxel has an intensity associated with this and
this is the basic measurement in fMRI.

So this sort of suggests that there's a hierarchy going from the experiment,
the subjects nested within this experiment,
the sessions nested within each subject, there's run within each session,
and then there's brain volumes within each run.
Each brain volume are measured sequentially over time,
and each voxel has a corresponding time series associated with it.
This is basically the final
level where we analyse the date on this time series level.
But again, we also want to be able to look at these time series and say,
what can they say about the population of subjects that we studied and what not?
Okay.
So that's the end of this module.
Here I've just attempted to kind of talk about some characteristics of fMRI data.
Talk about some nomenclature that's going to be important terminology,
and that will help you read method sections and what not.
Okay. I'll see you in the next lecture.
Bye.
------------------------------
-------------------------
Reattempt
Quiz 1
TOTAL POINTS 13
1.Question 1
Which of the following imaging modalities is not used to study brain function?
Computed axial tomography (CAT)
Functional magnetic resonance imaging (fMRI)
Electroencephalography (EEG)
Positron emission tomography (PET)
1 point
2.Question 2
Which technique has the best temporal resolution for fast events?
MEG
ASL fMRI
PET
BOLD fMRI
1 point
3.Question 3
Which technique has the best spatial resolution?
PET
fMRI
EEG
SPECT
1 point
4.Question 4
BOLD-fMRI

MEG
PET
1 point
5.Question 5
1 point
6.Question 6
a coronal
a saggital
an axial
1 point
7.Question 7
Matrix size
Voxel size
Field of view
1 point
8.Question 8
It is possible to scan someone’s brain and tell whether they love an object or person
Yes
No
1 point
9.Question 9
The probability of brain activity given a psychological state and the strength (effect size) of
task-related brain responses
1 point
10.Question 10
Which relaxation time is sensitive to differences across brain tissue types and therefore good for
anatomical imaging?

T1
T1 and T2
T2
1 point
11.Question 11
T2*
TR
TE
T1
1 point
12.Question 12
Fill in the blank: The MR signal is sampled in ________.
Outer-space
Image-space
K-space
1 point
13.Question 13
The outer portion of K-space samples ______ frequencies and takes the _____ time to sample
Low, least
High, least
High, most
Low, most
1 point

-------------------------------
Quiz 1
TOTAL POINTS 13
1.Question 1

1 point
2.Question 2
MRI can be used to measure
White-matter tracts
Task-related functional activity
Gray-matter density
1 point
3.Question 3
Which technique has the best spatial resolution?
SPECT
EEG
fMRI
PET
1 point
4.Question 4
MEG
PET
PET and BOLD-fMRI
BOLD-fMRI
1 point
5.Question 5
1 point
6.Question 6
a coronal
a saggital
an axial
1 point
7.Question 7
Matrix size
Field of view
Voxel size
1 point
8.Question 8
Yes
No
1 point
9.Question 9
The probability of brain activity given a psychological state and the strength (effect size) of task-
related brain responses
1 point
10.Question 10
anatomical imaging?
T2
T2*
T1 and T2
1 point
11.Question 11
TR
T2*
T1
TE
1 point
12.Question 12
Fourier transform

1 point
13.Question 13
The outer portion of K-space samples ______ frequencies and takes the _____ time to sample
Low, least
Low, most
High, most
High, least
1 point

-------------------------------------------------------------------------------------------------
Congratulations! You passed!

TO PASS 80% or higher
Keep Learning
GRADE
100%
Quiz 1
LATEST SUBMISSION GRADE
100%
1.Question 1
Correct
1 / 1 point
2.Question 2
MRI can be used to measure
Task-related functional activity
Gray-matter density
White-matter tracts
Correct
1 / 1 point
3.Question 3
Which of the following are common goals in the statistical analysis of fMRI data?
Determining networks corresponding to brain function

Making predictions about psychological or disease states
Localizing brain areas activated by a task
Correct
1 / 1 point
4.Question 4
PET
MEG
BOLD-fMRI
Correct
1 / 1 point
5.Question 5

Correct
1 / 1 point
6.Question 6
a coronal
a saggital
an axial
Correct
1 / 1 point
7.Question 7
What is a typical number of slices for a functional volume?
30
None, functional volumes are not acquired in slices
150
Correct
1 / 1 point
8.Question 8
Yes
No
Correct
1 / 1 point
9.Question 9
What cannot be inferred from standard brain maps?
How large the effect on brain activity is
What the psychological cause of the activation is
Whether the model used to estimate the activation is correct

Correct
1 / 1 point
10.Question 10
anatomical imaging?
T1 and T2
T2*
T2
Correct
1 / 1 point
11.Question 11
The strength of an MR scanner is measured in the following units.
Tesla
Ampere
Volt
Ohm
Correct
1 / 1 point
12.Question 12
Fourier transform
Correct
1 / 1 point
13.Question 13
One bad value (outlier/artifact) in K-space will appear as _________ in image space:
Bands at a particular angle and spatial frequency
Random Noise
A circular artifact
Ghosting/image wrap-around
Correct
---------------------------------------------------------------------------------------------------------------------------------

Principles of FMRI 1

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Principles of FMRI 1

Uploaded by

Copyright:

Available Formats

Principles of fMRI 1

by Johns Hopkins University & University of Colorado Boulder

1. Quiz, Quiz 1, Jun 8

1. Quiz, Quiz 2, Jun 15

1. Quiz, Quiz 3, Jun 22

1. Quiz, Quiz 4, Jun 29

Module 1: Introduction and Ground Rules

Estimated Time: 1h 54m

Estimated Time: 1h 40m

Do you want to receive emails from Johns Hopkins University?

 Acquisition and Reconstruction

 Physiology, Signal, and Noise

 Statistical Analysis: The General Linear Model

 Inference and Group Analysis

 Inference and Group Analysis (Continued)

 You must earn a score of at least 80% to pass a quiz.

 Each will be worth 25% of your final grade.

Principles of fMRI Book

Principles of fMRI Book - Click Here To Purchase

Play video starting at 28 seconds and follow transcript0:28

Play video starting at 1 minute 5 seconds and follow transcript1:05

Play video starting at 2 minutes 2 seconds and follow transcript2:02

Play video starting at 4 minutes 8 seconds and follow transcript4:08

Play video starting at 4 minutes 58 seconds and follow transcript4:58

Play video starting at 5 minutes 24 seconds and follow transcript5:24

Play video starting at 5 minutes 54 seconds and follow transcript5:54

Play video starting at 6 minutes 8 seconds and follow transcript6:08

Play video starting at 6 minutes 48 seconds and follow transcript6:48

Play video starting at 7 minutes 4 seconds and follow transcript7:04

Play video starting at 7 minutes 19 seconds and follow transcript7:19

Play video starting at 7 minutes 25 seconds and follow transcript7:25

Play video starting at 7 minutes 54 seconds and follow transcript7:54

Play video starting at 8 minutes 4 seconds and follow transcript8:04

Play video starting at 8 minutes 8 seconds and follow transcript8:08

Play video starting at 8 minutes 36 seconds and follow transcript8:36

Play video starting at 8 minutes 52 seconds and follow transcript8:52

Play video starting at 9 minutes 4 seconds and follow transcript9:04

Play video starting at 9 minutes 11 seconds and follow transcript9:11

Play video starting at 9 minutes 28 seconds and follow transcript9:28

Play video starting at 9 minutes 36 seconds and follow transcript9:36

Play video starting at 9 minutes 59 seconds and follow transcript9:59

Play video starting at 10 minutes 26 seconds and follow transcript10:26

Play video starting at 11 minutes 1 second and follow transcript11:01

Play video starting at 11 minutes 34 seconds and follow transcript11:34

Play video starting at 12 minutes 8 seconds and follow transcript12:08

Play video starting at 13 seconds and follow transcript0:13

Play video starting at 58 seconds and follow transcript0:58

Play video starting at 1 minute 23 seconds and follow transcript1:23

Play video starting at 1 minute 27 seconds and follow transcript1:27

Play video starting at 1 minute 35 seconds and follow transcript1:35

Play video starting at 1 minute 53 seconds and follow transcript1:53

Play video starting at 2 minutes 15 seconds and follow transcript2:15

Play video starting at 2 minutes 25 seconds and follow transcript2:25

Play video starting at 2 minutes 35 seconds and follow transcript2:35

Play video starting at 2 minutes 45 seconds and follow transcript2:45

Play video starting at 2 minutes 53 seconds and follow transcript2:53

Play video starting at 3 minutes 33 seconds and follow transcript3:33

Play video starting at 4 minutes 11 seconds and follow transcript4:11

Play video starting at 4 minutes 20 seconds and follow transcript4:20

Play video starting at 4 minutes 34 seconds and follow transcript4:34

Play video starting at 4 minutes 51 seconds and follow transcript4:51

Play video starting at 5 minutes 5 seconds and follow transcript5:05

Play video starting at 5 minutes 16 seconds and follow transcript5:16