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
Module 10.1: Spatial and Temporal Resolution of Bold Part 1
Loading...
From the course by Johns Hopkins University
Principles of fMRI 1
338 оценки
From the lesson
Week 2
This week we will discuss the fMRI signal, experimental design and pre-processing.
Meet the Instructors
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
This is Principles of fMRI.
In this module, we're going to talk about the spatial and
temporal resolution of BOLD.
This is a plot of both the spatial and
temporal resolution across different neuroimaging techniques.
On the x-axis we see the log time, ranging from milliseconds to days.
On the y-axis, we see the log of space,
ranging from a micrometer to the whole brain, 100 centimeters or so.
MEG and EEG have very good temporal resolution, but
very poor spatial resolution.
And in addition, it's very hard to know whether the signals are coming from
the part of the brain that you think they might be coming from.
PET imaging has reasonable spatial and temporal resolution.
Its temporal resolution is on the order of several minutes up to many minutes,
depending on what type of imaging one is doing.
If it's dopamine imaging, for example, that takes a long time.
And reasonable spatial resolution,
on the order of a centimeter to whole brain scale.
Among fMRI techniques, arterial spin labeling,
or ASL fMRI, can be quite good in terms of temporal stability.
So it can permit one to assess conditions that are different
across broad time scales.
So for example, you can scan somebody with a finger tapping task,
scan them again a month later, and compare the activation.
BOLD fMRI is in the middle of all these different techniques.
And it has spatial resolution on the order of whole brain down to,
as we'll see, below a millimeter.
And temporal resolution usually considered to be on the order of several seconds,
up to longer periods of seconds or a minute or two,
and also see if we can push that forward as well with new techniques.
This is a picture of those human neuroimaging techniques in a broader
context that includes other kinds of recording in animals and
humans, including optical imaging in humans and animals,
single-unit recording electrophysiology, and other techniques as well.
And as we can see, this plot shows both the spacial and
temporal resolution, along with the popularity of those techniques.
And as you can see from this, fMRI and
PET are the most popular techniques so far in humans.
So let's consider which types of information are actually contained
at which different spatial scales.
What we need to measure to extract information from the brain.
And first let's consider the spatial scale of large-scale networks.
Large-scale networks occupy patterns across the brain, across multiple systems
of the brain, down to about a centimeter or a little more, give or take.
And those are reflected in, for example, different functional connectivity across
large-scale networks, that's evident in recent papers from many groups.
The brain also contains many functional maps.
They're everywhere in the brain, and they are at both broad and fine spatial scales.
So for example, somatosensory topography is a relatively coarse,
spatially speaking, kind of topography.
And there are many finer grain types of maps as well.
We can also consider functional columns in the brain, different columns or subnuclei
that perform relatively different functions or dissociable functions.
We'll discuss several examples of those, and
those often occur at a finer grain spatial scale, somewhere less than a centimeter
on the order of several millimeters, down to sub-millimeter resolution.
And finally, much information in the brain is organized
in terms of cell assemblies or activation of individual neurons.
And those cell assemblies are things that we're unlikely to see directly with fMRI.
We'll look at examples of those as well.
In terms of large-scale networks,
what you're seeing here is one result among many.
This is from Randy Buckner's lab.
They looked at functional connectivity across 1000 human brains, and
grouped the brains into these broad, functional networks.
That's what you're seeing here.
There's many other examples.
We used these recently to look at the distribution of emotional
patterns across many different brain systems.
And we used the Buckner et al network maps to do that.
If you look at functional maps, in particular, somatosensory maps,
this is one example of many kinds of maps, we see,
on the left side we see human fMRI, mapping of somatosensory cortex
across different parts of the body, face, fingers, leg, etc.
And on the right, a human optical imaging map of similar things.
These are different digits, in this case.
So it shows you that at this relatively broad spatial scale,
there's quite a bit of specific information.
Moving to functional, oh sorry, moving to maps.
[LAUGH] Another kind of map is one, it's at a much finer spatial scale.
So this is a map collected using optical imaging in monkeys from Wang et al.
And what you're seeing here is different patches of the monkey cortex
that respond to different head orientations, human or monkey.
And the whole map is laid out within a couple of millimeters of brain tissue.
So, that would fit within one voxel in a standard imaging experiment.
If we move to functional columns, one example is the functional organization of
the PAG, or periaqueductal gray in the mid brain.
Stimulation of different PAG columns evokes different kinds of emotional and
motivated behaviors, fight or flight, or
a hyporesponsive sort of learned helplessness type of state, among others.
And with fMRI, particularly here, high-resolution fMRI,
we're able to see functional activity within different subregions of the PAG.
That occurs at a fairly fine spatial scale.
Another example is in the visual system, several kinds of maps.
One is ocular dominance columns, which have a fundamental spacial frequency
of about, on the order of half a millimeter to two millimeters,
and orientation columns, which follow pinwheel
patterns that occur at finer grains spatial scales of organization.
And using high field fMRI, at 7T, and to some degree at 3T,
it's possible to extract information about,which
eye something is coming from, that's ocular dominance columns, or
about which orientation a line is being presented to the subject.
So those are examples of finer grain maps.
And finally, this example comes not from human fMRI, but
from 2-photon imaging in rats in this case in the primary vision cortex,
which is a technique that allows single-cell resolution of fields of many,
many neurons and it's one of a very promising family of
techniques being developed in the neurosciences.
Among others are calcium imaging and light sheet microscopy.
And here what you can see is intermixed cell assemblies
of cells that respond to different orientations.
So, this is a challenge to see with human fMRI.
If we think about BOLD fMRI, it can cover this whole area here,
then, from very large-scale networks to relatively fine grained functional maps.
So, in principle, it's possible to capture many kinds of information.
And pushing the lower limits of spatial and
temporal resolution is an emerging and ongoing frontier.
Ознакомьтесь с нашим каталогом
Присоединяйтесь бесплатно и получайте персонализированные рекомендации, обновления и предложения.
Coursera делает лучшее в мире образование доступным каждому, предлагая онлайн-курсы от ведущих университетов и организаций.
© Coursera Inc., 2018 Все права защищены.
