1
Q
What are the two types of MRI we can do?
A
MRI comprises both structural and functional MRI.
- Structural MRI examines brain anatomy or structure. Something static that doesn’t change. Like an XRAY
- fMRI examines brain functions – task-related BOLD changes. This is over time – not static.
2
Q
WHat physical principles underscore both structural and functional MRI
A
- Brian contains a lot of protons. This is good it contains a lot of water
- Then put protons into a strong magnetic
- Lay a radio – send a high radio frequency pulse through the head
- Protons absorb this energy and begin to resonate
- Then stop emitting the radio frequency pulse
- → protons that absorbed the energy, emit the energy*
- This is the signal underlying MRI’s that we measure. The energy emitted by protons that you have excited within a magnetic field.*
3
Q
in MRI what do we use to construct the brain image?
A
- The energy emitted by protons that you have excited within a magnetic field.
- Using this emitted energy, we can construct a brain image
the reduction of the transverse magnetisation and increase in longitudinal magnetisation
4
Q
Describe the MRI set up
A
- Main magnet: where the magnetic field is constantly produced. Cannot switch this off. Range of 1.5 Tesla to 9.4 Tesla
- Radio frequency coil: surrounds your head, where the radio frequency pulse is transmitter, and where the measurements are taken
- Gradient coil: helps you aquire the signal at different spatial locations
5
Q
Discuss the purpose of the gradient coil in MRI
A
- Helps you aquire the signal at different spatial locations
- Changes the gradient of the magnetic field
- You only receive one signal from the coil – different to EEG using different locations acquiring data. You have one signal
6
Q
What are protons?
A
Sub atomic particles in atomic nucleus
7
Q
Why is it lucky for MRI that the brain has a low of water in it?
A
- Because hydrogen contains exactly 1 proton
- + very sensitive to the RF pulse (magnetic resonance)
8
Q
What do we mean by precession in MRI research?
A
related to the fashion in which protons spin
- Axis is tilted
- The head of this goes in a precise spin above
- The top (Circle) precession - has a certain orientation and frequency
9
Q
What does the RF pulse change with regards to the protons spinning?
A
- spin frequency itself is not changed – fixed.
- changes the precession frequency, and its orientation
10
Q
Without the magnetic field how are proton spins oriented normally?
A
Proton spins are random. Random orientation.
11
Q
How do protons spin when the magnetic field is on?
A
Fraction of proton spins align parallel or anti-parallel to the magnetic field BO
- The basic magnetic field = BO.
- It is a longitudinal field from bottom (neck) to top (top of head)
12
Q
How does the magnetic field in MRI affect protons? before and after
A
- protons have a certain orientation (precision orientation) and this goes around an oval.
- Normally it’s random but with a magnetic field, the precision orientation is affected.
- aligns parallel or anti-parallel with the magnetic field
Only a fraction of protons align – this is enough of a signal for us to pick up with fMRI
13
Q
How can we get longitudinal magnetisation of the signal?
A
- can sum up the vectors along the longitudinal axis
- bc of the alignment we get a longitudinal magnetisation
- thus sum of the alignment gives us a value graded signal
Longitudinal magnetisation produced by the magnetic field
14
Q
What’s the difference between the longitudinal and transverse magnetisation?
A
so we have a magnetic field, protons align, we measure the orientation of the protons
- longitudinal - measures magnetisation from the side, saggital view
- transverse - measures magnetisation from the top - no magnetisation in the transverse plane; get a vlaue of 0
15
Q
What value do we get with longitudinal vs transverse magnetisation?
A
Transverse magnetisation
- Mxy~0
longitudinal magnetisation
- value greater than 0
- because of the magnetic field
16
Q
how long after the RF pulse is delivered do the proton spins align with the magnetic field?
A
Trick question!
Second n enter the scanner the protons precession orientation is affected
17
Q
What happens when we play the RF pulse?
A
- Play the radio in the transverse plane (appaz left to right now)
- protons absorb RF energy
- spin axis gets flipped 90 degrees on their side
- align with the direction of the RF pulse
by sending the RF pulse we have reduced the longitudinal magnetisation while at the same time generating transverse magnetisation IF you look from the top of n head
18
Q
after playing the RF pulse and stopping what happens to the protons?
A
- Proton spins are on side but we still have the magnetic field along the longitudinal axis.
- the proton spins gradually return to that orientation
- they are re-radiating the absorbed energy from the RF pulse.
- time this takes it to return to this configuration = what MRI measure.
Transverse magnetisation decreases while longitudinal magnetisation decreases.
19
Q
how does MRI distinguish different tissues?
A
- The time it takes proton spins to get back to magnetic field orientation is different for different tissues.
- allows us to discern different tissues just looking at the MRI image
20
Q
What are T1, T2 and T2* effects
A
The reduction of the transverse magnetisation and increase in longitudinal magnetisation have different names
- T1 effect - the increase in longitudinal magnetisation (aka recovery). Underlies structural imaging.
- T2 effect - the decay/reduction of the transverse magnetisation because of spin-spin interactions
- T2* effect - the decay/reduction of the transverse magnetisation because the magnetic field is inhomogeneities - not the same at all spatial locations.
- reason it’s not the same everywhere because of differences in blood flow.
ONLY the T2* effect underlies fMRI. see this in papers - we acquired functional data using a T2* imaging protocol
21
Q
Will different tissues have the same T1 and T2 effect?
A
Molecules in different tissues have different proton structures, e.g., white matter, grey matter, skull, CSF
Thus have different speeds at which they re-radiate absorbed energy and
- return to their original configuration (T1 effects)
- and with which the transverse magnetisation reduces (T2 effects).
22
Q
Spin echo?
A
In most imaging studies we play both a
- 90 degree RF pulse in one direction
- then 180 degree RF pulse in other direction
- flips protons back
Produces an “echo”. this is the signal we measure, not the effect of the 90 degree initial pulse
23
Q
Spin echo sequence?
A
- bare 90 and 180 RF pulses
- of course influences the signal because the signal reflects how much energy protons re-radiate
24
Q
What two parameters do we need to clarify with MRI?
A
Echo time (TE)
- time (ms) between the original 90 degree and echo. Occurence of echo is the time you measure the signal. MR signal sapling.
- 180 degree pulse is applied at half the echo time
repetition time (TR)
- time between two 90 degree pulses. important as it indicates how often you cycle through the brain
25
What takes longer the T1 or T2 effect
T1
26
*Repetition time (TR)*
1 brain volume = 1 TR. ***Within TR** = Rapid RF pulses sent through brain to get the info*
* ***time betewen two 90 degree pulses -** _how often pulse cycles through the brain -_ usually within 2s*
* 1 screenshot of brain activation each 2s
* *then do it **again** for the next 2 seconds then **again and again***
* each time get 1 brain volume with 32 slices
* *measuring signal intensity*
*With fMRI you measure how much the signal has reduced there - transverse magnetization If ur measuring the magnetization plate*
27
How can we manipulate the parameters to get different things?
* Choose TE and TR - *manipulate and measure RF energy at specific points in time*
* do this to emphasise where diff brightness and contrast levels (signal intensities) are mainly arranged from T1 or T2 properties
28
Whats the difference beteween structural and functional MRI
T1 vs T2 weighted images
The times we play that radio to either produce images where diff brightness and contrast levels (signal intensities) are mainly arranged from T1 or T2 properties
* structural = look for changes in the size and integrity of brain structures
* functional = mostly a research method - which regions are activated when you do something
29
T1 weighted images
* *Structual MRI*
* *brightness/contrast – determined by T1 properties. Recovery of longitudinal magnetization - how ong protons returned to origional configuration*
* *Better quality – higher resolution than fMRI*
* *Here white matter structures are white. This bc they contain mainly myelinated axons. These are fatty so show up as white.*
* *grey matter is grey*
30
T2\* weighted images
* *fMRI*
* *Brightness/contrast determined by T2 properties*
* *the decay of transverse magnetization*
* *White matter is **grey** here. Ventricles are white and grey matter is white.*
* *Resolution is not as good*
31
RF coil - go on bih
*The cage thing surrounding n’s head. This is the coil.*
* *Both transmit pule and measure energy emitted by the protons*
* *also a receiver device*
* *12,32,16,78 channel options - don't pick up signals from different locations - just improves image quality.*
* *remember it's one signal in the transverse plane that is picked up by all these channels in the same way*
* *then calculate back using Fourier transformation to understand where within that plane the signal comes from*
32
Gradient coils - go on bih
Just below the main magnet
* *allow us to cycle through the brain and _measure the signal at different spatial locations._*
* changes the magnetic field locally - leads to differences in the strength of the magnetic field
* leads to differences in the resonance frequency in the slices
* gradient cold help you get data from different points in your head
*Within 1 TR the gradient moves around and you cycle through the head. This produces a brain image. pizza sprinkler.*
33
why do we need gradient coils
They help you get the data from different points of the head
34
TR - 1 pulse 90 degrees, then 180 degree pulse, then after 2s another 90 degree pulse
How long is the TR here? and what is the gradient doing during this
1 TR = 2s
within this, the gradient is moving around and cycles through your head to generate 1 brian volume.
35
Saftey and exclusion criteria in MRi?
* *No metal – in or outside body e.g., pacemaker*
* *Hair clips – dangerous and produce huge artifacts in the data*
* *No pregnant n*
* *Credit cards – will get wiped*
* *Underwire bras*
*But remember MRI non-invasive! Nothing to inject, no radiation – sick! There is however a magnetic field. in PET you have radioactive substances that you inject into people.*
36
Structural MRI
* Measures static brain anatomy*
* Mostly for clinical purposes but could also be useful in cognitive neuroscience research too.*
37
functional fMRI
Measures brain function
* activation maps elicited by a task
* bold signals
* look at something event related **over time**
38
2 main ways to use structural MRI in research for cognitive neuroscience?
1. *Use as an anatomical template on which functional activation patterns are overlaid for localisation*
* gives you a better ability to localise things
2. brain morphometry
* measures the shape, size and integrity of brain structures in healthy and pathological n. Includes:*
* ***Grey matter structures** - contains mostly cell bodies of neurons and glial cells*
* ***White matter structures** - contain mostly long-ranging nerve fibres (aka myelinated axons).*
* *White matter morphometry is also referred to as **tractography** as you are map brain tracks.*
* whole brain - could also measure the size of the whole thing
39
2 ways to measure grey matter morphology
1. mannual approach - go through with mouse on individual brain and identify grey matter structures using prior knowledge
2. *automated approach e.g., voxel based morphometry - first normalise image and then take prior inofrmation and compute likelihoods that a voxel is grey/white matter. Determines whether certain voxels are part of a structure*
*comparing sizes of different structures*
40
Describe the manual approach to studying grey matter morphology
* *Use computer mouse to identify grey matter structures*
* *based on prior knowledge – this is the amygdala, this is the hippocampus*
* *Then count the number of voxels within that area and compare it to a controls structure - standardise it*
* *n with bigger head will have more pixels in the area sso needs to be standardised*
41
*Describe the automated approach to studying grey matter morphometry*
*Voxel-based morphometry*
* *First normalise your image e.g. in TAL space -so you make lots of different brains look similar and comparable using specific algorithms*
* *Likelihood a voxel is grey or white matter is then determined based on prior knowledge*
* *then you can measure the size of the regions based on this*
42
how does white matter tractography assess the shape and size of white matter nerve fibres
One use of strucftural MRI's in cognitive nerusosicnece research
* measures the rate of water diffusion in each voxel
* *used for clinical reasons – stroke diagnosis*
* *but also used to know how different regions in the brain are connected*
43
What is fractonal *anisotropy*
Its somethign computed in tractography that tells us about teh flow of water in the brain. Index of how much diffusion is restricted. basically computes the flow of water.
* *0 = water is unrestricted – it can flow in all directions*
* *usually seen in ventricles which are filled with CSF or water*
* *in white matter structure the values closer to 1*
* *1 = fully restricted , water can only flow in one direction*
44
Give me a specific type of diffusion imaging (whie matter tractography)
*Diffusion Tensor Imaging (DTI), can both:*
* *measure rate of water diffusion*
* *Determine direction of fluid motion*
* produces very pretty 3D tract images instead of cross-sectional voxel images. Can have different colours to indicate the different directions of movment - e.g. red, left right fibre connection; green, posterior-anterior motion.*
* helps track thepath. ofconnections in your brain*
45
What are the two dimensions we can use to clssify structural MRI experimental designs
observational → expermiental
* observational - often ofer a long time because structural → need time for the changes. totake place
cross-sectional → longitudinal
* measure one time point in brain anatomy
* or have multiple scans seperated by days/weeks and compare these images
46
Example of cross sectional structural MRI study
* Maguire et al., 2000*
* Famous taxi driver study*
* *Got structural MRI of London taxi drivers and compared the volume of their hippocampus*
* *Found **right posterior hippocampal region** = **bigger** in taxi drivers*
* *This increase in volume correlated with the time n were taxi drivers*
* *In addition, **reduced anterior hippocampus** in controls*
47
Example of longitudinal observational structural MRI study
*Woollett and Maguire 2011*
* *Longitudinal observational study*
* *Measured hippocampal volume in London taxi driver trainees – once before and after the training (four years between)*
*Researchers showed – causal relationship. Only in trainees, n who passed the exam saw an increase in posterior hippocampal volume.*
48
What key biological constituent underlies fMRI
Hemoglobin
* ***certain molecule** – a part of red blood cells.*
* *Key thing is that haemoglobin can attach up to four oxygen atoms.*
* *The magnetic properties of this are very important*
* *when H attaches oxygen (oxyginated blood) = its weakly diamagnetic (repelled by magnetic field)*
* *when H not attached (Deoxyginated blood) = its paramagnetic ( attracted to magnetic field)*
49
Why is haemoglobin being attracted to or repelled by magnetic field impotant
* *bc if blood is oxyginated – the magnetic field is NOT distorted, **no signal loss***
* *but if blood deoxyginated - small distortion in the magnetic field, **signal loss, drop in signal***
*Could think of deoxyginated blood as an external object (like hairclip) that we bring into the magnetic field. It is metallic and introduces signal loss, causing artifacts. Not a massive amount but still enough for you to be able to detect it.*
50
How does oxygen in the blood affect the brain scan
* *by introducing deoxyginated H → produces some signal loss*
* *In the end affects T2\* time*
* *decay of transverse magnetisation is influenced by local inhomogeinities of the magnetic field*
* *thats what we measure*
51
What happens. atthe neural level with fMRI
*What happens at the neural level that causes these changes in the magnetic field*
* if neurons fire they consme oxygen (they also consume glucose) basically they need energy
* if they consume oxygen then we should have more deoxyginated haemogoblin in active regions
* consumption triggers a fast response from blood vessels
* compensatory influx of fresh blood that
* flushes out the **deoxyginated** hemoglobian and brings in **oxyginated** hemoglobin
* remmeber in **oxyginated** hemoglobin theres no signal loss
* this means the large influx of **oxyginated** hemoglobin generates a higher signal
52
When neurons fire they consume oxygen. This deoxygination should cause signal to drop. why then do we see signal increase in active regions.
* Because after they have consumed the oxygen (deoxygination) there is an influx of fresh blood.
* This flushes out deoxyginated haemoglobin (BOLD response)
* brings in oxyginated hemoglobin
53
What is the BOLD response?
* The influx of oxyginated blood
* and flushing of out deoxyginated hemoglobin
* the effects of this on the magnetic field picked up by the RF coil
* = the physiological basis of what we measure
54
history of the fMRI
*In 1992*
* *2 competing research groups (Ogawa et al; kwong et al.,)*
* *published the first fMRI paper using the bold signal*
* *So they realised that you can measure these tiny changes in blood flow with MRI*
* *before this was only structural MRI*
* *VERY simple experimental paradigm - checkerboards that went on and off*
* *found they could show these changes in signal intensity*
* could observe brain working in real time - wasnt possible before this unless you did invasive recording or scalp recording (like EEG)
55
time course of BOLD response (influx of fresh blood and how this affects magnetic field and MRI signal).
The % of signal change over time. lasts several seconds
* initial dip in signal (the time you have increased oxygen consumption)
* then have rise. insignal intensity which takes a while - this is really the onset of n processing the stimulus/doing something in the scanner when n
56
What do we mean when we say during the bold response we look at the “% of signal change”
(value - mean baseline value) /mean baseline value
* the intensity value at a given time point minus the baseline intensity value.
* think of it as the mean across different time points before this change when you present for example just the fixation cross
* then divide that difference by the baseline value again
* gives indication. of how much the signal has changed compared to baseline IN RESPONSE to something you askedn to do
this is a very small value - often between 0.5-3% . this means if you ask n to do something in the scanner - it will only induce a tinyyyy change in the BOLD response.
yes its small but enough to measure iw with. anMRI scan
57
initial dip in signal
* transient increase in oxygen consumption before influx of fresh blood
* *smaller amplitude than main BOLD signal*
* *More spatially specific*
* *Potentially a better measure, as oxygen utilisation may be more closely associated with neuronal activity*
* yes its more focal but its quite difficult to see this in your signal
58
Rise in signal
* begins soon after stimulus is shown
* influx of fresh blood - takes a while
* *inflection point can be used to index the onset of processing/task engagement*
59
signal peak , overshoot then saturation
* the % signal change peaks after 4-6 seconds
* after presenting stimulus
* called the hemodynamic delay
* the signal with fMRI is delayed by 4-6 seconds
* the blood resposne takes time
Overshoot
* overcompensatory influx of blood
Then have saturation
* signal doesnt change anymore
* the stimulus is still on
* acc the signal does not return to baselinei f you dont turn stimulus off
60
post stimulus undershoot
after switching sitmulus off
* signal falls
* supressed before baseline (undershoot)
* takes while to return to baseline
* undershoot is inconsistent across participants
* so usually we focus on the main signal in the analysis
61
what is the best spatial and temporal resolution we can achieve wittH fMRI technique?
* *Spatial resolution = voxel size = 1mm*
* *Temporal resolution = 1second*
62
what is a voxel?
* basic element of brain scan
* 1 cube within the brain scan
* 1 signal value (1 number)
* 1 shade of grey
* All those voxels form 1 slice and then you have different slices stuck on top of each other to form 1 volume.*
* The number of slices you have depend on the thickness of your voxel. You can play around with parameters. Based on how you set up imaging sequence - how you define your TR and TE – affects your voxel resolution*
63
What is “matrix size” in fMRI
* *simply the number of voxels in the x and y dimensions*
64
what do we mean by “field of view” in MRI
* Simply put it is the matrix size of your slice in cm.*
* normally you have about 200-300 slices*
65
why does structural and functional MRI look different?
structural - T1 weighted imaging; 1 volume
* *end up with a lotttt of voxels*
* *256 x 256 = 7, 077, 888 voxels*
* *voxel size of 1x1x1 mm , 1mm cubed*
* *this would give us a high spatial resolution*
* keep in mind this would all give us one time point, 1 volume only*
* takes about 5-10 mins to acquire. abrain volume like this with such high resolution*
* functional - T2\* weighted functional imaging; 1 volume aqcuired within 1 TR*
* matrix size = 64 x 64
* 32 slices
* 131, 072 voxels
* 3 x 3 x 3, aka 3mm cubed isovoxel resolution
* lower spatial resolution
66
What do we mean by 4th dimension in functional MRI
* Contrast to structural MRI, structural fMRI has a fourth dimension. Time.*
* For example you could have the whole thing last 5 minutes (300s) with a TR os 2s. owuld give us 150 volumes. Could have 5 runs of this. 5 x 150 volumes. fuck load of voxels.*
* Another way to think about time*
* each voxel has its own time course. Would have 131, 072 time courses
* dont need to look at time course of all voxels - some wont be interesting
* can exclude voxels outside the brain
67
how do. weactually construct the brain image?
Using text editor
* *its just a matrix of intensity values*
* *Basically each number is associated with a certain shade of greyness/whiteness/blackness value and that then creates the brain image.*
* *uses this to create brain images.*
68
We can manipulate the spatial resolution of the brain image we want. what affects the decision you choose?
Larger voxels - better
* better signal to noise ratio
* this is because small voxels are more suceptable to things like head motion
* smaller voxels take longer to acquire the same brain volume: more costly, v expensive to rent the scanner for 1 hour need to facture in the duration of scanning.
Smaller voxels - better
* larger the voxel the larger the risk of ***partial voluming***
* this is when the signal value obtained in the voxel reflects *also the signal of other functional regions/tissue - other things leak into the signal*
* BOLD signal only detectable in grey matter, if *signal does include white matter / fluid bc of partial voluming – signal is watered down. Becomes worse.*
69
in which tissue do we see the fMRI signal
*the fMRI only occurs in grey matter - BOLD response*
70
fMRI compared to other techniqeus
spatial resolution
* ranges between 1mm and 10cm
* better than EEG, TMS, PET
* determined by the strength of the magnetic field
* some scanners can produce a magnetic field of more than Tesla - here can go down to 0.5mm.
* with thank sort of resolution you could look at things like ocular dominance columns (fine structures within visua cortex telling us something about visual perception; Cheng et al., 2001)
Temporal resolution
* fMRI is much worse - in the supra-second range
* generally looks at processes with a resolution above 1s
* EEG - in sub-second range
* can track things below 1
71
What limits the temporal resolution of fMRI
72
*So, if we have a hemodynamic delay of 4-6 seconds. Then how is it possible to observe things below that rate.*
bc we can *We can disentangle those processes*
* bc BOLD signals from multiple sources of activation (stimuli)
* they summate linearly
* so we can calculate backwards from the BOLD signal
* and disentangle the different things that contributed to it
* evident in the paper by Dale and Buckner (1997)
73
What did Dale and Buckner show in 1997
that you could disentangel from the BOLD signal the different stimuli that contributed to it.
* *presented multiple stimuli close together (2s apart).*
* *after each trial added 1 more stimuli*
* *found that if you compare the BOLD response after each of these the activity collected stacks up on each other.*
* *They summate linearly.*
* *Good news for us because we can disentangle them using certain algorithms.*
74
what causes the limit in temporal resolution
* the physical properties of MRI scanning that limits temporal resolution
* the acquisition rate - how fast we can acquire the imagestakes time to covert he whole brain - need to change gradients and acquire multiple slices within a given TR
* this takes time
* also T1 - the longitinal magnetisation needs to recover before we can excite it again with our RF pulse
75
if you compare the bOLD response with the electrophysiological response (electrodes picing up action potentials)
if we compare data from the two
* time course of the two measuring neural firing is similar but this is purely coincidental
* bc fMRI measures population response
* not individual neural response captures by the BOLD response
76
what are the implications of fMRI picking up a populatin response as opposed to single neural response
* if you introduce a small change to something affecting a large number of neurons e.g. if you change attention/fatigue/expectation
* causes a bigger change in your bold signal
* compared to if you changed something quite dramatically in a small number of neurons e.g., stimulus colour
* most of the time this wouldnt produce a change in the BOLD response because this only affects a small number of neurons
77
what does and doesnt the BOLD signal measure
Does measure
* summed activity from PSP – like what you measure with local field potnetials with EEG.
* neural input in a region and the integration of signals rather than output – that we capture with the BOLD response
* both excitation and inhibition (contributing to input) but excitatory inputs have a larger effect
Doesn’t measure
* AP – the acc firing of neurons
* Neural output
78
Individual differences in the MRI signal
A lot of variability between subjects. Comparing BOLD signal within n over multiple occasions – more consistent however there is still some fluctuations still.
Aguire et al., 1998
* different n, same stimulus
* same n same stimulus different scans
* same n, same stimuli, different days
revealed
* Differences between n
* Differences within n – especially if done on different days
So if you have a repeated measures design – scan n on the same day.
TECHNIQUES
flashcards
Decks in class (7)
# Cards
