fMRI Flashcards
What is fMRI also known as?
Blood Oxygenation Level Dependent
Magnetic Resonance Imaging (BOLD MRI)
What are activation maps?
Use MRI to produce activation maps in reponse to a specific task condition = visualisation of which areas of the brain are activated
Aligns with the main goals of fMRI and neuroscience more broadly
What is the simple mechanism of fMRI?
fMRI BOLD signal is altered due to increase of blood flow in response to brain activity
This brain activity can be the result of a stimulus or due to resting state
Describe the basic process of fMRI
Stimulus/modulation in back ground activity
> Neuronal activity
>NVC
> Haemodynamic response
> Detection by MRI scanner
>fMRI BOLD response
Briefly describe Mosso’s experiment
Built a balance which individuals could lie on
> balance remained perfectly balanced until the individual engaged in some type of emotional or cognitive process
>here the balance tipped down on the side of the head as the blood flow redistributed in the brain
What did Mosso’s “human circulation balance” establish?
The relationship between blood flow and neuronal activity
How can a computer in the fMRI setup be helpful if you are using a response box to record participant responses?
Computer can receive feedback from the response box and this response will be synchronised in time with the scan
How does blood flow change the fMRI signal?
The BOLD sequence isn’t measuring the blood flow, but is sensitive to the different magnetic properties of oxygenated and deoxygenated red blood cells
The increased magnetic field in the deoxygenated blood causes a magnetic field gradient resulting in adjacent water molecules precessing at different frequencies.
As water molecules are precessing at different frequencies in the gradient field, they dephase (i.e. spread out in the x-y plane).
The variable dephasing results in signal loss around deoxygenated blood compared to oxygenated.
What is magnetic suceptibility?
Characterises the magnetic field
It’s the degree of magnetization in an object in response to a external magnetic field
What is the difference between diamagnetic and paramagnetic?
Diamagnetic = slightly reduces the magnetic field and this is caused by oxygenated blood
Paramagnetic = adds to the magnetic field and is caused by deoxygenated blood
What does it mean if there is an increase in the magnetic field?
Protons speed up
Related to the Lamour equation which shows that frequency of proton precession is proportional to the magnetic strength
How is the T2* signal affected at rest? Activation contrast
At rest oxygenated blood travels through the arteries exchanges in the capillary bed and deoxygenated blood leaves via the veins
The deoxygenated blood increases dephasing, reducing T2* signal.
How is the T2* signal affected during activation? Activation contrast
During activation an excess of oxygenated blood flows into the activated region, swamping the deoxygenated blood
The excess of oxygenated blood reduces dephasing, increasing the T2* signal.
Why is there an excess/oversupply of oxygenated blood during activation?
Still unclear but:
-could be related to supply of other nutrients e.g. glucose
-could be to provide a strong enough concentration gradient across the capillary wall so that the oxygenated blood can diffuse efficiently
-or could be related to blood flow being controlled at the arteriole level and this leads to an oversupply in smaller vessels
How can we get the optimum echo time when considering the task signal and rest signal in terms of exponential decay?
The longer the echo time is, the bigger the gap between the task signal and rest signal is, when this gap between these is too big, there is no longer any signal to detect
There is also a noise level as well which, once reached, means that the signal would be indistinguishable from the noise
Need a compromise between getting enough difference in signal between task and rest signal before they decay and not hitting the noise floor, so we can find an optimum echo time mathematically to do this
What are we imaging with fMRI?
Imaging voxels are an average signal over mm3, typically 4 x 4 x 4 = 64 mm3 in fMRI
An fMRI voxel will contain many different vessels, but these vessels still only make up around 3% of the voxel volume
Why is it important to consider macroscopic vs microscopic changes in fMRI?
Within imaging we tend to image at macroscopic levels so on the order of mms but theres a lot going on at the microscopic level too
We might not see an effect that’s expected- this could be that there is something going on at a microscopic level e.g. something could be happening to drown out the BOLD signal
Important to consider that we aren’t just imaging BOLD and vessels, other things also occur which may explain why you don’t see a positive result in an exp where you would expect there to be one
Where we move around K space is dictated by what?
The area under gradients
What is an issue with acquring the image through k-space?
Its too slow for fMRI
What method can we use to measure the BOLD response instead that is faster?
EPI- echo planar imaging
How can echo planar imaging help us quickly capture brain volumes quickly?
It means the whole of k-space is acquired after a single 90 degree pulse – single shot imaging
-we can image the BOLD effect with a spin echo sequence
So we can acquire a single slice in less than 100ms and so whole brain volume can be acquired within a second or two
What is an issue with BOLD EPI images?
They are low quality images and often have artefacts
Why are BOLD EPI images such low quality?
Because we acquire the echo signal after one excitation, we don’t get a lot of signal at the two extremes, and those spaces at the edge of k space are what give a sharper image with more detail
What is the order of acquiring the slices in whole brain coverage?
Slices are presented continuously but often not acquired in this way
Often we interleave slices when exciting them
What is the difference between sequential and interleaved slice aquisition?
Sequential slice acquisition acquires each adjacent slice consecutively, either bottom-to-top or top-to-bottom. Interleaved slice acquisition acquires every other slice, and then fills in the gaps on the second pass
Why do we interleave slices when acquiring images using the 90 degree pulse?
When we excite the slices with the 90 degree pulse, it doesn’t excite the whole slice, it bleeds a little into the adjacent slice so this means the first excitation effects the adjacent slice
So we interleave them so that by the time we return to that neighbouring slice the signal will have returned to a more normal baseline
What are some advantages of interleaved slice aquisition?
Increased signal to noise
Post processing is not required to remove distortion
Why do we normally acquire a few dummy scans?
Takes a few cycles for the magnetisation to become steady between acquisition
Tend to specify 3 or 4 dummy scans first to allow the signal to steady itself
What do we want to do once we’ve acquired these images and got the time-series for these images?
Plot how the signal varies voxel by voxel across the whole time series
What are the two main tequniques for modelling the BOLD signal?
Linear regressions (i.e. GLM)
Non-linear regression
Why do we want to model the data?
We want to fit a model to the data to analyse which voxels show significant activation
What is the easiest way to model the data?
Apply a simple model (e.g. boxcar- rest then activation then rest) to the data
If this is a poor fit we can always shift the timecourse a little to fit the data better
What is a better way to model the data than arbitrarily shifting the data?
Could try to model the physiology
Done using the haemodynamic response function which characterises the response to an impulse stimulus
Haemodynamic response function characterises what our BOLD signal is going to do in response to the impulse stimulus - these have been measured experimentally
What does a basic haemodynamic response look like?
We have an initial undershoot
Then we get the peak, followed by a decay, and another undershoot
Once we have the haemodynamic response function what do we do?
Convovle the HRF with our fMRI experimental model to produce a more realistic estimate of the expected signal change
If there is still a residual signal drift, what can we also do?
To account for the signal drift over time, we can add a linear ramp to our model
this then fits the voxel timecourse reasonably well
