Data Analysis - MRI/fMRI/DTI Flashcards

1
Q

What are examples of structural data quality?

A
  1. DTI - spin echo EPI
    - distortions (cm)
    - ghosting
    - resolution ~mm
  2. Structural MRI
    - distortions (mm or less) in readout direction
    - resolution 1mm (isotropic)
  3. CT
    - little distortion
    - faster to acquire than structural MRI
    - poor brain tissue visibility
    - radiation dose
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What needs to be considered, especially in clinical decision making?

A

When combining data from different modalities - limitations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Why is putting data together from different modalities potentially very powerful?

A

It allows the understanding of structural and functional data as a whole

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What does encoded FA map in DTI give?

A

The direction of white matter orientations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What regions does DTI suffer from distortion?

A
  1. Frontal region

2. Temporal regions

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Structural MRI

A

Much higher resolution - see details in the stratium and detail in the cortex and white matter

Good for localising different parts of the cortex

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What are examples of functional data quality?

A
  1. fMRI - gradient echo EPI
    - distortions (cm)
    - dropout
    - resolution ~mm
    - temporal resolution (seconds, HRF + MR limits)
  2. ESI -EEG source localisation
    - resolution (cm ‘inverse problem’)
    - resolution/ visibility location dependant
    - high temporal resolution (ms)
    - attenuation didn’t to head structure
  3. MEG
    - resolution poor (cm ‘inverse problem’)
    - resolution/ visibility location dependent
    - high temporal resolution (ms)
    - orientation dependent
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is gradient echo EPI data quality: dropouts?

A
  1. Tissue in B0 - magnetisation grows along z
  2. RF pulse > magnetisation in x/y plane
  3. Magnetisation rotates w=yB0
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is the gradient echo EPI data quality: dropouts?

A

Looking just after the application of the RF pulse that flips onto the x/y plane

Looking at different bunch of spins but have the same magnetic field - there is no Variation in the magnetic field

The spins for now we will assume they are not interacting with each other and is purely independently

Over time, they are processing at the same frequency - strong signal which will not decay

There will be a slow signal decay along T2*

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What do distortions affect?

A

Both gradient and spin eco - this acquires k-space

Alteration in k-space leads to local distortion
You can acquire field map in a separate image - that will help undo some of the distortions

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Where does the drop out occur?

A

Between RF pulse and acquisition

That is in the gradient echo

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Where does distortion occur?

A

During acquisition itself and that is true for both spin echo and gradient echo even though the gradient echo has the worst distortion

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What can be used for distortion correction?

A

In Anterior-posterior (AP) there is compression whereas on the other side it is stretching it out

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What are the two options for distortion correction, both of which require acquisition of additional Images?

A
  1. Field maps

2. Images acquired with negated phase encode direction

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is distortion correction?

A

It may not be possible to fully-correct regions where signal had been compressed into a small area

If possible, it is preferable to minimise distortion when setting up the acquisition protocol
- limit echo spacing to reduce Accrual of phase

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is used in invasive monitoring of epilepsy patients?

A

Depth electrode

The images were acquired on a phantom containing an electrode sometimes implanted in the brain of epilepsy patients for seizure localisation

17
Q

Why is gradient-echo EPI susceptible to drop out ?

A

Due to local inhomogeneities in the magnetic field

18
Q

What suffers from distortions caused by accumulation of phase during read-out?

A

Both gradient-echo EPU and spin-echo EPI

19
Q

What can both gradient-echo EPI and spin-echo EPI suffer from?

A
  1. Local distortion due to air-tissue boundaries (e.g. in frontal and temporal areas), inconsistencies in main magnetic field, foreign objects (e.g. dental work)
  2. Can minimise effect by minimising echo spacing (reducing time available for phase accumulation) during acquisition
  3. Can correct using a field map or second image with reverse encoding ( but cannot completely recover image where signal is compressed into a small region)
20
Q

What are the practical considerations for functional MRI?

A
  1. Tasks
    - task too easy/difficult?
    - task type
    - is the participant doing what you want them to?
  2. Accessing task performance
    - feedback
    - measure responses
    - video
    - eye tracking
21
Q

What do we not measure directly in fMRI?

A

Brain activity

We measure changes in blood oxygenation

22
Q

What are two considerations of functional MRI?

A
  1. Cardiac noise
    - heart rate (relative to TR)
    - BOLD changes due to cardiac cycle
    - pulsation-related effects
    - task related effects
    - task related changes to heart rate
  2. Respiratory noise
    - breathing changes blood oxygenation and flow
    - breathing changes local fields
    - task related changes in breathing
    - interaction between heart rate and breathing
23
Q

What is fMRI: processing and analysis?

A

Potentially get rid of any distortion on and realign images

Form normalisation - our subject into a template space - 2 step process: align them to their structural image and align that to a template m

You can smooth the data and use design/model to get parameter estimate

You get a lot of boobs - some of them would be real and some Aren’t - there will be plenty of false positives given the number of multiple comparisons performed

Few boobs to pass significant threshold

The main problems are normalisation and statistical inference