Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

Neuroimaging 1 > Structural Imaging > Flashcards

Structural Imaging Flashcards

1
Q

What does contrast allow for?

A

Lets us visually differentiate between different types of fluid, tissue and bone

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

What are the properties of high and low MR signal?

A

High MR signal shows up as bright
Low MR signal shows up as dark

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

What are the major types of structural brain imaging?

A

T1- weighted
T2-weighted
FLAIR

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

What are the characteristics of T1-weighted imaging?

A

Shows intensity of T1 signal (contrast comes from differences in tissues T1 relaxation times)

Highlights:
CSF = dark
White matter = very bright
Grey matter = quite bright

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

What are the characteristics of T2-weighted imaging?

A

Shows intensity of T2 signal (contrast comes from differences in tissues T2 relaxation times)

Highlights:
CSF = bright
White matter = darker grey
Grey matter = brighter grey

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

What are the characteristics of FLAIR?

A

FLAIR images are T2 weighted scans where an ‘inversion’ RF pulse in the scan additionally cancels signal from CSF (making it dark instead of bright)

This makes it particularly good for picking out certain types of brain pathology

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

What determines whether the scan is picking up more T1 or T2 signal?

A

Determined largely by the sequences “repetition time” and “echo time” which are written as “TR” and “TE” with a number following denoting that time in milliseconds

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What length of TR/TE times create which image?

A

Short TR/TE times will create heavier T1 weighting
Long TR/TE times will create heavier T2 weighting

Standard T1 and T2 scans do not literally show only T1 or T2 signal, they are just weighted much more heavily towards one or the other - the voxel values in these images are arbitrary

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Can scans be created which show actual T1 or T2 times?

A

Yes - these are called quantitative T1/T2 or T1/T2 mapping

Here the value in each pixel is the actual relaxation time for the tissue present in that location

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

What is quantitative T1/T2 mapping useful for?

A

Can help emphasise certain features like the evidence of a stroke

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

What other structural imaging methods are there which are not really used in clinical neuroimaging?

A

T2 star
Susceptibility-weighted imaging (SW)
Proton density

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

Why do we have different types of scan?

A
  1. T1 weighted images have good tissue contrast - tend to be the gold standard for showing healthy anatomy
  2. T1 scans have modest ability to show pathology - T1 scan does not differentiate well in colour from CSF
  3. Other types of scan are geared towards highlighting different kinds of pathology - T2 weighting is much better at highlighting white matter lesions as bright tissue, doing this via FLAIR image also makes differentiation from CSF easy
  4. FLAIRs are excellent for investigating a range of brain pathology
  5. Diffusion weighted imaging shows recent strokes before damage becomes visible on other scans
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What dimensions can structural scans be in?

A

2D
- Each slice is an image of a very thin section of tissue
- Anatomical information between slices is lost

3D
- Enabled by the inclusion of the third MR gradient
- All tissue now contributes to the data in the voxels (3D pixels)

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

What is partial volume?

A

Partial volume artefact occurs when tissues of widely different absorption are encompassed on the same CT voxel producing a beam attenuation proportional to the average value of these tissues

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

What is isotropic?

A

When a voxel is the same length in each axis

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

What tools are used for basic processing of structural images?

A

SPM central
FSL
FreeSurfer

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

What does FreeSurfer do?

A

Arguably the more unique option with a particular emphasis on cortical analysis

Delineates between where the white matter is and where the grey matter is

Can tell you how thick the grey matter is

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What is SPM central?

A

Statistical parametric mapping
Run out of matlab

19
Q

What is FSL?

A

FMRIB’s software library
Oxford centre for functional magnetic resonance imaging of the brain

Run using linux

20
Q

What is involved in MRI scan pre-processing?

A
  1. File conversion
  2. Ensuring correct image orientation data
  3. Intensity non-uniformity correction
21
Q

What tools are used for file conversion?

A

Many tools for this, but a popular one is “dcm2niix”

22
Q

How do you ensure correct image orientation?

A

Software tools like to know which way is up and parts of the nifti header can be modified to give this information

In 99% of cases, running the FSL tool “fslorient2std” on a fresh nifti image will add correct orientation information

23
Q

What is intensity non-uniformity correction?

A

The coil worn in an MRI scanner is an essential part of the scan, but creates distortions in the local magnetic field - so does the person’s head

These results in collected images being generally brighter/darker towards the centre which creates noise so correction is needed to create the real image
(Slow and smooth intensity variation across datasets)

FSL will do this as part of its FAST pipeline, for tissue segmentation - although it is a good step to have run on your structural scan before any analysis

24
Q

What are the structural MRI analysis methods?

A
  1. Brain extraction
  2. Tissue segmentation (and tissue volume)
  3. Masks
  4. Image registration
  5. Pulling it all together
25
Q

What is brain extraction?

A

The removal of all voxels which are not the brain
It is better to have the entire brain plus a few voxels of skull or other material than having no skull at all but missing brain

Brain extraction means imaging processing software can run more efficiently as it is not processing unwanted data

26
Q

What is tissue segmentation?

A

Aims at partitioning an image into segments corresponding to different tissue classes

FSL will read a structural brain scan and produce additional images in the same space which show where different major tissue types are

27
Q

What are TPM images?

A

Tissue probability maps or tissue partial volume maps

Good scan visualisation software lets us mount multiple images at once to create helpful collages

The voxel values in a TPM are no longer arbitrary - they now represent the % of the voxel which belongs to that type of tissue

28
Q

Can you measure tissue volume from segmentation?

A

A tissue segmentation image has all the information needed to know the overall volume of a person’s grey matter or white matter

  1. The total volume of all voxels identified as having some GM/WM
  2. The mean value of those voxels

Multiplying these together gives us the person GM/WM volume

29
Q

Why is it useful knowing the volume of someone’s brain?

A

Neurodegenerative conditions lead to brain atrophy - brain volume is a straightforward and intuitive way of measuring this, which structural MRI is perfect for

However, people naturally have different sized brains
Taking the raw brain volume straight from a scan is fine to look within one patient, e.g., comparing the size of their brain 5 years ago to now

However, comparing volume between people needs a normalising factor

30
Q

How do we normalise brain volume?

A

FSL has a pipeline that will give you grey and white matter volumes, normalised to the size of a person’s skull

We can imagine how this gives us a better idea of if a person’s brain has shrunk compared to their own, healthy baseline

31
Q

What is masking?

A

Brain masking of MRI images separates brain from surrounding tissue and its accuracy is critical for the further analyses of imaging data

Manual masking (MM) is generally considered the gold standard, but it is extremely time-consuming, taking up as much as several hours per case

The new image consists of only 1s or 0s (1 if there is white matter present, 0 if none present)

Can be used for white matter and grey matter

It is a way of identifying and defining voxels of particular interest for further analysis

32
Q

What is image space?

A

Images are in the same space when they are in alignment
This means that the same coordinates in each image refer to the same bit of brain anatomy

33
Q

What is image registration?

A

Scans from different people will always be in different spaces to begin with - that person’s ‘native space’
In an imaging study we often want to manipulate scans somehow so that they do share the same space

Image registration is a processing technique that moves, stretches and squishes one image to match the alignment of another

It always involves a target image and the image you are wanting to register to the target

34
Q

What is FLIRT?

A

FMRIBs linear registration

Used to register images from the same person
Up to 12 degrees of freedom - the number of ways the program can move, stretch and squish the image
Every manoeuvre introduces data noise so we want to be as conservative as possible

35
Q

What happens as you increase the degrees of freedom in FLIRT?

A

With 6 DOF = ‘rigid body’ registration - it will shift the brain around in the space of the image in all directions without deforming the actual shape

With 12 DOF - this may result in a scan being upscaled or downscaled - something to consider when thinking about how your data is being affected in the process

36
Q

What is FNIRT?

A

FMRIBs non-linear registration
Used to register different people’s brains together
Non-linear - so parts of the brain can be differently warped in uneven ways

37
Q

What technique does registration use?

A

A cost function to work out how to change an image to match another

This is the mathematical approach to determine how similar or dissimilar the images being registered are and therefore when a good registration has been achieved

Also need a method of interpolation to generate the newly-registered image

38
Q

How are all the methods pulled together?

A

We can start to imagine how combining these sorts of techniques let us process brain images in systematic ways to understand them better

39
Q

How would we measure the size of someone’s hippocampus who has Alzheimer’s?

A

This could be done manually, by literally drawing onto an MRI scan and then using the software to tell us the size of the objects we have just drawn.

But this has many pitfalls; it takes significant training to know where the hippocampus is, and there will be human error in the drawings meaning it is not very repeatable

40
Q

How do we overcome the potential human error when drawing manually on a scan?

A

A template is not a brain scan of a single person, but the average brain scan of multiple people. It can be created by scanning many people, registering their T1 scans into the same space, and then taking a mathematical average.

Templates are very powerful. They can often used as a “target” image in a study to register everyone’s scans into the same space.

They can also be used to pre-draw regions of interest (ROIs), like a hippocampus, so this only needs to be done once

41
Q

What is the most common brain template?

A

The MNI152 brain

42
Q

Explain the process using templates to measure the hippocampus

A
  1. Brain extraction
  2. Brain vs skull estimation (normalising factor)
  3. Automatic hippocampus identification (FSL’s “FIRST) -identifies a value which represents the hippocampus
    = From this you can use different tools to calculate the volume of voxels which have the corresponding value identified by FIRST
  4. Perform on all scans in a dataset and normalise all hippocampi volumes
  5. We can now compare everyone’s hippocampi size controlling for natural brain size variation
43
Q

What are TE and TR times?

A

TE = time to echo
Time between the delivery of the RF pulse and the receipt of the echo signal

TR = repetition time
The amount of time between successive pulse sequences applied to the same slice

Neuroimaging 1 (16 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions