An Overview of Medical Image Segmentation Flashcards

1
Q

What are many different types of objectives for segmentation?

A
  1. Identify grey matter, white matter and CSF
  2. Identify tumours, strokes etc
  3. Identify different brain structures, hippocampus etc
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are the many different types of iamges?

A

Segmentation quality depends on what is visible in the images

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

What is region growing?

A

Approach available within many software tools

Define an approximate region within a structure
Scribbles, points, etc
Maybe define some regions outside the structure

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

What does algorithm try to find?

A

A suitable boundary

Lots of different strategies involved

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

What is classification?

A

Rather than classify individuals (e.g. patient vs. control), the aim is to classify voxels (e.g. hippocampus vs. not hippocampus).

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

What are the information that can be combined probabilistically for classificatiion?

A

Voxel intensity
Position within the image
Classes of neighbouring voxels
Texture

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

What is the probability for classifitis (Fatal disease afflicting one in every thousand people)

A

Out of 1000 people, 999 will not have classifitis.
Of these, 5% will have a positive result – about 50.
One person will have classifitis, with a 99% chance of having a positive result.

About one positive result from someone with classifitis.
About 50 positive results from people without classifitis.
A positive result means about a 2% chance of having classifitis.

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

What does the Bayes rule say?

A

P(C+ | T+) = P(T+ | C+) P(C+) / P(T+)
= 0.99×0.001 / (0.99×0.001 + 0.05×0.999)
= 0.0194  2%

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

What is tissue classification?

A

Whereby voxels are assigned to a tissue class according to their intensities the intensity distribution ofeach tissue class needs to be characterised, often from voxelschosen to represent each class.

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

What is another approach?

A

involves some kind of registration, where atemplate brain is warped to match the brain volume to besegmented (Collins et al., 1995).

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

Why is automated and reliable intensity-based tissue classification complicated?

A

Spectral overlap of MR intensities if different tissue classes and by the presence of a spatially smoothly varying intensity inhomogeneity or bias field

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

What is iterative expectation-maximisation (EM) procedures?

A

Interleaved tissue classification with estimation of tissue-class-specific intensity models and bias field correction

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

What does initialisation of the iterative process use?

A

Digital brain atlas with a priori probability maps for the different tissue classes

Avoids all manual intervention

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

What is the Markov random field (MRF) designed to facilitate?

A

Discrimination between brain and non brain tissues, while preserving the detailed interfaces between the various tissue classes within the brain

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

What does the algorithm iteratively interleave?

A
  1. Voxel classification
  2. Intensity distribution parameter estimation
  3. MR bias field correction
  4. MRF parameter estimation
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What are MR images?

A

Theoretically piecewise constant with a small number of classes

Have a relatively high contrast between different tissues

17
Q

What does the contrast in an MR image dependent upon?

A

The way the image is acquired

18
Q

How is it possible to highlight different components in the object being images and produce high contrast images?

A

By altering radio frequency and gradient pulses, and by carefully choosing relaxation timings

19
Q

What two features facilitate segmentation?

A
  1. Altering radio frequency

2. Gradient pulses

20
Q

What is the piecewise constant property degraded considerably by?

A
  1. Electronic noise
  2. Bias field
  3. Partial volume effect
21
Q

What does the piecewise constant property cause?

A

Classes to overlap in the image intensity histogram

22
Q

What do many T2-weighted and proton density images have?

A

Low contrast between GM and WM

23
Q

What is widely employed for brain MR image segmentation?

A

Statistical approaches

Labels pixels according to probability values which are determined based on the intensity distribution of the image

24
Q

What does statistical approaches attempt to solve?

A

Problem of estimating the associated class label, given only the intensity for each pixel

25
Q

What is the most widely used models in segmentation?

A
  1. Finite mixture (FM) model

2. Finite Gaussian mixture (FGM) model

26
Q

What is the limitation of FM model?

A

Spatial information is not taken into account because all the data points are considered to be independent samples drawn from a population

Produce unreliable results

27
Q

What is the typical use of convolutional network?

A

Classification takes, where the output to an image is a single class label

28
Q

What does the desired output include?

A

Localisation I.e. a class label is supposed to be assigned to each piece

29
Q

what are the disadvantages of U-net?

A
  1. It is quite slow because the network must be run separately for each patch, and there is a lot of redundancy due to overlapping patches
  2. There is a trade-off between localisation accuracy and the use of context
30
Q

What is the main idea of fully convolutional network?

A

Supplement a usual contracting network by successive layers, where pooling operators are replaced by unsampling operators

These layers increase the resolution of the output

31
Q

What are the features of SPM12 method?

A
  1. Less accurate
  2. Intermediate speed (Mins)
  3. Generalises fairly well to new scan contrasts
  4. Involves image registration
  5. requires relatively little labelled training data
  6. Partially handles pathologies (lesion etc)
  7. A bit interpretable
32
Q

What are features of label-propagation?

A
  1. More accurate
  2. Slowest (hours)
  3. Generalise poorly to new scan contrasts
  4. Involves image registration
  5. Requires relatively little labelled training data
  6. Does not handle pathologies
  7. A bit interpretable
33
Q

What are features of CNN?

A
  1. Very accurate
  2. Very fast on GPU(secs)
  3. Generalised poorly to new scan contrasts
  4. Does not involve image registration
  5. Requires lots of labelled training data
  6. Handles pathologies well