Lecture 6- Basic Image Quality Flashcards

1
Q

what is image quality

A

it is the usefulness of an image in determining accurate diagnosis and it is highly dependent on the diagnostic task

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

what affects image quality

A

source, imaging subject, detection system, processing and or tomographic image reconstruction, display, obesrver

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

what is spatial resolution

A

level of detail that can be seen on an image. human perception of detail is based on edge detection, blurring effect of imaging system softens the edges of objects

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

factors determining spatial resoluton

A

focal spot size, magnification factors, detector sampling distance, detector aperture size, reconstruction filter, number of projections

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

point spread function

A

going from a point to a larger area

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

what are the requirements of linear shift invariant

A

PSF and further spatial frequency domain analysis can fully characterize a system only if the system is linear and shift-invariant

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

LSI system

A

a system that is both linear and shift invariant

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

why is the spatial frequency domain used

A

convinient and efficient, all real objects can be decomposed into sine waves of different ammplitudes, frequencies, and phases. A single analysis in spatial frequency domain can be used to predict perfomance of all possible objects. Computation in spatial frequency domain in easier than just the spatial domain

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

modulation trasnfer function

A

defines the ability of a system to reproduce the amplitude of an input signal at different frequencies

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

what are the methods of measuring MTF

A

point source method, slit method, edge method

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

point source method

A

phantom, small FOV reconstruction, to PSF

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

what is the slit method

A

slit represents a line impulse function, the response of the system to a line impulse function is line spread function.

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

what is the nyquist criteria

A

it is the sampling frequency needs to be at least twice the band limit of the signal to avoid aliasing or the band limit of the signal has to be less than the nyquist frequency to avoid aliasing

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

pre-sampled MTF

A

describes the system response up to and not including the stage of sampling

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

what is noise

A

varaitions in the spatial distribution of image signal that cannot be attributed to anatomic differences

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

what are the sources of noise

A

electronic noise, structure noise, quantam noise

17
Q

what is electronic noise

A

originated from electronic components of the detection system. it is additive to the original signal . usually low compared to the original signal. More significant contribution to image quality when detected number of photons is low.

18
Q

how do you reduce electronic noise

A

detector cooling by decreased thermal noise, noise reduction circuitry, electronics shielding

19
Q

what is structure noise

A

originated from spatial variation of detector pixels. Dark field signal or gain characteristics

20
Q

how can you correct structure noise

A

dark field calibration and gain calibration.

21
Q

does structure noise change over time

A

yes so need to perform routine calibration

22
Q

what is quantum noise

A

uncertainty in photon counting statistics. Xray coutning statistics obey the Poisson distribution. Standard deviation is the square root of the mean number of photons. When N is large, approaching Gaussian

23
Q

for photon counting statistics, noise level of the counted photons increases as the mean number of photons increases- T/F

A

true

24
Q

noise measurement

A

a simple practical measurement of noise is to calculate the standard deviation of a group of image pixels

25
Q

what is noise as standard deviation

A

appropirate for simple discussions of image quality

26
Q

standard deviation is a complete or incomplete measure of image nise

A

incomplete

27
Q

what is the noise power spectrum

A

more complete descriptor of image noise. defines both the magnitude and spatial frequency characteristics of image noise Noise power spectrum can be amount the various frequency components of the image. Integration of NPS over all frequencies yields variance.

28
Q

contrast

A

relative difference between two signal levels

29
Q

intrinsic factors for subject contrast

A

physical properties of the subject relative to background tissues- thickness: Z and P

30
Q

extrensic factors for subject contrast

A

how the image acquisition parameters can be optimized to enhance subject contrast.

31
Q

what is detector contrast

A

the shapre of the characteristic curve that shows how the detector maps input energy to output signal

32
Q

what is contrast to noise ratio

A

a simple measure of signal and noise. Useful metric to compare image quality hen other factors are fixed ( noise correlation, spatial resolution, and object size)

33
Q

what is a contrast detail diagram used for

A

used to subjectively determine the minimum detectable object size at different contrast levels

34
Q

what does CNR not consider

A

no consideration of object size, noise correlation, and spatial resolution. Not good metric for absolute detectability evaluation.

35
Q

what is better than CNR

A

SNR defined using statistical decision theory and task based approaches is more reusable metric for quantifying object