0:01

Measures of image quality can be looked in two ways.

0:04

One is the spatial resolution, the other one is contrast

0:07

resolution and noise.

0:09

This is how physicists measure the image quality on

0:13

CT from a physics standpoint of view.

0:16

Let me talk about the spatial resolution first.

0:20

Usually, we use a phantom, and this is an American College of

0:23

Radiology CT accreditation phantom.

0:27

Any site which is accredited by the ACR

0:31

uses this phantom, which has different modules and

0:34

different objects are embedded in the module.

0:38

A medical physicist scans this

0:41

phantom, and then he or she can visually

0:45

measure the image quality or assess the image quality.

0:48

I'm going to talk about just the spatial resolution aspect in this

0:52

module.

0:53

If we just image the phantom and look at the spatial resolution

0:57

module, this is how it appears. And

1:01

the way we define spatial resolution is the ability of the

1:04

imaging system to resolve small, independent

1:08

objects in close proximity to one another.

1:12

The best way to assess an MSCT scanner image quality in

1:16

terms of spatial resolution is to scan an object of

1:20

different line pair per mm. These are objects

1:24

embedded in the phantom, which has line pair means each

1:28

pair is one white and dark is called one line pair.

1:31

They are made up of solid attenuating material so that X-ray

1:35

passing through it create this white and black pattern.

1:39

The spacing depends on the spatial resolution.

1:42

So the larger objects are this is the highest,

1:46

like lowest spatial resolution. If you keep going down here,

1:51

this is four line pair per mm. Four

1:54

line pair per centimeter means

1:58

in 1 centimeter, there are four line pairs.

2:00

Means black and white is one line pair.

2:03

So one, two, three, four is line pair.

2:07

This is how we count. Since we know this module here, what

2:11

these line pairs are made up of, we can actually say

2:15

what is the spatial resolution result.

2:17

This is four line pair, five, six, seven, eight.

2:20

By eight, we begin to lose the capability of the resolution of this particular

2:24

scanner. Therefore, we can say this

2:27

scanner has a capability to resolve up to seven or

2:31

eight line pair per centimeter.

2:34

There are a number of factors which affect the spatial resolution, and they

2:38

are as follows. When I say detector aperture size,

2:42

this is the detector in the Z direction, what's the size of it?

2:46

And that's usually listed when we look at the protocol, the

2:50

number of detectors, the dash channel thickness is what the detector

2:54

aperture thickness. The second factor is the reconstructed

2:58

slice thickness. Again, this is one of the secondary factor

3:02

which influence the image quality.

3:04

So once you acquire data, we can reconstruct in different

3:08

slice thickness to vary the image quality.

3:11

The other factor is the focal spot size, number of projection,

3:15

and reconstruct algorithm. So if you look here, this

3:19

is the phantom spatial resolution, 4

3:23

to 12 line pair per centimeter. And you can see

3:26

correspondingly in this particular scanner, which is scanned around

3:30

the 120 kV, 300 mAs, a typical

3:34

abdominal CT protocol. This is resolving four,

3:38

five, six, seven line pairs per centimeter

3:42

perfectly. And the accreditation requires the

3:46

scanner to resolve at certain level for a different protocol by which

3:50

the scanner is deemed accredited or passes.

3:53

In the axial plane, the XY resolution in the

3:57

axial plane depends on the image matrix and the display

4:01

field of view and the pixel size. This has remained same throughout

4:05

from the time CT were developed back in 1974.

4:09

The main thing which has changed the past few years is with the multiple row

4:12

detector, the longitudinal Z direction is what is

4:16

determined by the detector array thickness, and that's measured either

4:20

by line pair per millimeter or line pair per centimeter.

4:24

Typically, in CT, we talk about line pair per centimeter,

4:28

and in mammography and in radiology, we measure in terms of line pair per

4:32

millimeter because in mammography, the spatial resolution is even more higher.

4:36

So

4:37

what is the trade-off between the image and the slice thickness is what

4:41

demonstrated here. In this particular one, ACR

4:45

phantom, there is also embedded materials which will tell

4:49

you how many materials you can detect in the axial plane.

4:53

This will tell you there are three objects can be seen, and this

4:57

is of an image which is of a slice thickness of

5:00

1.25.

5:02

1.25 slice thickness shows only three

5:05

object that is very noisier. And you can see here, as

5:09

the slice thickness is increased, that is the reconstructed slice thickness is

5:13

increased, we can see on the phantom more object.

5:16

We can simply count the resolution.

5:18

One millimeter resolution is considered as one set of these boxes can

5:23

one, two, two and a half. This is 1.25, and so

5:26

forth. So for a 10-millimeter slice thickness, you can see all these

5:30

object correspondingly slice thickness and image noise is different.

5:34

So with the multi-detector CT, the thinner

5:38

images always produce higher spatial resolution,

5:42

and it has more noise. That's the trade-off between.

5:45

And that it'll also have less partial volume effect.

5:49

The same thing can be looked here, how the spatial

5:52

resolution can be improved by reconstructed thinner and

5:56

thinner slices. That's where the CT technology

5:59

evolved from thick slices to thin slices, because

6:03

thinner the slices is higher the spatial resolution.

6:07

As you can see here, this is a 0.625-millimeter

6:11

slice thickness. You can resolve all these fine object.

6:15

Comparatively to 1.5, we are beginning to lose this

6:19

area. At two and a half, we can hardly notice the differences here.

6:23

So this spatial resolution will improve with

6:27

smaller slice thickness.

6:30

There are some trade-off in the spatial resolution.

6:33

One is if you keep all the material constant, like

6:37

same tube voltage and the tube current, the number of

6:41

detector photons varies linearly with slice

6:45

thickness. So the trade-off is here.

6:48

Thinner slices provides higher spatial resolution,

6:51

but image noise will increase as shown earlier.

6:56

The thicker slices provide higher contrast resolution, but

7:00

poor spatial resolution and less image noise. There is a trade-off.