Entrada y salida de datos

2. Increasing either the mA or the scan time.

Both of these involve increasing the patient dose. To minimize patient dose the radiologist must accept the noisiest picture consistent with good diagnosis.

•

Other factors increasing noise:

ƒ Zoom enlargement; "spreads available ray sum information over pixel matrix"

ƒ Narrower window width "each grey level covers a smaller range of CT numbers, i.e. derived from the absorption of fewer X-ray photons in each voxel" ƒ Reducing the scan time or reducing the slice thickness, unless the

mA is increased proportionally.

ƒ Deficiency of photons →; occurring with: 8 Thicker patients.

8 High-attenuation materials such as bone or prostheses in the slice.

Spatial resolution of high-contrast objects

High-contrast spatial resolution is good, being determined by Pixel Size.

• In CT terms 'high contrast' is between water and Perspex (about 12%). • Spatial resolution may be tested with BAR PHANTOM having a range of different

line pairs per millimeter.

ƒ As CT scanning smooths out the detail within each voxel → detail within a voxel is not imaged → ∴ 2 pixels are needed to define a line pair

∴ Resolution is about 1 lp\mm → much poorer than film- screen radiography.

High-resolution imaging

S By increasing matrix size or reducing field of view → decreasing pixel size. S Below a certain pixel size, spatial resolution is further limited by:

1) Size of the focal spot. 2) Collimators.

3) Number and size of detectors.

4) Spacing between detectors. 5) Patient movement.

Spatial resolution of low-contrast objects

Low-contrast spatial resolution is less good and is limited by the NOISE.

46

• A low-contrast structure may need to be 5-10 mm in diameter to be resolved. • Low-contrast spatial resolution is assessed by imaging a slice though a Perspex

phantom with water-filled holes of different diameters and different depths → providing different levels of contrast.

ƒ A graph is plotted "Detail-contrast diagram" showing the minimum contrast needed to see structures of different diameters. ƒ The solid curve in Fig. 4.18 shows how the spatial resolution depends on the contrast of the image.

ƒ The dashed curve shows the improvement in low-contrast perceptibility produced by increasing the mA, or dose per slice, and so reducing the noise.

Contrast resolution

'The ability to detect small differences in the attenuation coefficient of adjacent structures'

Contrast resolution is tied to the SNR

• The contrast between a structure and its surroundings is ONLY detectable if it is → 3-5 times greater than the noise in the image.

The more pixels a structure occupy → ↓ noise → better contrast resolution.

With a structure 10 mm in diameter, differences of 4-5 CT numbers "0.5% difference in attenuation coefficient" can be detected → at least 10 times better than can be achieved in film-screen radiography.

47

ƒ Not obscured by overlying bone. ƒ Smaller Scatter.

ƒ Windowing allows quite small differences of CT number to be selected from the full range and displayed over the whole grey scale.

Resolution compromise (trade-off)

• Although a spatial resolution of 1 lp mm-1 and a contrast resolution of 0.5% are quoted for CT → they cannot be achieved at the same time, as Fig. 4.18 makes clear.

• It is not possible to achieve excellent spatial and contrast resolution simultaneously, except by delivering an unacceptable dose to the patient.

• In fact: Well-established relationship among Noise (N), pixel dimensions (Δ), slice thickness (T), and radiation dose (D):

or alternatively,

Accordingly,

• To improve contrast delectability by a factor of 2 involves increasing the dose by a factor of 4.

• To improve spatial resolution by a factor of 2 involves increasing the dose by a factor of 8.

• To halve the slice thickness without impairing image quality involves increasing the dose by a factor of 2.

In conclusion:

Compared with x-ray radiography, CT has significantly worse spatial resolution and significantly better contrast resolution

• Limiting spatial resolution for screen-film radiography is about 7 lp/mm; for CT it is about 1 lp/mm

• Contrast resolution of screen-film radiography is about 5%; for CT it is about

0.5%

Dose

• The distribution of absorbed dose in the body section imaged is much more uniform than in conventional radiography.

ƒ With a single radiograph of the skull, if the entry skin dose is 100%, the exit skin dose might be 0.1% and the central dose 3%;

ƒ In CT, the skin dose is more or less uniform all round. The central dose in the

3 2

_.

.

1

Δ

∝

N

T

D

48

• Due to scatter from one slice into adjacent slices, the dose increases with the number of slices, but not proportionally.

• Although the detector collimators are set to the nominal slice thickness, the actual X-ray beams overlap, as their width is much greater, being determined by the collimation near the tube.

• effective dose is directly proportional to the tube current and total scan length (product of the slice thickness and total number of slices)

The CT dose index (CTDI):

• For calculating spread of dose outside a nominal slice.

• It is the integral of the dose along the axis of the patient from a single slice divided by the nominal thickness of the slice.

• It can be measured by inserting a 10 cm long, thin cylindrical ionization chamber dosemeter along the axis of a cylindrical Perspex phantom and imaging one slice through its middle.

• Organ doses from CT examinations estimated by multiplying CTDI by the appropriate conversion factors.

• Typical effective dosesare in the range 5-10 mSv per examination.

• Although CT scans account for only 2% of X-ray examinations, they contribute more than 20% of the radiation dose delivered to the UK population by medical X-rays.

ARTEFACTS

I- Motion artefacts

• Cardiac motion produces streak artefacts (black and white bands).

• The reconstruction process is misled by a moving structure occupying different voxels during the scan.

• Mechanical misalignment and movement of the patient have similar effects.

II- High-attenuation objects

• Neurosurgical clips, dental amalgam, Small areas of bone or contrast medium etc., give rise to star artefacts which may obscure the area of interest.

• The effect is accentuated by motion.

III- Defector malfunction

• In a third-generation scanner even a small imbalance in the sensitivity of the scintillation detectors can produce ring artefacts.

• Cause: the X-ray pencil associated with each

detector traces out a 'data ring' - a ring of tissue which is 'seen' by that detector alone.

• This ring can be seen on the image as an artefact if that detector malfunctions.

49

more easily matched.

IV- Beam hardening

• The reconstruction process assumes a homogeneous X-ray beam, with the result that CT numbers are lower in the center of the patient (known as 'cupping').

• This is corrected to some extent by a "beam-hardening algorithm".

V- Geometrical artefacts

Because of the diverging beam → CT slices are narrower at the center than at the edge → overlap at the edges or an unscanned region at the center.

VI- Aliasing

• A sharp and high contrast boundary (as at a bone edge) may produce a number of parallel streaks nearby in the image, for reasons explained in Section 9.2.

• Similarly, at the boundary between the lung and diaphragm, spurious increased density may appear in the base of the lung.

VII- Partial volume or volume averaging artefacts

Quality Assurance: summary

Each department should have a quality control protocol using appropriate test objects to verify the performance of their scanners.

The topics, already mentioned in this chapter include:

1.

Noise

water phantom, or other reference material;

2.

Reproducibility

3.

Uniformity

field for the reference material;

4.

Sensitivity

(different contrasts) (see Fig. 4.18);

5.

Contrast scale

materials;

6.

Resolution

7.

Alignment

example, a high-contrast pin; 8.

Slice thickness and spacing

9.

Light beam alignment

50