1. XI. National Turkish Medical Physics Congress
14-18 November Antalya
CT from past to future
Carlo Maccia
Medical Physicist
CAATS 43 Bd du Maréchal Joffre – Bourg-La-Reine – FRANCE

2. Content CT equipment and technology
Recall of basic physical principles of CT
Radiation protection rules and QC
CT dosimetry quantities
Reference Dose values and Quality criteria for CT images

3. INTRODUCTION
Computed tomography (CT) was commercially introduced into radiology in 1972 and was the first fully digital imaging device making it truly revolutionary in diagnostic imaging. In 1979, Godfrey Hounsfield and Allen Cormack were awarded the Nobel Prize in Physiology and Medicine for their contributions in the development of CT.
CT differs from conventional projection imaging in two significant ways:
CT forms a cross-sectional tomographic image, eliminating the superimposition of structures that occur in plane film imaging because of the compression of three-dimensional body structures into the two-dimensional recording system
the sensitivity of CT to subtle differences in x-ray attenuation is at least a factor of 10 higher than normally achieved by film-screen recording systems

4. THE BASIC PHYSICS PROBLEM
Under ideal conditions (monochromatic beam, ideal collimation, perfect detection, etc)
x-ray intensity observes an exponential decay law:
N = N0 e-x
where N0 and N are the intensities of the incident and exiting x-rays, respectively, x
is the path length through the attenuating material, and  is the linear attenuation
coefficient of the material along the path x.
ASIDE
If we had a block consisting of a single attenuating material with unknown , we
could measure its length (x) and the incident (N0) and exiting intensities (N) , and
then solve for .

5. Now suppose we have an object with unknown contents, we can make a measurement of x-ray attenuation along a straight line through it
but for all intense of purposes all that this will tell us is a single number representing
the total attenuation of the material in the path. What we really want is the attenuation coefficient at each position along the path.
So essentially we have

6. and thus
With a single transmission measurement, the separate attenuation coefficients cannot
be determined because there are too many unknown values of i where i = 1,2,3 ,…, n.
In order to solve this equation for the n values of i we will need n2 independent
transmission equations (the above equation would be one of the n2 required equations).
Consider the case for n = 4 and each block had a size x:
We can see from the above illustration that in order to solve for 1, 2, 3 and 4, we would need 4 independent equations (N1, N2, N3 and N4).

7. DATA ACQUISITION GEOMETRIES
A variety of geometry's have been developed to acquire the x-ray transmission data needed for image reconstruction in CT. Some geometry's have been tagged as a “generation” of CT scanner and these labels are useful in different scanner designs.
The following scanner geometry's, data acquisition modes and primary technologies have been used to date:
First Generation CT Scanner (EMI, 1973)
Second Generation CT Scanner (1974)
Third Generation CT Scanner (GE & Siemens, )
Fourth Generation CT Scanner (1977)
Low Voltage Slip Ring Technology (Siemens, 1982)
Fifth Generation CT Scanner (1984)
Spiral CT Scanner (Siemens, 1988)
Multi-slice CT Scanner (Dual-slice, Elscint, 1992)
Multi-slice CT Scanner (Quad-slice, 1998)
Dual source CT (64-slice with two X-ray tubes, Philips 2006, 256-slice Toshiba)

8. FIRST GENERATION CT SCANNER (Translate/Rotate)
First Head Scanner

10. NOTE
this method was theoretically immune to the effects of scattered x-ray (single
detector system)
because of the long scan times, this method of scanning was applicable to scanning
of parts of anatomy that could have been kept motionless, such as the head

11. Predominant design of current commercially
THIRD GENERATION CT SCANNER (Rotate/Rotate)
Predominant design of current commercially
available CT scanners

12. LOW VOLTAGE SLIP RING TECHNOLOGY
Some third and fourth generation CT scanners employ a slip ring to supply power and
receive signals from rotating parts. In the slip-ring method, an electrical conductive
brush moves along a ring-shaped electrically conductive rail. The use of a slip ring
permits high-speed continuous scanning, and dramatically increases both the
performance and range of clinical applications of CT scanning.
allows for 1 second ( or < 1 second or sub-second) scan times
allows for helical (or volumetric) scanning

13. A look inside a rotate/rotate CT
Detector Array
and Collimator
X-Ray
Tube

14. A Look Inside a Slip Ring CT
Note:
how most
of the
electronics is
placed on
the rotating
gantry
X-Ray
Tube
Detector
Array
Slip Ring

16. SPIRAL CT SCANNERS (Conventional Scanning Mode)

17. SPIRAL CT SCANNERS (Helical Scanning Mode)
If the x-ray tube can rotate constantly, the patient can then be moved continuously through the beam, making the examination much faster

18. MULTI-SLICE CT SCANNERS (Dual Slice)

20. MULTI-SLICE CT SCANNERS (Quad Slice)
To build quad-slice spiral CT scanners, manufacturers had to develop detector arcs with more than four elements in the longitudinal (z) axis direction, creating a curved two-dimensional detector arrays.
GE Scanners

22. Fast + Poor Image quality Fast + Improved Image quality
Single source CT
Dual Source CT
Fast + Poor Image quality
Fast + Improved Image quality

23. AXIAL IMAGE RECONSTRUCTION
The task of reconstruction is to compute an attenuation coefficient for each picture element (pixel) and then to assign a CT number to each of these elements.  
in order to create multiple projections in a single 360°
tube rotation, during a single projection the x-ray tube
is pulsed and the detector array is sampled after each
pulse

24. IMAGE RAY SUM A B
For a single detector, a ray sum consists of all the linear attenuation coefficient data
along the corresponding x-ray beam path (eg: path AB)
For a single x-ray beam path, the ray sum is not the simple summation of the
attenuation coefficients of the intercepted pixels.

25. I0   In I1 = I0 e-1w I2 = I1 e-2w = I0e-w(1 + 2)
Recall from previous lecture notes
I0 
 In
Pixel Position Output Intensity
I1 = I0 e-1w 1
I2 = I1 e-2w = I0e-w(1 + 2) 2 n
In = I0e-w(1 + 2 + … + n)
therefore
Ray Sum
Value
1 + 2 + 3 + … + n = ln (I0/ In) w
Actually, the ray sum value that is computed is proportional to the sum of the n
attenuation coefficients along the x-ray beam path

26. IMAGE PROJECTION Detector Position
a projection is defined as the set of ray sums measured in all detectors during a
single x-ray tube pulse
typically anywhere between projections are collected in one 360° tube
rotation to reconstruct a single axial image

27. Images slices are reconstructed into a matrix consisting of multiple volume elements (voxel) each with a unique value.

28. IMAGE INTERPOLATION (SPIRAL CT)
PROBLEM
The volume scanned in a single rotation differs between the conventional and helical scanning methods.
ANSWER
Interpolate desired axial image from volume data set prior to image reconstruction.

29. VOLUME ELEMENT (VOXEL)

30. New CT Features
The new helical scanning CT units allow a range of new features, such as :
CT fluoroscopy, where the patient is stationary, but the tube continues to rotate
multislice CT, where up to 64 ( ) slices can be collected simultaneously
3-dimensional CT and CT endoscopy
Cardiac image acquisition during relevant heart phases (ECG pulsing synchronization)

31. CT Fluoroscopy Real Time Guidance (up to 8 fps) Great Image Quality
Low Risk
Faster Procedures (up to 66% faster than non fluoroscopic procedures)
Approx. 80 kVp, 30 mA

32. Content CT equipment and technology
Recall of basic physical principles of CT
Radiation protection rules and QC
CT dosimetry quantities
Reference Dose values and Quality criteria for CT images

33. CT NUMBER CT Number = 1000 p - w = Hounsfield Unit (HU) w
The final result of the CT image reconstruction is an accurate estimate of the x-ray
absorption values characteristic of individual voxels.
CT Number = p - w = Hounsfield Unit (HU) w
where p is the linear attenuation value assigned to a given pixel and w is the linear
attenuation value of water.
ASIDE
w is obtained during calibration of the CT scanner
by definition, the HU of water is 0 and the HU value for air is -1,000
above equation defines 100 HU as equal to a 10 % difference in the linear
attenuation coefficient relative to water
the value 1000 in the numerator is a scale factor and determines the contrast scale

34. Because p and w are dependent on photon energy (keV), HU values depend on
the kVp and filtration. Therefore, HU values generated by a CT scanner are approximate and only valid for the effective kVp used to generate the image.

35. FIELD-OF-VIEW (FOV)
FOV is the diameter of the area being imaged (e.g: 25 cm Head and 35 cm Body scan)
CT pixel size is determined by dividing the FOV by the matrix size (typically
512 x 512 – 768 x 768 or 1024 x 1024)

36. IMAGE DISPLAY
reconstructed images are viewed on a CRT monitor or printed onto film using a laser printer
each pixel is normally represented by 12 bits, or 4096 gray levels, which is larger than the display range of monitors or film
window width and level are used to optimize the appearance of CT images by
determining the contrast and brightness levels assigned to the CT image data

37. IMAGE QUALITY Image quality may be characterized in terms of: contrast
noise
spatial resolution
ASIDE
in general, image quality involves tradeoffs between these three factors and patient
dose.
artifacts encountered during CT scanning can degrade image quality

38. IMAGE CONTRAST
CT contrast is the difference in the HU values between tissues. This contrast generally
increases as kVp decreases but is not affected by mAs or scan time.
CT Photon Energy Range
(120 or 140 kVp)
CT contrast may be artificially increased by adding a contrast medium such as
iodine
image noise may prevent detection of low-contrast objects such as tumors with a
density close to the adjacent tissue
the displayed image contrast is primarily determined by the CT window width and
window level settings.

39. LOW CONTRAST RESOLUTION
Measurement Technique
Catphan 500 (phantom)
Insert Diametre : 2 mm to 15 mm.
Contrast levels : 0.3, 0.5 and 1%
Supra slice (Periphery)
Z = 40 mm
Subs slice (centre)
Z= 3, 5, 7 mm 1%
0,5%
0,3%

40. IMAGE NOISE The sources of image noise in CT are:
quantum mottle (the number of photons used to make an image)
inaccuracies in the image reconstruction process (software filter phase); and
electronic noise introduced after detection
Noise in CT is usually defined as the standard deviation () of the CT numbers
calculated from pixel values in a predefined region-of-interest (ROI) using an image
of a uniform material (usually water). The selected ROI region should be void of
objects and cover a sufficiently large image area (circular diameter > 10 mm).
For GE scanners:
ROI CT number Average Value =  3.0 HU
ROI CT number Standard Deviation =  0.7 HU

41. ROI Area = 13.17 cm2 Mean = 1.75 Std. Dev. = 2.9 NOTE: Noise = 2.9
Scan Parameters: Small Scan FOV, 25 cm DFOV, 5122 Matrix, Standard Resolution, Peristaltic Option OFF, cm2 CROI, Normal Scan Type, 5 mm slice thickness, 170 mA and 2 sec scan time

42. ELECTRONIC NOISE
in modern CT scanners electronic noise is kept to a minimum
a CT scanner whose noise is dominated by the detection of a finite number of x-rays
(quantum mottle) is called quantum limited
in a quantum limited CT scanner
(noise) 
patient dose
a CT scanner can be shown to be quantum limited by plotting
(noise)2 vs
(any parameter that affects patient dose)
and determining the magnitude of the y-intercept of the interpolated linear curve fit

43. Since in a quantum limited CT scanner
(noise) 
patient dose/pixel
then
B • D • H • w3
where
B - is the fractional transmission of the patient
D - is the maximum surface dose ( mAs)
H - is the slice thickness
w - is the reconstructed pixel width
quantum mottle (and thus noise) decreases as the number of photons increases
CT noise is generally reduced by increasing the kVp, mA or scan time (if all other parameters are kept constant)
CT noise is also reduced by increasing voxel size (ie: by decreasing matrix size, increasing FOV or increasing the slice thickness)
typically noise with a modern CT scanner system is approximately 5 HU (or
0.5% difference in attenuation coefficient)

44. ASIDE (noise)2  1 B • D • H • w3 From the above equation :
A 4x increase in dose is required for a 2x decrease in σ. Reduced σ will provide better low contrast detectability
An 8x dose increase is required for a 2x decrease in W which maintains constant σ. Decreasing W will increase high contrast detectability.
A 2x dose increase is required to decrease H by a factor of 2 and keep σ constant.
Decreasing both W and H by 2x would require a dose increase of 16x in order to maintain constant σ.
Since B depends exponentially on patient thickness, increasing the thickness of the patient or body part being examined by a factor of n will result in a decrease in transmission from B to Bn. If B = 0.05 and n = 2, a dose increase of 20x would be required to maintain σ, H and W. This is one reason why larger pixels are often used for body scans than for head scans.

45. IMAGE RESOLUTION
Spatial resolution is the ability to discriminate between adjacent objects and is a
function of pixel size.
If the CT FOV is D and the matrix size is M, then pixel size is D/M.
Example:
For a typical head scan with a FOV of 25 cm and a matrix of 512 pixels, the pixel size is 0.5 mm
Because two pixels are required to define a line pair (lp), the best achievable spatial resolution is 1 lp/mm
typically resolution in CT scanning ranges from 0.5 to 1.5 lp/mm
the axial resolution may be improved by operating in a high resolution mode using a smaller FOV or a larger matrix size
factors that may also improve CT spatial resolution by reducing image blur include
smaller focal spots, smaller detectors and more projections
resolution perpendicular to the section is dependent on slice thickness and is important in Sagittal and Coronal image reconstruction

46. Measurement Technique
IMAGE RESOLUTION
Measurement Technique
MTF (Modulation Transfer Function) objective method
Assessment of a bar pattern – subjective method

47. IMAGE RESOLUTION
MTF can be considered as a reliable measure of the information transfer from the object to the image. It illustrates, for each individual spatial frequency, the progressive degradation of the signal due to the system in terms of % of contrast loss.

48. IMAGE RESOLUTION
The MTF is assessed from the Fourier Transform of the Linear Spread Function (LSF) which is a measure of the ability of a system to form sharp images; it is determined by measuring the spatial density distribution on film of the X-ray image of a narrow slit in a dense metal, such as lead.
The point spread function (PSF) describes the response of an imaging system to a point source or point object

49. IMAGE RESOLUTION
The image of the « point object » is not a single point but a set of different points representing the degradation of the signal.

50. IMAGE RESOLUTION
MTF curves at 50 %, 10 % and 2 %.
PQ 5000

51. IMAGE RESOLUTION Typical values Standard mode : 7 line pairs / cm .
Maximum values : 17 to 18 line pairs / cm (high resolution mode)

52. IMAGE RESOLUTION (influencing factors)
Acquisition
Number of projections

53. IMAGE RESOLUTION (influencing factors)
Acquisition
Number of projections (floating focal spot technique)

54. IMAGE RESOLUTION (influencing factors)
Acquisition
Actual detector aperture
The smaller detector aperture the better spatial resolution
Slice thickness (reduction of scattered radiation, improvement of image sharpness)

55. IMAGE RESOLUTION (Z-Axis)
Z-axis resolution is important for 3D reconstruction ==> Isotropic dimension of the pixel
Z-axis resolution
Slice thickness
Pitch
Abdomen, Pelvis
Chest
Angiography

56. IMAGE RESOLUTION (Z-Axis)
If, within the slice, the object shows a continuity along the Z-axis, the HU remain constant
If, within the slice, the object is not continuous, the partial volume effect would change the HU value

57. SLICE THICKNESS Measured at the isocentre of rotation
Allow to check the overlapping of adjacent slices
Expressed in terms of image profile at the Full Width at Half Maximum (FWHM) value
Note :
Θ = 45° magnification factor = 1
Θ = 63.5° magnification factor = 2

58. Content CT equipment and technology
Recall of basic physical principles of CT
Radiation protection rules and QC
CT dosimetry quantities
Reference Dose values and Quality criteria for CT images

59. SLICE THICKNESS
Catphan 500 Phantom
Θ = 23°

60. DOSIMETRIC QUANTITIES C.T.
CTDI (Computed Tomography Dose Index)
DLP (Dose-Length Product)
MSAD (Multiple Scan Average Dose)

61. RADIATION DOSE Single Slice Profile
The dose profile in a CT scanner is not uniform along the patient axis and may vary
within any irradiated section.
Single Slice Profile
Slice Thickness (T)
defined at FWHM
NOTE: because of scattered x-rays, the CT section dose profile is not perfectly square but has tails that extend beyond the section edges.
Tissues beyond the section are thus exposed to radiation

62. COMPUTED TOMOGRAPHY DOSE INDEX (CTDI)
The CTDI is the integral along a line parallel to the axis of rotation (z) of the dose profile (D(z)) for a single slice, divided by the nominal slice thickness T
In practice, a convenient assessment of CTDI can be made using a pencil ionization chamber with an active length of 100 mm so as to provide a measurement of CTDI100 expressed in terms of absorbed dose to air (mGy).

63. Measurement principle
COMPUTED TOMOGRAPHY DOSE INDEX (CTDI)
Measurement principle
Mean Dose.
Ionization Chamber
Aire= e x CTDI
CTDI Aire e e e
n x e
Z mm

64. COMPUTED TOMOGRAPHY DOSE INDEX (CTDI)
measurements of CTDI may be carried out free-in-air in parallel with the axis of rotation of the scanner (CTDI100, air)
or at the centre (CTDI100, c)
and 10 mm below the surface (CTDI100, p) of standard CT dosimetry phantoms.
the subscript `n' (nCTDI) is used to denote when these measurements have been normalised to unit mAs.

65. HETEROGENEITY OF DOSE PROFILES
Air
1 cm
Centre
Ideal

66. ) ( COMPUTED TOMOGRAPHY DOSE INDEX (CTDI)
On the assumption that dose in a particular phantom decreases linearly with radial position from the surface to the centre, then the normalised average dose to the slice is approximated by the (normalised) weighted CTDI: [mGy(mAs)-1]
where:
C is the tube current x the exposure time (mAs)
CTDI100,p represents an average of measurements at four different locations around the periphery of the phantom ) (
CTDI 3 2 + 1 C = p
100, c w n

67. C CTDI = × REFERENCE DOSE QUANTITIES
Two reference dose quantities are proposed for CT in order to promote the use of good technique:
CTDIw in the standard head or body CT dosimetry phantom for a single slice in serial scanning or per rotation in helical scanning : [mGy]
where:
nCTDIw is the normalised weighted CTDI in the head or body phantom for the settings of nominal slice thickness and applied potential used for an examination
C is the tube current x the exposure time (mAs) for a single slice in serial scanning or per rotation in helical scanning. C
CTDI = w n ×

68. REFERENCE DOSE QUANTITIES
CTDI(vol) for non adjacent slices : [mGy]
Axial mode CTDI(vol) = CTDI(w) x T
Slice interspace
Helical Mode CTDI(vol) = CTDI(w)
Pitch

69. å C N T CTDI = DLP × REFERENCE DOSE QUANTITIES
DLP Dose-length product for a complete examination : [mGy • cm]
where :
i represents each serial scan sequence forming part of an examination
N is the number of slices, each of thickness T (cm) and radiographic exposure C (mAs), in a particular sequence.
N.B.: Any variations in applied potential setting during the examination will require corresponding changes in the value of nCTDIw used. C N T
CTDI =
DLP w n × å i

70. å In the case of helical (spiral) scanning [mGy • cm] : t A T CTDI =
REFERENCE DOSE QUANTITIES
In the case of helical (spiral) scanning [mGy • cm] :
where, for each of i helical sequences forming part of an examination :
T is the nominal irradiated slice thickness (cm)
A is the tube current (mA)
t is the total acquisition time (s) for the sequence.
N.B. : nCTDIw is determined for a single slice as in serial scanning. å t A T
CTDI =
DLP w n i ×

71. Scan: 15 slices, 10 mm slice thickness and 10 mm slice increment
RADIATION DOSE
When continuous sections are scanned, the cumulative radiation dose in a section may be as high as twice the radiation dose associated with a single section.
Multiple Scan Average Dose (MSAD)
Scan: 15 slices, 10 mm slice thickness and 10 mm slice increment

72. ò (z)dz D = MSAD REFERENCE DOSE QUANTITIES
Multiple Scan Average Dose (MSAD) : The average dose across the central slice from a series of N slices (each of thickness T) when there is a constant increment I between successive slices:
where:
DN,I(z) is the multiple scan dose profile along a line parallel to the axis of rotation (z).
(z)dz D =
MSAD I N, 2 + - 1 ò Ι

73. Multiple Scan Average Dose
(z)dz D =
MSAD I N, 2 + - 1 ò Ι e
Z mm T
Pitch =1 ; CTDI=MSAD I
MSAD : dose delivered while scanning with non adjacent slices (axial mode)

74. CT MULTI-SLICE TECHNOLOGY
N detectors
Scanned area Larger collimation ==> 40 mm
« Important irradiated volume : overscan »
Speed  Rotation time 0.33 to 0.5 s – Matrix Size 512 x 512 to 1024 x 1024 (Philips).
Resolution  Detector width 20 mm ( 16 x 1.25 mm)
40 mm (64 X 0.625) or 40 x x 1.25
 More important applied mA values

75. CT MULTI-SLICE TECHNOLOGY
DOSE
Lower dose with multislice CT than with single slice CT.
X-ray beam width < detector width (80 to 90 %)
Dose reduction software
DLP values increase because of larger collimation (40 mm) ; L acquisition > L required
To compensate for the increase of noise due to the pitch values, the systems increase the mA station ==> constant dose.
effective mA concept
CTDI is measured in the same conditions than for single slice CT machine.

76. DOSE MODULATION
mA = function of (Image quality needed, tissues attenuation) Optimization of image noise
100%
55%
40%
mAs
mA Constants
Z Modulation - Auto mA
XYZ Modulation

77. IMAGE QUALITY FACTOR Quality of the image
Low noise
Good resolution
Sub-millimeter slices
Low dose
Image Q factor suggested by « Impact »
f spatial resolution (MTF pl/mm)
σ noise
Z slice thickness (mm)
D dose (CTDI vol)

78. PROPOSED REFERENCE DOSE VALUES
(CTDI) and effective dose for different CT examinations (EUR 16262)
Region
Head
Thorax
Abdomen
Pelvis
Length of examined Area (mm)
160
320
300
Slice thickness (mm) 5 10 3
Time (s) 32 40
Current (A)
210
165
Organ
Eye Lens
Lungs
Liver
Bladder
Organ dose (mSv)
28.1
23.3
12.9
13.3
Effective Dose (mSv)
1,1
6,7
4,3
2,7

79. mAs VARIATION (SLICE THIKNESS OF 5 mm)
French Survey carried out in 2004

80. mGy/mAs VARIATION (SLICE THIKNESS OF 5 mm)
French Survey carried out in 2004

81. EFFECTIVE DOSE COMPARISON (mGy)
French Survey carried out in 2004

82. EFFECTIVE DOSE (abdomen-pelvis)
French Survey carried out in 2004

83. PROPOSED REFERENCE DOSE VALUES
Routine CT examinations on the basis of absorbed dose to air (EUR )
Examination
Reference dose value
CTDIw (mGy)
DLP (mGy cm)
Routine heada 60
1050
Face and sinusesa 35
360
Vertebral traumab 70
460
Routine chestb 30
650
HRCT of lungb
280
Routine abdomenb
780
Liver and spleenb
900
Routine pelvisb
570
Osseous pelvisb 25
520 a.
Data relate to head phantom (PMMA, 16 cm diameter) b.
Data relate to body phantom (PMMA, 32 cm diameter)

84. QUALITY CONTROL
Example of QC Test periodicity :