Transcript Document

XI. National Turkish Medical Physics Congress
14-18 November 2007 - Antalya
CT from past to future
Carlo Maccia
Medical Physicist
CAATS 43 Bd du Maréchal Joffre – Bourg-La-Reine – FRANCE
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
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 twodimensional 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
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 .
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
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).
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, 1975-76)
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)
FIRST GENERATION CT SCANNER (Translate/Rotate)
First Head Scanner
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
THIRD GENERATION CT SCANNER (Rotate/Rotate)
Predominant design of
current commercially
available CT scanners
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
A look inside a rotate/rotate CT
Detector
Array
and
Collimator
X-Ray
Tube
A Look Inside a Slip Ring CT
X-Ray
Tube
Detector
Array
Slip Ring
Note:
how most
of the
electronics
is
placed on
the rotating
gantry
SPIRAL CT SCANNERS (Conventional Scanning Mode)
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
MULTI-SLICE CT SCANNERS (Dual Slice)
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
Single source CT
Fast + Poor Image quality
Dual Source CT
Fast + Improved Image quality
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
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.
Recall from previous lecture notes
I0 
 In
Pixel Position
Output Intensity
2
I1 = I0 e- w
I2 = I1 e- w = I0e-w(
n
In = I0e-w(
1
1
2
1
1
+ 2 + … + n)
therefore
1 + 2 + 3 + … + n =
+ 2)
1 ln (I0/ In)
w
Ray Sum
Value
Actually, the ray sum value that is computed is proportional to the sum of the n
attenuation coefficients along the x-ray beam path
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 800 - 1000 projections are collected in one 360° tube
rotation to reconstruct a single axial image
Images slices are reconstructed into a matrix consisting of multiple
volume elements (voxel) each with a unique value.
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.
VOLUME ELEMENT (VOXEL)
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 (128 - 256)
slices can be collected simultaneously
 3-dimensional CT and CT endoscopy
 Cardiac image acquisition during relevant
heart phases (ECG pulsing synchronization)
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
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
CT NUMBER
The final result of the CT image reconstruction is an accurate estimate of the x-ray
absorption values characteristic of individual voxels.
CT Number = 1000
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
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)
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
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
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.
LOW CONTRAST RESOLUTION
• Measurement Technique
Catphan 500 (phantom)
Insert Diametre : 2 mm
to 15 mm.
1%
0,5%
Contrast levels : 0.3, 0.5
and 1%
Supra slice (Periphery)
Z = 40 mm
0,3%
Subs slice (centre)
Z= 3, 5, 7 mm
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
= 0.0  3.0 HU
ROI CT number Standard Deviation = 3.5  0.7 HU
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, 13.17 cm2 CROI, Normal Scan Type, 5 mm slice thickness, 170 mA
and 2 sec scan time
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)2

1
patient dose
• a CT scanner can be shown to be quantum limited by plotting
(noise)2
vs
1
(any parameter that affects patient dose)
and determining the magnitude of the y-intercept of the interpolated linear curve fit
Since in a quantum limited CT scanner
(noise)2

1
patient dose/pixel
(noise)2

1
B • D • H • w3
then
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)
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
IMAGE RESOLUTION
• Measurement Technique
• MTF (Modulation Transfer Function) objective method
• Assessment of a bar pattern – subjective method
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.
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
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.
IMAGE RESOLUTION
• MTF curves at 50 %, 10 % and 2 %.
PQ 5000
IMAGE RESOLUTION
Typical values
• Standard mode : 7 line pairs / cm .
• Maximum values : 17 to 18 line pairs / cm (high resolution
mode)
IMAGE RESOLUTION (influencing factors)
• Acquisition
• Number of projections
IMAGE RESOLUTION (influencing factors)
• Acquisition
• Number of
technique)
projections
(floating
focal
spot
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)
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
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
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
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
SLICE THICKNESS
• Catphan 500 Phantom
• Θ = 23°
DOSIMETRIC QUANTITIES C.T.
• CTDI (Computed Tomography Dose
Index)
• DLP (Dose-Length Product)
• MSAD (Multiple Scan Average Dose)
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
1
CTDI =
T

+
D(z)dz
-
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).
COMPUTED TOMOGRAPHY DOSE INDEX (CTDI)
• Measurement principle
Ionization
Chamber
Mean Dose.
Aire= e x CTDI
CTDI Aire
e
ee
nxe
Z mm
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.
HETEROGENEITY OF DOSE PROFILES
Ideal
Air
1 cm
Centre
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]
1
n CTDI w =
C
(
1
2
CTDI100,c + CTDI100,p
3
3
)
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
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:
–
CTDI w =
n
CTDI w  C
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.
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
REFERENCE DOSE QUANTITIES
• DLP Dose-length product for a complete
examination : [mGy • cm]
DLP=  nCTDI w T  N  C
i
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.
REFERENCE DOSE QUANTITIES
In the case of helical (spiral) scanning [mGy • cm]
:
DLP=  nCTDIw  T  A t
i
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.
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:
MSAD =
1
Ι

+
-
I
2
I
2
D N, I (z)dz
where:
DN,I(z) is the multiple scan dose profile along a
line parallel to the axis of rotation (z).
Multiple Scan Average Dose
MSAD =
1

+
I
2
D N, I (z)dz
I
2
Pitch =1 ; CTDI=MSAD
Ι
-
I
e
e
T
T Z mm
MSAD : dose delivered while scanning with non adjacent slices (axial mode)
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 0.625 + 12 x 1.25
•  More important applied mA values
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.
DOSE MODULATION
mA = function of (Image quality needed, tissues attenuation)
Optimization of image noise
100%
mA Constants
55%
Z Modulation - Auto mA
40% XYZ Modulation
300
mAs
250
200
150
100
50
0
0.0
100.0
200.0
300.0
400.0
500.0
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)
Q
f
3
 zD
2
PROPOSED REFERENCE DOSE VALUES
(CTDI) and effective dose for different CT examinations (EUR 16262)
Region
Head
Thorax
Abdomen
Pelvis
Length of examined Area
(mm)
Slice thickness (mm)
160
320
300
160
5
10
5
3
Time (s)
32
32
40
40
Current (A)
210
210
165
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
mAs VARIATION (SLICE THIKNESS OF 5 mm)
French Survey carried out in 2004
500
400
300
200
100
0
mAs
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
mGy/mAs VARIATION (SLICE THIKNESS OF 5 mm)
French Survey carried out in 2004
0,5
0,4
0,3
0,2
0,1
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
EFFECTIVE DOSE COMPARISON (mGy)
French Survey carried out in 2004
12
Axial
helical
10
8
6
4
2
0
CHEST
ABDOMEN
EFFECTIVE DOSE (abdomen-pelvis)
French Survey carried out in 2004
30
mSv
max
mean
min
25
20
15
10
5
0
mean
helical
axial
PROPOSED REFERENCE DOSE VALUES
Routine CT examinations on the basis of absorbed dose to
air (EUR 16262 )
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
35
280
Routine abdomenb
35
780
Liver and spleenb
35
900
Routine pelvisb
35
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)
QUALITY CONTROL
Example of QC Test
periodicity :
QC Test
Mechanic
Noise
Uniformity
Low Contrast
detectability
Spatial Resolution
Contrast scale
linearity
Slice Thickness
Dose
Acceptance
*
*
*
*
*
*
*
*
Daily
Monthly
*
Annually
*
*
*
*
*
*
*