Computed Tomography Principles
A Little Bit History
Nobel prizesRoentgen (1901): Discovery of X-rays Hounsfield & Cormack (1979): Computed tomography
Source and Detectors
Source- Rotating anode disk- Small focal spot
down to 0.6 mm- Polychromatic beam
Detectors- Xenon (50-60%)- Scintillation (>90%)
(From Siemens)
Detector Collimation
• 1 mm: temporal bone• 2-3 mm: lung nodule, renal arteries• 5 mm: neck, kidney, pancreas• 8 mm: chest, liver
Source DetectorCollimation
Data Acquisition System (DAS)
Filter
Source Detector
Pre-Collimator Post-Collimator
Patient
Scattering
Exponential Attenuation of X-ray
∆x
µ1 µ2 µ3
µ
No
xio eNN ∆−= µ
xio eNN ∆++−= )( 321 µµµNi
NoNi
Ni: input intensity of X-rayNo: output intensity of X-rayµ: linear X-ray attenuation
x
X-rays
Attenuatedmore
Ray-Sum of X-ray AttenuationNoNi
o
i
kk N
Nx ln=∆∑µo
i
NNdxx ln)( =∫
∞
∞−
µ
x
iok
k
eNN∆−∑
=µ∆x
µκ
Ray-sum Line integral
E-Beam CT Scanner
• Speed: 50, 100 ms• Thickness: 1.5, 3, 6, 10 mm• ECG trigger cardiac images
(From Imatron)
Spiral/Helical/Volumetric CT
Continuous &Simultaneous
• Source rotation• Patient translation• Data acquisition
8 times faster than single-sliceSingle-Slice
One rotation / sec
Quad-Slice
Two rotations / sec+
4 slices / rotation
Image Analysis
Visualization & analysis• 3D, 4D• Networked, PC-based• Image fusion• Computer aided diagnosis• Image-based surgery
Projection & Sinogram
Sinogramt
θ
Sinogram:All projections
P(θ,t)
f(x,y)
t
θ
y
x
X-rays
Projection:All ray-sums in a direction
π
Computed Tomography
P(θ,t) f(x,y)P(θ,t)
f(x,y)
t
θ
y
x
X-rays
Computed tomography (CT):Image reconstruction fromprojections
Algebraic Reconstruction Technique(ART)
4 3
2 1
0 0
0 0
Guess 0
6 4 Error
4 3
2 1 −2
2Guess 2
Error
3 2
3 2
Update a guessbased on
data differences
Guess 1
Fourier Transformation
[ ]
[ ] dudvevuFvuFFyxf
dxdyeyxfyxfFvuF
vyuxj
vyuxj
∫ ∫
∫ ∫∞
∞−
∞
∞−
+−
∞
∞−
∞
∞−
+−
==
==
)(21
)(2
),(),(),(
),(),(),(
π
π
FourierTransform
f(x,y)
ImageSpace
F(u,v)
FourierSpace
Filtered Backprojection
f(x,y) f(x,y)
P(θ,t) P’(θ,t)
1) Convolve projections with a filter2) Backproject filtered projections
Image Display
CT Number- Hounsfield unit
• Air: -1024• Water: 0• Bone: +175 to +3071
Viewing Parameters• Window level (L)• Window width (W)• Zoom factor
water
waterHUµµµµ −
=1000)(
-1024 +3071
0 255
W
L
Comparison 3rd and 4th generation3rd generation has the X-ray tube in the apex (source fan)4th generation has a single detector in apex (detector fan)
High stability output of X-ray tube
High stability of detectors
One detector that is bad in the 3rd generation will create a ring artifact. A small ring diameter if central and a large diameter for a peripheral detector.
Beam-Hardening Artifacts
CauseEffective energy is shifted to higher value asthe X-rays pass through an object
Correction• Prefilter the X-ray beam near the focus• Avoid highly absorbing bony regions• Algorithms
Blurring Artifacts (Volume Averaging)
Causes• Large CT slice thickness and high contrast
structures only partially included• Finite source size• Finite sampling rates
Correction• Volume Artifact Reduction (VAR) mode• Deblurring
Blurring Artifacts (Volume Averaging)
Volume averaging Blurred Deblurred(Blurred data from GH Esselman at Wash U)
Stair-Step Artifacts (Helix)Associated with inclined surfaces inreformatted slices
Causes• Large reconstruction interval• Asymmetric helical interpolation
Correction• Collimation and feed less than feature sizes,
and small reconstruction interval• Adaptive interpolation
Metal Artifacts
CauseMetal blocks parts of projection data
Correction• Avoid metal parts• Algorithms
Metal Artifacts
Filteredbackprojection
Dentalphantom
FB with linearinterpolation
EM-like Iterativereconstruction
Motion Artifacts
CausesPatient motionOrgan motion
heart beatingbreathingswallowing
Correction• Fast scanning• Algorithms
Motion Artifacts
θ=0o
θ=90o
Filteredbackprojection
EM-like Iterativereconstruction
Time varying phantom
Radiation Dose
Dose - radiation energy transferred toan anatomic structure during X-ray scanning
The unit of dose is Gray (Gy)sometimes Rad (0.01 Gy)
Typical values for a CT transaxial scan are inthe range of 30 to 50 mGy
Radiation Profile: Single Scan
Ideal profile
Real profile
Radiation spreads outsidethe designated slice due to scattering
CT Dose Index
CTDI: CT dose indexT: slice thicknessD(z): local dosez: longitudinal coordinate
∫−
=T
T
dzzDT
CTDI7
7
)(1
T
Radiation Profile: Multiple Scans
Real profile
Radiation dose from multiple scans areaccumulated in the central slice
Multiple Scan Average Dose
MSAD: multiple scan average doseI: inter-slice distanceN: Number of scans
∫−
=2/
2/, )(1 I
IIN dzzD
IMSAD
I
Dose Measurement
• Cylindrical phantoms of 16 cm & 32 cm• Pencil ionization chamber• Dosimeter
16 or 32 cm
MSAD Estimation
MSAD is• directly proportional to mA• directly proportional to scan time• increases with kVp
as compared to dose at 120 kVp0.2-0.4 times less at 80 kVp1.2-1.4 times more at 140 kVp
• increases slightly with decreasing slice thickness• similar at the iso-center and near surface for head• significantly less at the iso-center than
near surface for body