ct physics registry review · in ct, µ needs to be calculated for each voxel in the patientpatient...
TRANSCRIPT
1
CT PHYSICSRegistry Review
Slide # 1
Registry Review
Wil Reddinger M.S. R.T. (R) (CT)
Slide # 2
X-ray tube/Focal spot
Filter (beam filter)Pre-patient Collimator (patient protection and slice thickness)
i
High frequency generator, now shown.
Makes beam more homogeneous.
Generates the radiation.
Slide # 3
Patient
Pre-detector Collimator
ADC
(post patient collimator and re-defines slice thickness)
DetectorDetects attenuated radiation.
Converts analog signal to digital.
• X-rays are produced, emitted, filtered, and collimated
• X-ray penetration occurs - data collection schemes or beam geometries
Data Acquisition
Filter
Tube
Pre-patient collimator
Slide # 4
g
• Transmission measurements from the patient are obtained and converted to digital signals (involves detectors detector electronics)
ADC
Detector collimator
Analog Signal
Digital Signal
Detector
Slide # 5 Slide # 6
2
Slide # 7
Volume (Helical - Spiral) CTAdvantages of Multi-Slice
• Speed of coverage• Slice reconstruction• Improved cooperation• Improved contrast
enhancement• Thinner slices
Slide # 8
Single-Slice & Multi-slice Volume CT Volume CT
Thinner slices Resulting in…“improved “
- MPR - VRdata sets
MPR = Multi-Planar ReformatsVR = volume rendered
Slide # 9
Multi-slice CT, Detector Evolution
Single Dual Quad
Slide # 10
Multi- slice systems = Multi – channel systemsWhereby, 64 channels produce 64 slices per rotation!
Then,8, 16, 32, 40, 64 slice systems
Scan Data to Image Data
• Path x-ray beam travels from the
tube to the detector = RAY
• Detector reads each ray &
Slide # 11
• Detector reads each ray &
measures beam attenuation
• Measurement = Ray Sum
Ray (Projection)
I0 Itissue attenuates the ray
Slide # 12
Each ray produces a single measurement of theEach ray produces a single measurement of thexx--ray attenuation along a path between the ray attenuation along a path between the
source and a detectorsource and a detector
3
View (Profile)Complete set of Ray Sums = VIEW“Like looking at an object from a particular angle.”Many views are needed to create an image
Slide # 13
Obtained from a set of measurements across the object at Obtained from a set of measurements across the object at one angular position. Views are made up of raysone angular position. Views are made up of rays
VIEWVIEW
Slide # 14
ProfileProfile
Image Reconstruction• CT system accounts for the attenuation
properties of each ray and correlates them with the position of the ray
• An ATTENUATION PROFILE is created• resulting shadow (analogy)
Slide # 15
Image Reconstruction• Attenuation Profile is obtained
for each view• All profiles are projected back
into a matrix
Slide # 16
into a matrix• a big rectangle divided into
smaller squares called pixels
Matrix
Slide # 17
6 x 66 x 6
Reconstruction• Describes method for
converting Scan Data to Image Data
• x-ray beam-absorbed x-ray
Slide # 18
• x-ray beam-absorbed x-ray info-to analog signal-converted to digital signal-placed on a matrix in the computer-matrix on TV
4
Back Projection Method• Oldest means of
reconstruction• An image is created by
reflecting attenuation profiles
Slide # 19
reflecting attenuation profiles in the same direction they were obtained
• Produces streak artifacts• 1st attempt at CT imaging
BPBP55
BackBack--ProjectionProjection
BPBP11
PP11
PP33
PP55
Slide # 20
BPBP22BPBP33 BPBP44
PP22 PP44
Back Projection Method• Artifacts are not in the scanned
image• Appear due to the reconstruction
process ( Back Projecting)
Slide # 21
p ( j g)• “placing” information on a matrix
ReconstructionReconstruction
Slide # 22
Scanning
• Movement of the
- Tube and
- Detectors with X-ray transmission
Scanning
Slide # 23
• Tube and detectors are
-Co-linear (in alignment)
- And at the same speed
Back Projection Scanning a “phantom” With two “cups of water”
Within a cube phantom
X-ray Beam
Water
Water
Phantom
X-ray Tube
X-ray Tube
Proj
ectio
n #1
Slide # 24
In CT, the x-ray tube rotates around the “phantom”
In this case the x-ray beam is attenuated by the water in the
phantom, and therefore “projects” a “shadow” within the detectors…
Detectors
When the tube and detectors are in this configuration
This Projection is sampled or “detected” or by the detectors
Projection #2
5
Back Projection - Reconstruction
Proj
ectio
n #1•If we draw lines
from the actual phantomto the projections
•Where the lines intersectthere should be a cup of water
Projection #1
Projection #2
Projection #3
•If, however, a third projection Is acquired
•Continue to connect “lines” from the
Slide # 25
Projection #2
•But, since only 2 projections are acquiredin this example
•It appears as thoughthere are 4 cups of water
•Two projections are not enough!
Projection #2 from the actual phantom to the projection
•Now, the locations where the lines intersect, represent the actual cups of water.
•Three projections ARE enough! (For a simple phantom), but many more projections are required for the complex anatomy to be displayed on a CT image!
Image Reconstruction• Artifacts minimized by changing
the shapes of the attenuation profiles before back projection to a matrix.
Slide # 26
• Reconstruction “filter” is applied to the Raw Data or Scan data
• 1st type of “recon” = Filtered back Projection
Filtered Back Projection• Reconstruction filter • Kernel , Algorithm, Math Filter• Process of applying
M h i l fil i h
Slide # 27
Mathematical filtration to the raw data = Convolution
Filtered Filtered BackBack--ProjectionProjection
Slide # 28
ProfileProfile Filtered ProfileFiltered Profile
Image Reconstruction
• Mathematical filtration• Kernel
Al ith
Slide # 29
• Algorithm
Algorithms
• Mathematical method for
solving a problem that
f
Slide # 30
involves repetition of an
operation.
6
Types of Reconstruction• 2nd type avail today
• Fourier Reconstruction
• any function or variation of a quantity
Slide # 31
any function or variation of a quantity
in time or space can be expressed as a
sum of sine and cosine waves
• frequencies instead of profiles
Reconstruction
• involves millions of data points• ARRAY PROCESSOR• dedicated to rapid calculations
Slide # 32
involved in reconstruction• large numbers of calculations
needed to convert data to image
Mathematical FiltersAlgorithms
• High Pass (Sharp, Bone)
• primarily used in high contrast regions
Slide # 33
contrast regions (temporal bones)
• areas of sudden “large” drops in CT numbers( extreme tissue density)
High Pass Algorithms
• Optimizes spatial resolution, Edge enhancement.
• Decreases
Slide # 34
• Decreases blurring of edges
• Bone , sharp, detail algorithm
Low Pass filters / algorithms
• Soft tissue algorithms
• standard, smooth• area of gradual
ti h
Slide # 35
tissue change• optimizes contrast
resolution• smoothness of
larger objects
Raw Data• Includes all
measurements obtained from detector array
• Within the SFOV
Slide # 36
Within the SFOV• specialized
reconstruction• attenuation
information
7
Image Data• Data displayed on
the monitor• operator chooses
to viewi l t d
Slide # 37
• manipulated using window level and window width
Image DisplayMATRIX
• CT Image is represented by a matrix of numbers
• Rows and Columns of pixels
Slide # 38
Rows and Columns of pixels• 512 X 512 Matrix = 262,144 Pixels• 256, 340, 512, 768, 1024• 80, 160, 180 not used today
THE IMAGE IS MADE OF BLOCKS
Slide # 39
MATRIX• The bigger the matrix, the
smaller the pixel size• reduces partial volume
i
Slide # 40
averaging• effects image quality• effects reconstruction time
VOXELVOXEL PIXEL
Slide # 41
3D3D 2D
Volume Element
• Voxel• Three Dimensional
representation of tissue
Slide # 42
representation of tissue• Pixel Area x Section Thickness• Voxel Depth determined by ST
8
Picture Element• Pixel• Two dimensional
representation of the tissue volume
Slide # 43
volume• Pixel Size = FOV divided by
Matrix size
Raw Data• Includes all
measurements obtained from detector arrayWithin the SFOV
Slide # 44
• Within the SFOV• specialized
reconstruction• attenuation
information
Image Data• Data displayed on
the monitor• operator chooses
to view
Slide # 45
• manipulated using window level and window width
THE IMAGE IS MADE OF BLOCKS
Slide # 46
MATRIX• The bigger the matrix, the
smaller the pixel size• reduces partial volume
i
Slide # 47
averaging• effects image quality• effects reconstruction time
Imaging Matrix
• Digital images are created with a matrix
S ll t it
Slide # 48
6 x 66 x 6 3 x 33 x 3
• Smallest unit of the digital image is a pixel
9
MATRIX & PIXEL SIZE
Size effects resolutionThe bigger the Matrix, The better
the resolution.The bigger the matri The more
Slide # 49
The bigger the matrix, The more pixels you have, smaller too.
The more “ little blocks” you have to make the image , the more detail…
Calculating Pixel & Voxel Size
PixelPixel SliceSlice
VoxelVoxel• Isotropic voxel
• Pixel sizeFOVMatrixMatrix
FOVMatrixMatrix
FOVMatrixMatrix== ==
ThicknessThickness
FOVMatrixMatrix
FOVMatrixMatrixxx
byby
equalsequals equalsequals
Slide # 50
Slice Slice ThicknessThickness• Area of the Pixel
• Voxel Volume
FOVMatrixMatrix
FOVMatrixMatrix
FOVMatrixMatrixxx = mm= mm22
Thickness = mmThickness = mm33FOVMatrixMatrix
FOVMatrixMatrixxx xx
timestimes
timestimes timestimes
VOXELVOXEL PIXEL
Slide # 51
3D3D 2D
Picture Element• Pixel• Two dimensional
representation of the tissue volume
Slide # 52
volume• Pixel Size = FOV divided by
Matrix size
Volume Element
• Voxel• Three Dimensional
representation of tissue
Slide # 53
representation of tissue• Pixel Area x Section Thickness• Voxel Depth determined by ST
PixelPixel
Voxel
Slide # 54
Slice Slice ThicknessThickness
PixelPixel
Voxel Volume = Pixel Size x Slice ThicknessVoxel Volume = Pixel Size x Slice Thickness
10
Pixels & Voxels
• In CT slices are acquired• The voxel is a 3D volume element
Slide # 55
PixelPixel
Slice Slice ThicknessThickness
Slice Slice ThicknessThickness
VoxelVoxel•The face of the voxel is the pixel
FOV, Matrix, Thickness & Voxels
• The size of the area imaged in CT is the field of view (FOV)• The number of pixels ( l ) i th
matrixmatrix FOVFOV
Slide # 56
Slice Slice ThicknessThickness
(rows x columns) is the matrix• The depth is the slice thickness
FOVFOV
matrixmatrix
larger
smaller
How Much “Meat” is in the Box? vs. Signal?
Slide # 57
The larger the pixel (voxel)The more “meat” tissues / protonsThe larger the CT signalBut… since signals get averaged togetherThe lower the spatial resolution
larger
smaller
How Much “Meat” is in the Box? vs. Resolution?
Slide # 58
The larger the pixel (voxel)The more “meat” tissues / protonsThe larger the CT signalBut… since signals get averaged togetherThe lower the spatial resolution
Partial Volume Averaging
Slide # 59
smaller
Signals signals get averaged together in large voxelsThe lower the spatial resolutionPartial Volume Averaging
larger smallest
Calculating Pixel Size
VoxelVoxel
• to calculate the pixel size• to calculate the voxel size FOVFOVmatrixmatrix
Slide # 60
Slice Slice ThicknessThickness
Slice Slice ThicknessThicknessFOV
MatrixMatrix
FOVMatrixMatrix
11
Slice Thickness & Voxel Size
Slide # 61
1.25 mm slice thickness 2.5 mm slice thickness
Partial Volume Averaging Artifact• PVA = The inability to distinguish separate structures within a single pixel or voxel …• To reduce PVA (Partial Volume Averaging)
- Reduce FOV- Slice Thickness
Slide # 62
- Increase Matrix
Matrix = 320 x 320 Matrix = 512 x 512FOV = 20 cm – better resolutionLess partial volume averaging
FOV = 40 cm
FOVFOV
Pixels
Slide # 63
6 x 6 Matrix6 x 6 Matrix
Pixel size = FOV / MatrixPixel size = FOV / Matrix
QuickTime™ and aQuickTime™ and a
Matrix and Spatial Resolution
Slide # 64
QuickTime and aPhoto - JPEG decompresso
are needed to see this picture
QuickTime and aPhoto - JPEG decompressor
are needed to see this picture
Scan FOV
Slide # 65
Scan vs. Reconstructed FOV
Reconstructed FOV
Scan FOV Scan FOV
Scan FOV• Large
Slide # 66
Reconstructing smaller FOV increases noise
Recon FOV
FOV• Large acquired FOV • Largereconstructed FOV
• Large acquired FOV • Smallreconstructed FOV
12
Scan FOV
Slide # 67
Scan FOV Reconstructed FOV
Large Scan FOV
Slide # 68
Scan FOV Reconstructed FOV
Small Reconstructed FOV
Slide # 69
Magnification vs. Reconstruction
Small FOV•Reconstructed FOV• Reconstructs pixels• Uses raw data• Reformats data• Better resolution
Slide # 70
Magnification• Stretches pixels• Uses Image Data• Poorer resolution
Targeting Vs Zooming
Slide # 71
The reconstruction process assigns
an integer value to each voxel that
is linearly related to the attenuation
CT Numbers
Slide # 72
is linearly related to the attenuation
coefficent (µ) of the tissue(s)
within a voxel
13
Linear Attenuation Coefficient - a quantitative measurement of attenuation per cm of an absorber ( i.e. bone or soft tissue)
Represented by the greek letter µ
I CT d t b l l t d f hf h ll i thi th
Key Terms - Linear Attenuation Coefficient
Slide # 73
In CT, µ needs to be calculated for each for each voxelvoxel in the in the patientpatient in order to obtain a CT number (Hounsfield Unit) which is the ultimate goal of reconstruction.
µ = (1/x) * = (1/x) * lnln(I(I00/I)/I)(Algebraic manipulation of the original formula)(Algebraic manipulation of the original formula)
CT NUMBERS• unit for attenuation values within
a pixel (brightness)• brightness value is determined
by the detector signal
Slide # 74
y g• linear attenuation coefficient• absorption measurement• Hounsfield Unit
attenuation > water = +attenuation of water = 0attenuation < water = -
CT Number ValuesCT Number Values
Bone1000
Water0
Slide # 75
A 1% difference in µ = 10 HUHounsfield Units (HU)
Fat- 100
Air-1000
Godfrey N. HounsfieldNobel Prize: Physiology or Medicine 1979
CT NUMBERS• Attenuation coefficients
dependent on photon energy (keV)
• HU values depend on kVp and
Slide # 76
• HU values depend on kVp and filtration
• µPE is energy dependent• µCompton not as energy dependent• As kVp is increased, µPE is decreased• A high kVp is used in CT to diminish the contribution of µPE and
increased the contribution of µCompton
Linear Attenuation Coefficient – Beam Energy & Z
µ = µPE + µComptonBeam Energy
Slide # 77
µCompton
Atomic Number (Z) –The probability of photo-electric effect goes by Z3
Example: Bone Z (13) and soft tissue (8)13 ÷ 8 = 1.625 or roughly 2 and 23 = 8
Therefore, the probability of photoelectric effect is 8 times more for bone
CT Numbers• HU values generated by a CT
scanner are only valid for the effective kVp used to generated the image
Slide # 78
14
The voxel values are normalized to the µ of water to obtain CT numbers
µmaterial - µwater
CT Numbers
A 1% difference in µ = 10 HUHounsfield Units (HU)
Bone1000
Water0
Fat- 100
Air-1000
Slide # 79
x 1000CT number =µmaterial µwater
µwater
CT numbers are called HOUNSFIELD UNITS (HU)
Godfrey N. HounsfieldNobel Prize: Physiology or Medicine 1979
CT Numbers
• Comparison of linear attenuation coefficient of tissue to linear attenuation coefficient of water
Slide # 80
coefficient of water• uwater-utissue• uwater• multiplied by 1000
CT Numbers
• Greek letter mu• linear attenuation coefficient• CT numbers
Slide # 81
• Hounsfield Unit• Absorption measurement
Region Of Interest … ROI
512
Slide # 82
Mean 54STD 2.4
attenuation > water = +attenuation > water = +attenuation of water = 0attenuation of water = 0attenuation < water =attenuation < water = --
CT Numbers
Slide # 83
attenuation < water = attenuation < water = --
A 1% difference in µ = 10 HUA 1% difference in µ = 10 HU
Since CT scanners operate at high xSince CT scanners operate at high x--ray ray energies, Compton interaction is energies, Compton interaction is predominate. The probability of a predominate. The probability of a
Compton interaction depends primarilyCompton interaction depends primarily
CT Numbers
Slide # 84
Compton interaction depends primarily Compton interaction depends primarily on the density of the tissue.on the density of the tissue.
CT numbers are closely related CT numbers are closely related to tissue densityto tissue density
15
Since CT scanners operate at high x-ray energies, Compton interaction is predominate.
The probability of a Compton interaction depends primarily on the density of the tissue.
Compton InteractionCompton Interaction
CT numbers are closely related to tissue density
Slide # 85
CT numbers are closely related to tissue density
FYI…“A Compton interaction is one in which only a portion of the energy is absorbed and a photon is
produced with reduced energy.”+
Compton
backscatter
CT Numbers
• Less than water = Negative Number
• Greater than water = Positive Number
Slide # 86
• 0 = water, 1000 = bone
• -100 = fat, -1000 air
Normal RangeNormal Range
WaterWater BoneBoneAirAir
CT Numbers
Slide # 87
--10001000 +1000+100000
The Hounsfield ScaleThe Hounsfield ScaleHounsfield Scale
60
40
Calcified Bone
Gray Matter
Congealed Blood
+1000 Hu
Slide # 88
-1000 Hu
-100 Hu
20
0 Hu
White Matter
Blood
Water
Fat
Air
Extended RangeExtended Range
WaterWater BoneBoneAirAir
CT Numbers
Slide # 89
--10001000 +3000+300000
11 bit (11 bit (--1000 to +2000)1000 to +2000)12 bit (12 bit (--1000 to +3000)1000 to +3000)
Window Width / LevelWindow Width / Level
width
possible range of pixel valuespossible range of pixel values
10001000
Slide # 90
--10001000 +3000+3000
levellevel
16
Introduction to “Windowing”WW (Window Width) = Range of Gray Shades
that the user chooses to display for a given clinical situation
WL (Window Level) = Center of Gray Shadesthat the user chooses to display for a given clinical situation
Slide # 91
Wide Narrow
Narrow Window Width
Slide # 92
Narrow Window Width – only two shades of gray (black & white)
Low Window level High Window Level
Wide Window Width
Slide # 93
Wide Window Width – Many shades of gray
Low Window level High Window Level
possible range of pixel valuespossible range of pixel values
--10001000 +3000+300000
width = 100
Slide # 94
level37
possible range of pixel valuespossible range of pixel values
--10001000 +3000+300000
width = 3000
level
+1875-1143
Slide # 95
357
Window Width and Level
Grayscale with Narrow Window Width (only black and white)
Grayscale with Wide Window Width (lots of grays)
Slide # 96
Low level (darker) settings on an Axial CT image of the chest
High level settings (whiter) on an Axial CT image of the chest
Wide Window Width Wide Window Width
17
Window Width & Level
Slide # 97
High Contrast & Density
Slide # 98
You do not have to have raw dataYou do not have to have raw datain order to alter window width / levelin order to alter window width / level
Slide # 99 Slide # 100
Slide # 101 Slide # 102
18
CT Numbers• Linear attenuation coefficients
are converted to CT Numbers so conversion to a visual gray scale is possible
Slide # 103
scale is possible
Windowing• Each Pixel is represented by
4096 gray levels• larger than display range of
f
Slide # 104
monitors or film
Signal Numbers
The numbers contained in each matrix pixel are assigned a shade of gray
Slide # 105
Giving each pixel a number allows for visualization of an image on a TV screen
Numbers and Shades
The number or shade of gray that is assigned to each pixel
Slide # 106
is determined and proportional to the electrical signal from the detector
Basics of Digital Images
0 1 2 3 4 4 60 152 120 22 215 34 1 0 1171 3 114 199 134 88 20 60 1992 234 72 65 17 145 185 235 1813 141 214 169 134 85 234 237 684 241 154 141 231 145 236 35 275 45 95 65 127 123 94 47 1665 45 95 65 127 123 94 47 1666 127 98 137 149 67 45 52 297 162 81 83 189 69 195 94 1718 64 123 130 100 58 226 214 34
102 189 174 35 169 203 243 135213 235 219 137 22 195 168 208227 103 192 243 102 220 187 21
1 2
IMAGE MATRIX
Slide # 108
3 4
19
Signals Pixel Brightness
1 2
Slide # 109
1 2
3 4
WINDOWING
Computer can “ see” more gray shades than the human eyemathematically brings density or
Slide # 110
contrast differences into the visual range.WINDOW WIDTH ( WW )WINDOW LEVEL ( WL )
Window Width (WW)
• Range of CT numbers for the gray scale
• Range of gray shades used to
Slide # 111
display the image• Defines contrast of the image
WINDOW WIDTH ( WW )
Range of Gray Shades Displayed
Contrast of the Image
Slide # 112
Contrast of the ImageWW increases, image
contrast decreases
Window Level (WL)
• Center of the gray Scale• Density of the image
Slide # 113
Wide Window Widths• Tissues of
greatly differing attenuation
• Lung air spaces
Slide # 114
Lung air spaces and vessels
• Bone
20
Narrow Window Widths• Display soft tissues• within structures containing different
tissues with similar densities• Brain & Liver
Slide # 115
Window Levels• Select Near
average attenuations of tissues of interest
Slide # 116
interest• Not too high,
obscure pathology (brain bleed)
WINDOW LEVEL ( WL )
Center of the Gray ScaleCenter of the Range of gray
shades used for viewing an image.
Slide # 117
Density of the imageWL increases, Density increases.