Digital radiography

RI

Digital Image Formation:

• Digital image is divided into
matrix of pixels
• each pixel is Assigned a value 194
based on density (high values
for dark pixels , and low
values for bright pixels)
• The Value is the average of 73
the pixel
– Any spatial resolution within
a pixel is lost
• Pixel values form the digital
image
22

125 25 311 111 182 222 176

199 192 85 69 133 149 112

77 103 118 139 154 125 120

145 301 256 223 287 256 225

Dr.Yossef Gamal Www.ScoreTraining.net 16
178 322 325 299 353 333 300 RI
 Digital Image Formation
digitalization Analogue Image

Dr.Yossef Gamal Www.ScoreTraining.net 17

RI

• Matrix size = number of pixels e.g. 512 x 512 matrix

• The more is the matrix size , the larger is the field of view (with fixed pixel size)
• The more is the matrix size , the smaller is the pixel (with fixed field of view)
• The finer the pixels , the more is the spatial resolution
• Objects smaller than the pixel size will not be seen

RI
Effect of changing matrix size

RI

Numerical and binary system

20
RI
 Numerical system is decimal (10 base numbers)

Binary system
• Fundamental method of
computer storage
• two base numbers i.e. Only 2
allowable values
–0
–1
• Bit = binary digit
– Smallest binary unit
– has value 0 or 1 only
• Computers do all operations
with 0’s & 1’s
• no –ve power , no values
Dr.Yosseb
f Ge
amtaw
l een 0 & 1 Www.ScoreTraining.net 21
RI

• Bit depth = bits per pixels

– Single memory bit: each pixel has
ONLY two possible values (0 or 1)
– 4 bit depth: 16 possible values (2Ϻ)
, minimum = 0000 , maximum =
1111
– 8 bit depth : 256 possible values
(2Ͼ) , i.e. 256 levels of gray could be
displayed
– The greater is the bit depth , the
greater is the potential to display
good contrast images
–Factors affecting bit depth required:
1)Image noise : noisy images e.g.
radionuclide, bit depth used is not
greater than 8
2)Dynamic range of the detectors
(directly proportional) 4 grade shades 256 grade RI
shades
 Effect of changing bit depth

RI

• Byte : method of expression of computer memory

– 1 bytes = 8 bits
• Example:
A CT image with matrix size of 512 x 512 pixels , and bit
depth of 12/pixels , how much is the memory
required to store this image?
total number of bits saved 512 x 512 x 12
= ------------------------------------ = -------------------
8 8
= 393216 bytes = 393 KB

N.B: IN 8 bit depth system , memory

required to store each pixel = ………………..

RI
 Analog vs. Digital Images
• Analog
– continuous gray
shade information
• Digital
– Discrete gray
shade information

RI

Compression
• Benefits:
1)Reduce storage requirements
2) Faster transmission times
• Types:
1)ossless (reversible):
-Image can be restored into identical version of original
-Allows reduction of size up to 2-3 times (depending on
the image quality)
2)Lossy (irreversible):
-Displayed image does not perfectly reproduce the
original
- Allow reduction of size up to 40 times
Dr.Yo-sM
sef Gayal not be acceptablWewwt.SocorebTraeininug.nseet d in primary diagnosis26
am
RI
 Principles of Digital Radiography
• Digital imaging comprises four
steps:
• Generation: energy absorbed by
the digital detector is
transformed into electrical
charges ї recorded and digitized
• Processing: organize raw data into
a meaningful image.
• Archiving: images are sent to a
digitized storage archive
• Presentation: digitally on a workstation (or hard copy), Image
can be manipulated e.g. zooming, windowing etc.
RI

Types of digital radiography

RI
 Computed radiography (CR)
•Depends on phosphor imaging
plates which:
9Replace film & screen in the
cassettes
9Re-usable
9Uses conventional bucky & x-
ray equipment

RI

Computed radiography (CR)

Detectors: Storage phosphor image plates:
-crystal of barium flourohalide doped in europium
-halide is 85% bromide , 15% iodide
(photostimulable phosphors)
-present in unstructured way
-laid down in a base (0.3 mm thickness)
-a surface coat is present to protect from damage

30
RI
 photostimu Iable phosphor

>-
.._
0 ')
Q,) Conduction band
c
w I
------------- ..- - - I
I
I
, I
2 ) Laser light
I
I
I
I
electron trap
I

1) X-ray
II
I
> 3) Light emission

.
I
I

Valence band

31
RI

Exposure process :
• X-ray absorbed and temporarily stored by bringing
electrons to a higher energy levels (electrons traps)

ELECTRON
TRAP

RI
Readout process
•Done By using flying spot
scanner
•Separate step
•Detective layer is scanned
pixel by pixel using high
energy laser beam of specific
wave length (using rotating
mirror)
•Stored energy is set free as
emitted light
•Intensity of emitted light
indicates amount of radiation
PMT
incident on phosphor at each
location
RI

• Light guide (optical fibers)

convey the light to array of
photomultiplier tubes which
collect the light and
measure its intensity and
then convert it into
electrical charge

RI
 • The emitted light should have wave length
different from that of the laser beam so that
Only color of light emitted by phosphor
measured by PMT
• Most phosphors emits blue light їŶĞĞĚƐůĂƐĞƌ
emitting red light

RI

scan direction
laser scanning
PMT mirror
tube

sub-scan
direction

plate translation

Read
CR plate

Position of light emitting center is determined from the

time at which the light is received

RI
Then The charge output of PMT is digitalized by analogue to digital
convertor

RI

• Following the read cycle , the residual plate

signal is erased by exposing it to bright light
(why?)
halogen lamp

Erase

RI
 Summary

RI

• Discharge fraction F
– Fraction of trapped
energy released by
stimulating laser
light
– Depends directly on
laser power incident on
the plate (not linear)
F

– F depends also on
• time to read a plate
• imaging plate type
• the plate size

–
Dr.YoT
ssey
f Gpical al figure for F is 0W.w5w.ScoreTraining.net
am 40
RI
 • Notes :
- The energy stored in the detective stays for several
hours depending on the physical properties of the
phosphor crystals, yet , readout process should be
done immediately before the stored energy decrease
over time
- Readout process of a 14“ x 17“ image plate takes 30-40
seconds (maximum workload of 90-120 image plate/h)
- Stacking reader are available in which several cassettes
is placed in queue for automatic feed in the reader

RI

CR image processing
• Photostimulable
phosphors have very wide
latitude (dynamic range)
= 10000:1
i.e. able to record photon
intensities varying by
factor of 10000:1 , with
linear relationship
between dose and
response (In film screen =
16:1)
i.e. very wide latitudeї
very low film ɶ(0.4)

RI
 Steps of image processing
1- image segmentation:
Detect the edges of intensities of x-ray beam
Outside that edgĞƐїsignal is ignored
i.e. Separate raw radiation from anatomy

RI

2) Histogram analysis of intensities

within the collimated area
• Plot of gray scale value vs. the
frequency of occurrence within
the defined area in the image
• Shape of the histogram is
anatomy specific
• Sets minimum and maximum
“useful” pixel values
• Very јƐignal (outside the body)
and vĞƌLJљ signal (e.g metal) are
rejected

RI
 3) Look up table transformation (Tone Scaling):
• Map useful intensity values to the linear portion of the
characteristic curve
• This process is optimized for each particular projection
• Determine the contrast in CR

RI

LUT Selected depends on

•Body part
•projection
User can also alter LUT manually

RI
 PREPROCESSING POSTPROCESSING

RI

CR spatial resolution
• Spatial resolution of CR is less than conventional
film screen radiology due to Scattering of laser
light in the phosphor layer
• To increase CR resolution (e.g. mammo):
1 use thinner phosphor
2 decrease diameter of scanning laser beam

RI
 In radiographs with large image plates (e.g. chest &
abdomen) low resolution plates are used to decrease
imaging time and file size

RI

Advantages of CR:
1- Wide dynamic range (latitude) ї decrease rates of failed x-ray exposures

Film screen radiography

CR

50
RI
 2-Cassette based , so that:
- Easily integrated into existing
radiographic devices
- Highly mobile (used in bed
side exams)
3- If single image plate shows defect
ї can be easily replaced by the
radiographer with no need to
change equipment

RI

• Disadvantages:
1 spatial resolution is lower than screen film , yet
equivalent diagnostic value
2 image quality and diagnostic value is less than
DR
• Used in
– General diagnostic & mobile radiography
– Mammography
– Dental (with small plates!)

RI
 Digilal Radiography

CR DR
Computed Radiography Direct Radiography

Indireet coflversion Indirect conversIon Direct conversion

Photoconductor
( Scintilla or-TFT
FPD
IStorage phosphors I Sefeniu m r u m

Image 1ntensifier

Dr.Yossef Gamal Www.ScoreTraining.net 53

RI

Direct radiography
• Receptor provides electrical charge output
• No need to reader, No intermediate steps (compare to CR)
• Images available in < 15 seconds
• Much less work for technologist
• Greater throughput

Dr.Yossef Gamal Www.ScoreTraining.net 54

RI
 Digilal Radiography
I
f
CR DR

*
Computed Radiography

t
Direct Radiography
I

,
11rtdi eel car1version K Ind rect conversIon .......

t> Direct conversion

/

, [ Scintillator-TFT I
Photoconductor
FPD
IStorage phosphors I S d ntJIIator..CCD I Selen um rum

I I mage 1ntensifier 1

Dr.Yossef Gamal Www.ScoreTraining.net 55

RI

Light

• Indirect DR: x-ray photon ї^Đŝntillator їůŝŐht

photons їƉŚŽtŽĚŝŽĚĞƐї produce electrons
(electrical chargĞͿїreadout process

Dr.Yossef Gamal Www.ScoreTraining.net 56

RI
 Digilal Radiography
I
f
CR DR

*
Computed Radiography

t
Direct Radiography
I

,
11rtdi eel car1version K Ind rect conversIon .......

t> Direct conversion

/

, [ Sci ntillator-TFT I
Photoconductor
FPD
IStorage phosphors I /
S d ntJIIator..CCD
' Selen u m r u m
/
""""'
I Image 1ntensifier 1

Dr.Yossef Gamal Www.ScoreTraining.net 57

RI

Indirect conversion using CCD (charge-

• Scintillator: coupled device )
-converts X-ray to light (phosphor)
- Scintillator used is T1-doped cesium iodide
• CCD:
- Made of crystalline silicon (light sensitive)
- Consists of integrated circuit containing array of linked capacitors
- Each capacitor represents a pixel
- Light falls on each pixel , electrons (charge) liberated and build up in each pixel
- i.e. convert light output of the Scintillator to electrical charge

RI
Two types:
1) lens coupled CCD system
• CCD chip can not be manufactured larger than 5 cm²
• ї Array of several CCD chips are combined to create larger detector area
• Optical lens is used to decrease area of projected light to fit CCD array
• Disadvantage:
-the number of photons reaching CCD is decreased їљ quantum
efficienĐLJїљ S/N ratio,

RI

2) Slot scan CCD:

•Collimated fan shaped X-ray beam
scan the patient
•CCD detector array of matching width
is simultaneously moving
•Advantages:
-љscatter radiation in the image
-љŶoise : compensate the relatively
low quantum efficiency of lens
coupled CCD system
•Disadvantages: needs fixed
installation ї used in chest , mammo
and dental films
•N.B: Lens coupled CCD is lower than
slot scan system in quantum
efficiency& S/N ratio

60
RI
 Digilal Radiography

CR DR
Computed Radiography Direct Radiography

Indireet coflversion Direct conversion

. - o r - 11
Photoconductor
FPD
IStorage phosphors I Sefenium rum

Image 1ntensifier

Dr.Yossef Gamal Www.ScoreTraining.net 61

RI

Indirect conversion using flat panel

detectors
• Flat panel detector is formed of:
1) Scintillator
2) Amorphous silicon photodiode circuit layer
3) TFT array (thin-film transistor)

light

RI
1) Scintillator layer:
Emit visible light when exposed to X-ray
Types:
A. Cesium iodide (CsI):
Crystals are shaped into 5-10 μm wide needles , arranged
perpendicular to the detector surface
This structured array decrease light diffusion in the
Scintillator layer їƚŚŝĐker scintillator layer can be used
with:
1 increase of absorption efficiency without loss
of spatial resolution
2 increase intensity of the emitted lighƚїј
quantum efficiency (highest) and better optical
properties
Disadvantages: highly vulnerable to mechanical load (fine
structure) RI

B- gadolinium oxi-sulfide:
• Unstructured Scintillator
• Advantages :
- Cheaper
- Resistant to mechanical stress їƵƐĞĚŝŶ portable flat
panel detector system
Disadvantages:
- Lower quantum efficiency

RI
 2) Amorphous silicon photodiode circuit layer connected
to TFT array:
• a made up of small (about 100 to 20ϬʅŵͿ detectors
(pixels). Each pixel contains a photodiode
• Amorphous silicon photodiode Converts light to
electrical charge
• During exposure charge is built in each TFT pixel
• After exposure is complete , charge in each detector
element is released by applying high potential (read
out)

RI

TFT Readout
• All switches are turned
off while the x-ray
exposure is made
• At the end of exposure,
the switches in the 1st readout digitzer
row are turned on charge amplifier
• Charges from the pixels
switch line 1
are transferred to the
charge amplifiers
switch line 2
• Charges are digitized
and stored
• row are switched off switch line 3
and the next row is
addressed
Dr.Yossef Gamal SwiWtcwhw.ScoreTraining.net pixel 66
RI
 switch

• Fill factor

– Not all of the pixel is sensitive to signal as a

portion of the pixel is taken up by the switch
and the storage capacitor

– The fraction of the pixel that is sensitive to the

signal is called the geometrical fill factor

RI

Advantages of Flat panel detector indirect DR:

• Small size : can be integrated in existing bucky tables
and thorax stands
• Image generation is almost real time process (time
between exposure and image display is less than 10
sec.)
• High throughput
• High image quality
Disadvantages:
• Any detector defect їcomplete system breakdown
(contingency image device is necessary)

RI
 I Dig ilal Radiograp hy

,CR DR
,Computed Radiography Direct Radiography'

Indlrect conversIon

Photoconductor
Scintilla or-TFT
FPD
IStorage phosphors I Scintilla or-cc D

Image intensffier
1

Dr.Yossef Gamal Www.ScoreTraining.net 69

RI

Idea of direct conversion:

Direct-to-Digital Radiography (DDR)

RI
 Photoconductor
- Type of detectors which convert X-ray photons into free electrons , which cause electrical
charge
- Most important : amorphous selenium
– Z= 34 : low compared to a typical phosphor ĺ need thick layers for good
absorption efficiency
– Uniform to a very fine scale (amorphous) ĺ The resolution is independent of its
thickness
– Free of the structure noise seen in most phosphors
- Other types: Lead iodide ,Lead oxide, Thallium bromide, Gadolinium compounds

high voltage incident x-rays

electrode

photoconductor
(-)
charge (+)
cloud

RI

- Process:
– X-ray generates electrical charge
(free carriers)

– electrode is attached to both

sides to apply Electric field
across the photoconductor to
guide the charge

– The greater the field strength, the

quicker the charges travel to the
electrodes, the less is the time for
lateral spread in the material the
more is the resolution
– The surface of the a-Se have small
transverse conductivity to maintain
spatial resolution

- The electrode contact must prevent

charge from the electrodes to enter a-Se
layer (blocking contact)

- Pixel size is not dependant on the detector ,

but on the read out only (highest spatial
resolution)

RI
 Types of direct conversion DR
1) selenium drum:
• Rotating selenium dotted drum
• +ve electrical surface charge
• X-ray exposure їƐƵƌface charge pattern ɲ X-ray
intensities
• Disadvantage: no mobility due to the mechanical
design

RI

2) Selenium based flat panel detectors:

Consists of:
Layer of selenium (convert X-ray to electrical charge)
Underlying array of TFT for readout
Advantages: can be mounted on thorax stands and bucky tables
Used in mammography

N.B: Digital detector function improves as radiation exposure is increased

RI
 Image post-processing

RI

Image post-processing
• Manipulating readout raw data , examples:
A) Low pass resolution filter:
• Each pixel is replaced by the weighted average of
the neighborhood pixels
• Effects : smoothen the image їĚĞĐrease the
noise
• Disadvantage: blur some details and edges

RI
B) High pass filter = edge enhancement:
• Increase the difference of grayscale values between
neighboring pixels
• Effects: exaggerate the contrast at boundary between
structures making the edges more visible
• Disadvantages: increase the noise , may generate false
structures in the image

RI

• Notes:
1)Spatial resolution can not be influenced by
processing , yet , by changing processing
variables (e.g. edge enhancement) low spatial
resolution can be partially counteracted
2)Processing algorithms are adapted for each
anatomical regions

RI
 Modulation transfer function (MTF)
• Definition: ratio of output and input modulation
i.e. capacity of the system to transfer modulations of the
input signal at a given spatial frequency to its output
i.e. A measure of information loss in the image compared
to the object

RI

i.e. MTF = amount of blur at certain spatial frequency

;јƐƉatial frequency їјďůƵƌͿ
• MTF varies with spatial frequency from 100% at low
spatial frequency , towards zero at higher spatial
frequencies
• Practical limits of MTF : 0-1 (when will it exceeds 1?)

80
RI
• MTF depends on the system and the
spatial frequency of the area
• MTF of the system = product of its
individual components
• Fine structures with sharp edges
corresponds to high spatial frequency
їrequire system with high MTF
• A common way to compare two
systems is quoting a figure at which
MTF = 10% for each
• In general radiology : relevant image
details are usually less than 2
cycles/mm
• In mammo: details are › 5 cycles/mm

RI

MTF of Direct flat panel is

the best as the electric
field in the selenium layer,
necessary to separate the
charges, inhibits the lateral
diffusion of the charge
cloud, thus preserving a
high resolution or MTF

RI
Fourier transform
• Technique of
converting from
spatial to
frequency domain
i.e. image signals are
converted to sine
waves in terms of
spatial frequency
and amplitude
• Fourier methods
are used in digital
image to enhance
the display of
image detail

RI

Sampling
• Image sampling is the process used to
digitalize the analogue spatial information in
digital image
– i.e. Process of creation of digital image
from the Fourier sine waves of many
frequencies
• Sampling frequency:
– Rate at which the analogue wave is
sampled to create digital image
– Determine the pixel size in the digital
images
– sampling frequency = 1/pixel size

RI
Nyquist law
• Signal must be sampled
at least twice/cycle
• i.e. sampling frequency
must be at least twice
of the highest
frequency present in
the image

RI

Good sampling under sampling

RI
 Nyquist frequency
• Maximum signal frequency that can be accurately sampled
= maximum achievable spatial resolution (in cycles/mm)
sampling frequency 1
= --------------------------- = --------------------------
2 2 x pixel size
e.g. pixel size = 200 ђŵї maximum detectable spatial frequency = 2.5 cycles/mm
• Aliasing: high frequency signal is faultily recorded low because sampling
frequency is ‹ Nyquist frequency

RI

Spatial resolution in digital radiology

• In digital system : it is limited by the pixel size
• Scattering of light photons decrease spatial
resolution so that:
1 spatial resolution of direct conversion DR ›
indirect conversion DR
2 spatial resolution of structured Scintillator ›
unstructured Scintillator

RI
 Spatial resolution of different digital systems

phosphor x-ray x-ray

x-ray x-ray grains
+V

(-)
(+)

light

charge

CR INDIRECT DR INDIRECT DR photoconductor

BaFBr0.85I0.15:Eu2+ GdO2S2 CsI:Tl a-Se

RI

Detector Dynamic range

• Range of X-ray over which meaningful X-ray image is obtained =
latitude
• Digital detectors have very wide dynamic range
• Results:
1 low risk of failed exposure
2 difference between images (e.g. bone, soft tissue and air)can be
seen in one image without need of additional images

90
RI
 Detective quantum efficiency (DQE)
(S/N ratio at detector output)²
= ---------------------------------------------------
(S/N ratio at detector input) ²
• Describes how effectively an x-ray imaging system can produce an
image with a high signal-to-noise ratio relative to an ideal detector
• Detector with high DQE =
1 has high photon detection Efficiency:
јDQE ї less radiation needed to achieve the same image quality ї ј
image quality with the same exposure
2 low noise is added to the detected signal
i.e. added to the quantum noise that is present in the X-ray beam
• If all X-ray photons are detected and no noise is added ї DQE = 100%

RI

• Factors affecting DQE:

1- Spatial frequency
јƐƉatial frĞƋƵĞŶĐLJїљY

RI
2) detector type:
• Depending on the absorption efficiency , structural noise and MTF of the
detector
• Indirect DR › direct DR › CR
• CR and film screen has comparable DQE
N.B. Low selenium Z are the cause of moderate X-ray absorption and
relatively low DQE (despite very high MTF)

3) Fill factor:
јĨŝůů factor їјĨraction
of absorbed photons
їјY

RI

4) Exposure levels:
• The DQE of the CR detectors and is best for
the low-exposure level.
• DQE of the indirect detector has no
dependency on exposure level.
• The best DQEs of the direct detector is
obtained at the highest exposure level

RI
 Radiation exposure with digital systems
• Most of digital detectors decrease the
exposure in comparison to film screen
because:
1) јY їůower exposure for the same image
quality (indirect DR has highest potential to
decrease the exposure
2) Wide dynamic rangĞїĚĞcreased number of
failed exposures

RI

Detector dose indicator

• Effect of wrong exposure in film screen system

RI
 • Effect of wrong exposure in digital system
• Almost impossible to under or overexpose CR / DR

RI

Detector dose indicator (DDI) = exposure index

• Each manufacturer provides exposure feedback to technologist
• Displayed on CR reader monitor
• Displayed on DR workstations
• If DDI is higher than recommended range, patient is overexposed
• Manufacturers provide normal range of DDI which is exam dependant
• Some manufacturers give DDI inversely proportional to dose ;ј DDI =
underexposure)
• Others give DDI = log exposure , i.e. doubling the dose = increase of DDI
15%
Receptor Kodak Fuji S
Exposure Number
EI
0.5 1700 400
1 2000 200
2 2300 100
4 2600
Www.Scor eTraining.net 50 RI
 Soft copy viewing
A) Cathode ray tube (CRT)
• Scanning electron beam (its intensity is modulated
according to pixel value)
• 1250 scan lines
• Resolution in the perpendicular plane of lines depends on
the number of lines
• Resolution parallel to the plane of lines is limited to
modulated signal frequency
• They have a big back and take up space on desk

99
RI

B) Flat panel monitor (Liquid crystal display = LCD)

• Screen is formed of pixels (2500 x 2000)
i.e. double the lines of CRT
• Zero geometric distortion

100
RI
 PACS (Picture archiving and
communication system)
• Networked group of
computers, servers, and
archives to store digital
images
• Uses of PACS:
1) Provides image access to
multiple users in multiple
locations at the same time
(not only in the hospital)
2) Secure storage of the
images (with elimination
of film stores)
3) Integration of the images
with other electronic
patient’s records
• Disadvantage: Cost
RI

• PACS workflow manager:

Computer in the heart of PACSs that controls flow of images and
information (send images from and to the archive)

• Types of the archives:

1 short term: for current cases (e.g. inpatients) їrapid access
on demand
2 long term: images that unlikely to be of immediate interest ї
access time is slower
Images are transferred between long term to short term
automatically
3 backup archive : for security

RI
PACS is connected to other information systems:
1 hospital information system: patient
demographics…etc.
2 radiology information system: appointments… etc.

PACS broker: interface between work flow manager and

these systems

RI

workstations that access the image:

1 Modality workstation: supplied with the image equipment , may
be able to display images from archives
2 Reporting workstation: used by radiologist , must have high quality
display monitors and software tools for image manipulation
-Display environment must be suitable: background lightening is
kept low (extraneous light reduce film perceived contrast )
3 Review workstations: with lower quality monitors and limited
software(reporting from these workstations must be done with
caution)
N.B: Access to archives is available through web browsers (with
passwords) , should not be used for diagnosis

RI
 Image file storage:
• Stored in the form of DICOM format
• Film file contains:
1 Basic digital data that allow display of the
image
2 Annotations
3 Display preferences ….etc.

RI

Quality assurance for

display monitors
SMPTE test pattern:
• Areas of varying
brightness from 0% to
100% in 10% steps
• Smaller (5%) contrast
squares within zero
and 100% squares (if
not visible їŵonitor
is out of calibration)
• Do not adjust the
image using contrast
and brightness
cDor.Yonssetf rGaomall s Www.ScoreTraining.net 106
RI
Dr.Yossef Gamal Www.ScoreTraining.net 107
RI