A |
|
Acquisition
and Reporting workstations |
|
The acquisition workstation
is where the data is initially received and pre-processing will
be undertaken. Ideally the image that appears should not require
further processing.
Reporting workstations can be a component of the PACS or dedicated
to a digital unit. In either case, these will be high specification
workstations with a comprehensive range of post- processing
facilities.
|
Air gap |
|
The gap between an object
to be imaged and the receptor; a, used in magnification examinations
or b, used instead of a grid to reduce scatter radiation reaching
the receptor.
|
Aliasing and the
Nyquist frequency |
|
The pixel
pitch (d), is of great importance in dictating the limiting resolution
of a digital system. According to sampling theory, only spatial
frequencies below a certain level can be imaged accurately. This
is known as the Nyquist frequency (Nv).
Frequencies higher than this
will appear as a lower frequency component of the image. This
is known as aliasing. When measuring high contrast spatial limiting
resolution, aliasing can result in the dark and light bars being
effectively reversed.
Theoretically, the first zero of the MTF is at twice the Nyquist
frequency (assuming that the active pixel dimension is equal to
the pixel pitch). If the signal between the Nyquist frequency
and the first zero of the MTF is kept as low as possible, the
amount of aliasing will be kept to a minimum.
|
a-Se/TFT flat panel detector |
|
The detector
consists of a layer of a-Se with a matrix of electrodes on each
face. The X-ray energy produces electron-hole pairs in the selenium
layer, which are attracted towards the electrodes by an electric
field. The charge is collected and readout using a TFT array.
The use of a-Se based photoconductor systems has the advantage
that there is only one energy conversion while a phosphor requires
both the conversion of X-ray energy to light and then light to
charge. Therefore, a-Se based systems do not suffer from light
spread prior to readout (figure 3). In common with a-Si based
detectors, the limiting resolution of these detectors is limited
by the effective pixel pitch.
|
Analogue to Digital converter (ADC) |
|
Converts
an analogue electrical signal to a discrete digital values so
it can be processed in a digital system.
|
Automatic Brightness Control (ABC) |
|
An ABC
system control the x-ray tube output of a fluoroscopic system
to maintain a constant image brightness at the image intensifier
output.
|
Automatic Exposure Control (AEC) |
|
An AEC
for a film/screen system is set up by ensuring the correct optical
density is achieved across a range of kVs. This method is not
practical for digital imaging as the image will always be displayed
optimally, irrespective of the exposure used. The AEC will need
to be set up in collaboration with the radiology, medical physics
departments and the supplier. The level of the exposure must be
optimised and the receptor dose measured.
One other consideration is that sometimes when film/screen systems
are replaced by a CR system then for simplicity the AEC is kept
at the same settings, which allows film/screen still to be used.
This may not be the optimal working level, in addition the CR
plate will have a different response to change in the kV and so
the AEC must be set up differently.
|
|
|
B |
|
Beam hardening |
|
Is the
increase in the mean energy of a x-ray beam as the lower energies
are absorbed. Beam hardening by the use of filters in the x-ray
tube reduce the lower energy x-rays that would be absorbed by
the patient, thereby adding noting to the image but increasing
the does to the patient.
|
Binning of data |
|
This
is useful for reducing the size of a data set into a manageable
amount, and/or averaging data. A set of bins is defined as a set
of evenly spaced ranges of values e.g. 2.5-3.5, 3.5-4.5 etc. Data
is then collated into the appropriate bin depending on the value
of the independent variable. Once all the data is collated into
the bins, the dependent variable is the average of all the dependent
variables in that bin. It can be a useful process to bin data
for example in producing a edge spread function. |
Bit |
|
The smallest
unit of computer memory, either on or off which equals one or
zero. Bit is from the words binary
digit.
|
Bit depth |
|
See
Image size and bit depth.
|
Bucky |
|
A device that moves
a grid so as to blur the grid lines in an image.
|
Bucky factor |
|
The
ratio of incident to transmitted radiation for a given grid (See
also grid factor).
|
Byte |
|
The smallest
unit of computer storage which can represent a character. Byte
is from the words binary
term.
|
|
|
C |
|
Charged couple device (CCD) |
|
A CCD is
an array of light sensitive elements which release electrons when
light falls on them. The electrical signal produced is proportional
to the intensity of the light.
|
Computed Aided Detection (CAD) |
|
This is a program designed to examine
an image and highlight areas of particular morphology. It should
be noted that this is detection and not diagnosis, the tool
is to help the radiologist identify areas of interest, the radiologist
then uses their clinical judgement on the importance of the
highlighted area. The main area of development for CAD is in
mammography.
A good CAD system will learn through use and become a more useful
tool. Ideally CAD should be used as a double check, as the combination
of CAD and a radiologist gives better results than either operating
singly.
|
Computed Radiography (CR) |
|
CR is in
first appearance similar to use as a film/screen system. The CR
plate is in a cassette which will fit the table and vertical Bucky
trays and can be used with a mobile unit on the wards. The plate
is then scanned in a reading system similar in size to a daylight
processor. This therefore makes the change to digital radiography
easier. A DR system entails more changes in x-ray couch and vertical
Bucky design and often changes to the x-ray tube assembly.
The X-ray photons interact with the image plate phosphor exciting
electrons to a higher energy level. Although, some of these instantly
de-excite and release light photons, some are stored within meta-stable
energy traps formed by the Europium dopant where, due to thermal
processes, they gradually de-excite. Therefore, it is important
that the image plate is read as soon as possible post irradiation.
The electron pattern in the phosphor makes up the image, as the
number of excited electrons is directly proportional to the number
of transmitted X-ray photons.
|
|
|
D |
|
Detective Quantum Efficiency (DQE) |
|
DQE is
a frequency dependent measure of the efficiency of an imaging
system. It is the square of the SNR of a detector compared to
that of an ideal detector, assuming that the noise of an ideal
detector is quantum limited. Therefore, DQE is a measure of the
SNR squared in the image compared to that in the transmitted X-ray
beam.
The image SNR squared of the system under test can be considered
to be the NEQ. Quantum limited noise obeys Poisson statistics
which states that the noise is proportional to the square root
of the number of photons. Therefore, the SNR squared for the ideal
detector becomes Q. The equation for DQE simplifies into the equation
below:
|
Detector calibration |
|
Digital
imaging devices have the potential for non-uniformity correction
using a pixel dependent correction matrix calculated from acquisitions
of nominally flat or uniform field exposures. The correction matrix
is usually stored separately to the image data and applied during
the image display process. The frequency and time required to
acquire the calibration data varies considerably. Some systems
may require calibration during service, whereas others may require
daily calibration by the users, or the system may self calibrate.
Thermal noise produces a dark signal, which manifests as pixel
values that are present without the application of X-rays. Some
systems may use dark field calibration to correct for these processes.
This involves the automatic acquisition of an image, without the
application of X-rays, with a time equal to that of the actual
diagnostic exposure. This is subtracted from the image data and
does not slow the process to an unacceptable level.
On certain systems, the digital detector can be directly cooled
to reduce noise. The cooling may be thermo-electric or by circulation
of a cooling fluid. In the latter case a heat exchange device
will have to be accommodated. A temperature interlock may be provided
such that the system will only operate within the correct temperature
range.
|
DICOM |
|
Digital
Imaging and Communications in Medicine (DICOM) Version 3.0 is
a protocol based standard to facilitate the transfer of digital
images and associated information between devices manufactured
by a range of vendors and to aid in the development of picture
archive and communications systems (PACS).
When purchasing a system, a DICOM conformance statement should
be received, this will inform how the device and software conform
to the standard. It must be noted that a conformance statement
indicates the particular sub set of DICOM conformance that is
used. If a system does not produce images in DICOM format then
a DICOM broker may be needed to convert the image to DICOM and
allow the system to be connected to the PACS.
|
Digital applications |
|
Digital
imaging is used in a number of imaging techniques. The following
table indicates what each is suitable for.
CR |
DDR |
general radiography |
chest radiography |
mobile radiography |
general radiography (not mobiles) |
full field mammography |
fluoroscopy |
|
small field mammography |
|
full field mammography |
|
Digital technology |
|
There are
a number of technologies used for digital imaging in planar radiography.
They can be divided into Computed Radiography (see
CR) and Direct Digital Radiography (see
DDR).
|
Direct Digital Radiography (DDR)
technology |
|
The
main X-ray technologies used in DDR are:
|
Dual energy subtraction |
|
This
allows the subtraction of bone or soft tissue from an image, allowing
soft tissue below bones to be easily visible. If an object containing
only one type of material is imaged then the thickness at any
point can be calculated providing the x-ray spectra, exposure
and energy response of the detector is known. The body can be
grouped into two sub groups of bone and soft tissue with similar
x-ray absorption properties. If a body is imaged twice with different
energy spectra, this can be undertaken at two kVp values or at
the same kV with and without a metal filter, this then becomes
a simultaneous equation, which can easily be solved for bone and
soft tissue. There should be no movement between the images otherwise
the images will need to be corrected, to reduce this problem,
the images will need to be taken in rapid succession or simultaneously. |
|
|
E |
|
|
|
F |
|
Fill factor |
|
For flat
panel detectors, a proportion of the detector contains the readout
circuitry and will be insensitive to the incoming light photons
or electrons. This leads to the concept of the fill factor (see
equation below), which is the ratio of the sensitive area of the
pixel to the effective area of the detector element itself.
Any improvements in resolution will require a reduced pixel pitch.
The fill factor will decrease with improved resolution as the
readout electronics will take up a larger proportion of the detector
element and decrease the detector sensitivity.
|
Fourier transform |
|
Any signal
can be produced by adding sine waves of different frequencies,
amplitudes and phases. The Fourier transform splits the real image
up into its component frequencies. Repeating the Fourier transform
on the spatial frequency image returns it to the real space. |
|
|
G |
|
Grids |
|
Low grid
densities can cause interference patterns in the image called
Moiré patterns. This can be solved by using moving grids
or high density grids of over 60 lines per cm. When using CR,
the grid lines must also be perpendicular to the scan lines in
the reader.
|
Grid factor |
|
The amount
an exposure needs to be increased by when a grid is used to maintain
the same image density (see also bucky factor).
|
Grid ratio |
|
Is the
ratio of the height of the lead grid slats to the distance between
the slats. The greater the grid ratio the more effective the grid
is at removing scatter radiation from the beam.
|
|
|
H |
|
HIS/RIS |
|
Hospital
Information Systems (HIS) and Radiology Information Systems (RIS)
contain patient details and examination information. If the digital
system and PACS is connected to a HIS/RIS this can speed up the
workflow of a department by bringing the information to the department
to give improved patient throughput. The patient details are then
connected to the image, for easier reporting and image retrieval.
The system uses HL7 (Health Level 7) language to transfer the
patient details.
|
|
|
I |
|
Image compression |
|
Compression
of images is useful, the smaller an image is the more that can
be stored and the faster the transmission. Compression can either
be lossless or lossy. Reconstruction of the image from a lossless
compression will be the same as the original image, while lossy
will have some changes. Some system while lossy will claim to
be “visually lossless”, i.e. there is no perceptible
differences between the original and the reconstructed image.
Compression ratio is defined as the ratio of the image size to
the compressed image size such that a 4:1 has reduced the image
size by a factor of 4.
|
Image optimisation |
|
Image quality
is related to the radiation exposure received by the detector.
Although a relatively low exposure will result in a noisy image,
it may still contain sufficient information to be diagnostically
acceptable. A high exposure will result in improved image quality
since quantum noise is reduced. However, image quality improvement
is not linear, it will eventually level off as the quantum noise
becomes less dominant and decrease as the plate becomes overexposed.
Ideally, a system should be set up to obtain adequate image quality
for the lowest possible dose.
|
Image phosphor coupled to CCD(s) |
|
A CCD comprises
a series of electrodes deposited on a semi-conductor substrate
to form an array of capacitors. Voltages are applied to each of
the electrodes to produce storage well locations. CCD technology
requires an image phosphor to convert the X-rays into light photons
which are read by the CCD camera. Photoelectric interactions within
the semi-conductor result in the light photons being converted
into electrical charge which resides in the charge storage wells.
Voltage switching is used to transfer the charge from the storage
wells to the read out circuitry.
For chest imaging systems, the dimensions of the CCD are typically
smaller than the required field size. Either fibre optic bundles
or lens/mirror systems are used to couple the image phosphor to
the CCD and produce the demagnification required. The coupling
efficiency decreases as the amount of demagnification increases,
potentially increasing noise and thereby reducing SNR. More than
one CCD can be used to cover the image phosphor. This is referred
to as tiling of the detectors.
|
Image processing |
|
Ideally, the acquisition
workstation should enable the following:-
- addition of anatomical markers
- demographics correction
- image annotation
- window and level adjustment
- electronic collimation
- magnification
- application of different LUTs.
The reporting workstation
will be a high specification workstation with a comprehensive
range of post-processing facilities, which include all of those
listed previously, plus other functions including:-
- edge enhancement
- noise reduction
- tools for measuring pixel values, distances
and angles
- zoom and roam.
|
Image retention or lag |
|
A proportion
of the signal from the previous image is added to the next image.
Image retention should be kept low for radiographic imaging, while
some fluoroscopy systems add lag to keep the noise low, this is
suitable for images which are not changing rapidly.
|
Image size and
bit depth |
|
The matrix
size is the number of pixels or memory locations into which the
image is divided. Thus the total number of pixels in a 1024 by
1024 matrix is 1,048,576, or approximately 1M.
Contrast resolution depends on the bit depth. The analogue value
of the output from each pixel is converted to digital form, and
the results stored at a separate location in a matrix. The number
of grey levels available equals 2 to the power of the number of
bits.
Number of grey levels per
bit size
Number of bits |
Grey Levels |
1 |
2 |
2 |
4 |
4 |
16 |
8 |
256 |
10 |
1024 |
12 |
4096 |
Clinical images require good
contrast resolution, which can be difficult to achieve due to
the noise inherent in a radiographic image. In order to achieve
good contrast resolution, high bit depths are required.
|
Image stitching |
|
Image stitching
software is being developed by several manufacturers. Image stitching
is designed to allow the visualisation of long areas of the body,
such as the legs or spine. In CR the images are acquired by using
several CR plates to cover the required anatomy and manufacturers
are now introducing special holders where multiple plates are
positioned vertically. In direct digital radiography the x-ray
tube and detector are moved in step across the anatomy. Individual
images are then electronically stitched together by the software.
|
|
|
J |
|
|
|
K |
|
k-space |
|
See Spatial
frequency |
|
|
L |
|
Look Up Table (LUT) |
|
An LUT converts each
pixel value into a new value. An LUT has two main uses:-
-
a digital detector generally
has a much wider dynamic range than the range of intensities
in a clinical image, therefore an LUT is used to compress
the data to cover only the clinically useful data
-
an LUT may not be a
linear relationship, but may be curved to enhance the contrast
in the clinically useful densities. There may be a range of
curves suitable for different clinical examinations.
The figure shows an example
LUT used to compress a 12 bit image to a 10 bit image and apply
a characteristic curve.
|
|
|
M |
|
Modulation Transfer Function (MTF) |
|
The MTF
describes how well an imaging system transmits the spatial frequency
information of the object being imaged. An ideal imaging system
would transmit all spatial frequency information and would have
an MTF of unity at all frequencies. The image would thus be a
perfect representation of the object. Blurring or unsharpness
introduced by the imaging system results in higher spatial frequencies
not being transmitted as well as lower spatial frequency information.
As a result, the MTF progressively decreases with increasing spatial
frequency. There are a number of established methods for deriving
the MTF of an imaging system. A common method is to calculate
the MTF as the Fourier Transform of the Line Spread Function (LSF)
or Point Spread Function (PSF).
|
Moiré pattern |
|
When two identical repetitive
patterns of lines, circles, or arrays of dots are overlapped
with imperfect alignment, a pattern of light and dark lines,
a moiré pattern, appears. The moiré pattern is
not a pattern in the screens themselves, but rather a pattern
in the image formed in your eye. In some places, black lines
on the front screen hide the clear lines on the rear screen,
creating a dark area. Where the black lines on the front screen
align with black lines on the rear, the neighbouring clear areas
show through, leaving a light region. The patterns formed by
the regions of dark and light are moiré patterns.
The American Heritage®
Dictionary of the English Language: Fourth Edition. 2000.
|
|
|
N |
|
National reference dose |
|
The health
protection agency (HPA) undertake surveys of doses appropriate
to a particular type of radiograph or x-ray examination on patients
of average size. The value is set at the level of the third quartile
of the data set. This can be considered the dose level that action
to reduce doses should be undertaken.
|
Networking |
|
So far
the digital technology has been discussed, however the great advantage
of digital imaging is to be able to integrate it into a hospital
wide system. Linking the digital x-ray into the HIS/RIS system
and storing the images on a PACS system enables images to be reviewed
at various reporting rooms, consultant’s offices, wards
etc. Images can also be reported off-site using teleradiology.
|
Noise Equivalent Quanta (NEQ) |
|
The NEQ
is a measure of the signal to noise ratio and can be considered
as the effective number of X-ray quanta used in acquiring an image
at a particular dose level, as a function of spatial frequency.
NEQ is dependent upon MTF and the normalised Noise Power Spectrum
(NNPS) of the imaging system (see equation below).
NPS – frequency dependant
noise power spectrum,
MTF – frequency dependant modulation transfer function,
LAS – Large area signal, average pixel value in image.
|
Noise Power Spectrum (NPS) |
|
The NPS
is essentially a measure of the noise level over the spatial frequencies.
It is normal to quote the normalised noise power spectrum (NNPS,
equation below).
LAS – Large area signal, average
pixel value in image.
|
Nyquist frequency |
|
See Aliasing
and the Nyquist frequency
|
|
|
O |
|
|
|
P |
|
Phosphors |
|
The X-ray
detector is normally a thallium doped CsI phosphor, although other
phosphors such as gadolinium oxysulphide (Gd2O2S) are also used.
The X-rays interact with the phosphor and release light photons.
Each type of phosphor will have advantages and disadvantages.
For example, CsI is columnar in structure and acts as a light
guide directing the emitted light photons towards the photodiode
array and readout circuitry. Gd2O2S is not columnar, however,
it does have the advantage that it releases more light per interacting
X-ray than CsI.
DDR Phoshors
Image phosphor coupled to
a-Si photodiode TFT array flat panel detector:
The light from the phosphor promotes charge in the photodiode
array, with the amount of charge being proportional to the intensity
of the incoming X-ray photons. Each TFT element is then sequentially
addressed so that the charge in the photodiodes is readout and
digitised.
The detector resolution is not only limited by the pixel pitch
(distance between centres of pixels), but may also be reduced
by the effect of light spread in the phosphor layer.
The light from the phosphor promotes charge in the photodiode
array, with the amount of charge being proportional to the intensity
of the incoming X-ray photons. Each TFT element is then sequentially
addressed so that the charge in the photodiodes is readout and
digitised.
|
Picture Archive and Communication
System (PACS) |
|
PACS
is an image management and communication system. It stores and
distributes the images around the system. It is connected to each
of the digital systems, the HIS/RIS, printers and the archive.
Images will be viewed at various points in the system and the
extent to which the these images can be manipulated will depend
on the type and function of the equipment. Typically image are
viewed on:-
- Acquisition and Reporting workstations
- Monitors
- Laser Printers
|
Pixel and pixel pitch |
|
A pixel
is the smallest element of a digitised picture. A smaller pixel
size will generally give an improved spatial resolution in the
image. The pixel pitch is the distance between the centres of
adjacent pixels.
|
Posterisation effects |
|
Posterisation
effects occur when there are insufficient grey levels in the image
to distinguish changes in density. These typically appear as bands
with the same pixel value through the image and may be caused
by having insufficient bits in an image or data being compressed
by a Look Up Table.
|
|
|
Q |
|
Quality Assurance |
|
A series of tests
undertaken on a piece of equipment to ensure that it is functioning
correctly.
|
|
|
R |
|
Reporting workstation |
|
See Acquisition
and Reporting workstations.
|
|
|
S |
|
Scanning systems |
|
CCDs are
generally only up to 5 cm wide, this means that the image needs
demagnifying and can result in loss of signal. An alternative
method for covering the full image area is to use slot-scanning
technology. A linear array of CCDs scans across the patient in
conjunction with a narrow fan X-ray beam. This method may result
in good scatter rejection and contrast differentiation but has
a number of disadvantages including a long exposure time and high
tube loading. Also, the alignment of the scanning radiation beam
and the detectors requires tight mechanical tolerances and mechanical
stability of the scanning mechanism.
This technology is generally used in mammography and chest imaging.
Generally the detector is a phosphor connected to a linear array
of CCDs.
The CCD is read using Time Delay Integration (TDI).
|
Signal transfer property (STP) |
|
This defines
the relationship between the pixel value and the exposure to the
detector. There are different relationships possible, such as
linear (with or without an offset) or logarithmic. The relationship
needs to be known to linearise the image for tests such as NPS,
MTF and uniformity. |
Spatial frequency |
|
Images
are viewed in real space. However an image can be converted into
frequency space using a Fourier transform. A spatial frequency
is related to the size of objects in an image, where high frequency
relates to rapid change (edges or noise) and low frequency relates
to slow changes e.g. slight density differences. Spatial frequencies
are useful for undertaking image processing and describing imaging
properties of a digital detector.
|
|
|
T |
|
Tiling |
|
A tiled
array consists of a number of detectors abutted together to sample
the whole image. However, there may be small areas on the read
out device which are not sensitive due to the gaps between the
detectors (typically about 100 microns).
|
Time Delay Integration
(TDI) |
|
During
exposure, the X-ray fan-beam is continuously directed at the CCD
detector. Charge is collected in each pixel of the CCD. The charge
is transferred to the next pixel by electronic means. Each pixel
accumulates charge for a set integration time before the charge
is transferred to the next pixel. Concurrently, the detector moves
in the opposite direction with the same velocity as the transfer
of charge. This permits the charge packet to accumulate charge
from a fixed point relative to the patient.
|
|
|
U |
|
|
|
V |
|
|
|
W |
|
Workstation |
|
See Acquisition
and Reporting workstations. |
|
|
X |
|
|
|
Y |
|
|
|
Z |
|
|
|