Transcript Question: How does a radiographic image get on a film?

Chapter 24: Appendix I
Digital Image Processing
What is Digital Imaging?
Digital Imaging is the transforming of energy:
(from light photon, sonic, magnetic, x-ray, or
gamma radiation sources) to electrical signals
that are measured and assigned discrete binary
values.
Binary data is processed into image information
which may be enhanced, printed, displayed on a
monitor, and stored as a computer file.
Digital Modalitites
Every imaging modality may be digital. CT and MRI
are only digital.
From an equipment standpoint, the major difference
between the modalities is the type of energy used,
how the energy is changed as is traverses the body,
and how the remnant energy is measured as it leaves
the body.
Computed Tomography (CT)
Is only digital
X-radiation passes through,
and is attenuated.
Cardiovascular Interventional
Technology (CVI) digital application started in the 1980s
X-radiation passes through,
and is attenuated
Magnetic Resonance Imaging (MRI)
Is only digital
Hydrogen atoms
excited by radio
frequencies (RF)
create magnetic
vectors that sweep
an antenna.
Nuclear Medicine Technology
An isotope is injected, ingested or
inhaled. After being metabolized,
concentrations of the isotope are collected
by the nuclear medicine camera, which
was originally viewed on a scope. Nucs
are now digital.
Diagnostic Medical Sonography
and Vascular Technology
Sound waves pass into, and
are reflected off of interfaces
of tissues and organs. Virtually
all ultrasound is digital today.
Digital Radiography (DR)
Digital applications were available in the early 1980s, but the
difficulties of displaying radiographic quality (in terms of spatial
resolution) limited its use. By 2000 high resolution systems became
Increasingly popular. By 2006 they have become the standard.
.
X-radiation passes through,
and is attenuated
Digital Mammography
X-radiation passes through, and is attenuated Like digital
radiography, highly dependant on excellent spatial resolution: even
more so.
Question: How is an analog
radiographic image created?
• Begin with photons
coming off the anode.
• Outline the process, as
each major step.
• Use the appropriate
terminology.
Answer: How is an analog
radiographic image created?
•
•
•
•
•
•
Incident beam leaves anode.
Attenuation in body.
Remnant radiation exits as the aerial image.
Photons interact with silver halide crystals.
Latent image is formed.
Latent image is manifest on development.
Question: What does a graphic
representation of density building
on a film look like, and what is it
called?
D log E (or)
H & D Curve (or)
Hurter & Driffield Curve (or)
Characteristic Curve (or)
Sensitometric Curve
Producing a digital radiograph is the
same as for analog film, up to the point
of the photons interacting with the film.
Digital imaging samples the remnant
radiation with (some kind of) a detector,
not film.
Analog (343): Information display that is
continually changing. Every value, to infinity,
exists. For examples: a sweep second hand on a
clock, a mercury thermometer, and the gray scale of
an x-ray film. (See digital)
Digital (343): Information (or a display of information)
that is discrete. Values are absolute, with nothing in
between. For examples: the second by second increments
of a digital clock, a digital thermometer, and the gray
scale of a digital image. (See analog)
Analog is continuous.
Digital is discrete.
Data (343): The smallest unit of information.
If a brick building were used for analogy,
a single brick would be data. (See information)
Digital computers store data as binary digits. Question: How do
they do that?
Computer circuitry is a series of switches
that store data in one of two elementary
discrete- states: on, or off.
On (Closed)
=1
Off (Open)
=0
Information: Processed data. If a brick building were used for
analogy, a single brick would be data, and the building itself would be
information. (See data)
Bit (Binary digit) (344): In the binary numbering system, two
symbols, 0 and 1, are used to represent any value from null to
infinity. Like base 10 numbering, binary (base 2), is derived
from the base number raised to every exponent of itself: 20, 21,
22, 23 etc. Bit depth refers to the dynamic range (gray scale) a
microprocessor can display. A 4 bit processor displays 32
shades of gray, an 8 bit processor 256.
0 and 1 are binary digits
or bits
Digital computers store
data as binary digits.
Question: How do they
do that?
Binary numbering system (344): In the
base 2 numbering system all values are
represented by 0 or 1. Zero = off when
representing an open switch. One = on.
4
10 10 3 10 2 10 1 10 0
Binary numbering
2
4
3
2
1
0
2 2 2 2
512 256 128 64 32 16 8 4 2 1
0
0
0
0
0
0
1
1 1
0
0
1
1 1 1 0 = 14
1 0 1 0 = 42
1 1 1 1 = 127
Binary numbering
512 256 128 64 32 16 8 4 2 1
1
0
1
0 1 0 0 =84
Binary numbering
512 256 128 64 32 16 8 4 2 1
1
1
0
0
1
0 0 0 0
384
Byte (344): A group of 8 bits, used in computer programming
for organization of data. An 8 bit processor (28) stores 256 bits
of data. When the data is an image, than image can have up to
256 shades of gray. (See appendix I)
=400
Machine language: (344) is a programming language that sets the configuration of switches in a
computer chip, which determines how data is processed. Though cryptic to humans (10010111),
machine language is the most efficient programming code, for it does not require interpretationfor
the computer uses it. (See bits, bytes, appendix I)
Three logic gates:
AND, OR, NOT,
form a half adder.
Any combination
of 0 and 1 can be
added in a half
adder
AND Gate
Off
+
Off = Off
On
+
Off = Off
On
+
On = On
OR Gate
Off
+
Off = Off
On
On
+
+
Off = On
On = On
NOT Gate
Off = On
On = Off
A digital computer processes and stores data by configuring circuit pathways: opening and closing
semiconductor switches of integrated circuits (IC). When a switch is closed a current will flow through that
branch of the circuit. When it is open current will not flow.
Operation of a Half Adder
0
+
0
0
0
=0
Operation of a Half Adder
1
+
0
0
1
=1
Operation of a Half Adder
0
+
1
0
1
=1
Operation of a Half Adder
1
+
1
1
0
=2
Program (343): Operating instructions in the form of operating systems or application
programs. (See software)
Software (343): Software is the program (machine language, BASIC, Visual BASIC,
Fortran, C, C+, C++, Magic) that directs a computer’s function by configuring the
switches in the semiconductor material of a computer chip. (See program, hardware)
Hardware (344): The computer chips or integrated circuits (IC), the mother
board they reside on, the DVD, hard drive, ports, plugs, box, and anything else
coffee can be spilled on.
Central Processing Unit (CPU)
ALU
CU
Primary
Memory
(RAM
&
ROM)
Arithmetic Logic
Unit (contains logic gates
in registers)
Control Unit (directs the
flow of input
and output)
Central processing unit (CPU) (344): The major
components of a digital computer: the arithmetic
logic unit (ALU), the control unit (CU), and
primary memory (Random access memory or
RAM), comprise the CPU. Input and output
devices, as well as secondary memory are
peripherals (CD, DVD, floppy drives, flash
drives) that communicate
with the CPU.
Random access memory (RAM) (345): Semiconductor switches in microchips have
addresses: locations which may be accessed directly (randomly): a quick, electronic
process. When a program (such as word processing) is loaded from a secondary storage
device (such as a hard drive) it is copied into a RAM chip. As changes are made they are
made in RAM. The program must be saved back to the hard drive to be retained. RAM
is also called volatile memory, which means that when power is turned off data in RAM
is lost. (See appendix I)
Read only memory (ROM) (345): Memory that can only be read (used)
and not written to (changed).
RAM: addressable memory
in an integrated circuit (IC)
Welcome to RAM
Population: 376,243,101,765
When input is analog, (as sampled by a conventional video
camera), an Analog to Digital Converter (ADC) digitizes the
signal before being sent to the control unit.
ALU
ADC
CU
Input
Primary
Memory
(RAM)
When input is digital (keyboard, CT, MRI, DR etc.)
the digital signal is sent directly to the contol unit
1. Raw data (digital signal) enters the CU from a digital imaging device or ADC
2. The signal is identified and sent to the ALU for processing
3. Processing complete, the CU routes the image data to RAM
ALU
Processing
& Saving
2
3
ADC
CU
Primary
Memory
1
4
(RAM)
Secondary memory
(floppy, hard, optical disk
or tape) is most often
measured in megabytes, or
gigabytes.
Secondary Memory
5
4. From Primary
memory the image
is most often sent
to a monitor, and,
in the case of
imaging equipment,
it is SAVEd in
secondary memory.
Output is routed from RAM to the CU, then to the output device.
If the image file had been saved to a secondary memory device and purged from
RAM, it must be loaded back to RAM before being sent to output.
Joe’s
colon
ALU
ADC
CU
Primary
Memory
DAC
When the output
device is analog,
it is sent to the
DAC
(RAM)
Secondary Memory
Output
When the output device
is digital the file is sent
directly to it
Pixel (Picture element) (346): Commonly thought of as
the dots on a monitor or TV screen, but pixels also refer
to the elements of certain digital detectors. The term
conveys the concept of reducing nformation into a matrix
of discreet elements (See matrix)
One Pixel
One bit of computer memory (on
or off) is all it takes to light up a
pixel (on), or not (off).
Matrix (346): An array of pixels arranged in rows and
columns. The pixels of a monitor or TV screen form a
matrix. A large matrix (more pixels) = better spatial
resolution (See pixels)
A conventional (not HDTV or
high resolution monitor) is
a matrix of 525 x 525 pixels.
Only two bits of data (2 bit processor) is
needed to control each pixel when the
dynamic range is 21: on or off.
MRI midsagittal head scan
displayed in 2 bits.
Printout of the data
in the matrix of a
CT image
250
The number 47 defines the
shade of gray for the pixel
in column 250, row 210.
Values of digits stored in bytes of computer memory directly
correspond to the illumination of pixels.
Column 250
Row 210
In this case, the pixel in column 250, row 210.
Voxel (Volume element) (346): Although each pixel on
a monitor displays a two dimensional representation of
data, the data did not come from a finite plane. For
example, a sectional image of a CT scan is sampled by a
fan beam which has a thickness from 1mm to a cm. The
volume of the tissue within the sampled tissue contributes
to the brightness of the pixels. This is the voxel.
Bandwidth (Bandpass) (350): The range of frequencies
in a signal. Conventional commercial TV has a 525 x 525
matrix (1-2 lp/mm), which is very poor spatial resolution.
Increasing the matrix size (more pixels) improves spatial
resolution but also requires the pixels to be scanned faster,
which requires the electron beam to modulate (change
from pixel to pixel) more rapidly. The frequency of
modulation is measured in Hz, and is referred to as the
bandwidth. Commercial TV has a bandwidth of 4 MHz.
A 1000 line (actually 1050 pixels vertically) high
definition monitor requires 20 MHz, and resolves 5-7
lp/mm.
Sampling the Voxel
in Cross Sectional Images
Cross sectional images have depth, which is selected prior
to a scan. When a two dimensional section is viewed, the
density of each pixel actually represents all the tissues in
the volume of the section. This density represents the
volume element, or the voxel.
The CT numbers from these
samples would represent an
average of the mass and
healthy tissue.
Three contiguous CT
sections with parts of
a frontal lobe mass in
each one
The CT number from
this sample would
represent the true
density of the mass.
Using the region of interest (ROI) cursor to sample
CT numbers from the voxels of pixels.
This ROI has been
sized to measure a
density in the right
kidney.
The mean density of
15.9 indicates a fluid
filled cyst.
Filtering (Convolution) (350-35): With image data stored as binary numbers,
mathematical algorithms (repeated applications of a formula, applied to raw data)
are used to enhance the appearance of the image. Digital filters do not add to the
data. They accentuate features by suppressing frequencies that may obscure
detail, like blowing dust from a written page enhances readability without altering
the print. This is done by suppressing spatial frequencies (which enhances
others). An example of anatomy with low spatial frequencies is the liver, for it is
a homogeneous shade of gray. High spatial frequencies occur in trabecular bone,
which demonstrates heterogeneous shades of gray. (See appendix I)
Band-pass filtering: Demonstration (display) of a selected range of
frequencies.
High-pass filtering: Suppression of all but high frequency signals.
Also called sharpening or edge enhancement.
Low-pass filtering: Suppression of all but low frequency signals.
Smoothing filter: A low-pass that averages adjacent pixels
Plain film x-ray of
the abdomen
demonstrating
low spatial frequencies
(few changes in density)
of water density
organs
An ankle with Paget’s
disease demonstrating
high spatial frequencies
(many changes in
density) of the diseased
trabecular bone
No filter
High
Ultra High
Ridiculously High
Repeated
applications
of a high pass
filter (also known
as sharpening
or edge
enhancement)
demonstrates
the effect of
suppressing low
spatial frequencies
Low pass or smoothing
filter
No filter
Edge sharpening filter
has an algorithm similar
to a high pass filter
Windowing (Window level and width) (349): Digital images may have a
dynamic range of thousands of shades of gray (depending on the bit depth of the
processor) but the human eye can only distinguish about 32. To utilize this much
information a feature called windowing allows selective display of the dynamic
range. (See appendix I)
Window width: If 256 shades of gray were acquired during imaging,
but only half of those were chosen to be displayed on the monitor, the
window width would be 128. Widow width can be described as
controlling density (or brightness).
Window level (or center): The displayed dynamic range could be
at the high or low end of the scale, or anywhere in between. The
window level (or center) is the number the display is centered on. For
example, if the window width were 105, and the window center was 70,
the display of grays would extend from 18 to 122. (52 shades of gray
below 70, and 52 above, which = 105 shades of gray. Window level
can be described as controlling contrast.
Windowing
It’s like contrast and brightness, but it’s not.
The human eye can
distinguish 32 shades
of gray.
5
2
32
Windowing
But a digitized image may
contain thousands of shades
of gray, known as the
dynamic range
5
2
32
2
10
1024
CT Numbers (Hounsfield Units)
Bone = + 1000
Based on
Water = 0
Air = - 1000
CT numbers (HU),
express attenuation
values relative to water
Window Width
The range of displayed pixel values
For example: A window
width of 5
A narrow window is the digital
equivalent of a short scale of
contrast on a radiograph
Window Width
The range of displayed pixel values
A wide window is the digital
equivalent of a long scale of
contrast on a radiograph
For example: A window
width of 500
Window Center (or level)
The middle of the range of
any given window width
7
0
-7
2
1
0
*
1
2
3
4
5
For example: A window
width of 5
At a center of 2
Window Center (or level)
7
A window width of 5
0
-7
0
-1
-2
At a center of - 2
*
This CT section
through the abdomen
was windowed with
a width of 300 shades
of gray
The median number
(the level or center)
is 2 HU below water,
-2
A soft tissue window set at
a width of 110, at a level
of 43
The same CT section of the
head set at a width of 2010,
at a level of 800
Note the fracture through
the frontal sinus
Width 2290
Level 907
Bone windows demonstrating
trauma to the left orbit,
maxilla, and sinuses.
Two windows of the same section through the thorax
Width 1269
Level 202
Width 1269
Level 2
Digital subtraction
angiography
(DSA) images use
narrow windows to
enhance the contrast
difference between
the vessel and the
surrounding tissue
Signal to noise ratio (SNR) (347): Along with a signal
(such as the video signal) there are stray electrical currents
that degrade image information. The higher the ratio the
better the signal. For example, a vidicon camera has a
SNR of 200:1 due the heat produced by an electron tube.
A digital signal must be no less than 1000:1 to be of
acceptable quality.
Question: How can a simple
on/off switch be used to store
complex information that
contains many shades of gray?
Answer: Many switches are
used in combination.
Question: If one bit of data is
enough to turn a pixel off or on,
what can a byte of data do for a
single pixel?
Answer: A byte of image data stores
values for 256 shades of gray.
How many KB of computer memory is
required for a monitor with a 512 x 512
matrix displaying a gray scale of 2?
512 x 512 = 262,144 bits
262,144 / 8 bits per byte = 32,768 bytes
32,768 bytes / 1024 bytes in a kilobyte = 32KB
Answer = 32KB (compared to the 3.2 KB file
for all the text of chapter 24)
How many bytes of computer memory is
required for a for a monitor with a 512 x 512
8
matrix, displaying 256 shades of gray (2 )?
512 x 512 = 262,144 bits
262,144(8bits)/8 bits per bytes = 262,144 bytes
262,144 bytes/1024 bytes in a kilobyte =262KB
Answer = 262KB
Conclusion: Images are
memory hogs.