Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

ELEC 471 / BMEG 420: Medical Imaging

Module 1: Introduction to Medical Imaging

Siamak Najarian, Ph.D., P.Eng.

Professor of Biomedical Engineering (retired)
Department of Electrical and Computer Engineering
University of British Columbia

© Siamak Najarian 2023 1

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Contents:

• Physical/engineering principles of ultrasound, magnetic

resonance, computed tomography, and x-ray projection
imaging.

• Applications in diagnostics, therapeutics, and interventions.

• Segmentation, registration, and visualization.

© Siamak Najarian 2023 2

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Overview
Medical Imaging is the use of various technologies to image human anatomy and function.

modality

emission transmission

projection tomography projection tomography

real-time off-line real-time off-line real-time off-line real-time off-line

NM PET Fluoro Radio US CT

SPECT Mammo MR

© Siamak Najarian 2023 3

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Overview

• Transmission imaging refers to imaging in which the energy source is

outside the body on one side, and the energy passes through the body and is
detected on the other side of the body by the detector (source-body-detector).
• Projection imaging refers to the case when each point on the image corresponds
to information along a straight line trajectory through the patient.
• Tomography imaging is a method of imaging by sections or sectioning that uses
any kind of penetrating wave.

• Emission imaging refers to the use of a special dye containing radioactive

tracers (these tracers are either swallowed, inhaled, or injected into a vein in
the patient’s arm depending on what part of the body is being examined).
• We also have projection imaging and tomography imaging for this modality.

© Siamak Najarian 2023 4

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging
modality

Radiography
emission transmission

projection tomography projection tomography

R-T O-L R-T O-L R-T O-L R-T O-L

NM PET Fluoro Radio US CT
SPECT Mammo MR
R-T : real-time
O-L : off-line

Andreas Hornig: CC-BY-SA

Radiography is a medical imaging modality

used for generating and recording an x-ray
pattern for the purpose of providing the user
with a static image(s) after termination of the
exposure.

U.S. Navy photo by Mike Leporati

© Siamak Najarian 2023 5
 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging
modality

Mammography
emission transmission

projection tomography projection tomography

R-T O-L R-T O-L R-T O-L R-T O-L

NM PET Fluoro Radio US CT
SPECT Mammo MR

NIH Senior Health

Mammography is an x-ray imaging method used

to examine the breast for the early detection of
cancer and other breast diseases.
Kieranmaher: Creative Commons

© Siamak Najarian 2023 6

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging
modality

Fluoroscopy
emission transmission

projection tomography projection tomography

R-T O-L R-T O-L R-T O-L R-T O-L

NM PET Fluoro Radio US CT
SPECT Mammo MR

Yarkob:Creative Commons

Fluoroscopy is a type of medical imaging modality Ben Stephenson:Creative Commons

that shows a continuous x-ray image on a monitor,
much like an x-ray movie.

© Siamak Najarian 2023 7

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging
modality

Computed Tomography (CT)

emission transmission

projection tomography projection tomography

R-T O-L R-T O-L R-T O-L R-T O-L

NM PET Fluoro Radio US CT
SPECT Mammo MR

Christaras: Creative Commons

Hellerhoff: Creative Commons

Computed Tomography refers to a computerized

x-ray imaging procedure in which a narrow beam
of x-rays is aimed at a patient and quickly NithinRao: Public Domain
rotated around the body, producing signals that
are processed by the machine’s computer to
generate cross-sectional images, or “slices.”

© Siamak Najarian 2023 8

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging
modality

Magnetic Resonance (MR)

emission transmission

projection tomography projection tomography

R-T O-L R-T O-L R-T O-L R-T O-L

NM PET Fluoro Radio US CT
SPECT Mammo MR

Helmut Januschka: Creative Commons

Magnetic resonance imaging (MRI) is a

medical imaging technique that uses a BrokenSphere: Wikimedia Commons
magnetic field and computer-generated radio
waves to create detailed images of the
organs and tissues in human body.

© Siamak Najarian 2023 9

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging
modality

Ultrasound Imaging (US)

emission transmission

projection tomography projection tomography

R-T O-L R-T O-L R-T O-L R-T O-L

NM PET Fluoro Radio US CT
SPECT Mammo MR

DebMomOf3: Creative Commons

Aseev Artem: Creative Commons

Ultrasound Imaging uses high-frequency

sound waves to create real-time pictures
Dr. Moroder: Wikimedia Commons or video of internal organs or other soft
tissues, such as blood vessels.

© Siamak Najarian 2023 10

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging
modality

Nuclear Medicine
emission transmission

• Planar NM projection tomography projection tomography

R-T O-L R-T O-L R-T O-L R-T O-L

NM PET Fluoro Radio US CT
SPECT Mammo MR

• SPECT Nuclear medicine imaging is a

Kieran Maher method of producing images by
detecting radiation from different
• PET (and CT)
parts of the body after a radioactive
tracer is given to the patient.

NM: Nuclear medicine

SPECT: Single Photon Emission
Computed Tomography
PET: Positron Emission Tomography

Myohan: Creative Commons

Christian Lackas: Creative Commons

© Siamak Najarian 2023 11

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

PACS • Picture Archiving and Communication Systems (PACS) are digital

network systems that are used to communicate and store medical
images in a hospital or clinic (see below for PACS workflow).
CT Modality

Quality Assurance (QA) Workstation

or PACS Gateway

MRI Modality
Doctors Reading
Workstation

US Modality

• The 4 main components are:

• (1) A high speed network that Archive

receives images from various • (3) Workstations for reviewing

scanners and transmits the the images.
images to servers for storage
• (4) Software to perform these
or review. DIc00-NJITWILL: tasks.
Creative Commons
• (2) Storage devices such as Digital Imaging and Communications in
RAID’s (redundant array Medicine (DICOM) is a standard for handling,
storing, printing, and transmitting information
of independent disks) and in medical imaging.
magnetic tape.
© Siamak Najarian 2023 12
 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Analysis
• On a workstation, we can perform many types of analysis including:
• Image segmentation is the process of partitioning a digital image into multiple segments
(sets of pixels, also known as superpixels).
• Image registration is the process of transforming different sets of data into one coordinate system.
• Image visualization is the process of converting (rendering) image voxel into 3D graphical
representation.

• Segmentation • Registration • Visualization

A B

C D Tatiana Alchueyr: Creative Commons

Breeuser, Ermes, Gerber: Creative Commons

Yadav Parihar: Creative Commons

© Siamak Najarian 2023 13

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Contrast

• The subject contrast is the difference in some aspect of the signal

(intensity, energy fluence, phase, etc.) prior to being recorded. An
image is created by processing these signals.
• Let us look at projection x-ray as an example. The subject contrast Cs
is calculated as the difference in signal divided by the initial signal:

• We will see in Module 2 that x-rays are attenuated exponentially, so

the following equations can be written for the two beams shown in the
figure: • No is incident x-ray fluence (#photons/cm2)
• A and B are x-ray fluence exiting the tissue
(#photons/cm2)
• μ is linear attenuation coefficient (cm-1)

© Siamak Najarian 2023 14

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Contrast

• Contrast in an image (image contrast) is simply the difference in gray-scales shown
for different types of tissue. In a high contrast image, there are large differences
between light and dark for different tissues (not many shades of gray). A low contrast
image is uniformly gray (many shades of gray).
• X-ray contrast (radiography, fluoroscopy, mammography and CT) is due to differences
in x-ray absorption among the various tissue types. As we will see, this is related
mainly to density and atomic number.
• MR contrast is due mainly to proton density and relaxation of proton spins due to a
magnetic field.
• Ultrasound contrast is due mainly to acoustic impedance of the tissue.

Black White
High contrast

Gray scale number Image contrast

(Number of shades of gray)
Low contrast

© Siamak Najarian 2023 15

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Contrast

• There is a difference between image contrast and contrast resolution.
➢ Contrast resolution is the ability to distinguish gray levels on an image.
✓ In the following diagram, the 1 bit has high image contrast (but low contrast
resolution), compared to the 4 bit.
✓ The 4 bit, has low image contrast but high contrast resolution.
• The higher the bit depth (dynamic range), the higher the contrast resolution.

Bit depth Contrast resolution

• Which modality shows higher contrast resolution in the image? Radiography or CT?

© Siamak Najarian 2023 16

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Contrast

A B

• When the x-ray is processed on film, the

radiographic contrast (image contrast)
is measured as the difference between
optical densities on the film.
• If the x-ray image is captured digitally, the
image contrast can be varied on the
screen through image processing.
B

Relative Brightness
• The simplest image processing is called
window-leveling (look-up table), where A

the range of x-ray values are linearly

mapped to screen values.
• A better measure of digital contrast is the
Contrast to Noise Ratio (CNR):
Gray Scale Value
Contrast
CNR =
Standard deviation of noise Conceptual lines of A and B are called
Display Curves.

© Siamak Najarian 2023 17

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Spatial Resolution

• Spatial resolution (shown by ∆) is the ability to resolve the fine details of the image.
• A high spatial resolution means the ability to resolve small details.
• A low spatial resolutions means the ability to only resolve large details.

Modality 𝚫 (mm) Comments

Screen film radiography (SFR) 0.08 Limited by focal spot and detector resolution
Digital radiography (DR) 0.17 Limited by size of detector elements
Fluoroscopy 0.125 Limited by detector and focal spot
Screen film mammography (SFM) 0.03 Highest resolution modality in radiology
Digital mammography (DM) 0.05-0.10 Limited by size of detector elements
Computed tomography (CT) 0.4 About ½-mm pixels
Nuclear medicine planar imaging (NMPI) 7 Spatial resolution degrades substantially with distance from detector
Single photon emission computed tomography (SPECT) 7 Spatial resolution worst toward the center of cross-sectional image slice
Positron emission tomography (PET) 5 Better spatial resolution than with the other nuclear imaging modalities
Magnetic resonance imaging (MRI) 1.0 Resolution can improve at higher magnetic fields
Ultrasound imaging, 3 MHz (US) 0.3 Limited by wavelength of sound

© Siamak Najarian 2023 18

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Spatial Resolution

• The spatial resolution (often just called resolution) refers to the ability of a
scanner to image small objects. Alternatively, it is the ability to image two
separate objects and visually distinguish one from the other.
• A scanner that can image very small objects is said to have high resolution (and
vice versa).
• The spatial resolution can be calculated by measuring how much a point object
is spread out in the final image. This is called the point spread function (PSF)
and is a typical scanner characteristic cited for image quality.

(A) Point Stimulus (B) Isotropic PSF (C) Non-Isotropic PSF

• The PSF can be expressed in the frequency domain as the modulation transfer
function (MTF).

© Siamak Najarian 2023 19

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Spatial Resolution

• The MTF is simply the Fourier Transform of the PSF. In some cases,
line spread function (LSF) is used instead of PSF.

Amplitude PSF MTF

MTF (f)
Amplitude

MTF (f)
Amplitude

MTF (f)

Position Spatial Frequency

© Siamak Najarian 2023 20

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Spatial Resolution

• One advantage of working with the MTF is when investigating the
sources of blurring at several stages of the image formation process.
• If each process can be modeled with an MTF, then the MTF of the
entire system is simply the product (multiplication) of the individual
MTF’s in the frequency domain. (This is equivalent to convolution in
the spatial domain).

A • For example, consider a

B system with three stages
C (A,B,C). The final MTF is the
System product of the individual MTF’s
MTF (f)

in the frequency domain (and

lower than the others since all
values are < 1.0).

Spatial Frequency (cycles/mm)

© Siamak Najarian 2023 21

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Spatial Resolution

• Spatial frequency has the units of cycles/mm while temporal frequency has the units of
cycles/sec.
➢ For objects on an image that are separated by shorter distances (measured in millimeters), these
objects correspond to high spatial frequencies (cycles/mm).
• Spatial frequency is just a different way of thinking about object size.
1
➢ That is, 𝑓spatial ∝ .
object size

• Low spatial frequencies correspond to larger objects in the image, and higher spatial
frequencies correspond to smaller objects.
➢ In the following diagram, if you know the spatial frequency (𝑓spatial , lp/mm or line pairs per millimeter),
1
you can convert it to the spatial resolution (Δ) using: Δ = .
2𝑓spatial

𝑓spatial
Δ
© Siamak Najarian 2023 22
 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Noise and Sampling

• The random or stochastic component that is injected into the image

is called noise.
• Noise is often described by the mean and standard deviation of its
distribution, but the distribution is not always Gaussian.
• The signal to noise ratio (SNR) is usually defined as the amplitude
of the desired signal divided by the amplitude of the noise (or by the
standard deviation of the noise).
• For example, when sampling a constant property of tissue multiple
times, the SNR can be calculated as:

signal mean
SNR =
standard deviation of noise
• Other definitions of SNR can be written, including logarithmic
functions that are expressed in dB.
• It is best to specify the meaning of “desired signal amplitude” and
“noise amplitude” when specifying SNR, to avoid ambiguity.

© Siamak Najarian 2023 23

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Noise and Sampling

• Noise can occur at all frequencies (white noise) or be limited to a certain range of
frequencies.
• The image measurements are usually measured at regular locations.
• If noise is at a higher frequency than the sampling spacing (also called sampling
frequency or digitizing frequency), the noise will be aliased to lower frequencies
according to the Nyquist sampling theorem.
➢ When an analog signal is digitized, any component of the signal that is above one-half
the sampling frequency will be “aliased.” This frequency limit is known as the
Nyquist frequency.
fNyquist = fN = ½ fSampling = ½ fS

• Aliasing is basically the misidentification of a signal frequency, introducing distortion

or error.
➢ This phenomenon occurs when a system is measured at an insufficient sampling rate.

© Siamak Najarian 2023 24

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Nyquist Criterion

• The Nyquist theorem is expressed as follows:
• Given a sampling rate of fS, the frequencies given by: fi = nfS ± fo = 2nfN ± fo ; where
0 ≤ fo ≤ fN and n = 1,2,3, … cannot be distinguished from each other.
• Note that: fS = 2fN, where fN is the Nyquist frequency.

Example:
Given a sampling rate of 100Hz, what frequencies cannot be distinguished from
20Hz?

Answer:
fS = 2fN → 100Hz = 2fN → The Nyquist frequency fN = 50Hz. The given frequency
(i.e., 20Hz) is lower than the Nyquist frequency, so the given frequency is fo. That is,
fo = 20Hz. The equation gives:
fi = 2nfN ± fo ; 2fN = 100Hz; fo = 20Hz → fi = 100n ± 20
n = 1 → fi = 100×1 ± 20 = 80 and 120 Hz
n = 2 → fi = 100×2 ± 20 = 180 and 220 Hz
and so forth.

Although the example is given with Hz, it can be used similarly for samples/mm.

© Siamak Najarian 2023 25

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Image Quality: Nyquist Criterion

• To avoid aliasing, the Nyquist theorem also states that the sampling frequency
should be at least twice the highest frequency in the signal.

aliased output signal samples

sampling pitch

analog input signal

Aliasing in CT: An insufficient sampling Normal Aliasing

frequency results in image aliasing. The

example shows a coronal image of the
thorax.

© Siamak Najarian 2023 26

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

Course overview

• We will investigate the major transmission modalities:

Radiography, CT, MR and Ultrasound
• Physics/engineering aspects of acquisition, including:
➢ Contrast Resolution (and Contrast)
➢ Spatial Resolution (also called Resolution)
➢ Temporal Resolution
➢ Clinical Uses
➢ Limitations
➢ Future Research Directions

© Siamak Najarian 2023 27

 Module 1: Introduction to Medical Imaging ELEC 471 / BMEG 420: Medical Imaging

End of Module 1

© Siamak Najarian 2023 28