Medical Imaging

Techniques

Dr. K. Adalarasu
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University
 2
Textbook and Materials
Rafael C. Gonzalez, Richard E. Woods,
“Digital Image Processing”, 2nd Edition,
Pearson Education, 2003
Digital Image Processing by Jayaraman,
Veerakumar, 2012
Khandpur R.S, Handbook of Biomedical
Instrumentation, 3/e, Tata McGraw
Hill,New Delhi, 2014

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 3
Reference
William K. Pratt, “Digital Image Processing” ,
John Willey ,2001
Steve Webb, The physics of medical imaging,
Adam Hilger, Bristol, England, Philadelphia,
USA, 1988
Jain A.K., “Fundamentals of Digital Image
Processing”, PHI, 1995.

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 4

Magnetic Resonance
(MR)

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 5

Basic MRI Components

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 6
Detection System
 Function
 To detect the nuclear magnetization and generate an output signal for
processing by the computer

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 7
Detection System
 Receiver coil usually surrounds the sample
 Acts as an antenna to pick up the fluctuating nuclear
magnetization of the sample
 Converts it to a fluctuating output voltage V(t)

 Where
 M(t, x) is the total magnetization in a volume
 Bc(x) -sensitivity of the receiver coil at different points in
space

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 8
Detection System
 Receiver coil design and placement is such that Bc(x) has the
largest possible transverse component
 Longitudinal component of Bc(x) contributes little to the output
voltage and can be ignored
 RF signals constitute the variable measured
 Extremely weak signals having an amplitude in the nV (nano-
Volt) range thus requiring specially designed RF antennas
 Sensitivity of an MR scanner
 Depends on the quality of its RF receiving antenna
 Sample magnetization
 Static magnetic field strengths and sample volume, the signal-to-
noise-ratio (SNR) of the RF signal at the receiver depends

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 9
Detection System
SNR≈ K(Q/Vc)
Where K is a numerical constant, specific to the
coil geometry
Q is the coil magnetization factor, and
Vc Receiver coil volume
Commonly available coils are
Body Coils
Head Coils
Surface coils
Organ-enclosing coils

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 10
Detection System
Body Coils
Constructed on cylindrical coils forms
Diameter ranging from 50 to 60 cm to entirely
surround the patient’s body
Head Coils
Designed only for head imaging
Typical diameter of 28 cm

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 11
Detection System
 Surface coils
 Orbit/ear coil
 Flat, planar ring-shaped coil with 10 cm diameter
 Neck coil
 Flexible, rectangular shaped surface coil (10 cm X 20 cm) capable of
adaptation to the individual patient anatomy
 Spine coil
 Cylindrical or ring-shaped coil with 15 cm diameter
 Organ-enclosing coils
 Breast coil
 Cylindrical or ring-shaped coil with 15 cm diameter
 Helmholtz-type coil
 A pair of flat ring coils each having 15 cm diameter with distance
between the two coils variable from 12 to 22 cm
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University
 12

Superconductive Magnet

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 13
Detection System
 Pre-amplifier is a low-noise amplifier
 Which amplifies the signal and feeds it to a quadrature
phase detector
 Detector accepts the RF NMR signal
 Detector circuit
 Accepts the inputs, the NMR signal V(t) and a reference
signal
 Multiplies them
 Output is the product of the two inputs
 Frequency of the reference signal
 Same as that of the irradiating RF pulse

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 14

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 15
Detection System
Output of the phase-sensitive detector
Narrow range of frequencies centred at 2w0
Low pass filter
Removes all components except those centred at
zero from the signal
ADC

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 16

Gradient System for Spatial Coding

 Resonance frequency depends on the magnetic field
strength
 Strength of the signal at each frequency
 Interpreted as the density of the hydrogen nuclei in the
plane
 One-dimensional projection of the structure of the
three dimensional object
 By taking a series of these projections at different
gradient orientations
 A two or even three-dimensional image can be produced

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 17

Gradient System for Spatial Coding

 Selecting a slice
 ‘Selective excitation’ method
 Radio wave pulses of the same frequency as the
resonant frequency are applied
 Resonant frequency changes in the same direction
 If a uniform static field
 Gradient magnetic field changing linearly in the
vertical direction of the layer

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 18

Gradient System for Spatial Coding

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 19
Gradient System for Spatial Coding
 Hardware can be broken down into four sub-
systems
 Interface between the computer and the
gradient control system
 Primary function is to allow the independent
positioning of the three time invariant planes (X, Y
and Z)
 Serial to parallel converter with independent reset
times
 Requires only six pulse lines from the host computer
 Maximum number of plane positions is limited to 256
which is adequate for most imaging applications
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University
 20

Gradient System for Spatial Coding

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 21
Gradient System for Spatial Coding

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 22

Gradient System for Spatial Coding

 Digital oscillator consists of a 555 timer followed by
shift registers
 Twelve registers shift out a sine
 While the other twelve shift out a cosine wave
 Digital oscillator facilitates varying the output
frequency over an extremely wide range through the
use of a single control
 First attenuator
 8-bit input from the interface circuit is used directly to
control
 Device consisting of an arrangement of resistors which
reduces the strength of a radio or audio signal.
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University
 23
Gradient System for Spatial Coding
 Second attenuator
 Same 8-bits are inverted to control
 Two complementary sine wave outputs which can
be stepped through 256 positions

 Current control requires through the shim coil

 That the control used to adjust the static field gradients be
available for setting the DC levels
 Typical scanners, gradient coils have an electric
resistance of about 1 Ω and an inductance of 1 mH

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 24
Gradient System for Spatial Coding
 Gradient fields are required to be switched from 0 to 10
mT/m in about 0.5 ms
 Power dissipation during the switching interval is about
20 kW
 Necessary to use water cooling to prevent overheating of the
gradient coils
 X and Y gradient drivers operate in the 4-6 A range
 Total requirement is at least 20 A.
 Z driver requires much less current due to the coil
geometry and positions with respect to the major fields
 Non-linear field gradients result in a geometrical
distortion of the image reconstructed from projections

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 25
Imager System
 Imager system
 Computer for image processing
 Display system
 Control console
 Computer
 Source of both the voltage waveforms of all gradient
pulses and the envelopes of the RF pulses
 Analog-to-digital convertors with 16 bits
 Data acquisition
 Information is acquired at the rate of about 800 kilobytes
per second
 Each image can contain upto a megabyte of digital data

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 26
Imager System
 Two-dimensional images are typically displayed as
256 X 256 or 512 X 512 pixel arrays
 Typically 32 bit machines equipped with upto 4
Mbyte of memory
 Display console is usually an intelligent console
 That can be used as an independent image processing
unit in an interactive system
 To simultaneously display proton density and T1, T2
distribution image

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 27
Imager System
Control console
Operation section
System control section
Display section
System control section
Microcomputer controls the gradient magnetic
field
High frequency pulse train (RF pulses)
Timing of A/D conversion of the signals received

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 28
Imager System
Display section
High resolution monitor
Keyboard
Image memory
Microcomputer for processing the image
Operating keys to set the scanning conditions
Control the patient couch from the operation
section
Various panel indicators to monitor the system
condition

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 29

Biological Effects of MRI

System

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 30

Biological Effects of MRI System

 Three aspects of NMR imaging which could cause
potential health hazard
 Heating due to the RF power
 Temperature increase produced in the head of NMR
imaging would be about 0.3°C
 Static magnetic field
 Slight decrease in cognitive skills
 Mitotic delay in slime moulds
 Delayed wound healing and elevated serum triglycerides

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 31
Biological Effects of MRI System
Electric current induction due to rapid change
in magnetic field
Oscillating magnetic field gradients may induce
electric currents strong enough to cause
ventricular fibrillation
No damage due to NMR from exposures has
been reported
Suggested that fields should not vary at a rate
faster than 3 tesla/s

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 32

Advantages of MRI System

 NMR image provides substantial contrast between
soft tissues that are nearly identical in existing
techniques
 While contrasts of only a few percent are possible
between soft tissues with X-rays
 Cross-sectional images with any orientation are
possible in NMR imaging systems
 Alternative contrast mechanisms of NMR provide
promising possibilities of new diagnostics for
pathologies that are difficult or impossible with
present techniques

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 33
Advantages of MRI System
 NMR imaging parameters are affected by chemical
bonding
 Therefore offer potential for physiological imaging
 NMR uses no ionizing radiation and has minimal
 If any, hazards for operators of the machines and for
patients
 Unlike CT, NMR imaging requires no moving parts,
gantries or sophisticated crystal detectors
 NMR permits imaging of entire three-dimensional
volumes simultaneously instead of slice by slice

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 34
Terms
 BOLD: blood oxygenation level-dependent
 CBF: cerebral blood flow
 CBV: cerebral blood volume
 CBv: cerebral blood velocity
 CMRO2: cerebral metabolic rate, oxygen
 CMRglu: cerebral metabolic rate, glucose

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 35

fMRI (Functional
Magnetic Resonance
Imaging) System

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 36
Basic Principles of MRI

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 37
Basic Principles of MRI

After the energy source is

Relaxation removed, some spins will return to
the low-energy state by giving off
that energy.

T1 (longitudinal) Relaxation T2* (transverse) Relaxation

Results from dephasing of the
Results from the net magnetization protons
returning to its aligned position

The basis of BOLD-fMRI signal

The basis of MRI anatomical signal

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 38

Why use fMRI ?

Localization – where in the brain do
we process information about ….?
Pattern of activation
Dissociation – Do we process X and
Y using the same brain mechanisms?
Same or different pattern of activation

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 39
History of fMRI
 BOLD signal
 Deoxygenated hemoglobin is paramagnetic (1936)
 Oxygen changes T2 relation (1982)
 Optical “intrinsic signals” depend upon blood
oxygenation (1986)
 Respiration produces “BOLD signal” in large veins in rats
 fMRI using BOLD signal (1992)
 CBF
 Demonstration of arterial spin labeling (1992)
 fMRI using Arterial spin labeling (ASL) (1992)

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 40
fMRI Characteristics
Good spatial resolution (1 mm)
But only good enough to study systems biology,
not biomolecular or mechanisms
Good temporal resolution (1 sec)
But only good enough to study the blood supply,
not electrical activity
Strongly affected by contrast agents
But only those within the blood stream, as MRI
agents do not cross the blood-brain barrier (BBB)
Good for activation studies
But less good for assessing basal physiology
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University
 41
fMRI: General Considerations
 Time & space
 a trade-off that is regulated by SNR

 Contrast mechanisms
 We have to encode physiology into MR
signal

Contrast-to-noise ratio (CNR) &

SNR All
 determine how we approach fMRI in terms related
of contrast mechanisms, averaging,

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 42
Tools Necessary for fMRI
High-field MRI (1.5T or greater) scanner
BOLD effect (fMRI signal) increases with field
strength
Fast imaging sequence
Echo Planar Imaging (EPI)
Stimulus presentation equipment
Projector to show visual stimuli
Response devices such as button box to record
subject’s response
Headphones for auditory stimuli (and hearing
protection)
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University
 43

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 44
fMRI Sensitivity
T1 contrast agents effect brain signal
weakly due to the BBB
T2* agents/methods effect brain signal
strongly through gradients extending from
the vessels into the tissue
T2 methods (spin echoes) refocus a large
portion of available signal, and so are weak
compared to T2*
“sensitivity” means CNR per unit time

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 45
Methods with High Spatial Resolution

PET - Positron emission tomography fMRI – functional Magnetic Resonance Imaging

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 46
MRI Vs. fMRI

T1
T2*
T1 + T2*

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 47

Spatial Resolution: Voxels

Voxel: A small rectangular prism that is the basic sampling unit of fMRI.
Typical anatomical voxel: (1.5mm)3. Typical functional voxel: (4mm)3.

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 48
Why T2* and not T1?
 MRI scanner is tuned to resonate and image hydrogen atoms
as in conventional MRI
 However, T2*-weighted images are performed which take advantage of
the fact that deoxygenated hemoglobin is magnetic whereas
oxygenated hemoglobin is not
 Because of the magnetic properties of the unflipped magnetic
deoxyhemoglobin molecule which causes rapid dephasing
 T2* signal is retained longer in a region when it has more
oxygenated blood
 Area with more oxygenated blood will show up more intense
on T2*-weighted images compared to when there is less
oxygenated blood around

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 49
Chain of Events in fMRI
Cognitive Localized Increased
Local vasodilation
task increase in blood flow
(widening of local blood
neuronal activity to active
vessels)
region

Increased
MR signal in
Decreased Ratio:
a specific
brain region

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 50
BOLD (Blood Oxygen Level Dependent)
 Basic idea behind the BOLD is that, since the brain
offers no storage space for energy, all energy usage
must be compensated immediately
 This means that active neural regions must be
provided with energy carriers via the blood flow
 Main carrier of energy in all metabolic cycles is
oxygen, which is transported via the blood by means
of hemoglobin
 Oxygenated and deoxygenated hemoglobin are
associated with different magnetic properties

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 51
BOLD fMRI
REST

Activation

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 52
BOLD fMRI
 fMRI signal is always a REST
contrast between rest and
task
 This contrast is mainly
driven by an increase in
oxyhemoglobin
 Causing protons to precess in Activation
phase
 Which consequently
produces a stronger signal as
compared to rest

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 53
fMRI acquisition technique
BOLD (Blood Oxygen Level Dependent), the most
common, provides a mixed signal dependant on
 Blood FLOW
 Blood VOLUME
 Blood OXYGENATION.
Variations on the technique can be used to
emphasize or de-emphasize one or another of these
components:
 Pefusion-based fMRI blood flow
 Injections of various compounds blood volume
 Diffusion-based fMRI cell swelling (after excitation)
So far, only little work has been done to quantify the
relationship between different fMRI techniques.
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University
 54
Advantages of fMRI
Non-invasive method with no known risks
that provides the best spatial scale in
studying the human brain
Repeat measurements within & across
subjects
Study brain function \ cognition in an awake
behaving human

KA – MIT – Unit III – March, 2019, Sastra Deemed to be University

 55
Disadvantages of fMRI
 Extremely sensitive to head movements
 Awkward environment for emotional paradigms
 Contraindications:
 Irremovable magnetic devices
 Extreme claustrophobia
 Loud sound from magnets
 Cannot perform receptor-ligand studies like PET and
SPECT
 New technique with less track record than PET and
SPECT
 There is a time lag of 3 to 6 s between when a brain
region is activated and blood flow increases to it.
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University
 56
Current Roles for fMRI Mechanisms
BOLD signal
fMRI work horse for human imaging

ASL (Arterial Spin Labeling)

Targeted studies of baseline physiology or
CMRO2 reactivity

IRON fMRI
Method of choice for animal fMRI; not yet
available for human studies (future clinical
role??)
KA – MIT – Unit III – March, 2019, Sastra Deemed to be University