Perry Sprawls

Chapter 1: MRI image characteristics

Introduction And Overview

Magnetic resonance imaging (MRI) is a medical imaging process that uses

a magnetic field and radio frequency (RF) signals to produce images of
anatomical structures, of the presence of disease, and of various biological
functions within the human body. MRI produces images that are distinctly
different from the images produced by other imaging modalities. A primary
difference is that the MRI process can selectively image several different
tissue characteristics. A potential advantage of this is that if a pathologic
process does not alter one tissue characteristic and produce contrast, it
might be visible in an image because of its effect on other characteristics.
This causes the MRI process to be somewhat more complex than most
imaging methods. In order to optimize an MRI procedure for a specific
clinical examination, the user must have a good knowledge of the
characteristics of the magnetic resonance (MR) image and how those
characteristics can be controlled.
In this chapter we will develop a basic knowledge and overview of
the MR image, how the image relates to specific tissue characteristics, and
how image quality characteristics can be controlled.

The MR Image
The MR image displays certain physical characteristics of tissue. Let us now
use Figure 1-1 to identify these characteristics and to see how they are
related.

Figure 1T-1. Physical characteristics of tissue and fluid

movement that can be displayed in the magnetic resonance image.
MRI can also provide certain chemical information by applying
spectroscopy analysis to the RF signals emitted by the tissue.

The MR image is a display of RF signals that are emitted by the tissue during
the image acquisition process. The source of the signals is a condition of
magnetization that is produced in the tissue when the patient is placed in the
strong magnetic field. The tissue magnetization depends on the presence of
magnetic nuclei. The specific physical characteristic of tissue or fluid that is
visible in the image depends on how the magnetic field is being changed
during the acquisition process. An image acquisition consists of an acquisition
cycle, like a heartbeat, that is repeated many times. During each cycle the
tissue magnetization is forced through a series of changes. As we will soon
learn in much more detail, all tissues and fluids do not progress through these
changes at the same rate. It is the level of magnetization that is present at a
special “picture snapping time” at the end of each cycle that determines the
intensity of the RF signal produced and the resulting tissue brightness in the
image.
MR images are generally identified with specific tissue characteristics
or blood conditions that are the predominant source of contrast. These
characteristics determine the level of tissue magnetization and contrast
present at the time the “picture is snapped.” The equipment operator, who
sets the imaging protocol, determines the type of image that is to be produced
by adjusting various imaging factors.
The characteristics that can be used as a source of image contrast fall
into three rather distinct categories. The first, and most widely used, category
is the magnetic characteristics of tissues. The second category is
characteristics of fluid (usually blood) movement. The third category is the
spectroscopic effects related to molecular structure.
At this time we will briefly introduce each of these characteristics to set
the stage for the much more detailed descriptions presented later.

Tissue Characteristics and Image Types

Proton Density (PD) Images
The most direct tissue characteristic that can be imaged is the concentration
or density of protons (hydrogen). In a proton density image the tissue
magnetization, RF signal intensity, and image brightness are determined by
the proton (hydrogen) content of the tissue. Tissues that are rich in protons
will produce strong signals and have a bright appearance.

Magnetic Relaxation Times — T1 and T2 Images

During an MRI procedure the tissue magnetization is cycled by flipping it into
an unstable condition and then allowing it to recover. This recovery process is
known as relaxation. The time required for the magnetization to relax varies
from one type of tissue to another. The relaxation times can be used to
distinguish (i.e., produce contrast) among normal and pathologic tissues.
Each tissue is characterized by two relaxation times: Tl and T2.
Images can be created in which either one of these two characteristics is the
predominant source of contrast. It is usually not possible to create images in
which one of the tissue characteristics (e.g., PD, T1, or T2) is the only pure
source of contrast. Typically, there is a mixing or blending of the
characteristics but an image will be more heavily weighted by one of them.
When an image is described as a T1-weighted image, this means that T1 is
the predominant source of contrast but there is also some possible
contamination from the PD and T2 characteristics.

Fluid Movement and Image Types

Vascular Flow
The MRI process is capable of producing images of flowing blood without the
use of contrast media. Although flow effects are often visible in all types of
images, it becomes the predominant source of contrast in images produced
specifically for vascular or angiographic examinations as described in Chapter
12.

Perfusion and Diffusion

It is possible to produce images that show both perfusion and diffusion within
tissue. These require specific imaging methods and are often characterized
as functional imaging.

Spectroscopic and Chemical Shift

The frequency of the RF signals emitted by tissue is affected to a small
degree by the size and characteristics of the molecules containing the
magnetic nuclei. These differences in frequencies, the chemical shift, can be
displayed in images. It is also the basis of MR spectroscopy. Spectroscopy is
the process of using magnetic resonance to analyze the chemical composition
of tissue. Spectroscopy makes use of the fact that different molecular
structures have different resonant frequencies. Typically, the MR signals from
a tissue specimen are sorted and displayed on a frequency scale. The signals
from different chemical compounds will appear as peaks along the frequency
scale. This leads to their identity and measure of relative abundance.

What Do You See In An MR Image?

We have discovered that an MR image can display a variety of tissue and
body fluid characteristics. However, there are several physical characteristics
that form the link between the image and the tissue characteristics described
above. Understanding this link gives us a better appreciation of how the tissue
characteristics are made visible. We will use Figure 1-2 to develop the link.

Figure 1-2. The physical characteristics that form the link

between the image and the three tissue characteristics

Radio Frequency Signal Intensity

The first thing we see in an image is RF signal intensity emitted by the
tissues. Bright areas in the image correspond to tissues that emit high signal
intensity. There are also areas in an image that appear as dark voids because
no signals are produced. Between these two extremes there will be a range of
signal intensities and shades of gray that show contrast or differences among
the various tissues.
Let us now move deeper into the imaging process and discover the
relationship between RF signal intensity and other characteristics.

Tissue Magnetization
The condition within the tissue that produces the RF signal is magnetization.
At this point we will use an analogy to radioactive nuclide imaging. In nuclear
medicine procedures it is the presence of radioactivity in the tissues that
produces the radiation. In MRI it is the magnetization within the tissues that
produces the RF signal radiation displayed in the image. Therefore, when we
look at an MR image, we are seeing a display of magnetized tissue.
We will soon discover that tissue becomes magnetized when the
patient is placed in a strong magnetic field. However, all tissues are not
magnetized to the same level. During the imaging process the tissue
magnetization is cycled through a series of changes, but all tissues do not
change at the same rate. It is this difference in rates of change of the
magnetization that makes the tissues different and produces much of the
useful contrast. This will be described in much more detail later when we will
learn that these rates of change are described as magnetic relaxation times,
T1 and T2.
It is the level of magnetization at specific “picture snapping” times
during the imaging procedure that determines the intensity of the resulting RF
signal and image brightness. The MR image is indeed an image of
magnetized tissue. Tissues or other materials that are not adequately
magnetized during the imaging procedure will not be visible in the image.
Protons (Magnetic Nuclei)
The next thing we see is an image of protons that are the nuclei of hydrogen
atoms. That is why an MRI procedure is often referred to as proton imaging.
The magnetization of tissue, which produces the RF signals, comes
from protons that are actually small magnets (magnetic nuclei) present in the
tissue. These small magnets are actually the nuclei of certain atoms that have
a special magnetic property called a magnetic moment. Not all chemical
substances have an adequate abundance of magnetic nuclei.

Hydrogen
The only substance found in tissue that has an adequate concentration of
magnetic nuclei to produce good images is hydrogen. The nucleus of a
hydrogen atom is a single proton. Therefore, the MR image is an image of
hydrogen. When tissue that contains hydrogen (small magnetic nuclei), i.e.,
protons, is placed in a strong magnetic field, some of the protons line up in the
same direction as the magnetic field. This alignment produces the
magnetization in the tissue, which then produces the RF signal. If a tissue
does not have an adequate concentration of molecules containing hydrogen,
it will not be visible in an MR image.

Tissue Characteristics
As we have moved deeper into the imaging process we arrive again at the
three tissue characteristics: PD, T1, and T2. It is these characteristics that we
want to see because they give us valuable information about the tissues.
These characteristics become visible because each one has an effect on the
level of magnetization that is present at the picture snapping time in each
imaging cycle. At this time we will briefly describe the effect of each and then
develop the process in more detail in Chapters 4 and 5.

PD (Proton Density)
PD has a very direct effect on tissue magnetization and the resulting RF
signal and image brightness. That is because the magnetization is produced
by the protons. Therefore, a tissue with a high PD can reach a high level of
magnetization and produce an intense signal.

T1
When the imaging protocol is set to produce a T1-weighted image, it is the
tissues with the short T1 values that produce the highest magnetization and
are the brightness in the image.

T2
When the imaging protocol is set to produce a T2-weighted image, it is the
tissues with the long T2 values that are the brightest. This is because they
have a higher level of magnetization at the picture snapping time.

Spatial Characteristics
Figure 1-3 illustrates the basic spatial characteristics of the MR image. MRI is
basically a tomographic imaging process, although there are some
procedures, such as angiography, in which a complete anatomical volume will
be displayed in a single image. The protocol for the acquisition process must
be set up to produce the appropriate spatial characteristics for a specific
clinical procedure. This includes such factors as the number of slices, slice
orientation, and the structure within each individual slice.

Figure 1-3. The spatial characteristics of MR images.

Slices

A typical examination will consist of at least one set of contiguous slices. In

most cases the entire set of slices is acquired simultaneously. However, the
number of slices in a set can be limited by certain imaging factors and the
amount of time allocated to the acquisition process.
The slices can be oriented in virtually any plane through the patient’s body.
The major restriction is that images in the different planes cannot generally be
acquired simultaneously. For example, if both axial and sagittal images are
required, the acquisition process must be repeated. However, there is the
possibility of acquiring 3-D data from a large volume of tissue and then
reconstructing slices in the different planes, as will be described in Chapter 9.

Voxels
Each slice of tissue is subdivided into rows and columns of individual volume
elements, or voxels. The size of a voxel has a significant effect on image
quality. It is controlled by a combination of protocol factors as described in
Chapter 10 and should be adjusted to an optimum size for each type of
clinical examination. Each voxel is an independent source of RF signals. That
is why voxel size is a major consideration in each image acquisition.

Image Pixels
The image is also divided into rows and columns of picture elements, or
pixels. In general, an image pixel represents a corresponding voxel of tissue
within the slice. The brightness of an image pixel is determined by the
intensity of the RF signal emitted by the tissue voxel.

Control Of Image Characteristics

The operator of an MRI system has tremendous control over the
characteristics and the quality of the images that are produced. The five basic
image quality characteristics are represented in Figure 1-4. Each of these
image characteristics is affected by a combination of the imaging factors that
make up the acquisition protocol.

Figure 1-4. Image quality characteristics that can be controlled

by the selection of protocol factors.

Not all types of clinical procedures require images with the same
characteristics. Therefore, the primary objective is to use an imaging protocol
in which the acquisition process is optimized for a specific clinical
requirement.
Although each of the image characteristics will be considered in detail
in later chapters, we will introduce them here.

Contrast Sensitivity
Contrast sensitivity is the ability of an imaging process to produce an image
of objects or tissues in the body that have relatively small physical differences
or inherent contrast. The contrast that is to be imaged is in the form of some
specific physical characteristic. In x-ray imaging, including CT (computed
tomography), difference in physical density is a principle source of contrast.
One of the major advantages of MRI is that it has a high contrast sensitivity
for visualizing differences among the tissues in the body because there are
several sources of contrast; that is, it has the ability to image a variety of
characteristics (PD, T1, T2) as described previously. Also, there is usually
much greater variation among these characteristics than among the tissue
density values that are the source of contrast for x-ray imaging. If a certain
pathologic condition does not produce a visible change in one characteristic,
there is the possibility that it will be visible by imaging some of the other
characteristics.
Even though MRI has high contrast sensitivity relative to most of the
other imaging modalities, it must be optimized for each clinical procedure.
This includes the selection of the characteristics, or sources of contrast, that
are to be imaged and then adjusting the protocol factors so that the sensitivity
to that specific characteristic is optimized. This is illustrated in Figure 1-5.

Figure 1-5. The images produced when the contrast sensitivity

is optimized for each of the three specific tissue characteristics.

Detail
A distinguishing characteristic of every imaging modality is its ability to image
small objects and structures within the body. Visibility of anatomical detail
(sometimes referred to as spatial resolution) is limited by the blurring that
occurs during the imaging process. All medical imaging methods produce
images with some blurring but not to the same extent. The blurring in MRI is
greater than in radiography. Therefore, MRI cannot image small structures
that are visible in conventional radiographs.
In MRI, like all modalities, the amount of blurring and the resulting
visibility of detail can be adjusted during the imaging process. Figure 1-6
shows images with different levels of blurring and visibility of detail. The
protocol factors that are used to adjust detail and the associated issues in
their optimization will be discussed in Chapter 10

Figure 1-6. Images with different levels of blurring. and visibility of

anatomical detail
Noise
Visual noise is a major issue in MRI. The presence of noise in an image
reduces its quality, especially by limiting the visibility of low contrast objects
and differences among tissues. Figure 1-7 shows images with different levels
of visual noise. Most of the noise in MR images is the result of a form of
random, unwanted RF energy picked up from the patient’s body.

Images with different levels of visual noise

The amount of noise can generally be controlled through a combination of
factors as described in Chapter 10. However, many of these factors involve
compromises with other characteristics.

Artifacts
Artifacts are undesirable objects, such as streaks and spots, that appear in
images which do not directly represent an anatomical structure. They are
usually produced by certain interactions of the patient’s body or body
functions (such as motion) with the imaging process.
There is a selection of techniques that can be used to reduce the
presence of artifacts. These will be described in Chapter 14.

Spatial
The general spatial characteristics of the MR image were described
previously. However, when setting up an imaging protocol the spatial
characteristics must be considered in the general context of image quality. As
we will discover later, voxel size plays a major role in determining both image
detail and image noise.

Image Acquisition Time

When considering and adjusting MR image quality, attention must also be
given to the time required for the acquisition process. In general, several
aspects of image quality, such as detail and noise, can be improved by using
longer acquisition times.

Protocol Optimization
An optimum imaging protocol is one in which there is a proper balance
among the image quality characteristics described above and also a balance
between overall image quality and acquisition time.
The imaging protocol that is used for a specific clinical examination has a
major impact on the quality of the image and the visibility of anatomical
structures and pathologic conditions.
Therefore, the users of MRI must have a good knowledge of the imaging
process and the protocol factors and know how to set them to optimize the
image characteristics.
The overall process of optimizing protocols will be described in Chapter 11.

Mind Map Summary

Magnetic Resonance Image Characteristics
 The magnetic resonance image is a display of radio frequency signal
intensities that are emitted by magnetized tissue during the imaging process.
The tissue becomes magnetized because it contains protons that are the
magnetic nuclei of hydrogen atoms. When placed in the strong magnetic field,
some of the protons align with the field producing the tissue magnetization.
The level of magnetization at the time during the procedure when the “picture
is snapped” is determined by a variety of tissue and fluid movement
characteristics. By adjusting the imaging process it is possible to produce
images in which these various characteristics are the principal sources of
contrast.
An advantage of MRI is the ability to selectively image a variety of
tissue and fluid characteristics. If a specific pathologic condition is not visible
when viewing one characteristic, there is the possibility of seeing it by imaging
some of the other characteristics.
During the imaging procedure a section of the patient’s body is divided
first into slices, and the slices are divided into a matrix of voxels. Each voxel is
an independent RF signal source. Voxel size can be adjusted and is what
determines image detail and also affects image noise.
The five major image quality characteristics—contrast sensitivity,
detail, noise, artifacts, and spatial—can be controlled to a great extent by the
settings of the various protocol factors.
MRI is a powerful diagnostic tool because the process can be
optimized to display a wide range of clinical conditions. However, maximum
benefit requires a staff with the knowledge to control the process and interpret
the variety of images.
Chapter  2.  MRi  system  components  
 
Introduction And Overview
The MRI system consists of several major components, as shown in Figure 2-
1. At this time we will introduce the components and indicate how they work
together to create the MR image. The more specific details of the image
forming process will be explained in later chapters.

Figure 2-1. The major components of the Magnetic Resonance Imaging

System.

The heart of the MRI system is a large magnet that produces a very strong
magnetic field. The patient’s body is placed in the magnetic field during the
imaging procedure. The magnetic field produces two distinct effects that work
together to create the image.
Tissue Magnetization
When the patient is placed in the magnetic field, the tissue becomes
temporarily magnetized because of the alignment of the protons, as described
previously. This is a very low-level effect that disappears when the patient is
removed from the magnetic field. The ability of MRI to distinguish between
different types of tissue is based on the fact that different tissues, both normal
and pathologic, will become magnetized to different levels or will change their
levels of magnetization (i.e., relax) at different rates.
Tissue Resonance
The magnetic field also causes the tissue to “tune in” or resonate at a very
specific radio frequency. That is why the procedure is known as magnetic
resonance imaging. It is actually certain nuclei, typically protons, within the
tissue that resonate. Therefore, the more comprehensive name for the
phenomenon that is the basis of both imaging and spectroscopy is nuclear
magnetic resonance (NMR).
In the presence of the strong magnetic field the tissue resonates in the
RF range. This causes the tissue to function as a tuned radio receiver and
transmitter during the imaging process. The production of an MR image
involves two-way radio communication between the tissue in the patient’s
body and the equipment.

The Magnetic Field

Figure 2-2 shows the general characteristics of a typical magnetic field. At any
point within a magnetic field, the two primary characteristics are field
direction and field strength.

Figure 2-2. The magnetic field produced by superconducting magnets.

Field Direction
It will be easier to visualize a magnetic field if it is represented by a series of
parallel lines, as shown in Figure 2-2. The arrow on each line indicates the
direction of the field. On the surface of the earth, the direction of the earth’s
magnetic field is specified with reference to the north and south poles. The
north-south designation is generally not applied to magnetic fields used for
imaging. Most of the electromagnets used for imaging produce a magnetic
field that runs through the bore of the magnet and parallel to the major patient
axis. As the magnetic field leaves the bore, it spreads out and encircles the
magnet, creating an external fringe field. The external field can be a source of
interference with other devices and is usually contained by some form of
shielding.
Field Strength
Each point within a magnetic field has a particular intensity, or strength. Field
strength is expressed either in the units of tesla (T) or gauss (G). The
relationship between the two units is that 1.0 T is equal to 10,000 G or 10 kG.
At the earth’s surface, the magnetic field is relatively weak and has a strength
of less than 1 G. Magnetic field strengths in the range of 0.15 T to 1.5 T are
used for imaging. The significance of field strength is considered as we
explore the characteristics of MR images and image quality in later chapters.

Homogeneity
MRI requires a magnetic field that is very uniform, or homogeneous with
respect to strength. Field homogeneity is affected by magnet design,
adjustments, and environmental conditions. Imaging generally requires a
homogeneity (field uniformity) on the order of a few parts per million (ppm)
within the imaging area.
 High homogeneity is obtained by the process of shimming, as
described later.

Magnets
There are several different types of magnets that can be used to produce the
magnetic field. Each has its advantages and disadvantages.
Superconducting
Most MRI systems use superconducting magnets. The primary advantage is
that a superconducting magnet is capable of producing a much stronger and
stable magnetic field than the other two types (resistive and permanent)
considered below. A superconducting magnetic is an electromagnet that
operates in a superconducting state. A superconductor is an electrical
conductor (wire) that has no resistance to the flow of an electrical current.
This means that very small superconducting wires can carry very large
currents without overheating, which is typical of more conventional conductors
like copper. It is the combined ability to construct a magnet with many loops or
turns of small wire and then use large currents that makes the strong
magnetic fields possible.
There are two requirements for superconductivity. The conductor or
wire must be fabricated from a special alloy and then cooled to a very low
temperature. The typical magnet consists of small niobium-titanium (Nb-Ti)
wires imbedded in copper. The copper has electrical resistance and actually
functions as an insulator around the Nb-Ti superconductors.
During normal operation the electrical current flows through the
superconductor without dissipating any energy or producing heat. If the
temperature of the conductor should ever rise above the critical
superconducting temperature, the current begins to produce heat and the
current is rapidly reduced. This results in the collapse of the magnetic field.
This is an undesirable event known as a quench. More details are given in
Chapter 15 on safety. Superconducting magnets are cooled with liquid helium.
A disadvantage of this magnet technology is that the coolant must be
replenished periodically.
A characteristic of most superconducting magnets is that they are in
the form of cylindrical or solenoid coils with the strong field in the internal
bore. A potential problem is that the relatively small diameter and the long
bore produce claustrophobia in some patients. Superconducting magnetic
design is evolving to more open patient environments to reduce this concern.
Resistive
A resistive type magnet is made from a conventional electrical conductor such
as copper. The name “resistive” refers to the inherent electrical resistance that
is present in all materials except for superconductors. When a current is
passed through a resistive conductor to produce a magnetic field, heat is also
produced. This limits this type of magnet to relatively low field strengths.
Permanent
It is possible to do MRI with a non-electrical permanent magnet. An obvious
advantage is that a permanent magnet does not require either electrical
power or coolants for operation. However, this type of magnet is also limited
to relatively low field strengths.
Both resistive and permanent magnets are usually designed to
produce vertical magnetic fields that run between the two magnetic poles, as
shown in Figure 2-3. Possible advantages include a more open patient
environment and less external field than superconducting magnets.

Figure 2-3. The magnetic field produced by typical resistive or

permanent magnets.

Gradients
When the MRI system is in a resting state and not actually producing an
image, the magnetic field is quite uniform or homogeneous over the region of
the patient’s body. However, during the imaging process the field must be
distorted with gradients. A gradient is just a change in field strength from one
point to another in the patient’s body. The gradients are produced by a set of
gradient coils, which are contained within the magnet assembly. During an
imaging procedure the gradients are turned on and off many times. This
action produces the sound or noise that comes from the magnet.
The effect of a gradient is illustrated in Figure 2-4. When a magnet is
in a “resting state,” it produces a magnetic field that is uniform or homogenous
over most of the patient’s body. In this condition there are no gradients in the
field. However, when a gradient coil is turned on by applying an electric
current, a gradient or variation in field strength is produced in the magnetic
field.
Figure 2-4. A magnetic field gradient produced by a current in the
gradient cell.

Gradient Orientation
The typical imaging magnet contains three separate sets of gradient coils.
These are oriented so that gradients can be produced in the three orthogonal
directions (often designated as the x, y, and z directions). Also, two or more of
the gradient coils can be used together to produce a gradient in any desired
direction.

Gradient Functions
The gradients are used to perform many different functions during the image
acquisition process. It is the gradients that create the spatial characteristics by
producing the slices and voxels that will be described in Chapter 9. The entire
family of gradient echo imaging methods uses a gradient to produce the echo
event and signal which will be described in Chapter 7. Gradients are also
used to produce one type of image contrast (phase contrast angiography) for
vascular imaging, as will be described in Chapter 12, and in the functional
imaging methods described in Chapter 1.3 Gradients also are used as part of
some of the techniques to reduce image artifacts, as will be described in
Chapter 14.
Gradient Strength
The strength of a gradient is expressed in terms of the change in field
strength per unit of distance. The typical units are millitesla per meter (mT/m).
The maximum gradient strength that can be produced is a design
characteristic of a specific imaging system. High gradient strengths of 20
mT/m or more are required for the optimum performance of some imaging
methods.
Risetime and Slew-Rate
 For certain functions it is necessary for the gradient to be capable of
changing rapidly. The risetime is the time required for a gradient to reach its
maximum strength. The slew-rate is the rate at which the gradient changes
with time. For example, a specific gradient system might have a risetime of
0.20 milliseconds (msec) and a slew-rate of 100 mT/m/msec.
Eddy Currents
Eddy currents are electrical currents that are induced or generated in metal
structures or conducting materials that are within a changing magnetic field.
Since gradients are strong, rapidly changing magnetic fields, they are capable
of producing undesirable eddy currents in some of the metal components of
the magnet assembly. This is undesirable because the eddy currents create
their own magnetic fields that interfere with the imaging process.
Gradients are designed to minimize eddy currents either with special
gradient shielding or electrical circuits that control the gradient currents in a
way that compensates for the eddy-current effects.

Shimming
One of the requirements for good imaging is a homogeneous magnet field.
This is a field in which there is a uniform field strength over the image area.
Shimming is the process of adjusting the magnetic field to make it more
uniform.
Inhomogeneities are usually produced by magnetically susceptible
materials located in the magnetic field. The presence of these materials
produces distortions in the magnetic field that are in the form of
inhomogeneities. This can occur in both the internal and external areas of the
field. Each time a different patient is placed in the magnetic field, some
inhomogeneities are produced. There are many things in the external field,
such as building structures and equipment, that can produce inhomogeneities.
The problem is that when the external field is distorted, these distortions are
also transferred to the internal field where they interfere with the imaging
process. Inhomogeneities produce a variety of problems that will be discussed
later.
It is not possible to eliminate all of the sources of inhomogeneities.
Therefore, shimming must be used to reduce the inhomogeneities. This is
done in several ways. When a magnet is manufactured and installed, some
shimming might be done by placing metal shims in appropriate locations.
Magnets also contain a set of shim coils. Shimming is produced by adjusting
the electrical currents in these coils. General shimming is done by the
engineers when a magnet is installed or serviced. Additional shimming is
done for individual patients. This is often done automatically by the system.

Magnetic Field Shielding

The external magnetic field surrounding the magnet is the possible source of
two types of problems. One problem is that the field is subject to distortions by
metal objects (building structures, vehicles, etc.) as described previously.
These distortions produce inhomogeneities in the internal field. The second
problem is that the field can interfere with many types of electronic equipment
such as imaging equipment and computers.
It is a common practice to reduce the size of the external field by
installing shielding as shown in Figure 2-5. The principle of magnetic field
shielding is to provide a more attractive return path for the external field as it
passes from one end of the magnetic field to the other. This is possible
because air is not a good magnetic field conductor and can be replaced by
more conductive materials, such as iron. There are two types of
shielding: passive and active.

Figure 2-5. The principle of magnetic field shielding.

Passive Shielding
Passive shielding is produced by surrounding the magnet with a structure
consisting of relatively large pieces of ferromagnetic materials such as iron.
The principle is that the ferromagnetic materials are a more attractive path for
the magnetic field than the air. Rather than expanding out from the magnet,
the magnetic field is concentrated through the shielding material located near
the magnet as shown in Figure 2-5. This reduces the size of the field.
Active Shielding
Active shielding is produced by additional coils built into the magnet
assembly. They are designed and oriented so that the electrical currents in
the coils produce magnetic fields that oppose and reduce the external
magnetic field.

The Radio Frequency System

The radio frequency (RF) system provides the communications link with the
patient’s body for the purpose of producing an image. All medical imaging
modalities use some form of radiation (e.g., x-ray, gamma-ray, etc.) or energy
(e.g., ultrasound) to transfer the image from the patient’s body.
The MRI process uses RF signals to transmit the image from the
patient’s body. The RF energy used is a form of non-ionizing radiation. The
RF pulses that are applied to the patient’s body are absorbed by the tissue
and converted to heat. A small amount of the energy is emitted by the body as
signals used to produce an image. Actually, the image itself is not formed
within and transmitted from the body. The RF signals provide information
(data) from which the image is reconstructed by the computer. However, the
resulting image is a display of RF signal intensities produced by the different
tissues.

RF Coils
The RF coils are located within the magnet assembly and relatively close to
the patient’s body. These coils function as the antennae for both transmitting
signals to and receiving signals from the tissue. There are different coil
designs for different anatomical regions (shown in Figure 2-6). The three
basic types are body, head, and surface coils. The factors leading to the
selection of a specific coil will be considered in Chapter 10. In some
applications the same coil is used for both transmitting and receiving; at other
times, separate transmitting and receiving coils are used.

Figure 2-6. The three types of RF coils (body, head, and surface) that are
the antennae
for transmitting pulses and receiving signals from the patient’s body.

Surface coils are used to receive signals from a relatively small anatomical
region to produce better image quality than is possible with the body and head
coils. Surface coils can be in the form of single coils or an array of several
coils, each with its own receiver circuit operated in a phased
array configuration. This configuration produces the high image quality
obtained from small coils but with the added advantage of covering a larger
anatomical region and faster imaging.
Transmitter
The RF transmitter generates the RF energy, which is applied to the coils and
then transmitted to the patient’s body. The energy is generated as a series of
discrete RF pulses. As we will see in Chapters 6, 7, and 8, the characteristics
of an image are determined by the specific sequence of RF pulses.
The transmitter actually consists of several components, such as RF
modulators and power amplifiers, but for our purposes here we will consider it
as a unit that produces pulses of RF energy. The transmitters must be
capable of producing relatively high power outputs on the order of several
thousand watts. The actual RF power required is determined by the strength
of the magnetic field. It is actually proportional to the square of the field
strength. Therefore, a 1.5 T system might require about nine times more RF
power applied to the patient than a 0.5 T system. One important component of
the transmitter is a power monitoring circuit. That is a safety feature to prevent
excessive power being applied to the patient’s body, as described in Chapter
15.
Receiver
A short time after a sequence of RF pulses is transmitted to the patient’s
body, the resonating tissue will respond by returning an RF signal. These
signals are picked up by the coils and processed by the receiver. The signals
are converted into a digital form and transferred to the computer where they
are temporarily stored.
RF Polarization
The RF system can operate either in a linear or a circularly polarized mode.
In the circularly polarized mode, quadrature coils are used. Quadrature coils
consist of two coils with a 90˚ separation. This produces both improved
excitation efficiency by producing the same effect with half of the RF energy
(heating) to the patient, and a better signal-to-noise ratio for the received
signals.
RF Shielding
RF energy that might be in the environment could be picked up by the
receiver and interfere with the production of high quality images. There are
many sources of stray RF energy, such as fluorescent lights, electric motors,
medical equipment, and radio communications devices. The area, or room, in
which the patient’s body is located must be shielded against this interference.
An area can be shielded against external RF signals by surrounding it
with an electrically conducted enclosure. Sheet metal and copper screen wire
are quite effective for this purpose.
The principle of RF shielding is that RF signals cannot enter an
electrically conductive enclosure. The thickness of the shielding is not a
factor—even thin foil is a good shield. The important thing is that the room
must be completely enclosed by the shielding material without any holes. The
doors into imaging rooms are part of the shielding and should be closed
during image acquisition.
Computer Functions
A digital computer is an integral part of an MRI system. The production and
display of an MR image is a sequence of several specific steps that are
controlled and performed by the computer.
Acquisition Control
The first step is the acquisition of the RF signals from the patient’s body. This
acquisition process consists of many repetitions of an imaging cycle. During
each cycle a sequence of RF pulses is transmitted to the body, the gradients
are activated, and RF signals are collected. Unfortunately, one imaging cycle
does not produce enough signal data to create an image. Therefore, the
imaging cycle must be repeated many times to form an image. The time
required to acquire images is determined by the duration of the imaging cycle
or cycle repetition time—an adjustable factor known as TR—and the number
of cycles. The number of cycles used is related to image quality. More cycles
generally produce better images. This will be described in much more detail in
Chapters 10 and 11.
Protocols stored in the computer control the acquisition process. The
operator can select from many preset protocols for specific clinical procedures
or change protocol factors for special applications.
Image Reconstruction
The RF signal data collected during the acquisition phase is not in the form of
an image. However, the computer can use the collected data to create or
“reconstruct” an image. This is a mathematical process known as a Fourier
transformation that is relatively fast and usually does not have a significant
effect on total imaging time.
Image Storage and Retrieval
The reconstructed images are stored in the computer where they are
available for additional processing and viewing. The number of images that
can be stored—and available for immediate display—depends on the capacity
of the storage media.
Viewing Control and Post Processing
The computer is the system component that controls the display of the
images. It makes it possible for the user to select specific images and control
viewing factors such as windowing (contrast) and zooming (magnification).
In many applications it is desirable to process the reconstructed
images to change their characteristics, to reformat an image or set of images,
or to change the display of images to produce specific views of anatomical
regions.
These post-processing (after reconstruction) functions are performed
by a computer. In some MRI systems some of the post processing is
performed on a work-station computer that is in addition to the computer
contained in the MRI system.

Mind Map Summary

Magnetic Resonance Imaging System Components

The magnetic resonance imaging system consists of several major

components that function together to produce images. During the image
acquisition process the patient’s body is placed in a strong magnetic field. At
each point, the magnetic field has a specific direction. This direction is used
as a reference for expressing the direction of tissue magnetization. The
strength of a magnetic field is determined by the type and design of the
magnet. Superconducting magnets can produce strong magnetic fields.
Resistive and permanent magnets are limited to relatively weak field
strengths. The homogeneity, or uniformity of field strength is necessary for
good imaging. Homogeneity is reduced by magnetically susceptible materials
that come into the field and produce distortions. This can occur in both the
external field and within a patient’s body. Shimming is the process of adjusting
the magnetic field to make it more homogeneous. This can be achieved by
passive shims that are added when a magnet is installed and with active
shimming produced by adjusting the currents in the shimming coils.
Shielding of the magnetic field reduces the size and strength of the
external magnetic field and also improves homogeneity by protecting from
interference caused by objects in the external field area.
A gradient is an intentional variation in magnetic field strength that is
produced by the gradient coils. There are three basic gradient coils that are
oriented to produce gradients in the three orthogonal directions. Gradients
perform several functions during the image acquisition process. An important
characteristic of a gradient, especially for some advanced image procedures,
is its strength and how fast it can be turned on and off.
The MRI process consists of an exchange of RF pulses and signals
between the equipment and the patient’s body. This is done through the RF
coils that serve as the antenna for transmitting the pulses and receiving the
signals. It is necessary to shield the imaging area by enclosing it in a
conductive metal (copper) room to block external RF interference.
The imaging process is controlled by information stored in a computer.
The protocols programmed into the computer and selected by the operator
guide the imaging process and determine the characteristics of the images.
The RF signals collected from the patient’s body during the acquisition
process are used by the computer to reconstruct the image.
 Chapter 3. Nuclear Magnetic Resonance

Introduction And Overview

When certain materials, such as tissue, are placed in a strong magnetic field,
two things happen. The materials take on a resonant characteristic and they
become magnetized. In this chapter we will consider the resonant
characteristic. In Chapter 4 we will study the magnetization
effect. Resonance means the materials can absorb and then re-radiate RF
radiation at a specific frequency, like a radio receiver-transmitter, as illustrated
in Figure 3-1. It is actually the nuclei of the atoms that resonate. The
phenomenon is generally known as nuclear magnetic resonance (NMR). The
resonant frequency of material such as tissue is typically in the RF range so
that the emitted radiation is in the form of radio signals. The specific resonant
frequency is determined by three factors as shown in the illustration and will
be described in detail later. The characteristics of the RF signals emitted by
the material are determined by certain physical and chemical characteristics
of the material. The RF signals produced by the NMR process can be
displayed either in the form of images (MRI) or as a graph depicting chemical
composition (MR spectroscopy).
 

Figure 3-1. The concept of Nuclear Magnetic Resonance (NMR).

Magnetic Nuclei

Materials that participate in the MR process must contain nuclei with specific
magnetic properties. In order to interact with a magnetic field, the nuclei
themselves must be small magnets and have a magnetic property or
magnetic moment, as shown in Figure 3-2. The magnetic characteristic of an
individual nucleus is determined by its neutron-proton composition. Only
certain nuclides with an odd number of neutrons and protons are magnetic.
Even though most chemical elements have one or more isotopes with
magnetic nuclei, the number of magnetic isotopes that might be useful for
either imaging or in vivo spectroscopic analysis is somewhat limited. Among
the nuclides that are magnetic and can participate in an NMR process, the
amount of signal produced by each nuclide varies considerably
 Figure 3-2. Magnetic and non-
magnetic nuclei.

Spins
Protons and neutrons that make up a nucleus have an intrinsic angular
momentum or spin. Pairs of protons and neutrons align in such a way that
their spins cancel. However, when there is an odd number of protons or
neutrons (odd mass numbers), some of the spins will not be canceled and the
total nucleus will have a net spin characteristic. It is this spinning
characteristic of a particle with an electric charge (the nucleus) that produces
a magnetic property known as the magnetic moment.
It is for this reason that magnetic nuclei, such as protons, are often
referred to as spins.
The magnetic property, or magnetic moment, of a nucleus has a
specific direction. In Figure 3-2, the direction of the magnetic moment is
indicated by an arrow drawn through the nucleus.
RF Signal Intensity
The intensity of the RF signal emitted by tissue is probably the most
significant factor in determining image quality and the time required to acquire
an image. This important issue is considered in Chapters 10 and 11. We now
begin to introduce the factors that contribute to signal intensity.
During the imaging process, the body section is divided into an array
of individual volume elements, or voxels. It is the signal intensity from each
voxel that determines image quality. The signal is produced by the magnetic
nuclei within each voxel. Therefore, signal intensity is, in general, proportional
to the quantity of magnetic nuclei within an individual voxel. We now consider
the factors that affect the number of magnetic nuclei within an individual voxel
Relative Signal Strength
The relative signal strength from the various chemical elements in tissue is
determined by three factors: (1) tissue concentration of the element; (2)
isotopic abundance; and (3) sensitivity of the specific nuclide.
In comparison to all other nuclides, hydrogen produces an extremely
strong signal. This results from its high values for each of the three
contributing factors.
Of the three factors, only the concentration, or density, of the nuclei
varies from point to point within an imaged section of tissue. The quantity is
often referred to as proton density and is the most fundamental tissue
characteristic that determines the intensity of the RF signal from an individual
voxel, and the resulting pixel brightness. In most imaging situations, pixel
brightness is proportional to the density (concentration) of nuclei (protons) in
the corresponding voxel, although additional factors, such as relaxation times,
modify this relationship.
Protons in solids, such as the tabletop and bone, do not produce
signals. Signals come only from protons in molecules that are free to move,
as in a liquid state.
Tissue Concentration of Elements
The concentration of chemical elements in tissue covers a considerable
range, depending on tissue type and such factors as metabolic or pathologic
state. The concentrations of elements in tissue are in two groups. Four
elements—hydrogen, carbon, nitrogen, and oxygen—typically make up at
least 99% of tissue mass.
The most abundant isotopes of the four elements are hydrogen-1,
carbon-12, nitrogen-14, and oxygen-16. Note that the mass number of
hydrogen (1) is odd while the mass numbers of the other three (12, 14, 16)
are even. Therefore, hydrogen is the only one of these four isotopes that has
a strong magnetic nucleus. The nucleus of the hydrogen-1 atom is a single
proton. Among all the chemical elements, hydrogen Spins is unique in that it
occurs in relatively high concentrations in most tissues, and the most
abundant isotope (H-1) has a magnetic nucleus.
Other elements, such as sodium, phosphorus, potassium, and
magnesium, are present in very low concentrations. Calcium is concentrated
in bone or localized deposits.
Within this group of elements with low tissue concentrations are
several with magnetic nuclei. These include fluorine-19, sodium-23,
phosphorus-31, and potassium-39.
Isotopic Abundance
Most chemical elements have several isotopes. When a chemical element is
found in a naturally occurring substance, such as tissue, most of the element
is typically in the form of one isotope, with very low concentrations of the other
isotopic forms. For the three elements— carbon, nitrogen, and oxygen—that
have a high concentration in tissue, the magnetic isotopes are the ones with a
low abundance in the natural state. These include carbon-l3, nitrogen-15, and
oxygen-17.
Relative Sensitivity and Signal Strength
The signal strength produced by an equal quantity of the various nuclei also
varies over a considerable range. This inherent NMR sensitivity is typically
expressed relative to hydrogen-1, which produces the strongest signal of all of
the nuclides. The relative sensitivities of some magnetic nuclides are shown in
Table 3-1.

Table 3-1. Relative Sensitivities of Some Magnetic Nuclides

Nuclide Sensitivity
Hydrogen-1 1.0
Fluorine-19 0.83
Sodium-23 0.093
Phosphorous-1 0.066

In summary, hydrogen has a lot going for it: 1) a high tissue concentration; 2)
the most abundant isotope (H-1) is magnetic; and 3) it produces a relatively
strong signal compared to an equal concentration of other nuclei. That is why
hydrogen is the only element that is imaged with conventional MRI systems.

Radio Frequency Energy

During an imaging procedure, RF energy is exchanged between the imaging
system and the patient’s body. This exchange takes place through a set of
coils located relatively close to the patient’s body as we saw in Chapter 2. The
RF coils are the antennae that transmit energy to and receive signals from the
tissue.
Pulses
RF energy is applied to the body in several short pulses during each imaging
cycle. The strength of the pulses is described in terms of the angle through
which they rotate or flip the magnetic nuclei and the resulting tissue
magnetization, as described later. Many imaging methods use both 90˚ and
180˚ pulses in each cycle.
Signals
At a specific time in each imaging cycle, the tissue is stimulated to emit an
RF signal, which is picked up by the coils, analyzed, and used to form the
image. The spin echo or gradient echo methods are generally used to
stimulate signal emission. Therefore, the signals from the patient’s body are
commonly referred to as echoes.

Nuclear Magnetic Interactions

The NMR process is a series of interactions involving the magnetic nuclei, a
magnetic field, and RF energy pulses and signals.
Nuclear Alignment
Recall that a magnetic nucleus is characterized by a magnetic moment. The
direction of the magnetic moment is represented by a small arrow passing
through the nucleus. If we think of the nucleus as a small conventional
magnet, the magnetic moment arrow corresponds to the south pole-north pole
direction of the magnet.
In the absence of a strong magnetic field, magnetic moments of nuclei
are randomly oriented in space. Many nuclei in tissue are not in a rigid
structure and are free to change direction. In fact, nuclei are constantly
tumbling, or changing direction, because of thermal activity within the
material; in this case, tissue.
When a material containing magnetic nuclei is placed in a magnetic
field, the nuclei experience a torque that encourages them to align with the
direction of the field. In the human body, however, thermal energy agitates the
nuclei and keeps most of them from aligning parallel to the magnetic field. The
number of nuclei that do align with the magnetic field is proportional to the
field strength. The magnetic fields used for imaging can align only a few of
every million magnetic nuclei present. However, this is sufficient to produce a
useful NMR effect.
Precession and Resonance
When a spinning magnetic nucleus aligns with a magnetic field, it is not fixed;
the nuclear magnetic moment precesses, or oscillates, about the axis of the
magnetic field, as shown in Figure 3-3. The precessing motion is a physical
phenomenon that results from an interaction between the magnetic field and
the spinning momentum of the nucleus.
 Figure 3-3. Magnetic nuclei
precession and resonance in a
magnetic field.

Precession is often observed with a child’s spinning top. A spinning top does
not stand vertical for long, but begins to wobble, or precess. In this case, the
precession is caused by an interaction between the earth’s gravitational field
and the spinning momentum of the top.
The precession rate (cycles per second) is directly proportional to the
strength of the magnetic field. It is this precessing motion that makes a
nucleus sensitive and receptive to incoming RF energy when the RF
frequency matches the precession rate. This precession rate corresponds to
the resonant frequency. It is the precessing nuclei, typically protons, that are
tuned to receive and transmit RF energy.
Excitation
If a pulse of RF energy with a frequency corresponding to the nuclear
precession rate is applied to the material, some of the energy will be absorbed
by the individual nuclei. The absorption of energy by a nucleus flips its
alignment away from the direction of the magnetic field, as shown in Figure 3-
4. This increased energy places the nucleus in an unnatural, or excited, state.

Figure 3-4. The excitation of a magnetic nucleus by the application

of a pulse of RF energy.

In MRI an RF pulse is used that flips some of the nuclei into the transverse
plane of the magnetic field. In this excited state the precession is now
transformed into a spinning motion of the nucleus around the axis of the
magnetic field. It should be noted that this spinning motion is an enhanced
precession and is different from the intrinsic spin of a nucleus about its own
axis.
The significance of a magnetic nucleus spinning around the axis of the
magnetic field is that this motion now generates an RF signal as shown in
Figure 3-5. It is this signal, from many nuclei, that is collected to form the MR
image.

Figure 3-5. RF signal production by magnetic nuclei spinning

in the transverse plane of a magnetic field.

Relaxation
When a nucleus is in an excited state, it experiences an increased torque
from the magnetic field, urging it to realign. The nucleus can return to a
position of alignment by transferring its excess energy to other nuclei or the
general structure of the material. This process is known as relaxation.
Relaxation is not instantaneous following an excitation. It cannot occur
until the nucleus is able to transfer its excess energy. How quickly the energy
transfer takes place depends on the physical characteristics of the tissue. In
fact, the nuclear relaxation rate (or time) is, in many cases, the most
significant factor in producing contrast among different types of tissue in an
image.
We are more interested in the collective relaxation of many nuclei that
produce the magnetization of tissue and will return to this point in the next
chapter.

Resonance
The significance of the nuclear precession is that it causes the nucleus to be
extremely sensitive, or tuned, to RF energy that has a frequency identical with
the precession frequency (rate). This condition is known as resonance and is
the basis for all MR procedures. NMR is the process in which a nucleus
resonates, or “tunes in,” when it is in a magnetic field.
Resonance is fundamental to the absorption and emission of energy
by many objects and devices. Objects are most effective in exchanging
energy at their own resonant frequency. The resonance of an object or device
is determined by certain physical characteristics. Let us consider two common
examples.
Radio receivers operate on the principle of resonant frequency. A
receiver can select a specific broadcast station because each station
transmits a different frequency. Tuning a radio is actually adjusting its
resonant frequency. Its receiver is very sensitive to radio signals at its
resonant frequency and insensitive to all other frequencies.
The strings of a musical instrument also have specific resonant
frequencies. This is the frequency at which the string vibrates to produce a
specific audio frequency, or musical note. The resonant frequency of a string
depends on the amount of tension. It can be changed, or tuned, by changing
the tension. This is somewhat analogous to the resonant frequency of a
magnetic nucleus being dependent on the strength of the magnetic field in
which it is located.
Larmor Frequency
The resonant frequency of a nucleus is determined by a combination of
nuclear characteristics and the strength of the magnetic field. The resonant
frequency is also known as the Larmor frequency. The specific relationship
between resonant frequency and field strength is an inherent characteristic of
each nuclide and is generally designated the gyromagnetic ratio. The Larmor
frequencies [in megahertz (MHz)] for selected nuclides in a magnetic field of 1
T are shown in Table 3-2.

Table 3-2. Larmor Frequencies for Selected Nuclides in a Magnetic Field

Nuclide Larmor Frequency
(MHz/T)
Hydrogen-1 42.58
Fluorine-19 40.05
Phosphorous-31 17.24
Sodium-23 11.26

The fact that different nuclides have different resonant frequencies means
that most MR procedures can “look at” only one chemical element (nuclide) at
a time.
Field Strength
For all nuclides, the resonant frequency is proportional to the strength of the
magnetic field. In a very general sense, increasing the magnetic field strength
increases the tension on the nuclei (as with the strings of a musical
instrument) and increases the resonant frequency. The fact that a specific
nuclide can be tuned to different radio frequencies by varying the field
strength (i.e., applying gradients) is used in the imaging process.
Chemical Shift
The resonant frequency of magnetic nuclei, such as protons, is also affected
by the structure of the molecule in which they are located.
When a proton, or other magnetic nucleus, is part of a molecule, it is
slightly shielded from the large magnetic field. The amount of shielding
depends on the chemical composition of the molecule. This means that
protons in different chemical compounds will be in slightly different field
strengths and will therefore resonate at different frequencies. This change in
resonant frequency from one compound to another is known as chemical
shift. It can be used to perform chemical analysis in the technique of MR
spectroscopy and to produce images based on chemical composition.
However, in conventional MRI the chemical-shift effect can be the source of
an unwanted artifact.
In tissue the chemical shift in resonant frequency between the fat and
water is approximately 3.3 ppm, as shown in Figure 3-6. At a field strength of
1.5 T the protons have a basic resonant frequency of approximately 64 MHz.
Multiplying this by 3.3 gives a water-fat chemical shift of approximately 210
Hz. At a field strength of 0.5 T the chemical shift would be only 70 Hz.

Figure 3-6. The chemical shift

effect on the relative resonant
frequency
of protons in fat and water

There are several imaging techniques that can be used to selectively image
either the water or fat tissue components. One approach is to suppress either
the fat or water signal with specially designed RF pulses. This technique is
known as spectral presaturation and will be described in Chapter 8. Another
technique makes use of the fact that the signals from water and fat are not
always in step, or in phase, with each other and can be separated to create
either water or fat images.
When a magnetic nucleus is located in a strong magnetic field, it resonates. In
effect, it becomes a tuned radio receiver and transmitter. The resonance
occurs because the spinning nucleus precesses at a rate that is in the radio
frequency range. The resonant frequency is determined by three factors.
Each specific nuclide has a unique resonant frequency. The resonant
frequency is affected to a small degree by the structure of the molecule
containing the magnetic nucleus. This, the chemical shift effect, is useful for
spectroscopy and to suppress fat signals in images. It can also lead to a
certain type of image artifact. The resonant frequency is directly proportional
to the strength of the magnetic field. This is useful because it makes it
possible to tune the various parts of a body to different frequencies by
applying magnetic field gradients.

When an RF pulse is applied to a magnetic nucleus oriented in the

longitudinal direction, it can be flipped into the transverse plane. There the
nucleus spins around the axis of the magnetic field and generates an RF
signal. It is the signals from many spinning nuclei that are collected and used
to form the image. It is necessary to have strong signals to produce good
images. Signal strength depends on three factors. Each magnetic nuclide has
a unique sensitivity or relative signal strength. All chemical elements have
several different isotopes, but all isotopes of an element are usually not in the
form of magnetic nuclei. Therefore, the abundance of the magnetic isotope for
a specific element has a major effect on signal strength. To produce strong
signals a tissue must have a relatively high concentration of a chemical
element and the most abundant isotope of that element must be magnetic.

Hydrogen is the only chemical element with a high concentration in

tissue and body fluids in the form of an isotope that has a magnetic nucleus.
Therefore, MR imaging is essentially limited to visualizing only one chemical
element, hydrogen.

Chapter 4: Tissue Magnetization And Relaxation

Introduction And Overview

We have considered the behavior of individual nuclei when placed in a

magnetic field. MRI depends on the collective, or net, magnetic effect of a
large number of nuclei within a specific voxel of tissue. If a voxel of tissue
contains more nuclei aligned in one direction than in other directions, the
tissue will be temporarily magnetized in that particular direction. This process
is illustrated in Figure 4-1. In the absence of a magnetic field, the nuclei are
randomly oriented and produce no net magnetic effect. This is the normal
state of tissue before being placed in a magnetic field. When the tissue is
placed in a magnetic field, and some of the nuclei align with the field, their
combined effect is to magnetize the tissue in the direction of the magnetic
field. A large arrow, the magnetization vector, is used to indicate the amount
and direction of the magnetization. When tissue is placed in a magnetic field,
the maximum magnetization that can be produced depends on three factors:
(1) the concentration (density) of magnetic nuclei, typically protons, in the
tissue voxel; (2) the magnetic sensitivity of the nuclide; and (3) the strength of
the magnetic field. Since an imaging magnetic field aligns a very small fraction
of the magnetic nuclei, the tissues are never fully magnetized. The amount of
tissue magnetization determines the strength of the RF signals emitted by the
tissue during an imaging or analytical procedure. This, in turn, affects image
quality and imaging time, as explained in Chapter 10.

Figure 4-1. The magnetization of

tissue produced by the
alignment of magnetic nuclei
(protons) in a magnetic field.

Let us recall that an MR image is an image of magnetized tissue and that the
contrast we see is produced by different levels of magnetization that exist in
the different tissues at the time when “the picture is snapped.” As we will see
in this chapter the level of magnetization at specific times during the imaging
process is determined by the three tissue characteristics: proton density (PD),
T1, and T2.

We will now see how these characteristics produce image contrast.

 Tissue Magnetization
When tissue is placed in a magnetic field, it reaches its maximum
magnetization within a few seconds and remains at that level unless it is
disturbed by a change in the magnetic field or by pulses of RF energy applied
at the resonant frequency. The MRI procedure is a dynamic process in which
tissue is cycled through changes in its magnetization during each imaging
cycle.

Magnetic Direction
The direction of tissue magnetization is specified in reference to the direction
of the applied magnetic field, as shown in Figure 4-2. There are two principle
directions that tissue is magnetized during the imaging process. Longitudinal
magnetization is when the tissue is magnetized in a direction parallel to the
direction of the field. Transverse magnetization is when the direction of tissue
magnetization is at a 90¡ angle with respect to the direction of the magnetic
field and is in the transverse plane.

Figure 4-2. Longitudinal and

transverse magnetization.

Magnetic Flipping
The direction of tissue magnetization can be changed or flipped by applying a
pulse of RF energy. This is done many times throughout the imaging process.

Flip Angle
The angle the magnetization is flipped is determined by the duration and
strength of the RF pulse. Pulses are characterized by their flip angles.

Pulses with 90° and 180° flip angles are the most common but smaller flip
angle pulses are also used in some imaging methods, such as gradient echo
imaging.

The 90° Pulse, Saturation and Excitation

When a 90° pulse is applied to longitudinal magnetization, it flips it into the
transverse plane as shown in Figure 4-3. This has two effects. First, it reduces
the longitudinal magnetization to zero, a condition called saturation. It also
produces transverse magnetization. As we will soon learn, transverse
magnetization is an unstable or excited condition. Therefore, when a 90°
pulse is applied to longitudinal magnetization, it produces both saturation of
the longitudinal magnetization and a condition of excitation (transverse
magnetization).

Figure 4-3. The application of a

90° RF pulse to longitudinal
magnetization produces
saturation of the longitudinal
magnetization and creates
transverse magnetization, an
excited condition.

The actual direction of magnetization is not limited to longitudinal or

transverse. It can exist in any direction. In principle, magnetization can have
both longitudinal and transverse components. Since the two components
have distinctly different characteristics, we consider them independently.

Longitudinal Magnetization And Relaxation

As we have seen, when tissue is placed in a magnetic field, it becomes

magnetized in the longitudinal direction. It will remain in this state until the
magnetic field is changed or until the magnetization is redirected by the
application of an RF pulse. If the magnetization is temporarily redirected by an
RF pulse, it will then, over a period of time, return to its original longitudinal
position. If we consider only the longitudinal magnetization, it regrows after it
has been reduced to zero, or saturated. This regrowth, or recovery, of
longitudinal magnetization is the relaxation process, which occurs after
saturation. The time required for the longitudinal magnetization to regrow, or
relax, depends on characteristics of the material and the strength of the
magnetic field.

Longitudinal magnetization does not grow at a constant rate, but at an

exponential rate, as shown in Figure 4-4. An important concept to remember
is that the MR image is an image of magnetized tissue with brightness
indicating the level of magnetization. During the relaxation process, the level
of magnetization is changing. Therefore, the brightness of tissue (if we could
see it) is also changing as indicated by the scale on the right of the illustration.
Saturation turns the tissue dark and then it recovers brightness during the
relaxation period.
Figure 4-4. The growth of longitudinal magnetization (and tissue
brightness) during the relaxation process following saturation.

The characteristic that varies from one type of tissue to another, and can be
used to produce image contrast, is the time required for the magnetization to
re-grow, or the relaxation time. Because of its exponential nature, it is difficult
to determine exactly when the magnetization has reached its maximum. The
convention is to specify the relaxation time in terms of the time required for
the magnetization to reach 63% of its maximum. This time, the longitudinal
relaxation time, is designated T1. The 63% value is used because of
mathematical, rather than clinical, considerations. Longitudinal magnetization
continues to grow with time, and reaches 87% of its maximum after two T1
intervals, and 95% after three T1 intervals. For practical purposes, the
magnetization can be considered fully recovered after approximately three
times the T1 value of the specific tissue. We will see later that this must be
taken into consideration when setting up an imaging procedure.

T1 Contrast
The time required for a specific level of longitudinal magnetization regrowth
varies from tissue to tissue. Figure 4-5 shows the regrowth of two tissues with
different T1 values. In this illustration we watch the intensity of brightness of a
voxel of tissue during the relaxation process. Let us recall that the brightness
of a tissue (RF signal intensity) is determined by the level of magnetization
existing in a voxel of tissue at any instant in time. What we see in an image
depends on when we “snap the picture” during the relaxation process. The
important thing to notice is that the tissue with the shortest T1 has the highest
level of magnetization at any particular time. The clinical significance of this is
that tissues with short T1 values will be bright in T1-weighted images.
Figure 4-5. The formation of contrast between two tissues with different
T1 values.

Table 4-1 lists typical T1 values for various tissues. Two materials establish
the lower and upper values for the T1 range: fat has a short T1, and fluid falls
at the other extreme (long T1). Therefore, in T1-weighted images, fat is
generally bright, and fluid [cerebrospinal fluid (CSF), cyst, etc.] is dark. Most
other body tissues are within the range between fat and fluid.

The longitudinal relaxation process involves an interaction between

the protons and their immediate molecular environment. The rate of relaxation
(T1 value) is related to the naturally occurring molecular motion. The
molecular motion is determined by the physical state of the material and the
size of the molecules. The relatively rigid structure of solids does not provide
an environment for rapid relaxation, which results in long T1 values. Molecular
motion in fluids, and fluid-like substances, is more inducible to the relaxation
process. In this environment molecular size becomes an important
characteristic.

Relaxation is enhanced by a general matching of the proton resonant

frequency and the frequency associated with the molecular motions.
Therefore, factors that change either of these two frequencies will generally
have an effect on T1 values

Molecular Size
Small molecules, such as water, have faster molecular motions than large
molecules, such as lipids. The frequencies associated with the molecular
motion of water molecules are both higher and more dispersed over a larger
range for the larger molecules. This reduces the match between the
frequencies of the protons and the frequencies of the molecular environment.
This is why water and similar fluids have relatively long T1 values. Larger
molecules, which have slower and less dispersed molecular movement, have
a better frequency match with the proton resonant frequencies. This enhances
the relaxation process and produces short T1 values. Fat is an excellent
example of a large molecular structure that exhibits this characteristic.
 Table 4-1. T2 and T1 Values for Various Tissues
T2 T1 (0.5 T) T1 (1.5 T)
(msec) (msec) (msec)
Adipose (Fat) 80 210 260
Liver 42 350 500
Muscle 45 550 870
White Matter 90 500 780
Gray Matter 100 650 920
CSF 100 1800 2400

Tissues generally contain a combination of water and a variety of

larger molecules. Some of the water can be in a relatively free state while
other water is bound to some of the larger molecules. In general, the T1 value
of the tissue is probably affected by the exchange of water between the free
and the bound states. When the water is bound to larger molecular structures,
it takes on the motion characteristics of the larger molecule. Factors such as a
pathologic process, which alters the water composition of tissue, will generally
alter the T1 values.

Magnetic Field Strength Effect

T1 values depend on the strength of the magnetic field. This is because the
field strength affects the resonant frequency of the protons. As field strength is
increased, the resonant frequency also increases and becomes less matched
to the molecular motion frequencies. This results in an increase in T1 values,
as indicated in Table 4-1.

Let us now combine two factors to create a T1 image as illustrated in

Figure 4-6. One factor is that different tissues have different T1 values and
rates of regrowth of longitudinal magnetization. This then causes the different
tissues to be at different levels of magnetization (brightness) when the picture
is snapped during the relaxation period. Here we see the order of tissue
brightness is inversely related to T1 values. In principle, the tissues with short
T1 values get brighter faster and are at a higher level when the picture is
snapped.

Figure 4-6. A T1 image showing the

relationship of tissue brightness (signal
intensity) to T1 values and level of
magnetization during the longitudinal
relaxation process.

Transverse Magnetization And Relaxation

Transverse magnetization is produced by applying a pulse of RF energy to

the magnetized tissue. This is typically done with a 90˚ pulse, which converts
longitudinal magnetization into transverse magnetization. Transverse
magnetization is an unstable, or excited, condition and quickly decays after
the termination of the excitation pulse. The decay of transverse magnetization
is also a relaxation process, which can be characterized by specific relaxation
times, or T2 values. Different types of tissue have different T2 values that can
be used to discriminate among tissues and contribute to image contrast.

Transverse magnetization is used during the image formation process

for two reasons: (1) to develop image contrast based on differences in T2
values; and (2) to generate the RF signals emitted by the tissue. Longitudinal
magnetization is an RF silent condition and does not produce any signal.
However, transverse magnetization is a spinning magnetic condition within
each tissue voxel, and that generates an RF signal. As we will see in the next
chapter, each imaging cycle must conclude with transverse magnetization to
produce the RF signal used to form the image.

The characteristics of transverse magnetization and relaxation are

quite different from those for the longitudinal direction. A major difference is
that transverse magnetization is an unstable condition and the relaxation
process results in the decay, or decrease, in magnetization, as shown in
Figure 4-7. The T2 value is the time required for 63% of the initial
magnetization to dissipate. After one T2, 37% of the initial magnetization is
present.

Figure 4-7. The decay of

transverse magnetization during
the relaxation process and the
associated tissue brightness.

T2 Contrast
The difference in T2 values of tissues is the source of contrast in T2-weighted
images. This is illustrated in Figure 4-8. Here we watch two tissues, with
different T2 values, during the relaxation process. We see that they are both
getting darker with time as the magnetization decays. However, they are not
getting darker at the same rate. The tissue with the shorter T2 becomes
darker faster leaving the tissue with the longer T2 to be bright at times during
the relaxation time.
 Figure 4-8. The formation of T2
contrast during the decay of
transverse magnetization.

What we will actually see in a T2-weighted image, as shown in Figure 4-9,

depends on the level of magnetization at the time when we snap the picture.
The important thing to observe here is that the tissues with long T2 values are
bright in T2 images.

Figure 4-9. A T2 image showing

the relationship of tissue
brightness
(signal intensity) to T2 values.

In general, a T2-weighted image appears to be a reversal of a T1-weighted

image. Tissues that are bright in one image are dark in the other image. This
is because of a combination of two factors. One factor is that T1 and T2
values are generally related. Even though T2 values are much shorter than T1
values, as shown in Table 4-1, they are somewhat proportional. Tissues with
long T1 values usually have long T2 values. The other factor is that the order
of brightness in a T2 image is in the same direction as the T2 values.
Remember, it was a reversed relationship for T1 images.

The decay of transverse magnetization (i.e., relaxation) occurs

because of a dephasing among individual nuclei (protons) within the individual
voxels, as shown in Figure 4-10.

Figure 4-10. The dephasing of

protons that produces
transverse magnetization decay.

   
Two basic conditions are required for transverse magnetization: (1) the
magnetic moments of the nuclei must be oriented in the transverse direction,
or plane; and (2) a majority of the magnetic moments must be in the same
direction, or in phase, within the transverse plane. When a nucleus has a
transverse orientation, it is actually spinning around an axis that is parallel to
the magnetic field.

After the application of a 90˚ pulse, the nuclei have a transverse

orientation and are rotating together, or in phase, around the magnetic field
axis. This rotation or spin is a result of the normal precession discussed
earlier. The precession rate, or resonant frequency, depends on the strength
of the magnetic field where the nuclei are located. Nuclei located in field areas
with different strengths spin (precess) at different rates. Even within a very
small volume of tissue, nuclei are in slightly different magnetic field strengths.
As a result, some nuclei spin faster than others. Also, there are interactions
(spin-spin interactions) among the spinning nuclei. After a short period of time,
the nuclei are not spinning in phase. As the directions of the nuclei begin to
spread and they dephase, the magnetization of the tissue decreases. A short
time later, the nuclei are randomly oriented in the transverse plane, and there
is no transverse magnetization.

Proton Dephasing
Two major effects contribute to the dephasing of the nuclei and the resulting
transverse relaxation. In the imaging process the spin echo technique is used
to separate the two sources of dephasing, as we will see in Chapter 6.

T2 Tissue Characteristics
One effect is the exchange of energy among the spinning nuclei (spin-spin
interactions), which results in relatively slow dephasing and loss of
magnetization. The rate at which this occurs is determined by characteristics
of the tissue. It is this dephasing activity that is characterized by the T2 values
as shown in Table 4-1.

T2* Magnetic Field Effects

A second effect, which produces relatively rapid dephasing of the nuclei and
loss of transverse magnetization, is the inherent inhomogeneity of the
magnetic field within each individual voxel. The field inhomogeneities are
sufficient to produce rapid dephasing. This effect, which is different from the
basic T2 characteristics of the tissue, tends to mask the true relaxation
characteristics of the tissue. In other words, the actual transverse
magnetization relaxes much faster than the tissue characteristics would
indicate. This real relaxation time is designated as T2*. The value of T2* is
usually much less than the tissue T2 value, as illustrated in Figure 4-11.
Several factors can contribute to field inhomogeneities and to T2* decay. One
is the general condition of the magnetic field. Some fields are more
homogeneous than others. Another factor is that different tissues or materials
in the body might have different magnetic susceptibilities. Susceptibility is a
characteristic of a material that determines its ability to become magnetized
when it is in a magnetic field. If a region of tissue contains materials with
different susceptibilities, this results in a reduction of field homogeneity.
 Figure 4-11. Comparison of
relaxation produced by the T2
characteristics of tissue and the
T2* effects associated with
magnetic field inhomogeneities.

Magnetic Susceptibility

The magnetization of tissue that we have been discussing is a nuclear

magnetic effect produced by the alignment of magnetic nuclei in a magnetic
field. Other materials can become magnetized by other, non-nuclear effects.

Many materials are susceptible to magnetic fields and become

magnetized when located in fields. The susceptibility of a material is
determined by the orbital electrons in the atom rather than the magnetic
properties of the nucleus. Significant susceptibility is present only when there
are unpaired electrons in the outer orbit.

There are three general types of materials with respect to magnetic

susceptibility: diamagnetic, paramagnetic, and ferromagnetic. The primary
characteristic of each type is the amount and direction of magnetization that
the material develops when placed in a magnetic field. There are situations
when each type plays a role in the MR imaging process.

Contrast Agents

The inherent tissue characteristics (PD, T1, and T2) do not always produce
adequate contrast for some clinical objectives. It is possible to administer
materials (i.e., contrast agents) that will alter the magnetic characteristics
within specific tissues or anatomical regions. There are several different types
of contrast agents, which will now be considered. Contrast agents used in
MRI are generally based on relaxation effects.

Diamagnetic Materials

Diamagnetic materials have negative and relatively low magnetic

susceptibility. This means that they develop only low levels of magnetization
and it is in a direction opposite to the direction of the magnetic field. Although
many biological molecules are diamagnetic, this is not a significant factor in
MR imaging.

Paramagnetic Materials
 Paramagnetic materials play an important role in contrast enhancement.
They are materials with unpaired electrons that give each atom a permanent
magnetic property. In paramagnetic materials each atom is magnetically
independent, which distinguishes it from other materials to be discussed later.

Paramagnetic substances include metal ions such as gadolinium,

manganese, iron, and chromium. Other substances such as nitroxide free
radicals and molecular oxygen also have paramagnetic properties.

Gadolinium has seven unpaired electrons in its orbit, which give it a

very strong magnetic property. It must be chelated to reduce its toxicity. An
example is gadolinium chelated to diethylene triamine penta-acetic acid
(GaDTPA).

When a paramagnetic substance, such as gadolinium, enters an

aqueous solution, it affects the relaxation rate of the existing protons. It does
not produce a signal itself. In relatively low concentrations, the primary effect
is to increase the rate of longitudinal relaxation and shorten the value of T1. In
principle, the fluctuating magnetic field from the individual paramagnetic
molecules enhances the relaxation rate. The primary result is an increase in
signal intensity with T1-weighted images. It is classified as a positive contrast
agent.

Signal intensity will generally increase with the concentration of the

paramagnetic agents until a maximum intensity is reached. This intensity is
very dependent on the imaging parameters. Higher concentrations will
generally produce a reduction of signal intensity. This occurs because the
transverse relaxation rate is also increased, which results in a shortening of
the T2 value.

Superparamagnetic Materials

When materials with unpaired electrons are contained in a crystalline

structure, they produce a stronger magnetic effect (susceptibility) in
comparison with the independent molecules of a paramagnetic substance.
The susceptibility of superparamagnetic materials is several orders of
magnitude greater than that of paramagnetic materials. These materials are in
the form of small particles. Iron oxide particles are an example.

The particles produce inhomogeneities in the magnetic field, which

results in rapid de-phasing of the protons in the transverse plane and a
shortening of T2.

Superparamagnetic materials in the form of large particles generally

reduce signal intensity and are classified as negative contrast agents. When
in the form of very small particles, they reduce T1 and increase signal
intensity.

Ferromagnetic Materials
Ferromagnetic is the name applied to iron and only a few other materials that
have magnetic properties like iron. These materials have a very high
susceptibility and develop a high level of magnetism when placed in a
magnetic field.

Mind Map Summary

Tissue Magnetization And Relaxation

When tissue containing magnetic nuclei, i.e., protons, is placed in a

strong magnetic field, the tissue becomes magnetized. It is initially
magnetized in the longitudinal direction. However, by applying a pulse of RF
energy the magnetization can be flipped into the transverse plane. Both
longitudinal and transverse magnetization have characteristics that can be
used to develop image contrast. An imaging procedure can be adjusted to
display the different types of contrasts.

When a 90° RF pulse is applied to longitudinal magnetization, it

produces two effects. First, it temporarily destroys the longitudinal
magnetization, a condition known as saturation. It also produces transverse
magnetization, a condition known as excitation because transverse
magnetization is an unstable excited state.

After a saturation pulse is applied, the longitudinal magnetization will

recover or regrow, a process known as relaxation. The rate of regrowth is a
characteristic of each specific tissue and is described by its T1 value, the
longitudinal relaxation time. A tissue with a short T1 will recover its
magnetization fast and will appear bright in a T1-weighted image. Tissues
with longer T1 values will recover magnetization somewhat slower and will be
relatively dark in T1-weighted images.

Following the production of transverse magnetization by the RF pulse

the magnetization begins to decay or relax. The rate of relaxation is a
characteristic of each specific tissue and is expressed by the T2 values, the
transverse relaxation time. A tissue with a short T2 will lose its transverse
magnetization rapidly and will appear relatively dark in T2-weighted images.
Tissues and body fluids with long T2 values will retain their transverse
magnetization longer and will appear bright in T2-weighted images.

Chapter 5: The Imaging Process

Introduction And Overview

The MR imaging process consists of two major functions as shown in Figure

5-1. The first is the acquisition of RF signals from the patient’s body and the
second is the mathematical reconstruction of an image from the acquired
signals.

Figure 5-1. The two

functions, acquisition and recons
truction, that make up the MR
image production process.

In this chapter we will develop a general overview of the imaging process and
set the stage for considering the different methods and techniques that are
used to produce optimum images for various clinical needs

k Space

During the acquisition process the signals are collected, digitized, and stored
in computer memory in a configuration known as k space. The k space is
divided into lines of data that are filled one at a time. One of the general
requirements is that the k space must be completely filled before the image
reconstruction can be completed. The size of k space (number of lines) is
determined by the requirements for image detail and will be discussed in
Chapters 9 thru 11.
Acquisition

The acquisition process consists of an imaging cycle that is repeated many

times. The time required for a complete acquisition is determined by the
duration of the cycle multiplied by the number of cycles. The duration of a
cycle is TR (Time of Repetition), the adjustable protocol factor that is used to
select the different types of image contrast. Also, the number of cycles used in
an acquisition is adjustable. The number of cycles depends on the quality of
the image that is required. The complete relationship between number of
imaging cycles and image quality characteristics is described in Chapter 10.

Reconstruction

The image reconstruction process is usually fast compared to the acquisition

process and generally does not require any decisions or adjustments by the
operator.

Imaging Protocol

Each imaging procedure is controlled by a protocol that has been entered into
the computer. Issues that must be considered in selecting, modifying, or
developing a protocol for a specific clinical procedure include:

The imaging method to be used

The image types (PD, T1, T2, etc.)
Spatial characteristics (slice thickness, number, etc.)
Detail and visual noise requirements
Use of selective signal suppression techniques
Use of artifact reduction techniques

In the following chapters we will address each of these issues and the
specific protocol factors that are used to produce the desired image
characteristics.

Imaging Methods

There are several different imaging methods that can be used to create MR
images. The principal difference among these methods is the sequence in
which the RF pulses and gradients are applied during the acquisition process.
Therefore, the different methods are often referred to as the different pulse
sequences. An overview of the most common methods is shown in Figure 5-
2. As we see, the different methods are organized in a hierarchy structure. For
each imaging method there is a set of factors that must be adjusted by the
user to produce specific image characteristics.
Figure 5-2. The principal spin echo and gradient echo imaging methods. GRandSE, or GRASE,
is a combination of the two methods.

The selection of a specific imaging method and factor values is generally

based on requirements for contrast sensitivity to a specific tissue
characteristic (PD, T1, T2) and acquisition speed. However, other
characteristics such as visual noise and the sensitivity to specific artifacts
might vary from method to method.

All of the imaging methods belong to one or both of the two major
families, spin echo or gradient echo. The difference between the two families
of methods is the process that is used to create the echo event at the end of
each imaging cycle. For the spin echo methods, the echo event is produced
by the application of a 180° RF pulse, as will be described in Chapter 6. For
the gradient echo methods the event is produced by applying a magnetic field
gradient, as described in Chapter 7. Each method has very specific
characteristics and applications.

The Imaging Cycle

A common characteristic of all methods is that there are two distinct phases of
the image acquisition cycle, as shown in Figure 5-3.

Figure 5-3. The longitudinal and transverse

magnetization phases of an imaging cycle.
T1 and PD contrast are produced during
the longitudinal phase and T2 contrast is
produced during the transverse phase.
One phase is associated with longitudinal magnetization and the other with
transverse magnetization. In general, T1 contrast is developed during the
longitudinal magnetization phase and T2 contrast is developed during the
transverse magnetization phase. PD contrast is always present, but becomes
most visible when it is not overshadowed by either T1 or T2 contrast. The
predominant type of contrast that ultimately appears in the image is
determined by the duration of the two phases and the transfer of contrast from
the longitudinal phase to the transverse phase.

The duration of the two phases (longitudinal and transverse) is

determined by the selected values of the protocol factors, TR (Time of
Repetition) and TE (Time to Echo).

TR
TR is the time interval between the beginning of the longitudinal relaxation,
following saturation, and the time at which the longitudinal magnetization is
converted to transverse magnetization by the excitation pulse. This is when
the picture is snapped relative to the longitudinal magnetization.

Because the longitudinal relaxation takes a relatively long time, TR is

also the duration of the image acquisition cycle or the cycle repetition time
(Time of Repetition).

TE
TE is the time interval between the beginning of transverse relaxation
following excitation and when the magnetization is measured to produce
image contrast. This happens at the echo event and is when the picture is
snapped relative to the transverse magnetization. Therefore, TE is the Time to
Echo event.

Excitation
The transition from the longitudinal magnetization phase to the transverse
magnetization phase is produced by applying an RF pulse. This is generally
known as the excitation process because the transverse magnetization
represents a more unstable or “excited” state than longitudinal magnetization.

The excitation pulse is characterized by a flip angle. A 90˚ excitation pulse

converts all of the existing longitudinal magnetization into transverse
magnetization. This type of pulse is used in the spin echo methods. However,
there are methods that use excitation pulses with flip angles that are less than
90˚. Small flip angles (<90˚) convert only a fraction of the existing longitudinal
magnetization into transverse magnetization and are used primarily to reduce
acquisition time with the gradient echo methods described in Chapter 7.

The Echo Event and Signals

The transverse magnetization phase terminates with the echo event, which
produces the RF signal. This is the signal that is emitted by the tissue and
used to form the image. The echo event is produced by applying either an RF
pulse or a gradient pulse to the tissue, as will be described in Chapters 6 and
7

Contrast Sensitivity
In MRI the usual procedure is to select one of the tissue characteristics (PD,
T1, T2) and then adjust the imaging process so that it has maximum, or at
least adequate, contrast sensitivity for that specific characteristic. This
produces an image that is heavily weighted by that characteristic. The
contrast sensitivity of the imaging process and the resulting image contrast is
determined by the specific imaging method and the combination of imaging
protocol factor values, which we will consider in much more detail in later
chapters. The discussion in this chapter will be based on the conventional
spin echo method that uses only two factors, TR and TE, to control contrast
sensitivity. However, it establishes some principles that apply to all methods.

T1 Contrast
During the relaxation (regrowth) of longitudinal magnetization, different
tissues will have different levels of magnetization because of their different
growth rates, or T1 values. Figure 5-4 compares two tissues with different T1
values.

Figure 5-4. The amount of T1 contrast

captured during the longitudinal
magnetization phase is determined by
the value of TR that is selected by the
operator

The tissue with the shorter T1 value experiences a faster regrowth of

longitudinal magnetization. Therefore, during this period of time it will have a
higher level of magnetization, produce a more intense signal, and appear
brighter in the image. In T1-weighted images brightness or high signal
intensity is associated with short T1 values.

t the beginning of each imaging cycle, the longitudinal magnetization is

reduced to zero (saturation) by an RF pulse, and then allowed to regrow, or
relax. This is what happens in the spin echo method. In some other imaging
methods, as we will see in the next two chapters, the cycle might begin with
either partially saturated or inverted longitudinal magnetization. In all cases,
T1 contrast is formed during the regrowth process. At a time determined by
the selected TR value, the cycle is terminated and the magnetization value is
converted to transverse, measured and displayed as a pixel intensity, or
brightness, and a T1-weighted image is produced.
 In principle, at the beginning of each imaging cycle all tissues are
dark. As the tissues regain longitudinal magnetization, they become brighter.
The brightness, or intensity, with which they appear in the image depends on
when during the regrowth process the cycle is terminated and the picture is
snapped. This is determined by the selected TR value. When a short TR is
used, the regrowth of the longitudinal magnetization is interrupted before it
reaches its maximum. This reduces signal intensity and tissue brightness
within the image but produces T1 contrast.

Increasing TR increases signal intensity and brightness up to the point

at which magnetization is fully recovered, which is determined by the PD of
each tissue. For practical purposes, this occurs when the TR exceeds
approximately three times the T1 value for the specific tissues. Although it
takes many cycles to form a complete image, the longitudinal magnetization is
always measured at the same time in each cycle as determined by the setting
of TR.

To produce a T1-weighted image, a value for TR must be selected to

correspond with the time at which T1 contrast is significant between the two
tissues. Several factors must be considered in selecting TR. If T1 contrast is
represented by the ratio of the tissue magnetization levels, it is at its
maximum very early in the relaxation process. However, the low
magnetization levels present at that time do not generally produce adequate
RF signal levels for many clinical applications. The selection of a longer TR
produces greater signal strength but less T1 contrast.

The selection of TR must be appropriate for the T1 values of the

tissues being imaged. If a TR value is selected that is equal to the T1 value of
a tissue, the picture will be snapped when the tissue has regained 63% of its
magnetization. This represents the time when there is maximum contrast
between tissues with small differences in T1 values.

Proton Density (PD) Contrast

The density, or concentration, of protons in each tissue voxel determines the

maximum level of magnetization that can be obtained. Differences in PD
among tissues can be used to produce image contrast, as illustrated in Figure
5-5.
 Figure 5-5. Proton density (PD)
contrast is captured by setting
TR to relatively long values. At
that time the magnetization is
determined by PD and is not
T1 as in the earlier part of the
cycle.

Here we see the growth of longitudinal magnetization for two tissues with the
same T1 values but different relative PDs. The tissue with the lowest PD (80)
reaches a maximum magnetization level that is only 80% that of the other
tissue. The difference in magnetization levels at any point in time is because
of the difference in PD and is therefore the source of PD contrast.

Although there is some PD contrast early in the cycle, it is generally

quite small in comparison to the T1 contrast.

The basic difference between T1 contrast and PD contrast is that T1

contrast is produced by the rate of growth (relaxation), and PD contrast is
produced by the maximum level to which the magnetization grows. In general,
T1 contrast predominates in the early part of the relaxation phase, and PD
contrast predominates in the later portion. T1 contrast gradually gives way to
PD contrast as magnetization approaches the maximum value. A PD-
weighted image is produced by selecting a relatively long TR value so that the
image is created or “the picture is snapped” in the later portion of the
relaxation phase, where tissue magnetizations approach their maximum
values. The TR values at which this occurs depend on the T1 values of the
tissues being imaged.

It was shown earlier that tissue reaches 95% of its magnetization in

three T1s. Therefore, a TR value that is at least three times the T1 values for
the tissues being imaged produces almost pure PD contrast.

T2 Contrast

Now let us turn our attention to the transverse phase. During the decay of
transverse magnetization, different tissues will have different levels of
magnetization because of different decay rates, or T2 values. As shown in
Figure 56, tissue with a relatively long T2 value will have a higher level of
magnetization, produce a more intense signal, and appear brighter in the
image than a tissue with a shorter T2 value.
 Figure 5-6 shows the decay of transverse magnetization for tissues
with different T2 values. The tissue with the shortest T2 value loses its
magnetization faster than the other tissues.

Figure 5-6. The formation

of T2 contrast during the
transverse magnetization
phase. The amount of T2
contrast captured depends
on the selected value of TE,
the Time to Echo event.

The difference in T2 values of tissue can be translated into image contrast.

For the purpose of this illustration we assume that the two tissues begin their
transverse relaxation with the levels of magnetization determined by the PD.
This is the usual case where the PD contrast present at the end of the
longitudinal phase carries over to the beginning of the transverse phase. In
effect, the transverse phase begins with PD contrast but adds T2 contrast as
time elapses. The decay of the magnetization proceeds at different rates
because of the different T2 values. The tissue with the longer T2 value
maintains a higher level of magnetization than the other tissue and will remain
bright longer. The difference in the tissue magnetizations at any point in time
represents contrast.

At the beginning of the cycle there is no T2 contrast, but it develops

and increases throughout the relaxation process. At the echo event the
magnetization levels are converted into RF signals that are displayed as
image pixel brightness; this is the time to echo event (TE) and is selected by
the operator. Maximum T2 contrast is generally obtained by using a relatively
long TE. However, when a very long TE value is used, the magnetization and
the RF signals might be too low to form a useful image. In selecting TE
values, a compromise must often be made between T2 contrast and good
signal intensity.

The transverse magnetization characteristics of tissue (T2 values) are,

in principle, added to the longitudinal characteristics carried over from the
longitudinal phase (e.g., T1 and PD) to form the MR image. Usually we do not
want to add T2 contrast to T1 contrast. That is because these two types of
contrast oppose each other. Remember in Chapter 1 we saw that tissues that
are bright in T1 images are dark in T2 images. This means that if we were to
mix T1 and T2 contrast in the same image, one would cancel the other. When
setting up a protocol for a T2 image it is necessary to use a long TR (in
addition to a long TE) so that no, or very little, T1 contrast carries over to the
transverse phase.
The MR imaging process is one of creating contrast among tissues based on
their magnetic characteristics. The primary characteristics are proton density
(PD), T1, and T2. It is a dynamic activity in which the magnetization levels of
the various tissues are undergoing almost constant change. During each
imaging cycle there are two distinct magnetization phases: longitudinal and
transverse. Different types of contrast are developed in each of these phases.

After application of a saturation pulse, which reduces the longitudinal

magnetization to zero, the magnetization begins to regrow, a process known
as relaxation. The rate of regrowth for a specific tissue is determined by that
tissue’s T1 value. Tissues with short T1 values grow faster than tissues with
long T1 values. During this regrowth, T1 contrast in the form of different levels
of magnetization is created among the tissues. This is the contrast that will be
displayed in an image if the protocol parameters are set to produce a T1-
weighted image. In a T1-weighted image, tissues with short T1 values will be
bright. Tissues and fluid with long T1 values will be darker.

When an RF pulse is applied to longitudinal magnetization, it converts

(flips) it to transverse magnetization, an unstable excited magnetic condition
that decays with time. This decay process is the transverse magnetization
relaxation process. The rate of decay of a specific tissue depends on that
tissue’s T2 value. Tissues with short T2 values decay faster than tissues with
longer T2 values. When the imaging protocol factors are set to produce a T2-
weighted image, tissues with short T2 values will be dark and tissues and
fluids with longer T2 values will be bright.
 Chapter 6: Spin Echo Imaging Methods

Introduction And Overview

Spin echo is the name of the process that uses an RF pulse to produce the
echo event. It is also the name for one of the specific imaging methods within
the spin echo family of imaging methods; all of which use the spin echo
process. We will first discuss the spin echo process and see how an RF pulse
can produce an echo event and signal and then consider the spin echo
methods.

The Spin Echo Process

The decay of transverse magnetization (i.e., relaxation) occurs because of

dephasing among individual nuclei, as described in Chapter 4.

Let us recall that two basic conditions are required for transverse
magnetization: (1) the magnetic moments of the nuclei must be oriented in the
transverse direction, or plane, and (2) a majority of the moments must be in
the same direction within the transverse plane. When a nucleus has a
transverse orientation, it is actually precessing or rotating around an axis that
is parallel to the magnetic field.

After the application of a 90˚ excitation pulse, the nuclei have a

transverse orientation and are precessing together, or in-phase, around the
magnetic field axis. This is the normal precession discussed earlier but flipped
into the transverse plane. However, within an individual voxel some nuclei
precess or spin faster than others. After a short period of time, the nuclei are
not spinning in-phase. As the directions of the nuclei begin to spread, the
magnetization of the tissue decreases. A short time later, the nuclei are
randomly oriented in the transverse plane; there is no transverse
magnetization.

The two factors that contribute to the de-phasing of the nuclei and the
resulting transverse relaxation will now be reviewed again here. One is an
exchange among the spinning nuclei (spin-spin interactions), which results in
relatively slow dephasing and loss of magnetization. The rate at which this
occurs is determined by characteristics of the tissue. It is this dephasing
activity that is characterized by the T2 values and the source of contrast that
we want to capture in T2 images. A second factor, which produces relatively
rapid dephasing of the nuclei and loss of transverse magnetization, is the
inhomogeneity of the magnetic field. Even within a small volume of tissue, the
field inhomogeneities are sufficient to produce rapid dephasing. This effect,
which is generally unrelated to the T2 characteristics of the tissue, tends to
mask the true relaxation characteristics of the tissue. In other words, the
actual transverse magnetization relaxes much faster than the tissue
characteristics would indicate. We remember that this real relaxation time is
designated as T2*. The value of T2* is always much less than the tissue T2
value. As a result, the transverse magnetization disappears before T2
contrast can be formed.

We are about to discover that spin echo is a process for recovering

the lost transverse magnetization and making it possible to produce images of
the three tissue characteristics, including T2.

An RF signal is produced whenever there is transverse magnetization.

Immediately after an excitation pulse, a so-called free induction decay (FID)
signal is produced. The intensity of this signal is proportional to the level of
transverse magnetization. Both decay rather rapidly because of the magnetic
field inhomogeneities just described. The FID signal is not used in the spin
echo methods. It is used in the gradient echo methods to be described in
Chapter 7.

The spin echo process is used to compensate for the dephasing and
rapid relaxation caused by the field inhomogeneities and to restore the
magnetization to the level that depends only on the tissue T2 characteristics.
The sequence of events in the spin echo process is illustrated in Figure 6-1.

Figure 6-1. The spin

echo process showing
the use of a 180° pulse
to rephase the protons
and to produce an
echo event.

Transverse magnetization is produced with a 90˚ RF excitation pulse that flips

the longitudinal magnetization into the transverse plane. Immediately following
the RF pulse, each voxel is magnetized in the transverse direction. However,
because of the local magnetic field inhomogeneities within each voxel, the
protons precess at different rates and quickly slip out of phase. This produces
the rapid decay characterized by T2* and the associated FID signal. At this
time the protons are still rotating in the transverse plane, but they are out of
phase.

If a 180° pulse is applied to the tissue containing these protons, it flips

the protons around an axis in the transverse plane; this reverses their
direction of rotation as illustrated in Figure 6-2.
 Figure 6-2. The 180° pulse
sets up the protons so that
they rephase.

This causes the fast protons to be located behind the slower ones. As the
faster protons begin to catch up with the slower ones, they regain a common
alignment, or come back into phase. This, in turn, causes the transverse
magnetization to reappear and form the echo event. However, the
magnetization does not grow to the initial value because the relaxation
(dephasing) produced by the tissue is not reversible. The rephasing of the
protons causes the magnetization to build up to a level determined by the T2
characteristics of the tissue. As soon as the magnetization reaches this
maximum, the protons begin to move out of phase again, and the transverse
magnetization dissipates. Another 180˚ pulse can be used to produce another
rephasing. In fact, this is what is done in multi-echo imaging and will be
described later in this chapter.

RF Pulse Sequence

The different imaging methods are produced by the type (flip angle) and time
intervals between the applied RF pulses. The basic pulse sequence for the
spin echo method is shown in Figure 6-3. Each cycle begins with a 90°
excitation pulse that produces the initial transverse magnetization and a later
180° pulse that rephases the protons to produce the echo event.

Figure 6-3. The

RF pulses and
time intervals in
a spin echo
imaging cycle.
The time between the initial excitation and the echo signal is TE. This is
controlled by adjusting the time interval between the 90˚ and the 180˚ pulses,
which is 1/2 TE.

The Spin Echo Method

This method can be used to produce images of the three basic tissue
characteristics: PD, T1, and T2. The sensitivity to a specific characteristic is
determined by the values selected for the two time intervals or imaging
factors, TR and TE.

The process of creating images with the three types of contrast (PD,
T1, and T2) described in the last chapter was a description of the spin echo
method. There we saw that the type of image that was produced depended on
the values selected for the two protocol factors, TR and TE. We will now
review that process with a few more details specifically as it applies to the
spin echo method.

Proton Density (PD) Contrast

PD contrast develops as the longitudinal magnetization approaches its

maximum, which is determined by the PD of each specific tissue. Therefore,
relatively long TR values are required to produce a PD-weighted image. Short
TE values are generally used to reduce T2 contrast contamination and to
maintain a relatively high signal intensity.

T1 Contrast

To produce image contrast based on T1 differences between tissues, two

factors must be considered. Since T1 contrast develops during the early
growth phase of longitudinal magnetization, relatively short TR values must be
used to capture the contrast. The second factor is to preserve the T1 contrast
during the time of transverse relaxation. The basic problem is that if T2
contrast is allowed to develop, it generally counteracts T1 contrast. This is
because tissues with short T1 values usually have short T2 values. The
problem arises because tissues with short T1s are generally bright, whereas
tissues with short T2s have reduced brightness when T2 contrast is present.
T2 contrast develops during the TE time interval. Therefore, a T1-weighted
image is produced by using short TR values and short TE values.

T2 Contrast

The first step in producing an image with significant T2 contrast is to select a

relatively long TR value. This minimizes T1 contrast contamination and the
transverse relaxation process begins at a relatively high level of
magnetization. Long TE values are then used to allow T2 contrast time to
develop.

The spin echo method is the only method that produces true T2 contrast. That
is because it is able to rephase the protons and remove the T2* effect.

Multiple Spin Echo

It is possible to produce a series of echo events within one cycle as illustrated

in Figure 6-4.

Figure 6-4. A
multiple spin
echo imaging that
produces both a
PD and T2 image
in the same
acquisition.

This is done by applying several 180° pulses after each 90° excitation pulse.
The advantage is that echo events with different TE values are produced in
one acquisition cycle. Separate images are formed for each TE value. This
makes it possible to create both a PD image (short TE) and a T2 image (long
T2) in the same acquisition.
Table 6-1 summarizes the combination of TR and TE values used to
produce the three basic image types with the spin echo method. Optimum
values of TR and TE for a specific protocol might vary because of
considerations for other factors such as image acquisition time, number of
slices, etc.

Table 6-1. Selection of TR and TE values to produce the three image types
with spin echo method. Values shown are typical but can be varied to some
extent to accommodate specific imaging conditions.

T1 Image PD Image T2 Image

TR Short Long Long
(500 msec) (2000 msec) (2000 msec)
TE Short Long Long
(15-20 msec) (15-20 msec) (120 msec)
Inversion Recovery
Inversion recovery is a spin echo imaging method used for several specific
purposes. One application is to produce a high level of T1 contrast and a
second application is to suppress the signals and resulting brightness of fat
and fluids. The inversion recovery pulse sequence is obtained by adding an
additional 180˚ pulse to the conventional spin echo sequence, as shown in
Figure 6-5.

Figure 6-5. The

inversion
recovery method
with TI set to
produce an image
with high T1
contrast.

The pulse is added at the beginning of each cycle where it is applied to the
longitudinal magnetization carried over from the previous cycle. Each cycle
begins as the 180˚ pulse inverts the direction of the longitudinal
magnetization. The regrowth (recovery) of the magnetization starts from a
negative (inverted) value, rather than from zero, as in the spin echo method.

The inversion recovery method, like the spin echo method, uses a 90°
excitation pulse to produce transverse magnetization and a final 180° pulse to
produce a spin echo signal. That is why it is classified as one of the spin echo,
rather than gradient echo, methods. An additional time interval is associated
with the inversion recovery pulse sequence. The time between the initial 180˚
pulse and the 90˚ pulse is designated the Time after Inversion (TI). It can be
varied by the operator and used as a contrast control.

T1 Contrast

The principal characteristic of many inversion recovery images is high T1

contrast. This occurs because the total longitudinal relaxation time is
increased because it starts from the inverted state. There is more time for the
T1 contrast to develop. A T1 image produced by the inversion recovery
method is compared to one produced by the spin echo method in Figure 6-6.
 Figure 6-6. Comparison of T1
images produced by spin echo
and inversion recovery
methods.

Notice the significant difference in contrast. The use of the inversion method
for other applications will be discussed in Chapter 8.

Spin Echo Imaging Methods

Spin echo is a technique used to produce an echo event by applying a 180˚

RF pulse to the dephased transverse magnetization. This compensates for
the dephasing produced by field inhomogeneities and makes it possible to
produce images that show the T2 characteristics of tissue. The time to the
echo event, TE, is a protocol factor that can be adjusted to produce different
weightings to the T2 contrast. When a short TE value is selected, the T2
effect is reduced, and the resulting image will be either a PD or T1-weighted
image, depending on the selected TR value.

It is possible to use a series of 180˚ RF pulses within one cycle to

produce multiple echo events, each with a different TE value. Both PD and
T2-weighted images can be acquired in the same acquisition.
 Chapter 7: Gradient Echo Imaging Methods

Introduction And Overview

It is possible to produce an echo event by applying a magnetic field gradient

without a 180˚ RF pulse to the tissue as in the spin echo methods. There are
several imaging methods that use the gradient echo technique to produce the
RF signals and these make up the gradient echo family of methods.

The primary advantage of the gradient echo methods over the spin
echo methods is that gradient echo methods perform faster image
acquisitions. Gradient echo methods are generally considered to be among
the faster imaging methods. They are also used in some of the angiographic
applications because gradient echo generally produces bright blood, as we
will see in Chapter 12, as well as for functional imaging, as described in
Chapter 13. One limitation of the gradient echo methods is they do not
produce good T2-weighted images, as will be described later in this chapter.
However, by combining the gradient and spin echo methods, this limitation
can be overcome.

At this time we will develop the concept of gradient echo and then
consider the specific gradient echo imaging methods and their characteristics.

The Gradient Echo Process

Transverse magnetization is present only when a sufficient quantity of protons

are spinning in-phase in the transverse plane. As we have seen, the decay
(relaxation) of transverse magnetization is the result of proton dephasing. We
also recall that an RF signal is being produced any time there is transverse
magnetization and the intensity of the signal is proportional to the level of
magnetization.

With the spin echo technique we use an RF pulse to rephase the

protons after they have been dephased by inherent magnetic field
inhomogeneities and susceptibility effects within the tissue voxel. With the
gradient echo technique the protons are first dephased, on purpose, by
turning on a gradient and then rephased by reversing the direction of the
gradient, as shown in Figure 7-1.
 Figure Figure 7-1.
The gradient echo
process using a
magnetic field
gradient to produce
an echo event during
the FID.

A gradient echo can only be created when transverse magnetization is

present. This can be either during the free induction decay (FID) period or
during a spin echo event. In Figure 7-1 the gradient echo is being created
during the FID. Let us now consider the process in more detail.

First, transverse magnetization is produced by the excitation pulse. It

immediately begins to decay (the FID process) because of the magnetic field
inhomogeneities within each individual voxel. The rate of decay is related to
the value of T2*. A short time after the excitation pulse a gradient is applied,
which produces a very rapid dephasing of the protons and reduction in the
transverse magnetization. This occurs because a gradient is a forced
inhomogeneity in the magnetic field. The next step is to reverse the direction
of the applied gradient. Even though this is still an inhomogeneity in the
magnetic field, it is in the opposite direction. This then causes the protons to
rephase and produce an echo event. As the protons rephase, the transverse
magnetization will reappear and rise to a value determined by the FID
process. The gradient echo event is a rather well-defined peak in the
transverse magnetization and this, in turn, produces a discrete RF signal.

The TE is determined by adjusting the time interval between the

excitation pulse and the gradients that produce the echo event. TE values for
gradient echo are typically much shorter than for spin echo, especially when
the gradient echo is produced during the FID.

Small Angle Gradient Echo Methods

The gradient echo technique is generally used in combination with an RF

excitation pulse that has a small flip angle of less than 90˚. We will discover
that the advantage of this is that it permits the use of shorter TR values and
this, in turn, produces faster image acquisition.

One source of confusion is that each manufacturer of MRI equipment

has given his gradient echo imaging methods different trade names. In this
text we will use the generic name of small angle gradient echo (SAGE)
method.

The SAGE method generally requires a shorter acquisition time than

the spin echo methods. It is also a more complex method with respect to
adjusting contrast sensitivity because the flip angle of the excitation pulse
becomes one of the adjustable protocol imaging factors.

Excitation/Saturation-Pulse Flip Angle

We recall that the purpose of the excitation/ saturation pulse applied at the
beginning of an imaging cycle is to convert or flip longitudinal magnetization
into transverse magnetization. When a 90° pulse is used, all of the existing
longitudinal magnetization is converted into transverse magnetization, as we
have seen with the spin echo methods. The 90° pulse reduces the
longitudinal magnetization to zero (i.e., complete saturation) at the beginning
of each imaging cycle. This then means that a relatively long TR interval must
be used to allow the longitudinal magnetization to recover to a useful value.
The time required for the longitudinal magnetization to relax or to recover is
one of the major factors in determining acquisition time. The effect of reducing
TR when 90° pulses are used is shown in Figure 7-2.

Figure 7-2. The effect

of reducing TR on the
recovery of
longitudinal
magnetization within
a cycle and the
resulting signal
intensity when using
90° pulses.

As the TR value is decreased, the longitudinal magnetization grows to a lower

value and the amount of transverse magnetization and RF signal intensity
produced by each pulse is decreased. The reduced signal intensity results in
an increase in image noise as described in Chapter 10. Also, the use of short
TR intervals with a 90° pulse (as in spin echo) cannot produce good PD or
T2-weighted images.

One approach to reducing TR and increasing acquisition speed

without incurring the disadvantages that have just been described is to use a
pulse that has a flip angle of less than 90°. A small flip-angle (<90°) pulse
converts only a fraction of the longitudinal magnetization into transverse
magnetization. This means that the longitudinal magnetization is not
completely destroyed or reduced to zero (saturated) by the pulse, as shown in
Figure 7-3.
Figure 7-3. The effect of using small flip angle pulses on longitudinal
magnetization.

Reducing the flip angle has two effects that must be considered together. The
effect that we have just observed is that the longitudinal magnetization is not
completely destroyed and remains at a relatively high level from cycle to
cycle, even for short TR intervals. This will increase RF signal intensity
compared to the use of 90° pulses. However, as the flip angle is reduced, a
smaller fraction of the longitudinal magnetization is converted into transverse
magnetization. This has the effect of reducing signal intensity. The result is a
combination of these two effects. This is illustrated in Figure 7-4.

Figure 7-4. The effect

of pulse flip angle on
the level of both
longitudinal and
transverse
magnetization after
the pulse is applied.

Here we see that as the flip angle is increased over the range from 0–90°, the
level of longitudinal magnetization at the beginning of a cycle decreases. On
the other hand, as the angle is increased, the fraction of this longitudinal
magnetization that is converted into transverse magnetization increases and
RF signal intensity increases. The combination of these two effects is shown
in Figure 7-5.

Figure 7-5. The relationship

of signal intensity to flip
angle.

Here we see how changing flip angle affects signal intensity. The exact shape
of this curve depends on the specific T1 value of the tissue and the TR
interval. For each T1/TR combination there is a different curve and a specific
flip angle that produces maximum signal intensity.

Let us now use Figure 7-6 to compare the magnetization of two

tissues with different T1 values as we change flip angle.

Figure 7-6. The effect of flip angle on contrast.

Contrast between the two tissues is represented by the difference in

magnetization levels. At this point we are assuming a short TE and
considering the contrast associated with only the longitudinal magnetization.
The flip-angle range is divided into several specific segments as shown.

Contrast Sensitivity

With the SAGE method the contrast sensitivity for a specific tissue
characteristic is controlled by three protocol factors. As with spin echo, TR
and TE have an effect. However, the flip angle becomes the factor with the
greatest effect on contrast. We will now see how changing flip angle can be
used to select specific types of contrast with a basic gradient echo method.

T1 Contrast

Relatively large flip angles (45°–90°) produce T1 contrast. This is what we

would expect because large flip angles (close to 90°) and short TR and TE
values are similar to the factors used to produce T1 contrast with the spin
echo method. Here, with gradient echo, we observe a loss of T1 contrast as
the flip angle is decreased significantly from 90°.

Low Contrast

There is an intermediate range of flip-angle values that produces very little, if

any, contrast. This is the region in which the PD and T1 contrast cancel each
other for many tissues, such as gray and white matter.

Proton Density (PD) Contrast

Relatively low flip-angle values produce PD contrast. As the flip angle is
reduced within this region, there is a significant decrease in magnetization
and the resulting signal intensity.

Up to this point we have observed generally how changing the flip angle of the
excitation pulse affects signal intensity and contrast. In the SAGE imaging
method the flip angle is one of the imaging factors that must be adjusted by
the user. However, it becomes somewhat complex because the specific effect
of flip angle is modified by the other imaging factors and techniques used to
enhance a specific type of contrast.

T2 and T2* Contrast

We recall that T2 contrast is produced by the decay characteristics of

transverse magnetization and that there are two different decay rates, T2 and
T2*. The slower decay rate is determined by the T2 characteristics of the
tissue. The faster decay is produced by small inhomogeneities within the
magnetic field often related to variations in tissue susceptibility differences.
This decay rate is determined by the T2* of the tissue environment. When a
spin echo technique is used, the spinning protons are rephased, and the T2*
effect is essentially eliminated. However, when a spin echo technique is not
used, the transverse magnetization depends on the T2* characteristics. The
gradient echo technique does not compensate for the inhomogeneity and
susceptibility effect dephasing as the spin echo technique does. Also, without
using a spin echo process the long TE values necessary to produce T2
contrast cannot be achieved. Therefore, a basic gradient echo imaging
method is not capable of producing true T2 contrast. The contrast will be
determined primarily by the T2* characteristics. The amount of T2* contrast in
an image is determined by the selected TE value. In general, longer TE
values (but short compared to those used in spin echo) produce more T2*
contrast.

Contrast Enhancement

In addition to using combinations of TR, TE, and flip angle to control the
contrast characteristics, some gradient echo methods have other features for
enhancing certain types of contrast.

When SAGE methods are used with relatively short TR values, there
is the possibility that some of the transverse magnetization created in one
imaging cycle will carry over into the next cycle. This happens when the TR
values are in the same general range as the T2 values of the tissue. SAGE
methods differ in how they use the carry-over transverse magnetization.

A typical SAGE sequence is limited to one RF pulse per cycle. If

additional pulses were used, as in the spin echo techniques, they would affect
the longitudinal magnetization and upset its condition of equilibrium. However,
because of the relatively short TR values it is possible for the repeating small-
angle excitation pulses to produce a spin echo effect. This can occur only
when the TR interval is not much longer than the T2 value of the tissue.

Associated with each excitation pulse, there are actually two

components of the transverse magnetization. There is the FID produced by
the immediate pulse and a spin echo component produced by the preceding
pulses. The spin echo component is related to the T1 characteristics of the
tissue. The FID component is related to the T2* characteristics. The contrast
characteristics of the imaging method are determined by how these two
components are combined. Different combinations are obtained by altering
the location of the gradient echo event relative to the transverse
magnetization and by turning the spin echo component on or off as described
below.

Mixed Contrast

When both the FID and spin echo components are used, an image with mixed
contrast characteristics will be obtained. This method produces a relatively
high signal intensity compared to the methods described below.

Spoiling and T1 Contrast Enhancement

An image with increased Tl contrast is obtained by suppressing the spin echo

component. This is known as spoiling. The spin echo component, which is a
carryover of transverse magnetization from previous cycles, can be destroyed
or spoiled by either altering the phase relationship of the RF pulses or by
applying gradient pulses to dephase the spinning protons.

The basic SAGE method discussed up to this point permits faster

(than spin echo) image acquisition because the TR can be set to shorter
values. However, the gradient echo process can be used in methods that
provide fast acquisition based on an entirely different principle. We will now
consider methods that achieve their speed by filling many rows of k space
during one acquisition cycle.

In Chapter 5 we saw that in the acquisition phase the signal data is

being directed into k space from which the image will be reconstructed. The k
space is filled one row at a time. The number of rows in the k space for a
specific image depends on the required image detail. The process that directs
the signals into a specific row of k space is the spatial encoding function
performed by one of the gradients. This will be described in Chapter 9. In
conventional spin echo and SAGE imaging only one row of k space is filled
with each imaging cycle. This is because there is only one echo signal
produced per cycle that can be encoded to go to a specific row of k space.
This means that the size of k space determines the minimum number of
cycles that an acquisition must have. We are about to see some gradient
echo methods that can fill many rows of k space in one imaging cycle. This is
achieved by using the gradient echo process to produce many echo events
from the transverse magnetization that is present during one cycle.

Echo Planar Imaging (EPI) Method

Echo planar is the fast gradient echo imaging method that is capable of
acquiring a complete image in a very short time. However, it requires an MRI
system equipped with strong gradients that can be turned on and off very
rapidly. All systems do not have this capability. The EPI method consists of
rapid, multiple gradient echo acquisitions executed during a single spin echo
event. The unique characteristic of this method is that each gradient echo
signal receives a different spatial encoding and is directed into a different row
of k space. The actual spatial encoding process will be described in Chapter
9. Here we are considering only the general concept of EPI and how it
achieves rapid acquisition.

The basic EPI method is illustrated in Figure 7-7.

Figure 7-8. The

use of the
GRASE method
to fill many rows
of k space and
produce a fast
acquisition

The basic cycle is a multiple spin echo as described in Chapter 5. The

difference is that in conventional multiple echo, each of the echo events have
different TE values and are used to form several images; typically, a PD and a
T2 image with the same acquisition. Here the multiple spin echo is used for a
different purpose. The multiple spin echoes are used to cover more of k
space. As we see, each of the spin echo events is cut into many gradient
echoes by the EPI process. This reduces the acquisition time by two factors:
the total speed factor is the number of multiple spin echoes multiplied by the
EPI speed factor.

Magnetization Preparation

Both SAGE with short TR values and EPI can produce very rapid acquisitions.
However, the short time intervals between the gradient echo events do not
provide sufficient time for good longitudinal magnetization contrast (T1 or PD)
to be formed. This problem is solved by “preparing” the magnetization and
forming the contrast just one time at the beginning of the acquisition cycle, as
shown in Figure 7-9. Two options are shown.

Figure 7-9. Using

preparation pulses to
produce longitudinal
magnetization contrast
prior to a rapid gradient
echo acquisition.

The longitudinal magnetization is prepared by applying either a saturation

pulse, as in the the inversion-recovery method or an inversion pulse, as in the
inversion-recovery method. As the longitudinal magnetization relaxes,
contrast is formed between tissues with different T1 and PD values. After a
time interval [TI or TS (Time after Saturation)] selected by the operator, a
rapid gradient echo acquisition begins.

The total acquisition time for this method l-is the time required by the
acquisition cycles plus the TI or TS time interval.

Gradient Echo Imaging Methods

 The common characteristic of the gradient echo imaging methods is
that a magnetic field gradient is used to produce the echo event rather than a
180° RF pulse, as is used in the spin echo methods. One of the principal
advantages of the gradient echo process is that it is a relatively fast imaging
method.

By using a gradient, and not an RF pulse, to produce the echo event,

it is possible to use saturation/excitation pulses with flip angles less than 90˚;
thereby all the longitudinal magnetization is not destroyed (saturated) at the
beginning of each cycle. Because some longitudinal magnetization carries
over from cycle to cycle, it is possible to reduce the TR value and still produce
useful signal levels. The reduced TR values result in faster imaging. The flip
angle of the RF pulse is an adjustable protocol factor that controls the type of
contrast produced.

Echo planar imaging is a gradient echo method in which many echo

events, each with a different phase encoding step, are created during each
imaging cycle. This makes it possible to fill multiple rows of k space, which
results in very fast imaging. GRASE is an imaging method that combines the
principles of echo planar and fast (turbo) spin echo to produce rapid imaging
acquisitions.

When very fast gradient echo methods are used, there is not sufficient
time between the echo events for significant tissue relaxation and contrast to
develop. Therefore, the desired contrast is developed at the beginning of the
acquisition by applying either inversion or saturation “magnetization
preparation” pulses. Then, when the desired contrast has developed, a rapid
acquisition is performed.
 Chapter 8:
Selective Signal Suppression

Introduction And Overview

There are many times when it is desirable to selectively suppress the signals
from specific tissues or anatomical regions. This is done for a variety of
reasons including the enhancement of contrast between certain tissues and
the reduction of artifacts. During the acquisition process signals can be
suppressed based on several properties of a tissue or fluid that make it
different from other surrounding tissues. These include differences in T1
values, resonant frequencies, and molecular binding properties. Also, signals
from specific anatomical regions can be suppressed or “turned off,” usually to
prevent interference with imaging in other areas. We will now see how these
techniques are used.

Fat and fluid are two materials in the body that can produce very
intense signals and brightness in images. This occurs with fat in T1 images
and with fluid in T2 images. A possible problem is that these bright regions
can reduce the visibility of other tissues and pathologic conditions in the area.

T1-Based Fat And Fluid Suppression

Let us recall that fat has very short T1 values (260 msec) and fluids have
very long T1 values (2000 msec). These values are outside of the range of
the T1 values of other tissues in the body and are separate and not mixed in
with the others. This makes it possible to use T1 as a characteristic for the
selective suppression of both fat and fluid.

STIR Fat Suppression

STIR is an inversion recovery method with the TI adjusted to selectively
suppress the signals from fat. This uses the fact that fat has a relatively short
T1 value and recovers its longitudinal magnetization faster than the other
tissues after the inversion pulse. The important point here is that the
magnetization of fat passes through the zero level before the other tissues, as
shown in Figure 8-1.
 Figure 8-1. The
use of STIR to
suppress signals
from fat by
setting TI to a
value (short) that
will image the
longitudinal
magnetization at
the time when fat
is relaxing
through the zero
level.

The TI interval is selected so that the “picture is snapped” by applying the

excitation pulse at that time. Because the fat has no magnetization at that
time, it will not produce a signal. Since this is achieved with relatively short
values for TI, this method of fat suppression is often referred to as Short Time
Inversion Recovery (STIR).

STIR is just the inversion recovery (IR) method with the TI set to a
relatively low value. The description of the basic IR method in Chapter 6
shows how the factor TI is used to select the time at which the longitudinal
magnetization “picture is snapped” and the magnetization is converted into
image contrast. The ability to use this method to suppress the signals from fat
is based on the fact that the longitudinal magnetization of fat passes through
zero at a time before and separated from the other tissues. Setting the TI to
measure the longitudinal magnetization at the time when fat is at zero
produces no signal and fat will be dark in the image.

The best TI value to suppress the signals from fat depends on the T1
value of fat, which depends on the strength of the magnetic field. It will
generally be in the range of 120 to 150 msec for field strengths in the 0.5 T to
1.5 T range.

Another consideration with STIR is that the TR must be set relatively

long (1500–2000 msec), compared to a T1 image acquisition with spin echo
using a TR value of approximately 500 msec. This additional time is required
for the longitudinal magnetization to more fully recover after the excitation
pulse and before the next cycle can begin.

Fluid Suppression

The suppression of signals from fluids can be achieved by using the IR

Method with the TI set to relatively long values as shown in Figure 8-2.
 Figure 8-2. The
suppression of
fluid by selecting
a long TI that
will image the
longitudinal
magnetization at
the time when
fluid is relaxing
through zero.

This works because the long T1 values of fluids are well separated from the
T1 values of other tissues. By setting the TI to a long value as shown, the
longitudinal magnetization is converted to transverse and the “picture is
snapped” when the fluid is at a zero value. Fluids appear as dark regions in
the image. When fluid suppression is used with a T2 image acquisition (long
TE), the usually bright fluid is suppressed but other tissues with long T2
values, such as pathologic tissue, remain bright.

Acquisition time is a special concern with this method. That is because

when long TI values are used, the TR values must also be long (5000–6000
msec) and that increases the acquisition time. For this reason, the practical
thing is to use this method with one of the fast acquisition techniques.

SPIR Fat Suppression

Spectral Presaturation with Inversion Recovery (SPIR) is a fat suppression

technique which makes use of the fact that fat and the water content of
tissues resonate at different frequencies (on the RF frequency spectrum) as
described in Chapter 3. We must be careful not to confuse the two fat
suppression methods, STIR and SPIR. As we have just seen, STIR uses the
difference in T1 values to selectively suppress the signals from fat. Now with
SPIR, we will use the differences in resonant frequency to suppress the fat
signals. This technique is illustrated in Figure 8-3.
 Figure 8-3.
Suppressing the
signals from fat by
applying an inversion
pulse tuned to the
resonant frequency of
fat so that it does not
affect the other
tissues.

The unique feature of this method is that the imaging cycle begins with an
inversion pulse that is applied at the fat resonant frequency. This selectively
inverts the longitudinal magnetization of the fat without affecting the other
tissues. The TI is set so that the spin echo excitation pulse is applied at the
time when the fat longitudinal magnetization is passing through zero. This
results in T1 and T2 images with the signals from fat removed.

The advantage of the SPIR method is that the contrast of tissues with
relatively short T1 values is not diminished as it might be with the STIR
method. For example, the use of gadolinium contrast media reduces the T1
value of the water component of tissue. These short T1 value signals would
be suppressed by STIR, but not by SPIR.

There are some precautions that must be observed when using SPIR.
They relate to having very good magnetic field homogeneity. Recall that the
resonant frequency is controlled by the field strength in each location.
Therefore, for the RF suppression pulse to accurately suppress the fat
magnetization over the image area, the fat must be resonating at precisely the
same frequency. This requires a very homogeneous (within just a few parts
per million) magnetic field. This is achieved by shimming the field before the
acquisition, removing metal objects that might distort the field, and by using a
relative small field of view.

An alternative to the SPIR method is to apply a saturation rather than

an inversion pulse tuned to the fat resonant frequency. This is sometimes
referred to as chemical saturation.

Magnetization Transfer Contrast (MTC)

Magnetization Transfer Contrast (MTC) is a technique that enhances image

contrast by selectively suppressing the signals from specific tissues. The
amount of suppression depends on a specific tissue’s magnetization transfer
characteristics. Maximum suppression is obtained for tissues that have a high
level of magnetization transfer.

The MTC technique is illustrated in Figure 8-4.

Figure 8-4. The use of magnetization transfer between different types of tissue to suppress
selective signals.

It is based on the principle that the protons in tissue are in different states of
mobility, which we will designate as the “free” pool and the “bound” pool.

Free Proton Pool

The protons that produce signals and are visible in MRI are not rigidly bound
and might be considered to be “free” and in a general “semi-solid” structure.
This environment produces relatively long T2 values (in comparison to the
bound state) and a relatively narrow resonant frequency.

Bound Proton Pool

Most tissues also contain protons that are more rigidly bound and associated
with more “solid” structures such as large macromolecules and membranes.
These structures have very short T2 values. This means that the transverse
magnetization decays before it can be imaged with the usual methods.
Therefore, these protons do not contribute to the image. An important
characteristic of these protons is that they have a much broader resonant
frequency spectrum than the “free” protons.

Magnetization Transfer

Magnetization transfer is a process in which the longitudinal magnetization of

one pool influences the longitudinal magnetization in the other pool. In other
words, the longitudinal magnetizations of the two pools are coupled together
but not to the same degree in all tissues. The MTC process makes use of this
difference in coupling to selectively suppress the signals from certain tissues.
This is how it is done.
Selective Saturation

The objective of this technique is to saturate and suppress selective

signals from specific tissues to increase the contrast.

Prior to the beginning of the imaging acquisition cycle a saturation

pulse is applied at a frequency that is different from the resonant frequency of
the “free” protons. Therefore, it does not have a direct effect on the protons
that are producing the signals. However, the saturation pulse is within the
broader resonant frequency of the “bound” protons. It produces saturation of
the longitudinal magnetization in the “bound” pool.

The effect of the saturation is now transferred to the longitudinal

magnetization of the “free” pool by the magnetization transfer process. The
key is that the transfer is not the same for all tissues. Only the tissues with a
relatively high magnetization transfer coupling and a significant bound pool
concentration will experience the saturation and have their signals reduced in
intensity.

Fluids, fat, and bone marrow have very little, if any, magnetization
transfer. Therefore, they will not experience the transferred saturation, and will
remain relatively bright in the images.

Most other tissues have some, but varying degrees of, magnetization
transfer. When the MTC technique is used, the saturation produced by the RF
pulse applied to the “bound” protons will be transferred to the “free” protons,
but only in those tissues that have a significant magnetization transfer
capability. The result is that these tissues will be saturated to some degree
and their signal intensities will be reduced.

Therefore, MTC is a way of enhancing contrast in an image by

suppressing the signals from tissues that have a relatively high magnetization
transfer. One example is to use MTC to reduce the brightness (signal
intensity) of brain tissue so that the vascular structures will be brighter in
angiography.

Regional Saturation

There are procedures in which it is desirable to suppress signals from

specific anatomical regions. The two major applications of this are to reduce
motion-induced artifacts, as described in Chapter 14, and to suppress the
signals from blood that is flowing in a specific direction, as discussed in
Chapter 12. At this time we will consider the general technique, which is
illustrated in Figure 8-5.
Figure 8-5. The application of a saturation pulse can be directed to a specific anatomical region
to suppress undesirable signals from moving tissues.

Let us recall that gradients are used to vary the magnetic field strength across
a patient’s body. In the presence of a gradient one region of the body is in a
different field strength from another and is therefore tuned to a different
resonant frequency. This makes it possible to apply RF pulses selectively to
specific regions without affecting adjacent regions.

In Chapter 14 we will see that a major source of artifacts in MRI is the

motion or movement of tissues and fluids. The motion produces errors in the
spatial encoding of the signals that causes them to be displayed in the wrong
location in the image. Signals from moving tissues and fluids are displayed as
streaks, which are undesirable artifacts.

With the regional saturation technique the objective is to suppress

selective signals originating from one region, usually the moving tissue or
fluid, without affecting these signals in the region that is being imaged. The
specific applications of this will be described in Chapter 14.

Prior to the imaging cycle pulse sequence, a saturation pulse is

selectively applied to the region that is to be suppressed. The saturation pulse
is given a frequency that is different from the frequency of the other imaging
pulses. This is so that it will be tuned to the resonant frequency of the region
that is to be suppressed. This region will have a resonant frequency different
from the imaged area because of the presence of the gradient as described
above.

The region that is saturated is a three-dimensional (3-D) volume or

slab of tissue. It is important that the slab be properly positioned in
relationship to the imaged area for best results.

The application of regional saturation to suppress artifacts will be discussed in

more detail when we consider artifacts in Chapter 14.
 Selective Signal Suppression

It is often desirable to suppress the signals and resulting brightness of

selected tissues or anatomical regions to improve visibility of other tissues or
general image quality. It is possible to selectively suppress signals from
specific tissues if the tissues are significantly different from the other tissues in
terms of some MR characteristic.

Signals from fat, generally very bright in T1 images, can be

suppressed with two techniques. Because fat has a very short T1 value
compared to other tissues, it can be suppressed with the STIR method, an
inversion recovery method in which the TI is set to snap the picture when the
magnetization of fat is passing through the zero level. The resonant frequency
of fat molecules is slightly different from water molecules because of the
chemical shift effect. The SPIR method makes use of this by applying an RF
pulse at the fat frequency to reduce the fat magnetization to the zero level at
the beginning of each imaging cycle.

Signals from fluid can be suppressed by using an inversion recovery

method with the TI set to a long value. This works because fluids have long
T1 values and the fluid’s magnetization passes through the zero level
significantly later and separate from that of tissues. The MTC technique can
be used to reduce signal intensity from tissues that have a relatively high
magnetization transfer characteristic. This can be used to enhance image
contrast.

Saturation pulses can be selectively applied to specific anatomical

regions to suppress any signals that could occur from tissues or fluids in that
region. This is useful for reducing motion artifacts and also for reducing the
signals from flowing blood in specific anatomical regions.

Chapter 9
Spatial Characteristics of the Magnetic Resonance Image
Introduction And Overview

The MR image formation process subdivides a section of the patient’s body

into a set of slices and then each slice is cut into rows and columns to form a
matrix of individual tissue voxels. This was introduced first in Chapter 1 and
illustrated in Figure 1-3. The RF signal from each individual voxel must be
separated from all of the other voxels and its intensity displayed in the
corresponding image pixel, as shown in Figure 9-1.

Figure 9-1. The relationship of tissue voxels to image pixels.

This is achieved by encoding or addressing the signals during the acquisition

phase and then, in effect, delivering the signal intensities to the appropriate
pixels which have addresses within the image during the reconstruction
phase. Because there are two dimensions, or directions, in an image, two
different methods of encoding must be used. This is analogous to mail that
must have both a street name and a house number in the address. We are
about to see that the two methods of addressing the signals are called
frequency-encoding and phase-encoding. One method is applied to one
direction in the image and the other method is used to address in the other
direction.

This two-step process consisting of the signal acquisition phase

followed by the image reconstruction phase is illustrated in Figure 9-2.

Figure 9-2. The two phases—signal acquisition and image reconstruction

—that are required to produce an MR image.
Different actions happen in these two phases that must be considered when
setting up an imaging procedure.

Signal Acquisition

During the acquisition phase the RF signals are emitted by the tissue and
received by the RF coils of the equipment. During this process the signals
from the different slices and voxels are given distinctive frequency and phase
characteristics so that they can be separated from the other signals during
image reconstruction. The acquisition phase consists of an imaging cycle that
is repeated many times. The time required for image acquisition is determined
by the time TR, which is the duration of one cycle or its repetition time, and
the number of cycle repetitions. The number of cycles is determined by the
image quality requirements. In general, the quality of an image can be
improved by increasing the number of acquisition cycles. This is considered in
much more detail in Chapter 10.

The result of the image acquisition process is a large amount of data

collected and stored in computer memory. At this point the data represent RF
signal intensities characterized by the two characteristics, frequency and
phase. The concept of frequency and phase will be developed later. At this
point in the process the data are not yet in the form of an image but are
located in k space. The data will later be transformed into image space by the
reconstruction process.

Image Reconstruction

Image reconstruction is a mathematical process performed by the computer. It

transforms the data collected during the acquisition phase into an image. We
can think of reconstruction as the process of sorting the signals collected
during the acquisition and then delivering them to the appropriate image
pixels. The mathematical process used is known as Fourier transformation.
Image reconstruction is typically much faster than image acquisition and
requires very little, if any, control by the user.

Image Characteristics

The most significant spatial characteristic of an image is the size of the

individual tissue voxels. Voxel size has a major effect on both the detail and
noise characteristics of the image. The user can select the desired voxel size
by adjusting a combination of imaging factors, as described in Chapter 10.

Gradients

The spatial characteristics of an MR image are produced by actions of the

gradients applied during the acquisition phase. Magnetic field gradients are
used first to select slices and then give the RF signals the frequency and
phase characteristics that create the individual voxels.

As we will see later, a gradient in one direction is used to create the

slices, and then gradients in the other directions are used to cut the slices into
rows and columns to create the individual voxels. However, these functions
can be interchanged or shared among the different gradient coils to permit
imaging in any plane through the patient’s body.

The functions performed by the various gradients usually occur in a

specific sequence. During each individual image acquisition cycle the various
gradients will be turned on and off at specific times. As we will see later, the
gradients are synchronized with other events such as the application of the
RF pulses and the acquisition of the RF signals.

Slice Selection

There are two distinct methods used to create the individual slices. The
method of selective excitation actually creates the slice during the acquisition
phase. An alternative method is to acquire signals from a large volume of
tissue (like an organ) and then create the slices during the reconstruction
process. These are often referred to as 2-D (volume) and 3-D (volume)
acquisitions. However, each produces data that are reconstructed into slice
images. Both methods have advantages and disadvantages, which will be
described later.

Selective Excitation

The first gradient action in a cycle defines the location and thickness of the
tissue slice to be imaged. We will illustrate the procedure for a conventional
transaxial slice orientation. Other orientations, such as sagittal, coronal, and
angled combinations, are created by interchanging and combining gradient
directions.

Slice selection using the principle of selective excitation is illustrated in

Figure 9-3.

Figure 9-3. The use of a gradient to tune a specific slice so that it can be
selectively excited by an RF pulse.

When a magnetic field gradient is oriented along the patient axis, each slice of
tissue is in a different field strength and is tuned to a different resonant
frequency. Remember, this is because the resonant frequency of protons is
directly proportional to the strength of the magnetic field at the point where
they are located. This slice selection gradient is present whenever RF pulses
are applied to the body. Since RF pulses contain frequencies within a limited
range (or bandwidth), they can excite tissue only in a specific slice. The
location of the slice can be changed or moved along the gradient by using a
slightly different RF pulse frequency. The thickness of the slice is determined
by a combination of two factors: (1) the strength, or steepness, of the
gradient, and (2) the range of frequencies, or bandwidth, in the RF pulse.

Multi-Slice Imaging

In most clinical applications, it is desirable to have a series of images (slices)

covering a specific anatomical region. By using the multi-slice mode, an entire
set of images can be acquired simultaneously. The basic principle is
illustrated in Figure 9-4.

Figure 9-4. Multiple slice imaging applies pulses to and produces signals from
different slices within one imaging cycle
The slices are separated by applying the RF pulses and detecting the signals
from the different slices at different times, in sequence, during each imaging
cycle.

When the slice selection gradient is turned on, each slice is tuned to a
different resonant frequency. A specific slice can be selected for excitation by
adjusting the RF pulse frequency to correspond to the resonant frequency of
that slice. The process begins by applying an excitation pulse to one slice and
collecting the echo signal. Then, while that slice undergoes longitudinal
relaxation before the next cycle can begin, the excitation pulse frequency is
shifted to excite another slice. This process is repeated to excite and collect
signals from the entire set of slices at slightly different times within one TR
interval.

The advantage of multi-slice imaging is that a set of slices can be

imaged in the same time as a single slice. The principal factor that limits the
number of slices is the value of TR. It takes a certain amount of time to excite
and then collect the signals from each slice. The maximum number of slices is
the TR value divided by the time required for each slice. This limitation is
especially significant for T1-weighted images that use relatively short TR
values.

A factor to consider when selecting the slicing mode is that multiple

slice selective excitation cannot produce the contiguous slices that the volume
acquisition technique can. With selective excitation there is the possibility that
when an RF excitation/saturation pulse is applied to one slice of tissue, it will
also produce some effect in an adjacent slice. This is a reason for leaving
gaps between slices during the acquisition.

Volume Acquisition

Volume (3-D) image acquisition has the advantage of being able to produce
thinner and more contiguous slices. This is because of the process used to
slice the tissue. Rather than producing each slice during the acquisition
phase, the slicing is done during the reconstruction phase using the process
of phase-encoding. The actual process of phase-encoding will be described
later in this chapter. At this time we only consider how it is used for slicing.
With this method, no gradient is present when the RF pulse is applied to the
tissue. Since all tissue within an anatomical region, such as the head, is tuned
to the same resonant frequency, all tissues are excited simultaneously. The
next step, as illustrated in Figure 9-5,
Figure 9-5. The 3-D volume acquisition process uses the phase-encoding
process to produce thin slices.
is to apply a phase-encoding gradient in the slice selection direction. In
volume imaging, phase-encoding is used to create the slices in addition to
creating the voxel rows as described below. The phase-encoding gradient
used to define the slices must be stepped through different values,
corresponding to the number of slices to be created. At each gradient setting,
a complete set of imaging cycles must be executed. Therefore, the total
number of cycles required in one acquisition is multiplied by the number of
slices to be produced. This has the disadvantage of causing 3-D volume
acquisitions to have a relatively long acquisition time compared to 2-D
multiple slice acquisitions. That is why this type of acquisition is often used
with one of the faster imaging methods.

The primary advantage of volume imaging is that the phase-encoding

process can generally produce thinner and more contiguous slices than the
selective excitation process used in 2-D slice acquisition. The primary
disadvantage is longer acquisition times.

Frequency Encoding

A fundamental characteristic of an RF signal is its frequency. Frequency is the

number of cycles per second of the oscillating signal. The frequency unit of
Hertz (Hz) corresponds to one cycle per second. Radio broadcast stations
transmit signals on their assigned frequency. By tuning our radio receiver to a
specific frequency we can select and separate from all other signals the
specific broadcast we want to receive. In other words, the radio broadcasts
from all of the stations in a city are frequency encoded. The same process
(frequency-encoding) is used to cause voxels to produce signals that are
different and can be used to create one dimension of the image.

Let us review the concept of RF signal production by voxels of tissue,

as shown in Figure 9-6.

Figure 9-6. The effect of field strength on the frequency of RF signals

produced by transverse magnetization.

RF signals are produced only when transverse magnetization is present. The

unique characteristic of transverse magnetization that produces the signal is a
spinning magnetic effect, as shown. The transverse magnetization spins
around the axis of the magnetic field. A spinning magnet or magnetization in
the vicinity of a coil forms a very simple electric generator. It generates one
cycle for each revolution of the magnetization. When the magnetization is
spinning at the rate of millions of revolutions per second, the result is an RF
signal with a frequency in the range of Megahertz (MHz).
Resonant Frequency

The frequency of the RF signal is determined by the spinning rate of the

transverse magnetization. This, in turn, is determined by two factors, as was
described in Chapter 3. One factor is the specific magnetic nuclei (usually
protons) and the other is the strength of the magnetic field in which the voxel
is located. When imaging protons, the strength of the magnetic field is the
factor used to vary the resonant frequency and the corresponding frequency
of the RF signals. In Figure 9-6 we see two voxels located in different strength
fields. The result is that they produce different frequency signals.

Figure 9-7 shows the process of frequency encoding the signals for a
row of voxels.

Figure 9-7. The frequency encoding of a row of voxels within a slice.

In this example, a gradient is applied along the row. The magnetic field
strength is increased from left to right. This means that each voxel is located
in a different field strength and is resonating at a frequency different from all of
the others. The resonant and RF signal frequencies increase from the left to
right as shown.

The frequency-encoding gradient is on at the time of the echo event

when the signals are actually being produced. The signals from all of the
voxels in a slice are produced simultaneously and are emitted from the body
mixed together to form a composite signal at the time of the echo event. The
individual signals will be separated later by the reconstruction process to form
the voxels.

Phase-Encoding

Phase is a relationship between one signal and another, as illustrated in

Figure 9-8.

Figure 9-8. The concept of phase between the signals from two voxels.
Here we see two voxels producing RF signals. The transverse magnetization
is spinning at the same rate and producing signals that have the same
frequency. However, we notice that one signal is more advanced in time or is
out of step with the other. In other words, the two signals are out of phase.
The significance of voxel-to-voxel phase in MRI is that it can be used to
separate signals and create one dimension in the image.

A phase difference is created by temporarily changing the spinning

rate of the magnetization of one voxel with respect to another. This happens
when the two voxels are located in magnetic fields of different strengths. This
can be achieved by turning on a gradient, as shown in Figure 9-9.
Figure 9-9. Phase-encoding produced by turning on a gradient for a short
time, and then turning it off. The phase difference remains.

Let us begin the process of phase encoding by considering the column

of voxels shown in the illustration. We are assuming that all voxels have the
same amount of transverse magnetization and that the magnetization is
spinning in-phase at the time just prior to the phase-encoding process.

When the phase-encoding gradient is turned on, we have the

condition illustrated with the center column of voxels. The strength of the
magnetic field is increasing from bottom to top. Therefore, the magnetization
in each voxel is spinning at a different rate with the speed increasing from
bottom to top. This causes the magnetization from voxel to voxel to get out of
step or produce a phase difference. The phase-encoding gradient remains on
for a short period of time and then is turned off. This leaves the condition
represented by the column of voxels on the right. This is the condition that
exists at the time of the echo event when the signals are actually produced.
As we see, the signals from the individual voxels are different in terms of their
phase relationship. In other words, the signals are phase-encoded. All of the
signals are emitted at the same time and mixed together as a composite echo
signal. Later, the reconstruction process will sort the individual signal
components.

Phase-encoding is the second function performed by a gradient during

each cycle, as shown in Figure 9-10.

Figure 9-10. The relationship of the three gradient actions—slice selection,

phase-encoding, and frequency-encoding—to each other and to the RF
pulses and signals. They are applied in different directions.

During each pass through an imaging cycle, the phase-encoding gradient is

stepped to a slightly different value.

The signals acquired with each phase-encoding gradient strength fills

one row of k space. This is a very important point that should be emphasized:
Each row of k space is reserved for signals with a specific degree of phase-
encoding. The degree of phase-encoding is determined by the strength and
duration of the phasing gradient applied during each cycle. Therefore, the
phase-encoding process must be repeated depending on the size of k space
and that is determined by the image matrix size in the phase-encoded
direction.

One MRI phase-encoding step produces a composite signal from all

voxels within a slice. The difference from one step to another is that individual
voxel signals have a different phase relationship within the composite signal.
 To reconstruct an image by the conventional 2-D Fourier
transformation method, one composite signal, or phase-encoded step, must
be collected for each voxel to be created in the phase-encoding direction.
Therefore, the minimum number of steps required to produce an image is
determined by the size of the image matrix and k space. It takes 256 phase-
encoding steps to produce an image with a 256 ´ 256 matrix.

The Gradient Cycle

We have seen that various gradients are turned on and off at specific times
within each imaging cycle. The relationship of each gradient to the other
events during an imaging cycle is shown in Figure 9-10. The three gradient
activities are:

1. The slice selection gradient is on when RF pulses are applied to the tissue.
This limits magnetic excitation, inversion, and echo formation to the tissue
located within the specific slice.

2. The phase-encoding gradient is turned on for a short period in each cycle

to produce a phase difference in one dimension of the image. The strength of
this gradient is changed slightly from one cycle to another to fill the different
rows of k space needed to form the image.

3. The frequency-encoding gradient is turned on during the echo event when

the signals are actually emitted by the tissue. This causes the different voxels
to emit signals with different frequencies.

Because of the combined action of the three gradients, the individual

voxels within each slice emit signals that are different in two respects—they
have a phase difference in one direction and a frequency difference in the
other. Although these signals are emitted at the same time, and picked up by
the imaging system as one composite signal at the time of the echo event in
each cycle, the reconstruction process can sort the signals into the respective
components and display them in the correct image pixel locations.

Image Reconstruction

The next major step in the creation of an MR image is the reconstruction

process. Reconstruction is the mathematical process performed by the
computer that converts the collected signals in k space into an actual image.
There are several reconstruction methods, but the one used for most clinical
applications is the 2-D Fourier transformation.
 It is a mathematical procedure that can sort a composite signal into
individual frequency and phase components. Since each voxel in a row emits
a different signal frequency and each voxel in a column a different phase, the
Fourier transformation can determine the location of each signal component
and direct it to the corresponding pixel.

Let us now use the concept illustrated in Figure 9-11 to summarize the
spatial characteristics of the MR image.

Figure 9-11. The concept of signal encoding (addressing) and image

reconstruction (sorting and delivery).

We will use a postal analogy for this purpose.

In the image each column of pixels has a phase address

corresponding to different street names. Each row of pixels has a frequency
address corresponding to house numbers. Therefore, each individual pixel
has a unique address consisting of a combination of frequency and phase
values analogous to a street name and house number.

The frequency- and phase-encoding process during acquisition

“writes” an address on the signal from each voxel. These signals are mixed
together and collected in a “post-office” called k space. The signals (“mail”)
are then sorted by the Fourier transform process and hopefully delivered to
the correct pixel address in the image.

In Chapter 14 we will see that if a voxel of tissue moves during the

acquisition process, it might not receive the correct phase address and the
signal will be delivered to the wrong pixel. This creates ghost images and
streak artifacts in the phase-encoded direction.

The chemical-shift artifact is caused by the difference in signal frequency

between tissues containing water and fat. When it is present in an image,
signals from the water components and fat will be offset by a few pixels. We
will see how this is controlled in Chapter 14.

Mind Map Summary

Spatial Characteristics of the Magnetic Resonance Image

During an MRI procedure a section of a patient’s body is first divided

into a set of slices, and then each slice is divided into a matrix of voxels.
These actions are produced by the gradients.

Two methods can be selected to produce the slices. The most

common method, 2-D multiple slice acquisition, applies a gradient so that an
individual slice is tuned to a resonant frequency different from the other slice
positions. This gradient is turned on when the RF pulses are applied.
Therefore, only the tissue in a specific slice is excited and goes through the
process to produce signals. An alternate method, 3-D volume acquisition,
uses phase-encoding to produce slices. It is generally capable of producing
thinner, more contiguous slices.

Two different methods are used to cut a slice into voxels. Phase-
encoding is used in one direction, and frequency-encoding in the other.
Phase-encoding is produced by applying a gradient to the transverse
magnetization during each imaging cycle. To produce sufficient phase-
encoding information to permit image reconstruction, many different phase-
encoding gradient strengths must be used. In the typical imaging procedure
the phase-encoding gradient strength is changed from cycle to cycle. The
strength of the phase-encoding gradient, in effect, directs the signal data into
a specific row of k space. All the rows of k space must be filled with data
before the image reconstruction can be performed. The number of rows of k
space is one of the factors that determine how many imaging cycles must be
used, which, in turn, affects image acquisition time.

Frequency-encoding is produced by applying a gradient at the time of the

echo event during each cycle
 Chapter 9
Image Details and Noise
Introduction And Overview

Two characteristics of the MR image that reduce the visibility of anatomical

structures and objects within the body are blurring and visual noise. These
were introduced in Chapter 1 as image quality characteristics. Both image
blurring and visual noise are undesirable characteristics that collectively
reduce the overall quality of an image and the objects in the image as
illustrated in Figures 1-6 and 1-7. In an image, the combined effects of blur
and noise produce a “curtain of invisibility” that extends over some objects
based on object characteristics. This is shown in Figure 10-1, where we see
objects arranged according to two characteristics. In the horizontal direction,
the objects are arranged according to size. Decreasing object size
corresponds to increasing detail.

Figure 10-1. The impact of

image noise and blurring on
object visibility. Noise reduces
visibility of low contrast objects.
Blur reduces visibility of small
objects.

In the vertical direction, the objects are arranged according to their contrast.
The object in the lower left is both large and has a high level of contrast. This
is the object that would be most visible under a variety of imaging conditions.
The object that is always the most difficult to see is the small, low contrast
object, which in Figure 10-1 would be located in the upper right corner.

In every imaging procedure we can assume that some potential

objects within the body will not be visible because of the blurring and noise in
the image. This loss of visibility is represented by the “curtain” or area of
invisibility indicated in Figure 10-1. The location of the boundary between the
visible and invisible objects, often referred to as a contrast-detail curve, is
determined by the amount of blurring and noise associated with a specific
imaging procedure. In general, blurring reduces the visibility of anatomical
detail or other small objects that are located in the lower right region. Visual
noise reduces the visibility of low contrast objects located in the upper left
region.
 The imaging protocol determines the boundary of visibility by altering
the amount of blurring and noise. These two characteristics are determined by
the combination of many adjustable imaging factors. It is a complex process
because the factors that affect visibility of detail (blurring) also affect noise,
but in the opposite direction. As we will see when a protocol is changed to
improve visibility of detail, the noise is increased. Another point to consider is
that several of the factors that have an effect on both image detail and noise
also affect image acquisition time, which will be discussed in Chapter 11.
Therefore, when formulating an imaging protocol one must consider the
multiple effects of the imaging factors and then select factor values that
provide an appropriate compromise and an optimized acquisition for a specific
clinical study with respect to detail (blurring), noise, and acquisition speed.

We will now consider the many factors that have an effect on the
characteristics of image detail and noise.

Image Detail

The ability of a magnetic resonance image to show detail is determined

primarily by the size of the tissue voxels and corresponding image pixels.
Pixel size can be changed without major tradeoffs. However, as we are about
to observe, there are significant effects of changing voxel size that must be
considered. The real challenge is selecting a voxel size that is optimum for a
specific clinical procedure.

In principle, all structures within an individual voxel are blurred

together and represented by the signal intensity representing that voxel. It is
not possible to see details within a voxel, just the voxel itself. When we view
an MR image, we are actually looking at an image of a matrix, or array, of the
voxels. We usually do not see the individual voxels because they are so small
and they might be interpolated into even smaller image pixels. However, even
if we do not see the individual voxels, their size determines the anatomical
detail that we can see. The amount of image blurring is determined by the
dimensions of the individual voxels.

Three basic imaging factors determine the dimensions of a tissue

voxel, as illustrated in Figure 10-2.
 Figure 10-2. Voxel size
and detail in MR images
is determined by the
values selected for the
three protocol factors:
FOV, matrix size, and
slice thickness.

The dimension of a voxel in the plan of the image is determined by the ratio of
the field of view (FOV) and the size of the matrix. Both of these factors can be
used to adjust image detail. The thickness of the slice is a factor in voxel
signal intensity.

The selection of the FOV is determined primarily by the size of the

body part being imaged. One problem that can occur is the appearance of
foldover artifacts when the FOV is smaller than the actual body section.
However, there are artifact suppression techniques that can be used to
reduce this foldover problem, as described in Chapter 14. The maximum
useful FOV is usually limited by the dimensions and characteristics of the RF
coil. The important thing to remember is that smaller image FOVs and smaller
voxels produce better visibility of detail.

Matrix size refers to the number of voxels in the rows or columns of

the matrix. The matrix size is a protocol factor selected by the operator before
the imaging procedure. Typical matrix dimensions are in the range of 128 to
512 mm.

Noise Sources

Random RF energy can be generated by thermal activity within electrical

conductors and circuit components of the receiving system. In principle, the
patient’s body is a component of the RF receiving system. Because of its
mass, it becomes the most significant source of image noise in most imaging
procedures. The specific noise source is the tissue contained within the
sensitive FOV of the RF receiver coils. Some noise might be generated within
the receiver coils or other electronics, but it is usually much less than the
noise from the patient’s body.

Many devices in the environment produce RF noise or signals that can

interfere with MRI. These include radio and TV transmitters, electrosurgery
units, fluorescent lights, and computing equipment. All MR units are installed
with an RF shield, as described in Chapter 2, to reduce the interference from
these external sources. External interference is not usually a problem with a
properly shielded unit. When it does occur, it generally appears as an image
artifact rather than the conventional random noise pattern.
Signal-To-Noise Considerations

Image quality is not dependent on the absolute intensity of the noise energy
but rather the amount of noise energy in relation to the image signal intensity.
Image quality increases in proportion to the signal-to-noise ratio. When the
intensity of the RF noise is low in proportion to the intensity of the image
signal, the noise has a low visibility. In situations where the signals are
relatively weak, the noise becomes much more visible. The principle is
essentially the same as with conventional TV reception. When a strong signal
is received, image noise (snow) is generally not visible; when one attempts to
tune in to a weak TV signal from a distant station, the noise (noise) becomes
significant.

In MRI, the loss of visibility resulting from the noise can be reduced by
either reducing the noise intensity or increasing the intensity of the signals.
This is illustrated in Figure 10-3.

Figure 10-3. Factors that affect the signal-to-noise ratios in MR images

Let us now see how this can be achieved.

Voxel Size

One of the major factors that affects signal strength is the volume of the
individual voxels. The signal intensity is proportional to the total number of
protons contained within a voxel. Large voxels, that contain more protons,
emit stronger signals and result in less image noise. Unfortunately, as we
have just discovered, large voxels reduce image detail. Therefore, when the
factors for an imaging procedure are being selected, this compromise
between signal-to-noise ratio and image detail must be considered. The major
reason for imaging relatively thick slices is to increase the voxel signal
intensity and it also allows shorter TE values.

Field Strength

The strength of the RF signal from an individual voxel generally increases in

proportion to the square of the magnetic field strength. However, the amount
of noise picked up from the patient’s body often increases with field strength
because of adjustments in the bandwidth factor for the higher fields. This is
described in Chapter 14. Because of differences in system design, no one
precise relationship between signal-to-noise ratio and magnetic field strength
applies to all systems. In general, MRI systems operating at relatively high
field strengths produce images with higher signal-to-noise ratios than images
produced at lower field strengths, when all other factors are equal.
Tissue Characteristics

Signal intensity, and the signal-to-noise ratio, depend to some extent on the
magnetic characteristics of the tissue being imaged. For a specific set of
imaging factors, the tissue characteristics that enhance the signal-to-noise
relationship are high magnetic nuclei (proton) concentration, short T1, and
long T2. The primary limitation in imaging nuclei other than hydrogen
(protons) is the low tissue concentration and the resulting low signal intensity.

TR and TE

Repetition time (TR) and echo time (TE) are the factors used to control
contrast in most imaging methods. We have observed that these two factors
also control signal intensity. This must be taken into consideration when
selecting the factors for a specific imaging procedure.

When a short TR is used to obtain a T1weighted image, the

longitudinal magnetization does not have the opportunity to approach its
maximum and produce high signal intensity. In this case, some signal strength
must be sacrificed to gain a specific type of image contrast. Also, when TR is
reduced to decrease image acquisition time, image noise can become the
limiting factor.

When long TE values are used, the transverse magnetization and the
resulting signal it produces can decay to very low values. This causes the
images to display more noise.

RF Coils
The most direct control over the amount of noise energy picked from the
patient’s body is achieved by selecting appropriate characteristics of the RF
receiver coil. In principle, noise is reduced by decreasing the amount of tissue
within the sensitive region of the coil. Most imaging systems are equipped
with interchangeable coils. These include a body coil, a head coil, and a set of
surface coils as shown in Figure 10-4.

Figure 10-4. Both the amount of noise and the intensity of the signal received
depend on the RF receiving coils. The body coil picks up the most noise and
the weakest signal, resulting in the highest noise level in the image.

The body coil is the largest coil and usually contains a major part of the
patient’s tissue within its sensitive region. Therefore, body coils pick up the
greatest amount of noise. Also, the distance between the coil and the tissue
voxels is greater than in other types of coils. This reduces the intensity of the
signals actually received by the coil. Because of this combination of low signal
intensity and higher noise pickup, body coils generally produce a lower signal-
to-noise ratio than the other coil types.

In comparison to body coils, head coils are both closer to the imaged
tissue and generally contain a smaller total volume of tissue within their
sensitive region. Because of the increased signal-to-noise characteristic of
head coils, relatively small voxels can be used to obtain a better image detail.

The surface coil provides the highest signal-to-noise ratio of the three
coil types. Because of its small size, it has a limited sensitive region and picks
up less noise from the tissue. When it is placed on or near the surface of the
patient, it is usually quite close to the voxels and picks up a stronger signal
than the other coil types. The compromise with surface coils is that their
limited sensitive region restricts the useful FOV, and the sensitivity of the coil
is not uniform within the imaged area. This non-uniformity results in very
intense signals from tissue near the surface and a significant decrease in
signal intensity with increasing depth. The relatively high signal-to-noise ratio
obtained with surface coils can be traded for increased image detail by using
smaller voxels.

Receiver Bandwidth

Bandwidth is the range of frequencies (RF) that the receiver is set to receive
and is an adjustable protocol factor. It has a significant effect on the amount of
noise picked up. This is because the noise is distributed over a wide range of
frequencies, whereas the signal is confined to a relatively narrow frequency
range. Therefore, when the bandwidth is increased, more noise enters the
receiver. The obvious question is: Why increase bandwidth? One reason is
that a wider bandwidth reduces the chemical shift artifact that will be
described in Chapter 14. Also, wider bandwidths are the result of short signal
sampling, or “picture snapping” times that are useful in some applications.

Averaging

One of the most direct methods used to control the signal-to-noise

characteristics of MR images is the process of averaging two or more signal
acquisitions. In principle, each basic imaging cycle (phase-encoding step) is
repeated several times and the resulting signals are averaged to form the final
image as illustrated in Figure 10-5.

Figure 10-5. an mage with reduced noise is created by averaging

the signals from four acquisitions.
The averaging process tends to reduce the noise level because of its
statistical fluctuation nature, from one cycle to another. You can think of it as
acquiring four images by repeating the basic acquisition four times. Then the
signal intensities in each pixel position in the four images are averaged to
produce an intensity value for the new averaged image.

The disadvantage of averaging is that it increases the total image

acquisition time in proportion to the number of cycle repetitions or number of
signals averaged (NSA). The NSA is one of the protocol factors set by the
operator. Typical values are 1 (no averaging), 2, or 4, depending on the
amount of noise reduction required. The general relationship is that the NSA
must be increased by a factor of 4 to improve the signal-to-noise ratio by a
factor of 2. The signal-to-noise ratio is proportional to the square root of the
NSA. Sometimes the noise contribution from independent acquisitions adds;
sometimes it cancels. Since the signals always add, adding or averaging
independently acquired images improves the signal-to-noise ratio.

Parallel Imaging

Parallel imaging is a technique that is useful for reducing acquisition time.

The signals are acquired with a set of coils in a phased-array configuration.
Making use of the geometric sensitivity of each coil within the array some
degree of anatomical spatial information is obtained and used in the image
reconstruction process. This makes it possible to reduce the number of
phase-encoding gradient cycles and the associated acquisition time.

Mind Map Summary

Image Detail and Noise

Two important image quality characteristics are blurring, which

reduces visibility of small objects or detail, and image noise, which reduces
visibility of low contrast objects. Both of these characteristics depend on
design characteristics of the imaging system and the combination of selected
protocol factors. The principal source of blurring in an MR image is the voxel
size. This is because all tissues within an individual voxel are blurred together
and represented by one signal. An image does not display any detail within
the individual voxels. It is just a display of a matrix of voxels. Voxel size, and
the resulting blurring, can be adjusted with the three protocol factors: FOV,
matrix, and slice thickness.

The level of noise that appears in an image depends on the

relationship (ratio) of the signal strength from the individual voxels and the
noise strength coming from a region of the patient’s body. The visible noise is
reduced by increasing signal strength. This can be done by increasing the
magnetic field strength, increasing voxel size, increasing TR, and decreasing
TE. The field strength is a design characteristic and cannot be changed by the
operator. Increasing voxel size to decrease noise has the adverse effect of
also increasing blurring. Voxel sizes must be chosen to provide an
appropriate balance between blurring and noise.

The noise strength picked up from the patient’s body is determined by

the mass of tissue contained within the sensitive pickup region of the RF coils.
Surface coils that cover a relatively small anatomical region and are also
close to the signal source (voxels) produce a high signal-to-noise relationship
that results in lower image noise. The RF receiver bandwidth can be adjusted
to block some of the noise energy from being received. However, decreasing
the bandwidth to reduce noise has the adverse effect of increasing the
chemical shift artifact.

Signal averaging is a useful technique for reducing noise but has the
adverse effect of increasing acquisition time.