Nuclear magnetic resonance (NMR) is the spectroscopic study of the magnetic

properties of the nucleus of the atom. The protons and neutrons of the nucleus have
a magnetic field associated with their nuclear spin and charge distribution. Resonance
is an energy coupling that causes the individual nuclei, when placed in a strong
external magnetic field, to selectively absorb, and later release, energy unique to
those nuclei and their surrounding environment. The detection and analysis of the
NMR signal has been extensively studied since the 1940s as an analytic tool in
chemistry and biochemistry research. NMR is not an imaging technique but rather
a method to provide spectroscopic data concerning a sample placed in the device.
In the early 1970s, it was realized that magnetic field gradients could be used to
localize the NMR signal and to generate images that display magnetic properties of
the proton, reflecting clinically relevant information. As clinical imaging applica-
tions increased in the mid-1980s, the "nuclear" connotation was dropped, and mag-
netic resonance imaging (MRI), with a plethora of associated acronyms, became
commonly accepted in the medical community.
MRI is a rapidly changing and growing image modality. The high contrast sen-
sitivity to soft tissue differences and the inherent safety to the patient resulting from
the use of nonionizing radiation have been key reasons why MRI has supplanted
many CT and projection radiography methods. With continuous improvements in
image quality, acquisition methods, and equipment design, MRI is the modality of
choice to examine anatomic and physiologic properties of the patient. There are
drawbacks, including high equipment and siting costs, scan acquisition complexity,
relatively long imaging times, significant image artifacts, and patient claustropho-
bia problems.
The expanding use of MRI demands an understanding of the underlying MR
principles to glean the most out of the modality. This chapter reviews the basic prop-
erties of magnetism, concepts of resonance, and generation of tissue contrast through
specific pulse sequences. Details of image formation, characteristics, and artifacts, as
well as equipment operation and biologic effects, are discussed in Chapter 15.

14.1 MAGNETIZATION PROPERTIES

Magnetism
Magnetism is a fundamental property of matter; it is generated by moving charges,
usually electrons. Magnetic properties of materials result from the organization and
 motion of the electrons in either a random or a nonrandom alignment of magnetic
"domains," which are the smallest entities of magnetism. Atoms and molecules have
electron orbitals that can be paired (an even number of electrons cancels the mag-
netic field) or unpaired (the magnetic field is present). Most materials do not
exhibit overt magnetic properties, but one notable exception is the permanent mag-
net, in which the individual domains are aligned in one direction.
Magnetic susceptibility describes the extent to which a material becomes mag-
netized when placed in a magnetic field. The induced internal magnetization can
oppose the external magnetic field and lower the local magnetic field surrounding
the material. On the other hand, the internal magnetization can form in the same
direction as the applied magnetic field and increase the local magnetic field. Three
categories of susceptibility are defined: diamagnetic, paramagnetic, and ferromag-
netic. Diamagnetic materials have slightly negative susceptibility and oppose the
applied magnetic field. Examples of diamagnetic materials are calcium, water, and
most organic materials (chiefly owing to the diamagnetic characteristics of carbon
and hydrogen). Paramagnetic materials have slightly positive susceptibility and
enhance the local magnetic field, but they have no measurable self-magnetism.
Examples of paramagnetic materials are molecular oxygen (02), some blood degra-
dation products, and gadolinium-based contrast agents. Ferromagnetic materials are
"superparamagnetic"-that is, they augment the external magnetic field substan-
tially. These materials can exhibit "self-magnetism" in many cases. Examples are
iron, cobalt, and nickel.
Unlike the monopole electric charges from which they are derived, magnetic
fields exist as dipoles, where the north pole is the origin of the magnetic field lines
and the south pole is the return (Fig. 14-1A). One pole cannot exist withour the
other. As with electric charges, "like" magnetic poles repel and "opposite" poles
attract. The magnetic field strength, B, (also called the magnetic flux density) can be

FIGURE 14-1. A: The magnetic field has two poles, with magnetic field lines emerging from the
north pole (N) and returning to the south pole (5), as illustrated by a simple bar magnet. B: A coiled
wire carrying an electric current produces a magnetic field with characteristics similar to those of a
bar magnet. Magnetic field strength and field density depend on the magnitude of the current and
the number of coil turns. Note: the flow of positive current is opposite that of the electron flow.
conceptualized as the number of magnetic lines of force per unit area. The magnetic
field drops off with the square of the distance. The 51 unit for B is the tesla (T), and
as a benchmark, the earth's magnetic field is about 1/20,000 T. An alternate unit is
the gauss (G), where 1 T = 10,000 G.
Magnetic fields can be induced by a moving charge in a wire (for example, see
the section on transformers in Chapter 5). The direction of the magnetic field
depends on the sign and the direction of the charge in the wire, as described by the
"right hand rule": The fingers point in the direction of the magnetic field when the
thumb points in the direction of a moving positive charge (i.e., opposite the direc-
tion of electron movement). Wrapping the current-carrying wire many times in a
coil causes a superimposition of the magnetic fields, augmenting the overall
strength of the magnetic field inside the coil, with a rapid falloff of field strength
outside the coil (see Fig. 14-lB). Direct current (DC) amplitude in the coil deter-
mines the overall magnitude of the magnetic field strength. In essence, this is the
basic design of the "air core" magnets used for diagnostic imaging, which have mag-
netic field strengths ranging from 0.3 to 2.0 T, where the strength is directly related
to the current.

Magnetic Characteristics of the Nucleus

The nucleus exhibits magnetic characteristics on a much smaller scale was described
in the previous section for atoms and molecules and their associated electron distri-
butions. The nucleus is comprised of protons and neutrons with characteristics
listed in Table 14-1. Magnetic properties are influenced by the spin and charge dis-
tributions intrinsic to the proton and neutron. For the proton, which has a unit
positive charge (equal to the electron charge but of opposite sign), the nuclear "spin"
produces a magnetic dipole. Even though the neutron is electrically uncharged,
charge inhomogeneities on the subnuclear scale result in a magnetic field of oppo-
site direction and of approximately the same strength as the proton. The magnetic
moment, represented as a vector indicating magnitude and direction, describes the
magnetic field characteristics of the nucleus. A phenomenon known as pairing
occurs within the nucleus of the atom, where the constituent protons and neutrons
determine the nuclear magnetic moment. If the total number of protons (P) and
neutrons (N) in the nucleus is even, the magnetic moment is essentially zero. How-
ever, if N is even and P is odd, or N is odd and P is even, the noninteger nuclear
spin generates a magnetic moment. A single atom does not generate a large enough
nuclear magnetic moment to be observable; the signal measured by an MRI system
is the conglomerate signal of billions of atoms.

Mass (kg) 1.674 X 10~27 1.672 x 10~27

Charge (coulomb) o 1.602 X 1O~19
Spin quantum number 112 '12
Magnetic moment Uouleltesla) -9.66 x 10-27 1.41 x 10-26
Magnetic moment (nuclear magneton) -1.91 2.79
 Biologically relevant elements that are candidates for producing MR images are
listed in Table 14-2. The key features include the strength of the magnetic moment,
the physiologic concentration, and the isotopic abundance. Hydrogen, having the
largest magnetic moment and greatest abundance, is by far the best element for gen-
eral clinical utility. Other elements are orders of magnitude less sensitive when the
magnetic moment and the physiologic concentration are considered together. Of
these, 23Na and 31P have been used for imaging in limited situations, despite their
relatively low sensitivity. Therefore, the proton is the principal element used for MR
Imagmg.
The spinning proton or "spin" (spin and proton are synonymous herein) is clas-
sically considered to be like a bar magnet with north and south poles (Fig. 14-2);
however, the magnetic moment of a single proton is extremely small and not
detectable. A vector representation (amplitude and direction) is helpful when con-
templating the additive effects of many protons. Thermal energy agitates and ran-
domizes the direction of the spins in the tissue sample, and as a result there is no
net tissue magnetization (Fig. 14-3A). Under the influence of a strong external
magnetic field, Bo, however, the spins are distributed into two energy states: align-
ment with (parallel to) the applied field at a low-energy level, and alignment against
(antiparallel to) the field at a slightly higher energy level (see Fig. 14-3B). A slight
majority of spins exist in the low-energy state, the number of which is determined
by the thermal energy of the sample (at absolute zero, 0 degrees Kelvin (K), all pro-
tons would be aligned in the low-energy state). For higher applied magnetic field
strength, the energy separation of the low and high energy levels is greater, as is the
number of excess protons in the low-energy state. The number of excess protons in
the low-energy state at 1.0 T is about 3 spins per million (3 X 10-6) at physiologic
temperatures and is proportional to the external magnetic field. Although this does
not seem significant, for a typical voxel volume in MRI there are about 1021 pro-
tons, so there are 3 X 10-6 X 1021, or approximately 3 X 1015, more spins in the low-
energy state! This number of protons produces an observable magnetic moment
when summed.
In addition to energy separation of the spin states, the protons also experience
a torque from the applied magnetic field that causes precession, in much the same

TABLE 14-2. MAGNETIC RESONANCE PROPERTIES OF MEDICALLY

USEFUL NUCLEI

Spin Quantum % Isotopic Magnetic Relative Physiologic Relative

Nucleus Number Abundance Moment Concentration * Sensitivity

'H '12 99.98 2.79 100 1

'60 0 99.0 0 50 0
170 512 0.04 1.89 50 9 x 10-6
19F '12 100 2.63 4 x 10-6 3 X 10-8
23Na 312 100 2.22 8 x 10-2 1 X 10-4
31p '12 100 1.13 7.5 x 10-2 6 X 10-5

*Note: all isotopes of a given element.

 Spinning proton with Correspond ing
dipole magnetic field "magnetic moment"

FIGURE 14-2. A: A spinning proton can classically be represented with a

dipole magnetic field. B: Magnetic characteristics of the proton are
described as a magnetic moment, with a vector that indicates direction
and magnitude.

way that a spinning top wobbles due to the force of gravity (Fig. 14-4A). Direction of
the spin axis is perpendicular to the torque's twisting. This precession occurs at an
angular frequency (number of rotations/see about an axis of rotation) that is propor-
tional to the magnetic field strength Bo. The Larmor equation describes the depen-
dence between the magnetic field, Bo, and the precessional angular frequency, 0)0:

O)o=yBo

Net
magnetic
moment

FIGURE 14-3. Simplified distributions of "free" protons without and with an external magnetic
field are shown. A: Without an external magnetic field, a group of protons assumes a random ori-
entation of magnetic moments, producing an overall magnetic moment of zero. B: Under the
influence of an applied external magnetic field, Bo, the protons assume a nonrandom alignment in
two possible orientations: parallel and antiparallel to the applied magnetic field. A slightly greater
number of protons exist in the parallel direction, resulting in a measurable sample magnetic
moment in the direction of Bo.
FIGURE 14-4. A: A single proton precesses about its axis with an
angular frequency. 0). that is proportional to the externally applied
magnetic field strength, according to the Larmor equation. B: A
group of protons in the parallel and anti parallel energy states gen-
erates an equilibrium magnetization. Mo. in the direction of the
applied magnetic field Bo. The protons are distributed randomly
over the surface of the cone and produce no magnetization in the
perpendicular direction.

fa = l Eo
21t
where y is the gyro magnetic ratio unique to each element, Bo is the magnetic field
strength in tesla, fis the linear frequency in MHz (where 0) = 21tf: linear and angu-
lar frequency are related by a 21t rotation about a circular path), and yl21t is the
gyro magnetic ratio expressed in MHz/T. Because energy is proportional to fre-
quency, the energy separation, ~E, between the parallel and anti parallel spins is pro-
portional to the precessional frequency, and larger magnetic fields produce a higher
precessional frequency. Each element has a unique gyro magnetic ratio that allows
the discrimination of one element from another, based on the precessional fre-
quency in a given magnetic field strength. In other words, the choice of frequency
allows the resonance phenomenon to be tuned to a specific element. The gyromag-
netic ratios of selected elements are listed in Table 14-3.
The millions of protons precessing in the parallel and anti parallel directions
results in a distribution that can be represented by two cones with the net magnetic
moment equal to the vector sum of all the protons in the sample in the direction of
the applied magnetic field (see Fig. 14-4B). At equilibrium, no magnetic field exists
perpendicular to the direction of the external magnetic field because the individual
protons precess with a random distribution, which effectively averages out any net
magnetic moment. Energy (in the form of a pulse of radio frequency electromag-
netic radiation) at the precessional frequency (related to ~E) is absorbed and con-
verts spins from the low-energy, parallel direction to the higher-energy, antiparallel
TABLE 14-3. GYROMAGNETIC RATIO FOR USEFUL
ELEMENTS IN MAGNETIC RESONANCE

y/2n (MHz/T)

lH 42.58
Be 10.7
170 5.8
19F 40.0
23Na 11.3
31p 17.2

direction. As the perturbed system goes back to its equilibrium state, the MR sig-
nal is produced.
Typical magnetic field strengths for imaging range from 0.1 to 4.0 T (1,000 to
40,000 G). For protons, the precessional frequency is 42.58 MHz in a I-T mag-
netic field (i.e., y/21t = 42.58 MHz/T for lH). The frequency increases or decreases
linearly with increases or decreases in magnetic field strength, as shown in the exam-
ple. Accuracy and precision are crucial for the selective excitation of a given nucleus
in a magnetic field of known strength. Spin precession frequency must be known
to an extremely small fraction (l0-12) of the precessional frequency for modern
imaging systems.

The differences in the precessional frequency allow the selective excitation of one
elemental species for a given magnetic field strength.

By convention, the applied magnetic field Bo is directed parallel to the z-axis of the
three-dimensional Cartesian coordinate axis system. The x and y axes are perpen-
dicular to the z direction. For convenience, two frames of reference are used: the
laboratory frame and the rotating frame. The laboratory frame (Fig. 14- 5A) is a sta-
tionary reference frame from the observer's point of view. The proton's magnetic
moment precesses about the z-axis in a circular geometry about the x-y plane. The
rotating frame (see Fig. 14-5B) is a spinning axis system whereby the angular fre-
quency is equal to the precessional frequency of the protons. In this frame, the spins
 FIGURE 14-5. A: The laboratory frame of reference uses stationary three-dimensional
Cartesian coordinates. The magnetic moment precesses around the z-axis at the Lar-
mor frequency. B: The rotating frame of reference uses Cartesian coordinate axes that
rotate about the z-axis at the Larmor precessional frequency, and the other axes are
denoted x' and y'. When precessing at the Larmor frequency, the sample magnetic
moment appears stationary.

appear to be stationary when they rotate at the precessional frequency. If a slightly

higher precessional frequency occurs, a slow clockwise rotation is observed. For a
slightly lower precessional frequency, counterclockwise rotation is observed. A
merry-go-round exemplifies an analogy to the laboratory and rotating frames of ref-
erence. Externally, from the laboratory frame of reference, the merry-go-round
rotates at a specific angular frequency (e.g., 15 rotations per minute [rpm)). Indi-
viduals riding the horses are observed moving in a circular path around the axis of
rotation, and up-and-down on the horses. If the observer jumps onto the merry-go-
round, everyone on the ride now appears stationary (with the exception of the up-
and-down motion of the horses)-this is the rotating frame of reference. Even
though the horses are moving up and down, the ability to study them in the rotat-
ing frame is significantly improved compared with the laboratory frame. If the
merry-go-round consists of three concentric rings that rotate at 14, 15, and 16 rpm
and the observer is on the 15-rpm section, all individuals on that particular ring
would appear stationary, but individuals on the 14-rpm ring would appear to be
rotating in one direction at a rate of 1 rpm, and individuals on the 16-rpm ring
would appear to be rotating in the other direction at a rate of 1 rpm. Both the lab-
oratory and the rotating frame of reference are useful in explaining various interac-
tions of the protons with externally applied static and rotating magnetic fields.
The net magnetization vector, M, is described by three components. Mz is the
component of the magnetic moment parallel to the applied magnetic field and is
known as longitudinal magnetization. At equilibrium, the longitudinal magnetiza-
tion is maximal and is denoted as Mo, the equilibrium magnetization, where Mo =
MZ) with the amplitude determined by the excess number of protons that are in the
low-energy state (i.e., aligned with Bo). Mxy is the component of the magnetic
moment perpendicular to the applied magnetic field and is known as transverse
magnetization. At equilibrium, the transverse magnetization is zero, because the vec-
tor components of the spins are randomly oriented about 360 degrees in the x-y
FIGURE 14-6. Longitudinal magnetization, Mz, is the vector component of the mag-
netic moment in the z direction. Transverse magnetization, Mxy, is the vector compo-
nent of the magnetic moment in the x-y plane. Equilibrium magnetization, Mo, is the
maximal longitudinal magnetization of the sample; in this illustration it is shown dis-
placed from the z-axis.

plane and cancel each other. When the system absorbs energy, Mz is "tipped" into
the transverse plane; why this is important is explained in the next section. Figure
14-6 illustrates these concepts.
The famous Bloch equations are a set of coupled differential equations that
describe the behavior of the magnetization vectors under any conditions. These
equations, when integrated, yield the x', y', and z components of magnetization as
a function of time.

14.2 GENERATION AND DETECTION OF THE MAGNETIC

RESONANCE SIGNAL

Application of radiofrequency (RF) energy synchronized to the precessional fre-

quency of the protons causes displacement of the tissue magnetic moment from
equilibrium conditions (i.e., more protons are in the antiparallel orientation).
Return to equilibrium results in emission of MR signals proportional to the num-
ber of excited protons in the sample, with a rate that depends on the characteristics
of the tissues. Excitation, detection, and acquisition of the signals constitute the
basic information necessary for MR spectroscopy and imaging.

The displacement of the equilibrium magnetization occurs when the magnetic

component of the RF pulse, also known as the B1 field, is precisely matched to the
precessional frequency of the protons to produce a condition of resonance. This RF
frequency (proportional to energy) corresponds to the energy separation between
the protons in the parallel and anti parallel directions, as described by either a quan-
tum mechanics or a classical physics approach. Each has its advantages in the under-
standing of MR physics.
The quantum mechanics approach considers the RF energy as photons
(quanta) instead of waves. Spins oriented parallel and antiparallel to the external
magnetic field are separated by an energy gap, L1E. Only when the exact energy is
applied do the spins flip ( i.e., transition from the low- to the high-energy level or
from the high- to the low-energy level). This corresponds to a specific frequency of
the RF pulse, equal to the precessional frequency of the spins. The amplitude and
duration of the RF pulse determine the overall energy absorption and the number
of protons that undergo the energy transition. Longitudinal magnetization changes
from the maximal positivevalue at equilibrium, through zero, to the maximal neg-
ative value (Fig. 14-7). Continued RF application induces a return to equilibrium
conditions, as an incoming "photon" causes the spontaneous emission of two pho-
tons and reversion of the proton to the parallel direction. To summarize, the quan-
tum mechanical description explains the exchange of energy occurring between
magnetized protons and photons and the corresponding change in the longitudinal
magnetization. However, it does not directly explain how the sample magnetic
moment induces a current in a coil and produces the MR signal. The classical
physics model better explains this phenomenon.

Mz -- Longitudinal Magnetization Resonance frequency

z 42.58 MHzl T

666
,/J
Equilibrium --

69696
more spins parallel
than anti-parallel ......................
y
~ V\ANW~
Mz positive RF pulse
z (81 field)

966 Equal numbers of

6969er
parallel and Time of 81 field
anti-parallel spins y
increasing
x Mz = 0
z

80 Excited spins
96699 More anti-parallel occupy anti-
spins than parallel
er 6 9 x··················r···················y parallel energy
levels
Mz negative

FIGURE 14-7. A simple quantum mechanics process depicts the discrete energy absorp-
tion and the time change of the longitudinal magnetization vector as radiofrequency
(RF) energy, equal to the energy difference between the parallel and anti parallel spins,
is applied to the sample (at the Larmor frequency). Discrete quanta absorption changes
the proton energy from parallel to antiparallel. With continued application of RF
energy at the Larmor frequency, Mz is displaced from equilibrium, through zero, to the
opposite direction (high-energy state).
 From the classical physics viewpoint, the BI field is considered the magnetic
component of an electromagnetic wave with sinusoidally varying electric and mag-
netic fields. The magnetic field variation can be thought of as comprising two mag-
netic field vectors of equal magnitude, rotating in opposite directions around a
point at the Larmor frequency and traveling at the speed oflight (Fig. 14-8). At 0,
180, and 360 degrees about the circle of rotation, the vectors cancel, producing no
magnetic field. At 90 degrees, the vectors positively add and produce the peak mag-
netic field, and at 270 degrees the vectors negatively add and produce the peak
magnetic field in the opposite direction, thus demonstrating the characteristics of
the magnetic variation. Now consider the magnetic vectors independently; of the
two, only one vector will be rotating in the same direction as the precessing spins
in the magnetized sample (the other vector will be rotating in the opposite direc-
tion). From the perspective of the rotatingframe, this magnetic vector (the applied
BI field) is stationary relative to the precessing protons within the sample, and it is
applied along the x' -axis (or the y' -axis) in a direction perpendicular to the sample
magnetic moment, Mz (Fig. 14-9A). The stationary BI field applies a torque to M"
causing a rotation away from the longitudinal direction into the transverse plane.
The rotation of Mz occurs at an angular frequency equal to WI = yBI• Because the
tip angle, e, is equal to WI X t, where t is the time of the B1 field application, then,
by substitution, e
= yB1t, which shows that the time of the applied B1 field deter-
mines the amount of rotation of Mz• Applying the BI field in the opposite direction
(l80-degree change in phase) changes the tip direction of the sample moment. If
the RF energy is not applied at the precessional (Larmor) frequency, the B1 field is
not stationary in the rotating frame and the coupling (resonance) between the two
magnetic fields does not occur (see Fig. 14-9B).
Flip angles describe the rotation through which the longitudinal magnetization
vector is displaced to generate the transverse magnetization (Fig. 14-10). Common
angles are 90 degrees (rrl2) and 180 degrees (rr), although a variety of smaller (less
than 90 degrees) and larger angles are chosen to enhance tissue contrast in various

~tt--'"
~.....•

t
/
IL
Clockwise Counter-
rotating clockwise
vector rotating
vector

FIGURE 14-8. A classical physics description of the magnetic field component of the
radiofrequency pulse (the electric field is not shown). Clockwise (solid) and counter-
clockwise (dotted) rotating magnetic vectors produce the magnetic field variation by
constructive and destructive interaction. At the Larmor frequency, one of the mag-
netic field vectors rotates synchronously in the rotating frame and is therefore sta-
tionary (the other vector rotates in the opposite direction).
 81 at Larmor Frequency 81 off resonance

z z

Mot:/81
B1\
...................... ::.::~: . . :,•........................ y'
............~ y' ....... ~
.... :~
x''· ;
x' ~
~ ~ B1
. ---
FIGURE 14-9. A: In the rotating frame, the radiofrequency pulse (B1 field)
is applied at the Larmor frequency and is stationary in the x'-y' plane. The
B1 field interacts at 90 degrees to the sample magnetic moment and pro-
duces a torque that displaces the magnetic vector away from equilibrium.
B: The B1 field is not tuned to the Larmor frequency and is not stationary
in the rotating frame. No interaction with the sample magnetic moment
occurs.

x···········
.....r------
Direction of torque on M,

....................
!.1./
::.
180 flip
0

x·
.....
•••••••
'
-Mz
B1

y'

FIGURE 14-10. A: Flip angles describe the angular displacement of the longitudinal
magnetization vector from the equilibrium position. The rotation angle of the mag-
netic moment vector depends on the duration and amplitude of the B1 field at the
Larmor frequency. Small flip angles (-30 degrees) and large flip angles (-90 degrees)
produce small and large transverse magnetization, respectively. B: Common flip angles
are 90 degrees, which produces the maximum transverse magnetization, and 180
degrees, which inverts the existing longitudinal magnetization Mz to -Mz.
ways. A 90-degree angle provides the largest possible transverse magnetization. The
time required to flip the magnetic moment is linearly related to the displacement
angle: For the same Bl field strength, a 90-degree angle takes half the time to pro-
duce that a 180-degree angle does. The time required to implement a rotation is on
the order of tens to hundreds of microseconds. With fast MR imaging techniques,
30-degree and smaller angles are often used to reduce the time needed to displace
the longitudinal magnetization and generate the transverse magnetization. For flip
angles smaller than 90 degrees, less signal in the Mxy direction is generated, but less
time is needed to displace M" resulting in a greater amount of transverse magnetiza-
tion (signal) per excitation time. For instance, a 45-degree flip takes half the time of
a 90-degree flip yet creates 70% of the signal, because the projection of the vector
onto the transverse plane is sin 45 degrees, or 0.707. In instances where short exci-
tation times are necessary, small flip angles are employed.

Free Induction Decay: 12 Relaxation

The 90-degree RF pulse produces phase coherence of the individual protons and
generates the maximum possible transverse magnetization for a given sample vol-
ume. As Mxy rotates at the Larmor frequency, the receiver antenna coil (in the labo-
ratory frame) is induced (by magnetic induction) to produce a damped sinusoidal elec-
tronic signal known as thefree induction decay (FID) signal, as shown in Fig. 14-11.

Equilibrium 90° RF pulse Dephasing Dephased

Mxy = zero Mxy large Mxy decreasing Mxy = zero
Time

Antenna
receiver

+ ··.~FID
'.

FIGURE 14-11. A: Conversion of longitudinal magnetization, Mz, into transverse mag-

netization, Mxy, results in an initial phase coherence of the individual spins of the sam-
ple. The magnetic moment vector precesses at the Larmor frequency (stationary in the
rotating frame) and dephases with time. B: In the laboratory frame, Mxy precesses and
induces a signal in an antenna receiver that is sensitive to transverse magnetization. A
free induction decay (FID) signal is produced, oscillating at the Larmor frequency, and
decays with time due to the loss of phase coherence.
 The "decay" of the FID envelope is the result of the loss of phase coherence of
the individual spins caused by magnetic field variations. Micromagnetic inhomo-
geneities intrinsic to the structure of the sample cause a spin-spin interaction,
whereby the individual spins precess at different frequencies due to slight changes in
the local magnetic field strength. Some spins travel faster and some slower, resulting
in a loss of phase coherence. External magnetic field inhomogeneities arising from
imperfections in the magnet or disruptions in the field by paramagnetic or ferro-
magnetic materials accelerate the dephasing process. Exponential relaxation decay,
T2, represents the intrinsic spin-spin interactions that cause loss of phase coherence
due to the intrinsic magnetic properties of the sample. The elapsed time between the
peak transverse signal and 37% of the peak level (1 I e) is the T2 decay constant (Fig.
14-12A). Mathematically, this exponential relationship is expressed as follows:
Mxy (t)= Moe -tlT2

where Mxy is the transverse magnetic moment at time t for a sample that has Mo
transverse magnetization at t = O. When t = T2, then e-1 = 0.37, and Mxy = 0.37
Mo. An analogous comparison to T2 decay is that of radioactive decay, with the
exception that T2 is based on lie decay instead of half-life (1/2) decay. This means
that the time for the FID to reach half of its original intensity is given by t = 0.693
xT2.
T2 decay mechanisms are determined by the molecular structure of the sam-
ple. Mobile molecules in amorphous liquids (e.g., cerebral spinal fluid [CSF])
exhibit a long T2, because fast and rapid molecular motion reduces or cancels
intrinsic magnetic inhomogeneities. As the molecular size increases, constrained
molecular motion causes the magnetic field variations to be more readily manifested
and T2 decay to be more rapid. Thus large, nonmoving structures with stationary
magnetic inhomogeneities have a very short T2.
. In the presence of extrinsic magnetic inhomogeneities, such as the imperfect
main magnetic field, Bo, the loss of phase coherence occurs more rapidly than from

Mxy Mxy Mxy

maximum decreasing zero

\
\
\
\
\

" T2* decay

....•..

FIGURE 14-12. A: The loss of Mxy phase coherence occurs exponentially and is caused
by intrinsic spin-spin interactions in the tissues, as well as extrinsic magnetic field inho-
mogeneities. The exponential decay constant, T2, is the time over which the signal
decays to 37% of the maximal transverse magnetization (e.g., after a 90-degree pulse).
B: T2 is the decay time that results from intrinsic magnetic properties of the sample. T2*
is the decay time resulting from both intrinsic and extrinsic magnetic field variations. T2
is always longer than T2*.
spin-spin interactions by themselves. When Bo inhomogeneity is considered, the
spin-spin decay constant T2 is shortened to T2*. Figure 14-12B shows a compari-
son of the T2 and T2* decay curves. T2* depends on the homogeneity of the main
magnetic field and susceptibility agents that are present in the tissues (e.g., MR con-
trast materials, paramagnetic or ferromagnetic objects).

The loss of transverse magnetization (T2 decay) occurs relatively quickly, whereas
the return of the excited magnetization to equilibrium (maximum longitudinal
magnetization) takes a longer time. Individual excited spins must release their
energy to the local tissue (the lattice). Spin-lattice relaxation is a term given for the
exponential regrowth of M" and it depends on the characteristics of the spin inter-
action with the lattice (the molecular arrangement and structure). The Tl relaxation
constant is the time needed to recover 63% of the longitudinal magnetization, M"
after a 90-degree pulse (when Mz = 0). The recovery ofMz versus time after the 90-
degree RF pulse is expressed mathematically as follows:
Mz (t)= MoO - e -tlTI)
where Mz is the longitudinal magnetization that recovers after a time t in a mater-
ial with a relaxation constant Tl. Figure 14-13 illustrates the recovery of Mz• When
t = Tl, then 1 - e-1 = 0.63, and Mz = 0.63 Mo. Full longitudinal recovery depends
on the Tl time constant .. For instance, at a time equal to 3 X Tl after a 90-degree
pulse, 95% of the equilibrium magnetization is reestablished. After a period of 5 X
Tl, the sample is considered to be back to full longitudinal magnetization. A

FIGURE 14-13. After a gO-degree pulse, longitudinal magnetization (Mz) is

converted from a maximum value at equilibrium to zero. Return of Mz to
equilibrium occurs exponentially and is characterized by the spin-lattice T1
relaxation constant. After an elapsed time equal to n, 63% of the longitu-
dinal magnetization is recovered. Spin-lattice recovery takes longer than
spin-spin decay (T2).
 method to determine the T1 time of a specific tissue or material is illustrated in Fig.
14-14. An initial 90-degree pulse, which takes the longitudinal magnetization to
zero, is followed by a delay time, ~ T, and then a second 90-degree pulse is applied
to examine the Mz recovery by displacement into the transverse plane (only Mxy
magnetization can be directly measured). By repeating the sequence with different
delay times, ~T, between 90-degree pulses (from equilibrium conditions), data
points that lie on the T1 recovery curve are determined. The T1 value can be esti-
mated from these values.
T1 relaxation depends on the dissipation of absorbed energy into the sur-
rounding molecular lattice. The relaxation time varies substantially for different tis-
sue structures and pathologies. From a classical physics perspective, energy transfer
is most efficient when the precessional frequency of the excited protons overlaps
with the "vibrational" frequencies of the molecular lattice. Large, slowly moving
molecules exhibit low vibrational frequencies that concentrate in the lowest part of
the frequency spectrum. Moderately sized molecules (e.g., proteins) and viscous flu-
ids produce vibrations across an intermediate frequency range. Small molecules
have vibrational frequencies with low-, intermediate-, and high-frequency compo-
nents that span the widest frequency range. Therefore, T1 relaxation is strongly
dependent on the physical characteristics of the tissues, as illustrated in Fig. 14-15.
Consequently, for solid and slowly moving structures, low-frequency variations exist
and there is little spectral overlap with the Larmor frequency. A small spectral over-
lap also occurs for unstructured tissues and fluids that exhibit a wide vibrational fre-
quency spectrum but with low amplitude. In either situation, the inability to release

z
Equilibrium ....t. y
x'"
~
90° pulse :
........
--..y
~ x
Delay time , .

'7 ~ z FIGURE 14-14. Spin-lattice

~
Longitudinal recovery ......
L y ......
J 'y x
....
J 'y relaxation for a sample can be
measured by using various
x x delay times between two 90-
~
90° pulse :
........'+- 'y
:
--- 'y
:-- ..'y
degree radiofrequency pulses.
After an initial 90-degree
pulse, longitudinal magnetiza-

::::~~
x~.
~ tion (Mz) is equal to zero; after
a known delay, another 90-
degree pulse is applied, and
the longitudinal magnetiza-
tion that has recovered during
the delay is converted to trans-
verse magnetization. The max-
.... :
imum amplitude of the resul-
tant free induction decay (FID)
is recorded as a function of
delay time, and the points are
fit to an exponential recovery
function to determine n.
 /:~Med~;:rtV~~OUS:

Small, aqueous:

~ ~~ LongT1
............... Large,

:)(....=.~
/
stationary:
- - - - •••• '"=" L,9ngest T1
•.•... .
'. .•..
.............. ' ...•..

FIGURE 14-15. A classical physics explanation of spin-lattice relaxation is based on

the vibration of the molecular lattice of a sample material and its frequency spec-
trum. Large, stationary structures exhibit little motion with mostly low-frequency
vibrations (solid curve). Medium sized, proteinated materials have increased fre-
quency amplitudes (short dash curve), and small sized, aqueous materials have
frequencies distributed over a broad range (long dash curve). The overlap of the
Larmor precessional frequency (vertical bar) with the molecular vibration spectrum
indicates the probability of spin-lattice relaxation.

energy to the lattice results in a relatively long T1 relaxation. One interesting case
is that of water, which has an extremely long T1, but the addition of water-soluble
proteins produces a hydration layer that slows the molecular vibrations and shifts
the high frequencies in the spectrum to lower values that increase the amount of
spectral overlap with the Larmor frequency and result in a dramatically shorter T1.
Moderately sized molecules, such as lipids, proteins, and fats, have a more struc-
tured lattice with a vibrational frequency spectrum that is most conducive to spin-
lattice relaxation. For biologic tissues, T1 ranges from 0.1 to 1 second in soft tis-
sues, and from 1 to 4 seconds in aqueous tissues (e.g., CSF) and water.
T1 relaxation increases with higher field strengths. A corresponding increase in
the Larmor precessional frequency reduces the spectral overlap of the molecular
vibrational frequency spectrum, resulting in longer T1 times. Contrast agents (e.g.,
complex macromolecules containing gadolinium) are effective in decreasing T1
relaxation time by allowing free protons to become bound and create a hydration
layer, thus providing a spin-lattice energy sink and a rapid return to equilibrium.
Even a very small amount of gadolinium contrast in pure water has a dramatic effect
on T1, decreasing the relaxation from a couple of seconds to tens of milliseconds!

T1 is significantly longer than T2. For instance, in a soft tissue, a T1 time of 500
msec has a correspondingT2 time that is typically 5 to 10 times shorter (i.e., about
 50 msec). Molecular motion, size, and interactions influence T1 and T2 relaxation
(Fig. 14-16). Molecules can be categorized roughly into three size groups-small,
medium, and large-with corresponding fast, medium, and slow vibrational fre-
quencies. For reasons described in the previous two sections, small molecules
exhibit long T1 and long T2, and intermediate-sized molecules have short T1 and
short T2; however, large, slowly moving or bound molecules have long T1 and
short T2 relaxation times. Because most tissues of interest in MR imaging consist
of intermediate to small-sized molecules, a long T1 usually infers a long T2, and a
short T1 infers a short T2. It is the differences in Tl, T2, and T2* (along with pro-
ton density variations and blood flow) that provide the extremely high contrast in
MRI.
Magnetic field strength influences T1 relaxation but has an insignificant
impact on T2 decay. This is related to the dependence of the Larmor frequency on
magnetic field strength and the degree of overlap with the molecular vibration
spectrum. A higher magnetic field strength increases the Larmor frequency (0)0 =
yBo), which reduces the amount of spectral overlap and produces a longer T 1. In
Table 14-4, a comparison ofT1 and T2 for various tissues is listed. Agents that dis-
rupt the local magnetic field environment, such as paramagnetic blood degrada-
tion products, elements with unpaired electron spins (e.g., gadolinium), or any fer-
romagnetic materials, cause a significant decrease in T2*. In situations where a
macromolecule binds free water into a hydration layer, T1 is also significantly
decreased.
To summarize, T1 > T2 > T2*, and the specific relaxation times are a function
of the tissue characteristics. The spin density, T1, and T2 decay constants are fun-
damental properties of tissues, and therefore these tissue properties can be exploited
by MRI to aid in the diagnosis of pathologic conditions such as cancer, multiple
sclerosis, or hematoma. It is important to keep in mind that T1 and T2 (along with
spin density) are fundamental properties of tissue, whereas the other time-depen-
dent parameters are machine-dependent. Mechanisms to exploit the differences in
T1, T2, and spin density (proton density) are detailed in the next section.

FIGURE 14-16. Factors affect-

ing T1 and T2 relaxation times
of different tissues are gener-
ally based on molecular motion,
Molecular size: large
size, and interactions. The re-
laxation times (vertical axis) are
Molecular interactions: bound different for T1 and T2.
Fat 210 260
Liver 350 500
Muscle 550 870
White matter 500 780
Gray matter 650 900
Cerebrospinal fluid 1,800 2,400

Emphasizing the differences among spin density, Tl, and T2 relaxation time con-
stants of the tissues is the key to the exquisite contrast sensitivity of MR images. Tai-
loring the pulse sequences-that is, the timing, order, polarity, and repetition fre-
quency of the RF pulses and applied magnetic field gradients-makes the emitted
signals dependent on T1, T2 or spin density relaxation characteristics. MR relies on
three major pulse sequences: spin echo, inversion recovery, and gradient recalled
echo. When these used in conjunction with localization methods (i.e., the ability to
spatially encode the signal to produce an image, discussed in Chapter 15), "con-
trast-weighted" images are obtained. In the following three sections, the salient
points and essential considerations of MR contrast generated by these three major
pulse sequences are discussed.

Spin echo describes the excitation of the magnetized protons in a sample with an RF
pulse and production of the FID, followed by a second RF pulse to produce an
echo. Timing between the RF pulses allows separation of the initial FID and the
echo and the ability to adjust tissue contrast.

An initial90-degree pulse produces the maximal transverse magnetization, MxY'and

places the spins in phase coherence. The signal exponentially decays with T2* relax-
ation caused by intrinsic and extrinsic magnetic field variations. Mter a time delay
ofTEI2, where TE is the time of echo (defined below), a ISO-degree RF pulse is
applied, which inverts the spin system and induces a rephasing of the transverse
magnetization. The spins are rephased and produce a measurable signal at a time
equal to the time of echo (TE). This sequence is depicted in the rotating frame in
Fig. 14-17. The echo reforms in the opposite direction from the initial transverse
magnetization vector, so the spins experience the opposite external magnetic field
inhomogeneities and this strategy cancels their effect.
The Bo inhomogeneity-canceling effect that the spin echo pulse sequence
produces has been likened to a foot race on a track. The racers start running at
 TE
I
180 0
'I
Echo

Rotating frame
./

FID signal gradually After 1800 pulse, signal Spin echo peak amplitude
decays with decay rate T2* reforms at same rate depends on T2

FIGURE 14-17. The spin echo pulse sequence starts with a 90-degree pulse and produces a free
induction decay (FID) that decays according to T2* relaxation. After a delay time equal to TE/2,
a 180-degree radiofrequency pulse inverts the spins that reestablishes phase coherence and pro-
duces an echo at a time TE. Inhomogeneities of external magnetic fields are canceled, and the
peak amplitude of the echo is determined by T2 decay. The rotating frame shows the evolution
of the echo vector in the opposite direction of the FID.

the 90-degree pulse, but quickly their tight grouping at the starting line spreads
out (dephases) as they run at different speeds. After a short period, the runners
are spread out along the track, with the fastest runners in front and the slower
ones in the rear. At this time (TE/2), a 180-degree pulse is applied and the run-
ners all instantly reverse their direction, but they keep running at the same speed
as before. Immediately after the 180-degree rephasing RF pulse, the fastest run-
ners are the farthest behind and the slowest runners are in front of the pack.
Under these conditions, the fast runners at the end of the pack will catch the slow
runners at the front of the pack as they all run past the starting line together (i.e.,
at time TE). Even in a field of runners in which each runs at a markedly differ-
ent speed from the others, they all will recross the starting line at exactly TE. The
MR signal is at a maximum (i.e., the peak of the FIO envelope) as the runners are
all in phase when they cross the starting line. They can run off in the other direc-
tion, and after another time interval ofTE/2 reverse their direction and run back
to the starting line. Again, after a second TE period, they will all cross the start-
ing line (and the FIO signal will be at its third peak), then head off in the other
direction. This process can be repeated (Fig. 14-18). The maximal echo amplitude
depends on the T2 constant and not on T2*, which is the decay constant that
includes magnetic field inhomogeneities. Of course all MR signals depend on the
proton density of the tissue sample, as well. Just before and after the peak ampli-
tude of the echo (centered at time TE), digital sampling and acquisition of the
 0 0
90
0
180 0
180 180
pulse pulse pulse pulse

T2 decay

FIGURE 14-18. "True" T2 decay is determined from multiple 180-degree refocusing pulses.
While the free induction decay (FID) envelope decays with the T2* decay constant, the peak
echo amplitudes of subsequent echoes decay exponentially with time, according to the T2
decay constant.

signal occurs. Spin echo formation separates the RF excitation and signal acqui-
sition events by finite periods of time, which emphasizes the fact that relaxation
phenomena are being observed and encoded into the images. Contrast in the
image is produced because different tissue types relax differently (based on their
Tl and T2 characteristics).
Multiple echoes generated by 180-degree pulses after the initial excitation allow
the determination of the "true T2" of the sample. Signal amplitude is measured at
several points in time, and an exponential curve is fit to this measured data (see Fig.
14-18). The T2 value is one of the curve-fitting coefficients.

Time of Repetition and Partial Saturation

The standard spin echo pulse sequence uses a series of 90-degree pulses separated
by a period known as the time of repetition (TR), which typically ranges from about
300 to 3,000 msec. A time delay between excitation pulses allows recovery of the
longitudinal magnetization. During this period, the FID and the echo produce the
MR signal (explained later). After the TR interval, the next 90-degree pulse is
applied, but usually beftre the complete longitudinal magnetization recovery of the
tissues. In this instance, the FID generated is less than the first FID. After the sec-
ond 90-degree pulse, a steady-state longitudinal magnetization produces the same
FID amplitude from each subsequent 90-degree pulse (spins are rotated through
360 degrees and are reintroduced in the transverse plane). Tissues become partially
saturated (i.e., the full transverse magnetization is decreased from the equilibrium
magnetization), with the amount of saturation dependent on the Tl relaxation
time. A short-Tl tissue has less saturation than a long-Tl tissue, as illustrated in
Fig. 14-19. For spin echo sequences, partial saturation of the longitudinal magne-
tization depends on the TR and Tl of the tissues. Partial saturation has an impact
on tissue contrast.
 --_. unsaturated

partially saturated

::t )::< ):':(, -..c ¥

TR TR TR TR TR
90· 90· 90· 90· 90· 90·

FIGURE 14-19. Partial saturation of tissues occurs because the repetition time between
excitation pulses does not completely allow for full return to equilibrium, and the longi-
tudinal magnetization (Mz) amplitude for the next radiofrequency pulse is reduced. After
the first excitation pulse, a steady-state equilibrium is reached, wherein the tissues become
partially saturated and produce a constant transverse magnetization of less amplitude. Tis-
sues with long T1 experience a greater saturation than do tissues with short n.

Contrast in an image is proportional to the difference in signal intensity between

adjacent pixels in the image, corresponding to two different voxels in the patient.
The details of how images are produced spatially in MRI are discussed in Chapter
15. However, the topic of contrast can be discussed here by understanding how the
signal changes for different tissue types and for different pulse sequences. The sig-
nal, S, produced by an NMR system is proportional to other factors as follows:
Sex: PHf(v)[1 - e-TRlTI] e-TEIT2

where PH is the spin (proton) density,f(v) is the signal arising from fluid flow (dis-
cussed in Chapter 15), Tl and T2 are physical properties of tissue, and TR and TE
are pulse sequence controls on the MRI machine (Fig. 14-20).

TR
_-~---
I- _.~ •..
(~
TE
I- -I
90° 180° 90°

RF
~ ~ (~
pulses

Signal
out
------+-(---
FIGURE 14-20. Spin echo pulse sequence timing is initiated with a gO-degree pulse, fol-
lowed by a 180-degree pulse applied at time TE/2. The peak echo signal occurs at the
echo time, TE. The signal is acquired during the evolution and decay of the echo. After a
delay time allowing recovery of Mz, (the end of the TR period) the sequence is repeated
many times (e.g., 128 to 256 times) to acquire the information needed to build an image.
The repetition time, TR, is the time between gO-degree initiation pulses.
 The equation shows that for the same values ofTR and TE (i.e., for the same
pulse sequence), different values ofTl or T2 (or of PH or f(v)) will change the sig-
nal S. The signal in adjacent voxels will be different when Tl or T2 changes
between those two voxels, and this is the essence of how contrast is formed in MR!.
Importantly, by changing the pulse sequence parameters TR and TE, the contrast
dependence in the image can be weighted toward Tl or toward T2.

11 Weighting
A "Tl-weighted" spin echo sequence is designed to produce contrast chiefly based
on the Tl characteristics of tissues by de-emphasizing T2 contributions. This is
achieved with the use of a relatively short TR to maximize the differences in longi-
tudinal magnetization during the return to equilibrium, and a short TE to mini-
mize T2 dependency during signal acquisition. In the longitudinal recovery and
transverse decay diagram (Fig. 14-21A), note that the TR time on the abscissa of

Image
Longitudinal recovery (T1)
Mz intensity
1 ,----- -..--------------

2000 3000 4000 5000

Time (ms)

FIGURE 14-21. A: Longitudinal recovery (left)

and transverse decay (right) diagrams (with
quite different time scales) show four brain tis-
sues and their corresponding T1 and T2 relax-
ation constants. T1-weighted contrast requires
the selection of a TR that emphasizes the differ-
ences in the T1 characteristics of the tissues (e.g.,
TR = 500 msec) and reduces the T2 characteristics
(e.g., TE :0; 15 msec). B: The T1-weighted spin
echo axial brain image obtained with TR = 549
msec and TE = 11 msec demonstrates bright
image intensity for short-T1 tissues (white mat-
ter and fat) and dark intensity for long-T1 tis-
sues (cerebrospinal fluid).
the figure on the left (longitudinal recovery) intersects the individual tissue curves
and projects over to the figure on the right (transverse decay). These values repre-
sent the amount of magnetization that is available to produce the transverse signal,
and therefore the individual tissue curves on right-hand figure start at this point at
time T = O.The horizontal projections (arrows) graphically demonstrate how the TI
values modulate the overall MRI signal. When TR is chosen to be 400 to 600 msec,
the difference in longitudinal magnetization relaxation times (Tl) between tissues
is emphasized. Four common cerebral tissues-fat, white matter, gray matter,
CSF-are shown in the diagrams. The amount of transverse magnetization (which
gives rise to a measurable signal) after the 90-degree pulse depends on the amount
of longitudinal recovery that has occurred in the tissue of the excited sample. Fat,
with a short TI, has a large signal, because the short TI value allows rapid recovery
of the Mz vector. The short TI value means that the spins rapidly reassume their
equilibrium conditions. White and gray matter have intermediate TI values, and
CSF, with a long TI, has a small signal. For the transverse decay (T2) diagram in
Fig. 14-2IA, a 180-degree RF pulse applied at time TE/2 produces an echo at time
TE. A short TE preserves the TI signal differences with minimal transverse decay,
which reduces the signal dependence on T2. A long TE is counterproductive in
terms of emphasizing TI contrast, because the signal becomes corrupted with T2
decay. Tl-weighted images therefore require a short TR and a short TE for the spin
echo pulse sequence. A typical Tl-weighted axial image of the brain acquired with
TR = 500 msec and TE = 8 msec is illustrated in Fig. 14-21B. Fat is the most
intense signal (shortest Tl); white matter and gray matter have intermediate inten-
sities; and CSF has the lowest intensity (longest TI). A typical spin echo TI-
weighted image is acquired with a TR of about 400 to 600 msec and a TE of 5 to
20 msec.

Image contrast with spin density weighting relies mainly on differences in the num-
ber of magnetizable protons per volume of tissue. At thermal equilibrium, those tis-
sues with a greater spin density exhibit a larger longitudinal magnetization. Very
hydrogenous tissues such as lipids and fats have a high spin density compared with
proteinaceous soft tissues; aqueous tissues such as CSF also have a relatively high
spin density. Fig. 14-22A illustrates the longitudinal recovery and transverse decay
diagram for spin density weighting. To minimize the TI differences of the tissues,
a relatively long TR is used. This allows significant longitudinal recovery so that the
transverse magnetization differences are chiefly those resulting from variations in
spin density (CSF > fat> gray matter> white matter). Signal amplitude differences
in the FID are preserved with a short TE, so the influences of T2 differences are
minimized. Spin density-weighted images therefore require a long TR and a short
TE for the spin echo pulse sequence. Fig. 14-22B shows a spin density-weighted
image with TR = 2,400 msec and TE = 30 msec. Fat and CSF display as a relatively
bright signal, and a slight contrast inversion between white and gray matter occurs.
A typical spin density-weighted image has a TR between 2,000 and 3,500 msec and
a TE between 8 and 30 msec. This sequence achieves the highest overall signal and
the highest signal-to-noise ratio (SNR) for spin echo imaging; however, the image
contrast is relatively poor, and therefore the contrast-to-noise ratio is not necessar-
ily higher than with a TI- orT2-weighted image.
 Image
Longitudinal recovery (T1 )
intensity
Mz 1,

E'
~
~':1
0'7j
Fat .,._.-._.,;-.:,,:--."7':-
,Whit, , .
c:: . , ..••
=iij 0.6 ' /Gray ....·····
" " CSF
~ 0.5 . , .
~ 0.4
~
! /
I I
.'
...-

Qj o.3li / .
0:: 0.2 '! ,' .
" .
0.1 .

o
o 1000 2000 3000 4000 5000 200 300
TR Time (ms) Time (ms)

FIGURE 14-22. A: Proton density-weighted

contrast requires the use of a long TR (e.g., more
than 2,000 msec), to reduce T1, and a short TE
(e.g., less than 35 msec), to reduce T2 influence
in the acquired signals. B: The proton density-
weighted axial spin echo brain image, obtained
with TR = 2,400 msec and TE = 30 msec, shows
reduced contrast compared with the T1-
weighted images (but an overall higher signal
amplitude). Tissues with higher spin density
(e.g., fat, (SF) have higher image intensity.

T2 Weighting
T2 weighting follows directly from the spin density weighting sequence: Reduce Tl
effects with a long TR, and accentuate T2 differences with a longer TE. The T2-
weighted signal is usually the second echo (produced by a second ISO-degree pulse)
of a long- TR spin echo pulse sequence (the first echo is spin density weighted, with
short TE, as explained in the previous section). Generation ofT2 contrast differ-
ences is shown in Fig. 14-23A. Compared with a Tl-weighted image, inversion of
tissue contrast occurs (CSF is brighter than fat instead of darker), because short- Tl
tissues usually have a short T2, and long-Tl tissues have a long T2. Tissues with a
longT2 (e.g., CSF) maintain transverse magnetization longer than short-T2 tissues,
and thus result in higher signal intensity. A T2-weighted image (Fig. 14-23B),
demonstrates the contrast inversion and high tissue contrast features, compared
with the Tl-weighted image (Fig.14-21B). As TE is increased, more T2 contrast is
 Image
intensit)

Gray
White
Fat

1000 2000 3000 4000 5000 300

Time (ms) Time (ms)

FIGURE 14-23. A: T2-weighted contrast

requires the use of a long TR (e.g., more than
2,000 msec), to reduce T1 influences, and a long
TE (e.g., more than 80 msec), to allow for T2
decay differences to evolve. B: The T2-weighted
spin echo axial brain image, obtained with TR =
2,400 msec and TE = 90 msec, has bright image
intensity for long- T2 tissues such as cere-
brospinal fluid and dark intensity for short-T2
tissues such as white matter and fat.

achieved, at the expense of a reduced transverse magnetization signal. Even with low
signal, window width and window level adjustments (see Chapter 4) remap the sig-
nals over the full range of the display, so that the overall perceived intensity is sim-
ilar for all images. The typical T2-weighted sequence uses a TR of approximately
2,000 to 4,000 msec and a TE of 80 to 120 msec.

Spin Echo Parameters

Table 14-5 lists the typical contrast weighting values ofTR and TE for spin echo
imaging. For situations intermediate to those listed, mixed-contrast images are the
likely outcome. For conventional spin echo sequences, both a spin density and a
T2-weighted contrast signal are acquired during each TR by acquiring two echoes
with a short TE and a long TE (Fig. 14-24).
TABLE 14-5. SPIN ECHO PULSE SEQUENCE CONTRAST
WEIGHTING PARAMETERS •.

TR (msec) 400-600 1,500-3,500 1,500-3,500

TE (msec) 5-30 5-30 60-150

I, TR
0 10 ms 45 ms 2500 ms

I I 1 1

0
90
0
180 0
180 900

TE,= 20 ms TE2= 90 ms
,I ,I

FIGURE 14-24. Multiple echo pulse sequence can produce spin density and T2-weighted
contrast from the first and second echoes during the same TR. In this diagram, TEl = 20
msec and TE2 = 90 msec.

Inversion recovery, also known as inversion recovery spin echo, emphasizes Tl

relaxation times of the tissues by extending the amplitude of the longitudinal recov-
ery by a factor of two. An initial ISO-degree RF pulse inverts the Mz longitudinal
magnetization of the tissues to -Mz• Mter a delay time known as the time a/inver-
sion (TI), a 90-degree RF pulse rotates the recovered fraction of Mz spins into the
transverse plane to generate the FID. A second ISO-degree pulse at time TE/2 pro-
duces an echo signal atTE (Fig. 14-25). TR in this case is the time between lBO-degree
initiation pulses. As with the spin echo pulse sequence, a TR less than about 5 X TI
of the tissues causes a partial saturation of the spins and a steady-state equilibrium
of the longitudinal magnetization after the first three to four excitations in the
sequence. The echo amplitude of a given tissue depends on TI, TE, TR, and the
magnitude (positive or negative) of Mz•
The signal intensity at a location (x,y) in the image matrix is approximated as
follows:

5 DC PH/(V)(l - 2e-TIITJ + e-TRlTl)

where j(v) is a function of motion (e.g., blood flow) and the factor of 2 arises from
the longitudinal magnetization recovery from -Mz to Mz. Note the lack of T2
dependence in this approximation of S (in fact, there is some T2 decay.) The time
 TR
TE

90 0
180 0

readout refocusing pulse

RF
pulses
~~~~~~--n~
Signal
out

FIGURE 14-25. Inversion recovery spin echo sequence begins with a 1aO-degree "exci-
tation" pulse to invert the longitudinal magnetization, an inversion time (TI) delay,
and then a gO-degree "readout" pulse to convert recovered Mz into Mxy• A refocusing
1aO-degree pulse creates the echo at time TE for signal acquisition, and then the
sequence is repeated after a time TR between 1aO-degree excitation pulses.

of inversion, TI, controls the contrast between tissues (Fig. 14-26). TheTE must be
kept shon to avoid mixed-contrast images and to minimize T2 dependency. TR
determines the amount of longitudinal recovery between 180-degree excitation
pulses.
The inversion recovery sequence produces "negative" longitudinal magnetiza-
tion that results in negative (in phase) or positive (out of phase) transverse magne-

Image
intensity
Mxy
~-----
. -' -' -' :: •• -~~ =./ -' -
;·;!!~

....
I
Fat
---

White
--~
Gray

_ .. ~

Signal intensity is dependent on T1 (with short TE)

TI Contrast is dependent on TI
~t
-Mz
180
pulse
0
l 90
pulse
0

Inversion Time

FIGURE 14-26. The inversion recovery longitudinal recovery diagram shows the 2x Mz range pro-
vided by the 1aO-degree excitation pulse. A gO-degree readout pulse at time TI and a 1aO-degree
refocusing pulse at time TE/2 from the readout pulse form the echo at time TE. The time scale is not
explicitly indicated on the x-axis, nor is the 1aD-degree refocusing pulse. A short TE is used to reduce
T2 contrast characteristics.
tization when a short TI is used. The actual signals are encoded in one of two ways:
(a) a magnitude (absolute value) signal flips negative Mz values to positive values,
or (b) a phase signal preserves the full range of the longitudinal magnetization from
-Mz to +M" and the relative signal amplitude is displayed from dark to bright
across the dynamic range of the image display. The first (magnitude) method is
used most often, for reasons described later.

Short Tau Inversion Recovery

Short tau inversion recovery, or STIR, is a pulse sequence that uses a very short TI
and magnitude signal processing (Fig. 14-27A). In this sequence, there are two
interesting outcomes: (a) materials with short Tl have a lower signal intensity (the
reverse of a standard Tl-weighted image), and (b) all tissues pass through zero
amplitude (Mz = 0), known as the bounce point or tissue null. Because the time
required to reach the bounce point is known for a variety of tissue types, judicious
TI selection can suppress a given tissue signal (e.g., fat).
Analysis of the equations describing longitudinal recovery with this excita-
tion reveals that the signal null (Mz = 0) occurs at TI = In(2) X Tl, where In is
the natural log and In(2) = 0.693. For fat suppression, the null occurs at a time
equal to 0.693 X 260 msec = 180 msec (where Tl "" 260 msec at 1.5 T). A typi-
cal STIR sequence uses a TI of 140 to 180 msec and a TR of2,500 msec. Com-
pared with a Tl-weighted examination, STIR "fat suppression" reduces distract-
ing fat signals (see Fig. 14-27B) and eliminates chemical shift artifacts (explained
in Chapter 15).

Fluid Attenuated Inversion Recovery

The use of a longer TI reduces the signal levels of CSF and other tissues with long
Tl relaxation constants, which can be overwhelming in the magnitude inversion
recovery image. Fluid attenuated inversion recovery, the FLAIR sequence, reduces
CSF signal and other water-bound anatomy in the MR image by using a TI selected
at or near the bounce point of CSF. This permits a better evaluation of the anatomy
that has a significant fraction of fluids with long Tl relaxation constants. Nulling
of the CSF signal requires TI = In(2) X 3,500 msec, or about 2,400 msec. A TR of
7,000 msec or longer is typically employed to allow reasonable Mz recovery (see Fig.
14-27C).

MR contrast schemes for clinical imaging use the spin echo or a variant of the spin
echo pulse sequences for most examinations. Comparisons of the contrast charac-
teristics are essential for accurate differential diagnoses and determination of the
extent of pathology. This is illustrated with brain images obtained with Tl, proton
density, T2, and FLAIR contrast sequences (Fig. 14-28).
 Transverse decay (T2) Image
intensit~

FIGURE 14-27. A: The short tau inversion

recovery (STIR) and fluid attenuation inversion
recovery (FLAIR) techniques use inversion recov-
ery with magnitude longitudinal magnetization
signals. This allows tissues to be nulled at the
"bounce point" and is a method by which to
apply selective tissue suppression based on
inversion time (TI) and T1 values. Note the
reduced fat signal for STIRand the reduced cere-
brospinal fluid «(SF) signal for FLAIR. B: A STIR
sequence (TR = 5,520 msec, TI = 150 msec, TE =
27.8 msec) and comparison T1 sequence (TR =
750 msec, TE = 13.1 msec) shows the elimination
of fat signal. C: A FLAIR sequence (TR = 10,000
= =
msec, TI 2,400 msec, TE 150 msec) shows the
suppression of (SF in the image, caused by selec-
tion of a TI with which (SF is de-emphasized.
FIGURE 14-28. Magnetic resonance sequences produce variations in contrast for delineation of
anatomy and pathology, including n, proton density (PD), T2, and fluid attenuation inversion recov-
ery (FLAIR) sequences. The images illustrate the characteristics of n with high fat signal; PD with
high signal-to-noise ratio (SNR) yet low contrast discrimination; T2 with high signal and contrast for
long-T2 tissues; and FLAIR, with fluid-containing tissues clearly delineated as a low signal. All images
assist in the differential diagnosis of disease processes.

The gradient recalled echo (GRE) technique uses a magnetic field gradient to
induce the formation of an echo, instead of the ISO-degree pulses discussed ear-
lier. A gradient magnetic field changes the local magnetic field to slightly higher
and slightly lower strength and therefore causes the proton frequencies under its
influence to precess at slightly higher and lower frequencies along the direction of
the applied gradient. For an FID signal generated under a linear gradient, the
transverse magnetization dephases rapidly as the gradient is continually applied.
If the gradient polarity is instantaneously reversed after a predetermined time, the
spins will rephase and produce a gradient echo; its peak amplitude occurs when the
opposite gradient (of equal strength) has been applied for the same time as the
initial gradient. Continually applying this latter (opposite) gradient allows the
echo to decay, during which time the signal can be acquired (Fig. 14-29).
The GRE is not a true spin echo technique but a purposeful dephasing and
rephasing of the FID. Magnetic field (Bo) inhomogeneities and tissue susceptibili-
ties (magnetic field inhomogeneities caused by paramagnetic or diamagnetic tis-
sues or contrast agents) are emphasized in GRE, because the dephasing and
rephasing occur in the same direction as the main magnetic field and do not can-
cel the inhomogeneity effects, as the ISO-degree refocusing RF pulse does. Fig-
ures 14-17 and 14-29 (the rotating frame diagram) may be compared to illustrate
the direction of the Mxy vector of the FID and the echo. With spin echo tech-
niques, the FID and echo vectors form in opposite directions, whereas with GRE
they form in the same direction. Unlike the image produced by spin echo tech-
niques, the GRE sequence has a significant sensitivity to field nonuniformity and
magnetic susceptibility agents. The decay of transverse magnetization is therefore
a strong function ofT2*, which is much shorter than the "true" T2 seen in spin
echo sequences. The "echo" time is controlled either inserting a time delay
between the negative and positive gradients or by reducing the amplitude of the
 Rotating frame
../

FIGURE 14-29. Gradient imaging techniques use a magnetic field gradient instead of a 180-
degree radiofrequency pulse to induce the formation of an "echo." Dephasing the transverse mag-
netization spins with an applied gradient of one polarity and rephasing the spins with the gradient
reversed in polarity produces a "gradient recalled echo." Note that the rotating frame depicts the
magnetic moment vector of the echo in the same direction as the free induction decay (FID) relative
to the main magnetic field, and therefore extrinsic inhomogeneities are not cancelled.

reversed (usually positive) gradient, thereby extending the time for the rephasing
process to occur.
A major variable determining tissue contrast in GRE sequences is the flip angle.
Depending on the desired contrast, flip angles of a few degrees to more than 90
degrees are applied. With a short TR, smaller flip angles are used because there is
less time to excite the spin system, and more transverse magnetization is actually

Qi c
""0 0
:J ._
~m
C1.N
E+::o
(1l (])
c
~~
OlE
'00 (])
""0 (/)
•...
(])
N (])
.- >
- (/)
(1l C
E (1l
o.=:
z
0.0 0.0
0.0 1.0 1.5 2.0 0.0
TR (sec)

FIGURE 14-30. Transverse magnetization as a function of TR and the flip angle. For small flip
angles and short TR, the transverse magnetization is actually higher than for larger flip angles.
generated with smaller flip angles than with a larger flip angle. This occurs because
magnetization quickly builds up in the tissue sample. Figure 14-30 shows a plot of
the signal amplitude of the transverse magnetization versus TR as a function of flip
angle. Notice the lower signal amplitude for the larger flip angles with TR less than
200 msec. Tissue contrast in GRE pulse sequences depends on TR, TE, and the flip
angle, as discussed in the following sections.

Gradient Recalled Echo Sequence with Long TR

(>200 msec)

For the GRE sequence with "long" TR (>200 msec) and flip angles greater than 45
degrees, contrast behavior is similar to that of spin echo sequences. The major dif-
ference is the generation of signals based on T2* contrast rather than on T2 con-
trast, because the gradient echo does not cancel Bo inhomogeneities, as a spin echo
technique does. In terms of clinical interpretation, the mechanisms ofT2* contrast
are different from those of T2, particularly for MRI contrast agents. A long TE
tends to show the diffirences between T2* and T2 rather than the improvement of
T2 contrast, as would be expected in a spin echo sequence. Tl weighting is achieved
with a short TE (less than 15 msec).
For flip angles less than 30 degrees, the small transverse magnetization ampli-
tude reduces the Tl differences among the tissues. Proton density differences are
the major contrast attributes for short TE, whereas longer TE provides T2*
weighting. In most situations, GRE imaging is not useful with the relatively long
TR, except when contrast produced by magnetic susceptibility differences is
desired.

Steady-state precession refers to equilibration of the longitudinal and transverse

magnetization from pulse to pulse in an image acquisition sequence. In all stan-
dard imaging techniques (e.g., spin echo, inversion recovery, generic GRE
sequences), the repetition period TR of RF pulses is too short to allow recovery
of longitudinal magnetization equilibrium, and thus partial saturation occurs.
When this steady-state partial saturation occurs, the same longitudinal magneti-
zation is present for each subsequent pulse. For very short TR, less than the T2*
decay constant, persistent transverse magnetization also occurs. During each pulse
sequence repetition a portion of the transverse magnetization is converted to
longitudinal magnetization and a portion of the longitudinal magnetization is
converted to transverse magnetization. Under these conditions, steady-state lon-
gitudinal and transverse magnetization components coexist at all times in a
dynamic equilibrium. Acronyms for this situation include GRASS (gradient
recalled acquisition in the steady state), FISP (fast imaging with steady-state pre-
cession), FAST (Fourier acquired steady state), and other acronyms given by the
MRI equipment manufacturers.
The user-defined factors that determine the amount of transverse magnetiza-
tion created by each RF pulse and the image contrast in steady-state imaging
include TR, TE, and flip angle. Steady-state imaging is practical only with short and
 very shortTR. In these two regimes, flip angle has the major impact on the contrast
"weighting" of the resultant images.

Small Flip Angles. With small flip angles from 5 to 30 degrees, the amplitude of
longitudinal magnetization is very small, resulting in a very small transverse mag-
netization. Therefore, Tl relaxation differences between tissues are small. For the
same reasons, T2* decay differences are small, particularly with a short TE. Spin
density weighting is the single most important factor for short TR, small flip angle
techniques.

Moderate Flip Angles. With moderate flip angles of 30 to 60 degrees, a larger lon-
gitudinal magnetization produces a larger transverse magnetization, and tissues with
a long Tl have more signal saturation. This produces some Tl weighting. For the
same reason, tissues with long T2* generate a higher signal amplitude. Because Tl
and T2 for most tissues are correlated (i.e., long Tl implies long T2), the contrast
depends on the difference in T2/Tl ratio between the tissues. For example,
parenchyma tissue contrast is reduced (e.g., white versus gray matter), because the
T2/Tl ratios are approximately equal. On the other hand, the T2/Tl ratio for CSF
is relatively high, which generates a large signal difference compared with the
parenchyma and produces a larger signal.

Large Flip Angles. For flip angles in the range of 75 to 90 degrees, almost com-
plete saturation of the longitudinal magnetization occurs, reducing the Tl tissue
contrast. Conversely, a large transverse steady-state magnetization produces greater
T2* weighting and reduces spin density differences. The contrast for short TR, large
flip angle techniques depends on T2* and Tl weighting.
The echo time, TE, also influences the contrast. The descriptions of flip angle
influence assume a very short TE (e.g., less than 5 msec). As TE increases, a greater
difference in T2* decay constants occurs and contributes to the T2* contrast in the
Image.

Steady-State Gradient Recalled Echo Contrast Weighting

Table 14-6 indicates typical parameter values for contrast desired in steady state
and GRE acquisitions. The steady-state sequences produce T2* and spin density
contrast. A GRASS/FISP sequence using TR = 35 msec, TE = 3 msec, and flip
angle = 20 degrees shows unremarkable contrast, but blood flow shows up as a

Flip angle (degrees) 45-90 30-50 5-15 5-15 5-30

TR (msec) 200-400 10-50 200-400 100-300 100-300
TE (msec) 3-15 3-15 30-50 10-20 5-15
FIGURE 14-31. A steady-state gradient recalled echo (GRASS) sequence (TR = 24 msec,
TE = 4.7 msec, flip angle = 50 degrees) of two slices out of a volume acquisition is shown.
Contrast is unremarkable for white and gray matter because of a T2fT1 weighting depen-
dence. Blood flow appears as a relatively bright signal. MR angiography depends on
pulse sequences such as these to reduce the contrast of the anatomy relative to the vas-
culature.

bright signal (Fig. 14-31). This technique enhances vascular contrast, from which
MR angiography sequences can be reconstructed (see Chapter 15).

With very short TR steady-state acquisitions, Tl weighting cannot be achieved to

any great extent, owing to either a small difference in longitudinal magnetization
with small flip angles, or dominance of the T2* effects for larger flip angles pro-
duced by persistent transverse magnetization. The T2* influence can be reduced
by using a long TR (usually not an option), or by "spoiling" (destroying) the
steady-state transverse magnetization by adding a phase shift to successive RF
pulses during the acquisition. (Both the transmitter and receiver must be phase
locked for this approach to work.) By shifting the residual transverse components
out of phase, the buildup of the transverse steady-state signal does not occur,
effectively eliminating the T2* dependency of the signal. Although small T2*
contributions are introduced by the gradient reversal, mostly Tl-weighted contrast
is obtained. Short TR, short TE, moderate to large flip angle, and spoiled trans-
verse magnetization produces the greatest Tl contrast. Currently, spoiled trans-
verse magnetization gradient recalled echo (SPGR) is used in three-dimensional
volume acquisitions because of the extremely short acquisition time allowed by
the short TR of the GRE sequence and the good contrast rendition of the
anatomy provided by Tl weighting (Fig. 14-32).
MR contrast agents (e.g., gadolinium) produce greater contrast with Tl-
weighted spoiled GRE techniques than with a comparable Tl-weighted spin echo
sequence because of the greater sensitivity to magnetic susceptibility of the con-
trast agent. The downside of spoiled GRE techniques is the increased sensitivity
to other artifacts such as chemical shift and magnetic field inhomogeneities, as
well as a lower SNR.
 FIGURE 14-32. A spoiled transverse magnetization gradient recalled echo
(SPGR)sequence (TR = 8 msec, TE = 1.9 msec, flip angle = 20 degrees) of two
slices out of a volume acquisition is shown. The ability to achieve T1 contrast
weighting is extremely useful for rapid three-dimensional volume imaging.
Bright blood (lower portion of each image) and magnetic susceptibility arti-
facts are characteristic of this sequence.

Moving fluid (vascular and CSF) has an appearance in MR images that is compli-
cated by many factors, including flow velocity, vessel orientation, laminar versus
turbulent flow patterns, pulse sequences, and image acquisition modes. Flow-
related mechanisms combine with image acquisition parameters to alter contrast.
Signal due to flow covers the entire gray scale of MR signal intensities, from "black-
blood" to "bright-blood" levels, and flow can be a source of artifacts. The signal
from flow can also be exploited to produce MR angiographic images.
Low signal intensities (flow voids) are often a result of high-velocity signal loss
(HVSL), in which protons in the flowing blood move out of the slice during echo
reformation, causing a lower signal (see discussion of slice selection methods in
Chapter 15). Flow turbulence can also cause flow voids, by causing a dephasing of
spins in the blood with a resulting loss of the magnetic moment in the area of tur-
bulence. With HVSL, the amount of signal loss depends on the velocity of the mov-
ing fluid. Pulse sequences to produce "black blood" in images can be very useful in
cardiac and vascular imaging. A typical black-blood pulse sequence uses a "double
inversion recovery" method, whereby a nonselective ISO-degree RF pulse is initially
applied, inverting all spins in the body, and is followed by a selective ISO-degree RF
pulse that restores the magnetization in the selected slice. During the inversion
time, blood outside of the excited slice with inverted spins flows into the slice, pro-
ducing no signal; therefore, the blood appears dark.
Flow-related enhancement is a process that causes increased signal intensity due
to flowing spins; it occurs during imaging of a volume of tissues (see Chapter 15).
Even-echo rephasing is a phenomenon that causes flow to exhibit increased signal
on even echoes in a multiple-echo image acquisition. Flowing spins that experience
two subsequent ISO-degree pulses (even echoes) generate higher signal intensity
due to a constructive rephasing of spins during echo formation. This effect is
prominent in slow laminar flow (e.g., veins show up as bright structures on even-
echo images).
Flow enhancement in gradient echo images is pronounced for both venous and
arterial structures, as well as CSF. The high intensity is caused by the wash-in
(between subsequent RF excitations) of fully unsaturated spins (with full magneti-
zation) into a volume of partially saturated spins caused by the short TR used with
gradient imaging. During the next excitation, the signal amplitude resulting from
the moving unsaturated spins is about IO times greater than that of the nonmoving
saturated spins. With gradient echo techniques, the degree of enhancement depends
on the velocity of the blood, the slice or volume thickness, and the TR. As blood
velocity increases, unsaturated blood exhibits the greatest signal. Similarly, a thin-
ner slice or decreased repetition time results in higher flow enhancement. In arter-
ial imaging of high-velocity flow, it is possible to have bright blood throughout the
imaging volume of a three-dimensional acquisition if unsaturated blood can pene-
trate into the volume without experiencing an RF pulse.

PerfUsion of tissues via the capillary bed permit the delivery of oxygen and nutrients to
the cells and removal of waste (e.g., CO2) from the cells. Conventional perfusion mea-
surements are based on the uptake and wash-out of radioactive tracers or other exoge-
nous tracers that can be quantified from image sequences using well-characterized
imaging equipment and calibration procedures. For MR perfusion images, exoge-
nous and endogenous tracer methods are used. Freely diffusible tracers using nuclei
such as 2H (deuterium), 3He, 170, and 19Fare employed in experimental procedures
to produce differential contrast in the tissues. More clinically relevant are intravas-
cular blood-pool agents such as gadolinium-diethylenetriaminepentaacetic acid
(Gd-DTPA), which modify the relaxation of protons in the blood in addition to
producing a shorter T2*. Endogenous tracer methods do not use externally added
agents, but instead depend on the ability to generate contrast from specific excita-
tion or diffusion mechanisms. For example, labeling of inflowing spins ("black-
blood" perfusion) uses protons in the blood as the contrast agent. Tagged
(labeled) spins outside of the imaging volume perfuse into the tissues, resulting in
a drop in signal intensity, a time course of events that can be monitored by quan-
titative measurements.
Blood oxygen level-dependent (BOLD) and "functional MR" imaging rely on
the differential contrast generated by blood metabolism in active areas of the brain.
In flow of blood and conversion of oxyhemoglobin to deoxyhemoglobin (a para-
magnetic agent) increases the magnetic susceptibility in the localized area and
induces signal loss by reducing T2*. This fact allows areas of high metabolic activ-
ity to produce a correlated signal and is the basis of functional MRI (fMRI). A
BOLD sequence produces an image of the head before the application of the stim-
ulus. The patient is subjected to the stimulus and a second BOLD image is
acquired. Because the BOLD sequence produces images that are highly dependent
on blood oxygen levels, areas of high metabolic activity will be enhanced when the
prestimulus image is subtracted, pixel by pixel, from the poststimulus image. These
fMRI experiments determine which sites in the brain are used for processing data.
Stimuli in fMRI experiments can be physical (finger movement), sensory (light
flashes or sounds), or cognitive (repetition of "good" or "bad" word sequences). To
improve the SNR in the fMRI images, a stimulus is typically applied in a repetitive,
periodic sequence, and BOLD images are acquired continuously. Areas in the brain
that demonstrate time-dependent activity and correlate with the time-dependent
application of the stimulus are coded using a color scale and are overlaid onto a
gray-scale image of the brain for anatomic reference.
Diffusion depends on the random motion of water molecules in tissues. Inter-
action of the local cellular structure with the diffusing water molecules produces
anisotropic, directionally dependent diffusion (e.g., in the white matter of brain
tissues). Diffusion-weighted sequences use a strong MR gradient (see Chapter 15)
applied to the tissues to produce signal differences based on the mobility and direc-
tionality of water diffusion. Tissues with more water mobility have a greater signal

FIGURE 14-33. Diffusion-

weighted image contrast illus-
trates gray scale-encoded dif-
fusion coefficients of the
brain. (This image corresponds
to the images in Figs. 14-21
through 14-23 and 14-27C).
loss than those of lesser water mobility. The in vivo structural integrity of certain
tissues (healthy, diseased, or injured) can be measured by the use of diffusion-
weighted imaging (DWl), in which water diffusion characteristics are determined
through the generation of apparent diffusion coefficient (ADC) maps (Fig. 14-33).
ADC maps of the spine and the spinal cord have shown promise in predicting and
evaluating pathophysiology before it is visible on conventional Tl- orT2-weighted
images. DWl is also a sensitive indicator for early detection of ischemic injury. For
instance, areas of acute stroke show a drastic reduction in the diffusion coefficient
compared with nonischemic tissues. Diagnosis of multiple sclerosis is a potential
application, based on the measurements of the diffusion coefficients in three-
dimensional space.
Various acquisition techniques are used to generate diffusion-weighted con-
trast. Standard spin-echo and echo planar image (EPr, see Chapter 15) pulse
sequences with diffusion gradients are common. Obstacles to diffusion weighting
are the extreme sensitivity to motion of the head and brain, which is chiefly caused
by the large pulsed gradients required for the diffusion preparation, and eddy cur-
rents, which reduce the effectiveness of the gradient fields. Several strategies have
been devised to overcome the motion sensitivity problem, including common elec-
trocardiographic gating, motion compensation methods, and eddy current com-
pensation.

Magnetization transfer contrast is the result of selective observation of the inter-

action between protons in free water molecules and protons contained in the
macromolecules of a protein. Magnetization exchange occurs between the two
proton groups because of coupling or chemical exchange. Because the protons
exist in slightly different magnetic environments, the selective saturation of the
protons in the macromolecule can be excited separately from the bulk water by
using narrow-band RF pulses (because the Larmor frequencies are different). A
transfer of the magnetization from the protons in the macromolecule partially sat-
urates the protons in bulk water, even though these protons have not experienced
an RF excitation pulse. Reduced signal from the adjacent free water protons by
the saturation "label" affects only those protons having a chemical exchange with the
macromolecules and improves image contrast in many situations (Fig. 14-34). This
technique is used for anatomic MR imaging of the heart, the eye, multiple scle-
rosis, knee cartilage, and general MR angiography. Tissue characterization is also
possible, because the image contrast in part is caused by the surface chemistry of
the macromolecule and the tissue-specific factors that affect the magnetization
transfer characteristics.
MR pulse sequences allow specific contrast differences between the tissues to
be generated by manipulating the magnetization of the tissues in well-defined ways.
The next chapter (Chapter 15) discusses the methods of signal acquisition and
image reconstruction, as well as image characteristics, artifacts, quality control, and
magnetic field safety and bioeffects.
412 Section II: Diagnostic Radiology

Macromolecule Hydration
Bulk Water
.
Layer .
I I H.,H
:~,H H.,H: 0
10 --I:
I )H.Ji
OHI H. /{ I H.,H
1 0 0H.,HI H.,H
1 0 1 0
IH.,HI-(,H I 0
~,HO H. n
H.,H
0
H.,H
0
A: ,H • 01I H.,H
)
H3~,H I 0
1 11. ,H0 H.I,H
H.,H
O~,H~ H.,H H.,H
IH.,H° I 0
0 0
10H.,H1
: H.,H O.
I 0 H.,H
:
I
»~,H H.,H
0
H.,H
0
IH.,HO I
Protons bound to 1 0 I
0

macromolecule I I

RF Excitation
(off-resonance pulse)

FIGURE 14-34. Magnetization transfer contrast is implemented with an off-

resonance radiofrequency pulse of about 1,500 Hz from the Larmor fre-
quency. Excitation and dephasing of hydrogen atoms on the surface of macro-
molecules is transferred via the hydration layer to adjacent "free water"
hydrogen atoms. A partial saturation of the tissues reduces the signals that
would otherwise compete with signals from blood flow.

Axel L, et al. Glossary of MR terms, 3rd ed. Reston, VA: American College of Radiology, 1995.
Brown MA, Smelka RC. MR: basic principles and applications. New Yotk: John Wiley & Sons,
1995.
Hendrick RE. The AAPM/RSNA physics tutorial for residents. Basic physics of MR imaging: an
introduction. Radiographies 1994; 14:829-846.
NessAiver M. All you really need to know about MRl physics. Baltimore: Simply Physics, 1997.
Plewes DB. The AAPM/RSNA physics tutorial for residents. Contrast mechanisms in spin-echo
MR imaging. Radiographies 1994; 14: 1389-1404.
Price RR. The AAPM/RSNA physics tutorial for residents. Contrast mechanisms in gradient-
echo imaging and an introduction to fast imaging. Radiographies 1995;15:165-178.
Smith H], Ranallo FN. A non-mathematical approach to basic MR!. Madison, WI: Medical Physics
Publishing, 1989.
Smith RC, Lange RC. Understanding magnetic resonance imaging. Boca Raton, FL: CRC Press,
1998.
Wherli FW. Fast-scan magnetic resonance: principles and applications. New York: Raven Press,
1991.