MRI made easy 5

Let us start with a

general overview of MRI . . .
The single steps of an MRI examination can be described quite simply:

1. the patient is placed in a magnet,

2. a radio wave is sent in,
3. the radio wave is turned off,
4. the patient emits a signal, which is received and used for
5. reconstruction of the picture.
6 MRI made easy The proton

Fig. 1: Protons possess a positive charge. Like the earth, they are constantly
turning around an axis and have their own magnetic field.

Now, may be you remember from

Let’s take a look at your physics at school that an electri-
these steps in detail cal current induces, causes a magnetic
What happens, when we put a patient force, or magnetic field. So, where there
into the magnet of an MR machine? is an electrical current, there is also a
To understand this, it is necessary to magnetic ield.
at least know some very basic physics This can be demonstrated very easily.
– even though this may seem to be bor- Take a rusty nail and approach an elec-
ing. trical outlet – closer, closer. Do you feel it
As we all know, atoms consist of a being repelled by the magnetic force, so
nucleus and a shell, which is made up you do not put the nail into the outlet?
of electrons. In the nucleus – besides
other things – there are protons, little
particles, that have a positive electri-
cal charge (whatever that may actually
Let’s review what
be). These protons are analogous to lit- we have read
tle planets. Like the earth, they are con- A proton has a spin, and thus
stantly turning, or spinning around an the electrical charge of the
axis (igure 1); or – as one says, protons proton also moves. A moving
possess a spin. The positive electrical electrical charge is an electri-
charge, being attached to the proton, cal current, and this is accompanied by a
naturally spins around with it. And magnetic ield. Thus, the proton has its
what is a moving electrical charge? It is own magnetic ield and can be seen as a
an electrical current. little bar magnet (igure 1C).
 The proton MRI made easy 7

What happens to the of the external ield, parallel to it. Or

they may point in the completely op-
protons, when we put posite direction, anti-parallel. These
them into an external types of alignment are on different
energy levels. To explain this: a man
magnetic field? can align himself parallel to the mag-
The protons – being little magnets – netic ield of the earth, i.e. walk on his
align themselves in the external mag- feet, or he can align himself anti-par-
netic ield like a compass needle in the allel, in the opposite direction. Both
magnetic ield of the earth. However, states are on different energy levels,
there is an important difference. For i.e. they need different amounts of en-
the compass needle there is only one ergy.
way to align itself with the magnet- Walking on one’s feet is undoubt-
ic ield, for the protons, however, there edly less exhausting, takes less ener-
are two (igure 2): gy than walking on one’s hands. (In
The protons may align with their the igures, this will be illustrated as
South and North Poles in the direction pointing up or down, see igure 2).

Fig. 2: Normally, protons are aligned in a random fashion. This, however, changes when
they are exposed to a strong external magnetic field. Then they are aligned in only two
ways, either parallel or anti-parallel to the external magnetic field.
8 MRI made easy The proton

Fig. 3: When there

are two possible
states of alignment,
the one that takes
less energy, is on a
lower energy level,
is preferred.

Naturally, the preferred state of align- of the applied magnetic ield. To get a
ment is the one that needs less energy. rough idea: for about 10 million pro-
So more protons are on the lower ener- tons “walking on their hands”, there
gy level, parallel to the external mag- are about 10,000,007 “walking on their
netic ield, walking on their feet, so to feet”. The difference “007” is probably
speak. A smaller number is on the high- easy to remember, isn’t it?
er energy level, anti-parallel, “walking It may be obvious at this point al-
on their hands”. ready that for MRI the mobile protons
The difference in number is, however, are important (which are a subset of all
very small and depends on the strength protons that are in the body).
 The proton MRI made easy 9

this is a very fast movement, as we will

The movement of protons – precession
see below. For the sake of simplicity, we
will just make “freeze frame” pictures,
as if we were taking a fast lashlight
Let us take a closer photograph of the situation at a specif-
look at these protons ic moment in time.
We will see that the protons do not just For reasons we will learn later, it
lie there, aligned parallel or anti-paral- is important to know how fast the
lel to the magnetic ield lines. Instead, protons precess. This speed can be
they move around in a certain way. The measured as precession frequency,
type of movement is called precession that is how many times the pro-
(igure 4A). tons precess per second. This pre-
cession frequency is not constant.
It depends upon the strength of the
What type of movement magnetic field (for magnetic field
is “precession”? strength, see page 94), in which the
Just imagine a spinning top. When protons are placed.
you hit it, it starts to “wobble” or tum- The stronger the magnetic field,
ble around. It does not, however, fall the faster the precession rate and the
over. During the precession, the ax- higher the precession frequency.
is of the spinning top circles forming This is like a violin string: the strong-
a cone shape (igure 4B). It is hard to er the force exerted upon the string,
draw such a precessing proton, because the higher its frequency.

Fig. 4: A spinning top which is hit, performs a wobbling type of motion.

Protons in a strong magnetic field also show this type of motion, which is
called precession.
10 MRI made easy Precesssion

It is possible and necessary to precise- and anti-parallel. The state that needs
ly calculate this frequency. This is done less energy is preferred, and so there are
by using an equation called the Larmor a few more protons “walking on their
equation: feet” than “on their hands” (igure 3).

w 0 = γB 0 7 The protons precess along the ield

lines of the magnetic ield, just like a

spinning top that precesses along the
w0 is the precession frequency – in Hz ield lines of the magnetic ield of earth.
or MHz, 7 The precession frequency can be cal-

B0 is the strength of the external mag- culated by the Larmor equation, and is
netic ield, which is given in Tesla (T) higher in stronger magnetic ields. Why
(see page 94), and is this precession frequency important?
γ is the so-called gyro-magnetic ratio. It has something to do with the “reso-
nance” in magnetic resonance imaging.
The equation states that the precession But to understand this will take a few
frequency becomes higher when the more minutes.
magnetic ield strength increases. The After the break you should go over
exact relationship is determined by the this last summary again, and then con-
gyro-magnetic ratio γ. This gyro-mag- tinue . . .
netic ratio is different for different ma-
terials (e.g. the value for hydrogen pro-
tons is 42.5 MHz/T).
It can be compared to an exchange
Introducing the
rate, which is different for different coordinate system
currencies. To make communication (and drawing
of illustrations) easier, let us start using
a coordinate system like the one used
Time to take a break at school (igure 5). As you can see, the
However, let us briely re- z-axis runs in the direction of the mag-
view what we have learned netic ield lines, and thus can represent
up to now: them. So we can stop drawing the ex-
7 Protons have a positive ternal magnet in all other illustrations.
electrical charge, which is constantly From here on we will also illustrate
moving, because the protons possess a the protons as vectors, as little arrows.
spin. Maybe you remember: a vector rep-
7 This moving electrical charge is resents a certain force (by its size) that
nothing more than an electrical cur- acts in a certain direction (direction of
rent, and the latter always induces a the arrow). The force that is represent-
magnetic ield. ed by vectors in our illustrations, is the
7 So every proton has its own little magnetic force.
magnetic ield, and can thus be seen as
a little bar magnet.
7 When we put a patient in the MR

magnet, the protons, being little mag-

nets, align with the external magnetic
ield. They do this in two ways: parallel
 Precesssion MRI made easy 11

Fig. 5: Using
a coordinate
system makes
the description
of proton motion
in the magnetic
field easier, and
we can also stop
drawing the ex-
ternal magnet.
12 MRI made easy Precession

Fig. 6: The five protons, which “point” down, cancel out the magnetic effects of the same number of protons,
which “point” up (A). So in effect it is sufficient to look only at the four unopposed protons (B).

Now, let us look at igure 6. Here we tons in your body precessing this fast.
have 9 protons pointing up, precessing It is easy to imagine that at a certain
parallel to the external magnetic ield moment, there may be one proton (A in
lines, and 5 protons pointing down, the illustration) pointing in one direc-
precessing anti-parallel to the external tion, and another proton (A’) pointing
magnetic ield. exactly in the opposite direction. The
What we see in the igure is just a pic- result is very important; the magnetic
ture taken at a speciic point in time. A forces in the opposing directions cancel
picture taken just a little later would each other out, like two persons pulling
show the protons in different posi- at the opposite ends of a rope. Finally,
tions, because they precess. The preces- for every proton pointing down, there
sion actually goes very fast, the preces- is one pointing up, cancelling its mag-
sion frequency for hydrogen protons is netic effect. But as we have learned:
somewhere around 42 MHz in a mag- there are more protons pointing up
netic ield strength of 1 Tesla (see page than down, and the magnetic forces of
94); this means that the protons pre- these protons are not cancelled by oth-
cess around the “ice cream cone” more ers. So we are left – in effect – with some
than 42 million times per second. Now protons (4 in our example) pointing up
there are millions and millions of pro- (igure 6).
 Precession MRI made easy 13

Fig. 7: The magnetic force of

proton A, illustrated as an arrow,
a vector, can be seen as result-
ing from two components: one
pointing up along the z-axis, and
one in direction of the y-axis.
The component along the y-axis
is cancelled out by proton A’,
the magnetic force of which
also has a component along
the y-axis, but in the opposite
direction. The same holds true
for other protons, e.g. B and B’,
which cancel their respective
magnetic vectors along the x-
axis. In contrast to the magnetic
vectors in the x-y-plane, which
cancel each other out, the vec-
tors along the z-axis point in the
same direction, and thus add up
to a new magnetic sum vector
pointing up.

However, not only magnetic forces What we end up with in effect is a

pointing up and down can cancel or magnetic vector in the direction of the
neutralize each other. As the protons external magnetic ield (the orange ar-
that are pointing up, precess, there may row on the z-axis in igure 7); and this
be one pointing to the right, while an- vector is a sum vector made up by add-
other one is pointing to the left; or for ing the magnetic vectors of the protons
one pointing to the front, there is one pointing upwards.
pointing backwards, and so on (the Now – what does this mean? This
corresponding protons in igure 7 are means that by placing a patient in the
marked A and A’, B and B’ for example). magnet of the MR unit (or in any other
This means that the opposing magnetic strong magnetic ield), the patient him-
forces of the remaining protons cancel self becomes a magnet, i.e. has his own
each other out in these directions. This magnetic ield. Why? Because the vec-
is true for all but one direction, the di- tors of the protons that do not cancel
rection of the z-axis, along the external each other out, add up (igure 8).
magnetic ield (igure 7). In this direc- As this magnetization is longitudi-
tion, the single vectors, the single mag- nal to the external magnetic ield, it is
netic forces add up, like people pulling called longitudinal magnetization.
on the same end of a rope.
14 MRI made easy Magnetic vectors

As we have seen, the resulting new

magnetic vector of the patient points
in the direction of the external ield,
along its ield lines. This is described
as longitudinal direction. And it is ac-
tually this new magnetic vector that
may be used to get a signal. It would
be nice if we could measure this mag-
netization of the patient, but there is
a problem: we cannot measure this
magnetic force, as it is in the same di-
rection, parallel to the external mag-
netic ield (igures 7 and 8).
To illustrate this: Imagine that you
are sitting on a boat, loating down
a river. You have a water hose in your
hand and squirt water into the river.
For somebody who is watching you
from the shore, it is impossible to tell
how much water you pour out (i.e. how
much new magnetization is added in
the old direction).
However, when you point the water
hose at the shore, change the direction
of the new magnetic ield, then the wa-
ter may perhaps be directly picked up
and measured by an impartial observer
on the shore (igure 9). What we should
learn from this is: magnetization lon-
gitudinal to the external magnetic ield
cannot be measured directly. For this,
we need a magnetization which is not
longitudinal, but transversal to the ex-
ternal magnetic ield.

Fig. 8: In a strong external magnetic field, a new

magnetic vector is induced in the patient, who becomes a
magnet himself. This new magnetic vector is aligned
with the external magnetic field.
Magnetic vectors MRI made easy 15

Fig. 9: Magnetization along an

external magnetic field cannot be
measured. To achieve this, a mag-
netization transverse to the external
magnetic field is necessary.
16 MRI made easy Magnetic vectors

Time to take a break Sending in a radio wave

Before you walk away, let us

just sum up. And when you What happens after
come back, start out with
this summary again.
we put the patient into
7 Protons have a positive charge and the magnet?
possess a spin. Due to this, they have a We send in a radio wave. The term radio
magnetic field and can be seen as little wave is used to describe an electromag-
bar magnets. netic wave that is within the frequency
7 When we put them into a strong ex- range of the waves, which are received
ternal magnetic field, they align with by your radio. Well, you can imagine it
it, some parallel – pointing up –, some is not quite like this kind of radio wave.
anti-parallel – pointing down. What we actually send into the patient
7 The protons do not just lie there, but is not a wave of long duration, but a
precess around the magnetic field lines. short burst of some electromagnetic
And the stronger the magnetic field, wave, which is called a radio frequen-
the higher the precession frequency, a cy- or RF pulse. The purpose of this RF
relationship that is mathematically de- pulse is to disturb the protons, which
scribed in the Larmor equation. are peacefully precessing in alignment
7 Parallel and anti-parallel protons can with the external magnetic field.
cancel each others forces out. But as We will hear about the details later.
there are more parallel protons on the But not every RF pulse disturbs the
lower energy level (“pointing up”), we alignment of the protons: For this, we
are left with some protons, the mag- need a special RF pulse, one that can ex-
netic forces of which are not cancelled. change energy with the protons.
All of these protons pointing up, add up This is as if someone were looking
their forces in the direction of the ex- at you. You may not notice it, because
ternal magnetic field. And so when we there is no exchange of energy, so you
put the patient in the MR magnet, he do not change your position / align-
has his own magnetic field, which is ment. However, if someone were to
longitudinal to the external field of the pound you in the stomach, exchange
MR machine’s magnet (figures 7 and 8). energy with you, your alignment
Because it is longitudinal, however, it would be disturbed. And this may ex-
cannot be measured directly. plain why we need a certain RF pulse
that can exchange energy with the pro-
tons to change their alignment.
But when can an RF pulse exchange
energy with the protons?
For this, it must have the same fre-
quency; the same “speed” as the pro-
tons, so to speak. Just imagine that you
are driving down a race track on your
bike, and someone in the lane next to
you wants to hand you a hamburger,
i.e. exchange energy with you – as you
 Resonance MRI made easy 17

are hungry, the hamburger would give

you new energy.
This energy transfer is possible when
both bikers have the same speed, move
around the race track with the same
frequency:
With differences in speed/frequency …
… little or no energy transfer is possible.

Fig. 10:
Energy exchange is The “speed” of protons and resonance
possible when protons
and the radio frequency
pulse have the same
frequency.
What speed, or better,
what frequency did the
protons have?
They had their precession frequency
which can be calculated by the Larmor
equation (see page 10). So the Larmor
equation gives us the necessary fre-
quency of the RF pulse to send in. On-
ly when the RF pulse and the protons
have the same frequency, can protons
pick up some energy from the radio
wave, a phenomenon called resonance
– this is where the “resonance” in mag-
netic resonance comes from.
The term resonance can be illustrated
by the use of tuning forks. Imagine that
you are in a room with different kinds
of tuning forks, tuned e.g. to a, e, and d.
Somebody enters the room with a tun-
ing fork with “a”-frequency that was
struck to emit sound. Of all the tun-
ing forks in the room, all of a sudden
the other “a”-forks, and only those, pick
up energy, start to vibrate and to emit
sound, they show a phenomenon called
resonance.
18 MRI made easy RF pulse

What happens to the

protons when exposed
to the RF pulse?
Some of them pick up energy, and go
from a lower to a higher energy level.
Remember, some, which were walk-
ing on their feet, start walking on
their hands. And this has some effect
on the patient’s magnetization, as you
can see in figure 11. Let us assume that
from the net sum of 6 protons point-
ing up, after the RF pulse is sent in, 2
point down.
The result is that these 2 protons can-
cel out the magnetic forces of the same
number of protons, that point up.
So in effect then, the magnetization
in the longitudinal direction – being 6
before the RF pulse – decreases to 2.

What else happens

to the protons
when exposed to the
RF pulse?
As we have just learned, some of the
protons pick up energy, and go from a
lower to a higher energy level, thus de-
creasing magnetization in longitudi-
nal direction.
But something else happens. Do you
remember what drawings of radio
waves look like? Just look at figure 12;
they resemble a whip.
And what does a whip make the
bears do? It makes them walk in line,
in step, in synch – they are in phase.
Believe it or not, the RF pulse also Fig. 12: The drawing of
has a whip-like action – not on bears radio waves normally
resembles a whip, and radio
but on protons: When the protons ran- waves in MRI also have
domly point left/right, back/forth and a whip-like action.
so on, they also cancel their magnetic
forces in these directions (as we read
 RF pulse MRI made easy 19

Fig. 11: The radio

frequency pulse
exchanges energy
with the protons
(A), and some of
them are lifted to
a higher level of
energy, point-
ing downward in
the illustration
(B). In effect, the
magnetization
along the z-axis
decreases, as the
protons which point
down, “neutralize”
the same
number of protons
pointing up.

Fig. 13: The radio

wave has two effects
on the protons: it
lifts some protons
to a higher level of
energy (they point
down), and it also
causes the protons
to precess in step,
in phase. The former
results in decreasing
the magnetization
along the z-axis, the
so-called longitudinal
magnetization. The
latter establishes a
new magnetization
in the x-y-plane (J),
a new transversal
magnetization,
which moves around
with the precessing
protons.

on page 13). Due to the RF pulse, the add up in this direction. This results
protons do not point in random direc- in a magnetic vector pointing to the
tions any more, but move in step, in side to which the precessing protons
synch – they are “in phase”. They now point, and this is in a transverse direc-
point in the same direction at the same tion (figure 13). This is why it is called
time, and thus their magnetic vectors transversal magnetization.
20 MRI made easy RF pulse

Fig. 14: Protons

precessing in phase
cause a new transversal
magnetization.
 RF pulse MRI made easy 21

So – what were – Some protons pick up energy, start

to walk on their hands, and thus de-
the new things that crease the amount of longitudinal
we have learned? magnetization.
– AND: The protons get in synch, start
Repeat them using figure to precess in phase. Their vectors now
15. also add up in a direction transverse
7 When we put the patient to the external magnetic field, and
in the MR machine, a mag- thus a transversal magnetization is
netic field in the patient, longitudinal established.
to the external field, results.
7 Sending in an RF pulse that has the In summary: The RF pulse causes lon-
same frequency as the precessing fre- gitudinal magnetization to decrease,
quency of the protons causes two ef- and establishes a new transversal
fects: magnetization (figures 13 and 15).

Fig. 15: In a strong external magnetic field, a new magnetic vector along the external field is
established in the patient (A). Sending in an RF pulse causes a new transversal magnetiza-
tion while longitudinal magnetization decreases (B). Depending on the RF pulse, longitudinal
magnetization may even totally disappear (C).
22 MRI made easy The MR signal

ing magnetic field causes an electrical

The transversal vector – a closer look
current, e.g. in an antenna. And this
electrical current induced by the mov-
Let us have a look at ing magnetic field is the MRI signal. As
the transversal magnetic vector moves
the newly established around with the precessing protons, it
transversal magnetiza- moves with the precession frequency.
The resulting MR signal therefore also
tion vector. has the precession frequency (figure
This moves in phase with the precess- 16): But . . . how can we make a picture
ing protons (figure 16). The new mag- out of this electrical current, which is
netic vector comes towards you, goes actually our MR signal?
away from you, comes again towards For this we have to know, where in
you, and so on. the body the signal came from. How
And this is important: the magnetic can we know that? The trick is real-
vector, by constantly moving, constant- ly quite simple: we do not put the pa-
ly changing, induces an electric cur- tient into a magnetic field which has
rent. We have talked about the opposite the same strength all over the section
already: the moving electrical charge of of the patient, which we want to exam-
the proton, the electric current, induc- ine.
es the proton’s magnetic field. This al- Instead we take a magnetic field,
so is true the other way around: a mov- which has a different strength at each

Fig. 16: The new transver-

sal magnetization moves
around with the precessing
protons (see figure 13).
Thus for an external
observer, transversal
magnetization constantly
changes its direction, and
can induce a signal in an
antenna.
 The MR signal MRI made easy 23

signal from different locations also has

a different frequency. And by the fre-
quency we can assign a signal to a cer-
tain location.
Why make a It is like with your TV: when you are

complex image of my in the kitchen (where you probably do

not have a TV) and hear a sound from
husband? your favorite TV show, you know where
He’s actually quite the sound is coming from. It comes

simply structured! from the spot in your apartment where

the TV stands. What you subconscious-
ly do, is to connect a certain sound to a
certain location in space.
That is enough about spatial infor-
mation for now, we will go into more
detail about this on page 87.

Further details about

the MR signal
If our protons rotated around in synch,
in phase, and nothing changes, then we
would get a signal as illustrated in fig-
ure 16.
This, however, is not what happens.
As soon as the RF pulse is switched off,
the whole system, which was disturbed
by the RF pulse, goes back to its original
quiet, peaceful state, it relaxes.
The newly established transverse
magnetization starts to disappear – a
process called transversal relaxation,
and the longitudinal magnetization
grows back to its original size – a proc-
point of the patient's cross-section. ess called longitudinal relaxation.
What does this do? Why is that?
We have heard that the precession The reason why the longitudinal
frequency of a proton depends on the magnetization grows back to its nor-
strength of the magnetic field – as the mal size is easier to explain, so let us
frequency of a violin string depends on start with that (see figure 17).
the strength with which you pull it. No proton walks on its hands long­er
If this strength differs from point to than it has to – a sort of human trait.
point in the patient, then protons in The protons that were lifted onto a
different places precess with different higher energy level by the RF pulse go
frequencies. And as they precess with back to their lower energy level, and
different frequencies, the resulting MR start to walk on their feet again.
24 MRI made easy Longitudinal magnetization

Fig. 17: After the RF pulse is switched off, protons go back from their higher to the lower state of energy, i.e. point up again.
This is illustrated “one-by-one”. The effect is that longitudinal magnetization increases and grows back to its original value. Note that for
simplicity the protons were not depicted as being in phase: this subject is covered in more detail in figures 20 and 26.

After the RF pulse is switched off, pro- And this is why this process is not
tons go back to the lower state of ener- only called longitudinal relaxation, but
gy, i.e. point up again, but not all pro- also spin-lattice relaxation.
tons do this at exactly the same time. By going back on their feet, pointing
Instead it is a continuous process, as if upwards again, these protons no longer
one proton after the other goes back to cancel out the magnetic vectors of the
its original state. This is illustrated in same number of protons pointing up, as
figure 17 for a group of protons. The ef- they did before. So, the magnetization
fect is that longitudinal magnetization in this direction, the longitudinal mag-
increases and grows back to its original netization increases, and finally goes
value. back to its original value (figure 17).
What happens to the energy which If you plot the longitudinal mag­
they had picked up from the RF pulse? netization vs. time after the RF pulse is
This energy is just handed over to switched off, you get a curve like figure
their surroundings, the so-called lat- 18. It increases with time. This curve is
tice. also called a T1-curve.
 Longitudinal magnetization MRI made easy 25

Fig. 18: If one plots

the longitudinal
magnetization vs. time
after the RF pulse was
switched off, one gets
a so-called T1-curve.
26 MRI made easy Transversal magnetization

The time that it takes for the longitu-

dinal magnetization to recover, to go
back to its original value, is described
by the longitudinal relaxation time,
also called T1. This actually is not the
exact time it takes, but a time constant,
describing how fast this process goes.
This is like taking time for one circuit
round at a car race.
The time gives you an idea of how
long the race may take, but not the ex-
act time. Or more scientifically, T1 is a
time constant comparable to the time
constants that for example describe ra-
dioactive decay.
That T1 is the l ongitudinal relax-
ation time, can easily be remembered:
If you turn the “1” upside down, it
looks very much like an “l” as in longi-
tudinal.

About T2
Fig. 19: T1 is the longitudinal relaxation time
that has something to do with the exchange of
thermal energy. Enough about the longitu-
dinal magnetization – what
Reminding you also that it describes happens with the trans-
the spin-“1”attice relaxation.
But there are more hidden hints to versal magnetization?
this: the “1” also looks like a match. Let us assume that this is the situation just
And this match should remind you of before the RF pulse is switched off.
something, which we also have men- When the RF pulse is switched off, the
tioned already: longitudinal relaxation protons get out of step, out of phase again,
has something to do with exchange of as nobody is telling then to stay in step.
energy, thermal energy, which the pro- For the sake of simplicity, this has been
tons emit to the surrounding lattice illustrated for a group of protons which
while returning to their lower state of all “point up” in figure 20. We heard ear-
energy. lier that protons precess with a frequency
 Transversal magnetization MRI made easy 27

Fig. 20: After the RF pulse is switched off, protons lose phase coherence, they get out of step. When you look at these
dephasing proton ensembles from the top (which is illustrated in the lower part of the figure), it becomes obvious, how they
fan out. Fanning out, they point less and less in the same direction, and thus transversal magnetization decreases.

which is determined by the magnetic field different precession frequencies – as we

strength that they are in. And all the pro- have just learned –, they will be soon out
tons should experience the same magnetic of phase.
field. This, however, is not the case: It is interesting to see, how fast the pro-
Firstly, the field of the MR magnet, in tons get out of phase: just suppose that one
which the patient is placed, is not totally proton – p1 – is rotating at 10 million rev-
uniform, not totally homogeneous, but olutions per second, i.e. with a precessing
varies a little, thus causing different pre- frequency of 10 megahertz.
cession frequencies. Due to inhomogeneities, a neighboring
Secondly, each proton is influenced by proton – p2 – is in a magnetic field, which
the small magnetic fields from neighbor- is 1% stronger; this proton has a precession
ing nuclei that are also not distributed frequency of 10.1 megahertz, 1% more. In
evenly, thus causing different precession 5 microseconds (0.000005 sec or 5x10-6), p2
frequencies, too. These internal magnet- will have made 50.5 turns or revolutions,
ic field variations are somehow charac- while proton p1 will have made only 50.
teristic of a tissue. So, after the RF pulse So in this short time span, the protons will
is switched off, the protons are no longer be 180° out of phase, cancelling their mag-
forced to stay in step; and as they have netic moments in the respective plane.
28 MRI made easy Transversal magnetization

Similar to what we did for the longi-

tudinal magnetization, we can plot
transversal magnetization versus
time. What we get is a curve like in
figure 21. This curve is going down-
hill, as transversal magnetization
disappears with time. And as you
probably expect: there is also a time
constant, describing how fast trans-
versal magnetization vanishes, goes
downhill. This time constant is the
transversal relaxation time T2 .

How to remember, what “T2” is?

Easy:

T 2 is T x 2 is TT is Tt
and this means, it describes the “T
transversal”, thus the relaxation of
the transversal magnetization. The
resulting curve in figure 21 thus is one is the T1- and which the T2 -curve?
called a T2 -curve. Another term for Just put both curves together. You can
transversal relaxation is spin-spin- see something like a mountain with a
relaxation, reminding us of the un- ski slope. You first have to climb to the
derlying mechanism, a spin-spin in- top – T1-curve ­–, before you ski down –
teraction. How to remember, which T2 -curve (figure 22).

Fig. 21: If one plots

transversal magnetiza-
tion vs. time after the
RF pulse is switched
off, one gets a curve
as illustrated, which is
called a T2-curve.
Transversal magnetization MRI made easy 29

Fig. 22: Coupling

of a T1- and a
T2-curve resembles
a mountain with a
slope. It takes longer
to climb a mountain
than to slide or jump
down, which helps
to remember that T1
is normally longer
than T2.
 Relaxation MRI made easy 31

quite an overlap of time ranges. What is is naturally crowded, as it is a popular

a long, what is a short relaxation time, place, and order a long drink, you have
and which tissues have long or short re- to wait quite a while to get your drink
laxation times? – T1 is long. When you finally have your
Look at figure 23 – what do you see? long drink, it also takes you a long time
You see somebody having a long drink, to drink it, so T2 is also long. And we
something liquid (representing water). want to remember: water/liquids have
When you go to your favorite bar, which a long T1 and a long T2.

Fig. 23: Liquids

have a long T1 and a
long T2.
32 MRI made easy Relaxation

Now, look at this gentleman in figure

24 getting a hamburger. These normal-
ly contain much fat, and will represent
fat for us.
The hamburger is fast food, you get it
fast, thus fat has a short T1.
What about T2? It takes some time to
eat fast food because of the fat; however,
you normally spend more time with
your long drink, so fat has a shorter T2
than water.
As water has a long T1 and a long T2 , it
is easy to imagine that “watery tissues”,
tissues with a high water content, can
also have long relaxation times.
Interestingly enough, pathological/
diseased tissues often have a higher
water content than the surrounding
normal tissues.
Relaxation MRI made easy 33

Fig. 24: Compared to

liquids/water, fat has a short
T1 and short T2.
34 MRI made easy Relaxation

energy – from one bicycle – proton – to

What is T1 influenced by? the other – lattice – is easy and efficient,
Actually, T1 depends on tissue composi- when both move with the same speed.
tion, structure and surroundings. With a difference in speed, the energy
As we have learned, T1-relaxation transfer will be less efficient.
has something to do with the exchange
of thermal energy, which is handed Why does fat have a short T1?
over from the protons to the surround- The carbon bonds at the ends of the
ings, the lattice. The precessing protons fatty acids have frequencies near the
have a magnetic field that constant- Larmor frequency, thus resulting in
ly changes directions, and which con- effective energy transfer.
stantly fluctuates according to the Lar-
mor frequency. The lattice also has its And why is T1 longer in stronger
own magnetic fields. magnetic fields?
The protons now want to hand ener- As we heard in the beginning, the pre-
gy over to the lattice to relax. cession frequency depends on magnetic
This can be done very effectively, field strength, a relationship described
when the fluctuations of the magnetic by the Larmor equation.
fields in the lattice occur with a fre- If we have a stronger magnetic field,
quency that is near the to Larmor fre- then the protons precess faster.
quency. And when they precess faster, they
When the lattice consists of pure liq- have more problems handing down
uid/water, it is difficult for the protons their energy to a lattice with more
to get rid of their energy, as the small slowly fluctuating magnetic fields.
water molecules move too rapidly.
And as the protons, which are on the
higher energy level, cannot hand their What influences T2?
energy over to the lattice quickly, they T2 -relaxation comes about when pro-
will only slowly go back to their lower tons get out of phase, which – as we al-
energy level, their longitudinal align- ready know – has two causes: inhomo-
ment. geneities of the external magnetic field,
Thus it takes a long time for the lon- and inhomogeneities of the local mag-
gitudinal magnetization to show up netic fields within the tissues (see page
again, and this means that liquids/wa- 27). As water molecules move around
ter have long T1s. very fast, their local magnetic fields
When the lattice consists of medium- fluctuate fast, and therefore kind of av-
size molecules – most body tissues can erage each other out, so there are no big
be looked at as liquids containing vari- net differences in internal magnetic
ous-sized molecules, kind of like a soup fields from place to place. And if there
– that move and have fluctuating mag- are no big differences in magnetic field
netic fields near the Larmor frequency strength within a tissue, the protons
of the precessing protons, energy can be stay in step for a long time, before they
transferred much faster, thus T1 is short. dephase, and so T2 is longer. With im-
This can again be illustrated by our pure liquids, e.g. those containing some
hamburger and bicycle example: (see larger molecules, there are bigger vari-
page 17) handing over hamburgers – i.e. ations in the local magnetic fields. The
 Relaxation MRI made easy 35

larger molecules do not move around ences in the surroundings (the mag-
as fast, so their local magnetic fields netic field variations) influence you
do not cancel each other out as much. considerably. When you drive very fast
These larger differences in local mag- (which is the same as if the surround-
netic fields consequently cause larger ings move very fast), you do not feel
differences in precessing frequencies, each single pot hole anymore. Before
thus protons get out of phase faster, T2 they have a major effect on you, you
is shorter. are already back at normal street lev-
This can be illustrated by the follow- el; thus their effect is averaged out, you
ing example: imagine that you drive are much less influenced by differences
down a street with many pot holes. in the surroundings (the variations in
When you drive slowly (which is equal magnetic field strength).
to the surroundings moving slowly What does all this have to do with
and you are standing still), you will be what we want to know? All these proc-
bumping up and down in your car as esses influence how your MR picture
it drives over each pot hole. The differ- will finally look!

Well, in this
case we’ve got
a problem.
36 MRI made easy Relaxation

A brief review might An experiment ...

be advisable:
7 T1

7
is longer than T2 .
T1 varies with the mag-
Now let us perform an
netic field strength; in experiment
stronger magnetic fields it Look at figure 25, where you can see
is longer. two protons, precessing around the z-
7 Water has a long T1, fat has a short T1. axis. I hope you recall that the z-axis
7 T2 of water is longer than the T2 of im- indicates the direction of a magnetic
pure liquids containing larger mole- field line (see page 9). Instead of on-
cules. ly these two protons, in reality there
may be 8 pointing up and 6 pointing
down, or 82 up and 80 down – there

Fig. 25: If after the RF pulse the

number of protons on the higher
energy level equals the number of
protons on the lower energy, lon-
gitudinal magnetization has disap-
peared, and there is only transversal
magnetization due to phase coher-
ence. The magnetic vector seems to
have been “tilted” 90° to the side.
The corresponding RF pulse is thus
also called a 90° pulse.
 Relaxation MRI made easy 37

shall only be two more protons point- will decrease, in our example to zero
ing up. (one pointing up is neutralized by one
As we know, these are the ones that pointing down). But: as both protons
have a net magnetic effect because are in phase, there is now a transver-
their effects are not cancelled out. sal magnetization which had not been
Now let us send in an RF pulse, there before.
which has just the correct strength The RF pulse seemingly “tilts” the
and duration, so that one of the two longitudinal magnetic vector 90° to
protons picks up energy to go into the side.
a higher state of energy, i.e. points Such an RF pulse is called a 90°
down/walks on its hands. pulse. Naturally, other RF pulses are
What will happen? The longitudi- also possible, and are named accord-
nal magnetization (up to now result- ingly, e.g. 180° pulse.
ing from two protons pointing up)
38 MRI made easy Relaxation

To really understand this, let us look to their lower state of energy, and
at another example. In figure 26A, we they lose phase coherence.
have 6 protons pointing up; we send in It is important to note that both
an RF pulse, which lifts 3 of them onto processes occur simultaneously and
a higher energy level (B). independently. For the sake of sim-
The result: we no longer have a lon- plicity, let us look at what happens
gitudinal, but a transversal magneti- step by step, and first focus on the lon-
zation (again having used a 90° pulse). gitudinal magnetization:
What happens, when the RF pulse is 7 After the RF pulse is switched off

switched off? (figure 26C), one proton goes back

Two things happen: protons go back to the lower energy state, result-
 Relaxation MRI made easy 39

ing in 4 protons pointing up, and two You surely have already noticed that
pointing down. The net effect: we the transversal magnetization de-
now have a longitudinal magnetiza- creases at the same time (figure 26C-
tion of “2”. E). Why? You should be able to answer
7 Then the next proton goes back up; this: After the RF pulse is switched off,
now 5 protons are pointing up, and one the precessing protons lose phase co-
down, resulting in a net longitudinal herence.
magnetization of “4” (figure 26D). In figure 26B, all protons point in the
7 After the next proton goes up, longi- same direction, but then increasingly
tudinal magnetization equals “6” (fig- get out of phase, and thus kind of fan
ure 26E). out (figure 26C-E).

Fig. 26: (A) shows the

situation before and
(B) immediately after
an RF pulse is sent in.
The RF pulse causes
the longitudinal mag-
netization (I) to de-
crease, and with a 90°
pulse as illustrated,
it becomes zero (B).
Protons also start to
precess in phase (B),
which causes the new
transversal magnetiza-
tion (J). After the
RF pulse is switched
off (C-E), longitudinal
magnetization in-
creases, recovers, and
transversal magnetiza-
tion disappears, de-
cays. Both processes
are due to entirely
different mechanisms
and occur indepen-
dently, even though at
the same time.
40 MRI made easy Relaxation

In figure 27, only the longitudinal and of force in the original directions. If we
transversal magnetic vectors are de- do the same with the longitudinal and
picted at corresponding times as in fig- the transversal vector, we get the sum
ure 26. These magnetic vectors add up vector.
to a sum vector ( J). This sum vector is very important,
As you remember, vectors represent as it represents the total magnetic mo-
forces of a certain size and a certain di- ment of a tissue in general, and thus
rection. If you add up vectors pointing can be used instead of the two single
to different directions, you will come vectors, representing longitudinal and
up with a direction that is somewhere transversal magnetization separately.
in between, depending on the amount Our magnetic sum vector during relax-
 Relaxation MRI made easy 41

ation goes back to a longitudinal direc-

tion, in the end equaling the longitudi-
nal magnetization.
What we have to remember is that
this whole system is actually precess-
ing, including the sum magnetic vector.
And thus – after the RF pulse is switched
off – the sum vector will actually per-
form a spiraling motion (figure 27F).

Fig. 27: In this illustration

only the longitudinal and
transversal magnetization
vectors from our experiment
in figure 26 are depicted.
In (A) – before the RF pulse
– there is only longitudinal
magnetization. Immediately
after the 90° RF pulse there
is no longitudinal but new
transversal magnetization
(B), and this transversal
magnetization vector is spin-
ning around. With time this
transversal magnetization
decreases, while longitudinal
magnetization increases
(C-D), until the starting point
with no transversal but full
longitudinal magnetization is
reached again (E). Transversal
and longitudinal magnetiza-
tion vectors add up to a
sum vector (J). This sum
vector performs a spiraling
motion (F) when it changes
its direction from being in the
transversal (x-y) plane (no
longitudinal magnetization)
to its final position along
the z-axis (no transversal
magnetization).
42 MRI made easy Relaxation

I hope that you recall that a changing mag- magnetic vector as having some kind of a
netic force/moment can induce an electric sound-emitting device like a steam pipe
current, which is the signal that we receive at its tip. The further the vector goes
and use in MR. away from the microphone, the less
So if we put up an antenna somewhere loud the sound. The frequency of the
(figure 28), we will get a signal as illustrat- sound, however, remains the same be-
ed. This is easy to imagine, if you think of cause the sum vector spins with the
the antenna as a microphone, and the sum precessing frequency (figure 29). So the

Fig. 28: For an

external observer,
the sum vec-
tor of figure 27F
constantly changes
its direction and
magnitude, while it
performs its spiral-
ing motion. The sum
vector induces an
electrical current
in an antenna, the
MR signal. This is of
greatest magni-
tude, immediately
after the RF pulse
is switched off, and
then decreases.
 Relaxation MRI made easy 43

signal from our experiment disappears inducing electrical currents in the an-
with time, however, it has a constant tenna. Instead of the terms “longitudi-
frequency. nal” or “transversal magnetization”, we
This type of signal is called a free in- can also use the term “signal or signal
duction decay signal, or FID signal. intensity” at the axis of our T1- and T2 -
By now it should be obvious that the curves.
magnetic vector directly determines This will hopefully become clearer, as
the MRI signal and signal intensity by you continue reading.

Fig. 29: The signal

from our experiment
in figures 26 to 28
disappears with
time, however, it
has a constant
frequency. This type
of signal is called a
FID (free induction
decay) signal.
44 MRI made easy Time to repeat

Another experiment pulse is switched off. At TR short however,

tissue A has regained more of its longitudi-
Let us perform another experiment simi- nal magnetization than tissue B. When the
lar to the one illustrated in figure 30. As second 90° pulse now “tilts” the longitudi-
we want to concentrate purely on the nal magnetization 90 degrees, the trans-
magnetization, we can leave out the co- versal magnetic vector of tissue A is larger
ordinate system. In figure 30A, we have than that of tissue B.
two tissues, A and B, which have different And when this vector of A is larger, it
relaxation times as we will see later. We will reach closer to our antenna; thus the
send in a 90° RF pulse, switch it off and imaginary steam pipe at the tip of vector
wait a certain time TR long (we will explain A will cause a louder, stronger signal in
later, why we use the term TR). Then we our “microphone”, the antenna, than the
send in a second 90° pulse. What will hap- vector of B.
pen? The difference in signal intensity in this
As after the time TR long tissue A and tis- experiment depends on the difference
sue B have regained all of their longitudi- in longitudinal magnetization, and this
nal magnetization (frame 5), the transver- means on the difference in T1 between the
sal magnetization after the second pulse tissues. Using these two pulses, we can now
will be the same for both tissues, as it was differentiate tissue A from tissue B, which
after the first RF pulse (frame 1). Tissue A in our experiment was impossible, choos-
cannot be differentiated from tissue B. ing only one 90° pulse or two 90° pulses
What if we do not wait so long from pulse that are a long time apart (after a long time,
to pulse? Let us look at figure 30B: the differences in T1 between tissue A and B
After the first pulse, an equally sized no longer play a role in our experiment, be-
transversal magnetization is established cause after that time the tissue B with the
in both tissues, which decreases after the longer T1 is back to its original state, too).

Fig. 30A

When we do not wait

as long as in figure
30A, but send in
the RF pulse after a
shorter time (TRshort),
like in figure 30B,
longitudinal mag-
netization of tissue B,
which has the longer
T1, has not recovered
as much as that of
tissue A with the
shorter T1. The trans-
versal magnetization
 Time to repeat MRI made easy 45

When you use more than one RF pulse – a The resulting image is called a T1-weight-
succession of RF pulses – you use a so-called ed image. This means that the difference
pulse sequence. As you can use different of signal intensity between tissues in that
pulses, e.g. 90° or 180° pulses, and the time image, the tissue contrast, is mainly due
intervals between successive pulses can be to their difference in T1. However, there is
different, there can be many different pulse always more than one parameter influenc-
sequences. As we saw in our experiment, ing the tissue contrast; in our example, T1 is
the choice of a pulse sequence will deter- just the most outstanding one.
mine, what kind of signal you get out of a
tissue. So it is necessary to carefully choose What is a short, what is a long TR?
and also describe the pulse sequence for a A TR of less than 500 msec is considered to
specific study. be short, a TR greater than 1,500 msec to be
The pulse sequence that we used was long. As you may imagine or know already,
made up of one type of pulse only, the 90° we cannot only create T1-weighted images,
pulse. This was repeated after a certain but also T2 -weighted images, and so-called
time, which is called TR = time to repeat. proton density (-weighted) images.
This proton density, which is also called
How did TR influence the signal in our spin density, influences tissue contrast and
experiment? can be explained quite simply: where there
With a long TR we got similar signals from are many protons, we will have “lots” of
both tissues, both would appear the same signals. Where there are no protons, there
on an MR picture, since the transversal will be no signal. We will read more about
magnetization was the same for both tis- this later. The point is that by using certain
sues. Using a shorter TR, there was a dif- pulse sequences, we can make certain tis-
ference in signal intensity between the tis- sue characteristics to be more or less im-
sues, determined by their difference in T1. portant in the resulting image.

Fig. 30B

of the two tissues

after the second RF
pulse will then be
different (frame 5).
Thus, by changing
the time between
successive RF
pulses, we can in-
fluence and modify
magnetization and
the signal intensity
of tissues.
46 MRI made easy Time to repeat

By choosing a pulse sequence, the doc-

tor can be compared to a conductor
Let us go back to our
of an orchestra (figure 31): he can in- experiment once more
fluence the overall appearance of the
sound (signal) by making certain in-
for a short repetition:
struments (parameters) influence the 7 With a certain type of RF
sound more than others. All instru- pulse, we can cause the lon-
ments (parameters), however, always gitudinal magnetization
play some role in the final sound (sig- to disappear, while a trans-
nal). versal magnetization appears. The “net
magnetization” (the sum vector of lon-
gitudinal and transversal magnetiza-
tion) is “tilted” 90° in this case (when we
started, we only had longitudinal mag-
netization). The corresponding RF pulse
is therefore called a 90° pulse.
7 The transversal component of the net

magnetization can induce a measurable

signal in an antenna.

Fig. 31: The MRI doctor

can be compared to a
conductor: by choosing
certain pulse sequences,
he can modify the result-
ing signal, which is itself
influenced by different
parameters.
 Time to repeat MRI made easy 47

7 Immediately after the RF pulse relax-

ation begins: transversal magnetiza-

The difference in
tion starts to disappear and longitudi- signal intensity
nal relaxation begins to reappear. The In figure 32, you can see the T1-curves
signal disappears. for brain and for cerebrospinal fluid
7 When we send in the second 90° pulse, (CSF). Brain has a shorter longitudinal
the net magnetization is again tilted relaxation time than CSF.
90°, and we again receive a signal. At the time 0, we have no longitudi-
7 The strength of this signal depends nal magnetization at all, and this can be
(among other things) on the amount of the time immediately after our first 90°
longitudinal magnetization we start out pulse. When we wait a long time before
with. Do you remember the T1-curve? we repeat the 90° pulse (TR long), longitu-
The T1-curve described the relationship dinal magnetization has pretty much
between time (after an RF pulse) and the recovered. The longitudinal magnetic
amount of longitudinal magnetization vectors that will be “tilted” 90°, differ
(figure 18). only to a small degree, so there will only
When we wait a long time before be a small difference in signal intensi-
sending in our second RF pulse, longi- ty, i.e. tissue contrast between brain and
tudinal magnetization will have recov- CSF is small. If we, however, send in the
ered totally. second pulse after a shorter time, TR short,
The signal after the second RF pulse the difference in longitudinal magneti-
will thus be the same as the one after zation is rather large, so there will be a
the first pulse. However, when the sec- better tissue contrast.
ond pulse comes in earlier, the signal And as we can see from the distance
will be different, since the amount of between the two curves, there is a time
longitudinal magnetization at that time span where tissue contrast is most pro-
is less. nounced.

Fig. 32: Brain

has a shorter
longitudinal
relaxation time
than CSF. With
a short TR, the
signal intensities
of brain and CSF
differ more than
after a long TR.
48 MRI made easy Time to echo

Why are the signals tons . . . lose phase coherence, as we

learned earlier.
after a very long time This is illustrated in figure 33 for
TR between pulses not three protons, which are almost ex-
actly in phase as seen in (A) but in-
identical? creasingly spread out, as they have
We have heard the explanation already. different precession frequencies (see B
The signal intensity depends on many and C). The loss of phase coherence re-
parameters. When we wait a long time, sults in decreasing transversal mag-
T1 does not influence the tissue contrast netization and thus loss of signal.
any more, however, there may still be a Now we will do something new: af-
possible difference in the proton densi-
ty of the tissues in question.
And when we wait a very long time
TR in our experiment from figure 32,
the difference in signal is mainly due
to different proton densities, we have a
so-called proton density- (or spin den-
sity-) weighted image. Fig. 33: After the RF
pulse is switched
Now we have heard about T1- and off, the protons
proton density-weighted images. dephase (A-C). The
180° pulse causes
them to precess in
the opposite direc-
T2-weighted images tion, and so they
rephase again (D-F).

How do we obtain a
T2-weighted image?
This is a little more difficult to under-
stand. Let us perform another experi-
ment, which is a little different from
the ones before. First, we use a 90°
pulse. The longitudinal magnetiza-
tion is tilted, we get a transversal
magnetization. What happens after
this pulse, when we wait a short
time?
You can surely answer this question
without difficulty – if not, go back to
page 27 before you continue to read.
After the pulse is switched off, lon-
gitudinal magnetization starts to re-
appear, the transversal magnetiza-
tion, however, starts to disappear.
Why does the transversal magnetiza-
tion disappear? It is because the pro-
 Time to echo MRI made easy 49

ter a certain time (which we call TE/2, er ones. If we wait another time TE/2,
half of TE, for reasons you will under- the faster ones will have caught up
stand in a few minutes), we will send with the slower ones (see figure 33F).
in a 180° pulse. What happens? At this point in time, the protons
The 180° pulse acts like a rub- are nearly in phase again, which re-
ber wall; it makes the protons turn sults in a stronger transversal mag-
around, so that they precess in ex- netization, and thus in a stronger
actly the opposite direction, which is signal again. A little later, however,
clockwise. the faster precessing protons will be
The result is that the faster precess- ahead again, with the signal decreas-
ing protons are now behind the slow- ing again.
50 MRI made easy The echo

Fig. 34: When a hare and a tortoise

run in one direction for a certain time,
then turn around and run in the oppo-
site direction with the same speed for
the same time, they will arrive at the
starting point at the same time.
 The echo MRI made easy 51

To illustrate this: think about a race be- bounce back, like a mountain reflecting
tween a tortoise and a hare starting at sound waves as echoes. This is why the
the same line (figure 34). After a certain resulting strong signal is also called an
time (TE/2), the hare is ahead of the tor- echo, or spin echo.
toise. When you make the competitors After we have our signal, our spin
run in the opposite direction for the echo, the protons lose phase coherence
same length of time, they will both be again, the faster ones getting ahead, as
back at the starting line at exactly the we have seen.
same time (assuming, that they run at We naturally can perform the experi-
constant speed). ment again with another 180° pulse,
and another and another  . . .  
In our experiment, the 180° pulse acts If we now plot time vs. signal intensi-
like a wall, from which the protons ty, we get a curve like in figure 35.

Fig. 35: The 180° pulse refocuses the dephasing protons, which results in a stronger signal,
the spin echo after the time TE. The protons then dephase again and can be refocused anoth-
er time by a 180° pulse, and so on. Thus it is possible to obtain more than one signal, more
than one spin echo. The spin echoes, however, differ in intensity due to so-called T2-effects.
A curve connecting the spin echo intensities is the T2­-curve. If we did not use the 180° pulse,
the signal intensity would decay much faster. A curve describing the signal intensity in that
case is the T2*-(T2 star) curve, which is described in a little more detail on page 52.
52 MRI made easy The echo

From this curve we can see that the spin same speed also for the time TE/2. After
echo, the resulting signal, decreases with 2 x TE/2 = TE, the buses will be back at
time. Responsible for this is the fact that the starting line. The signal intensity that
our 180° pulse only “neutralizes” effects you record with your microphone then
that influence the protons in a constant depends only on inherent properties, i.e.
manner, and these are the constant inho- how tired the crowds have become.
mogeneities of the external magnetic field. Let us have a look at our curve, when
Inconstant inhomogeneities from lo- we plotted time vs. signal intensity send-
cal magnetic fields inside the tissue can- ing in several 180° pulses (see figure 35).
not be “evened out”, as they may influence If you do not use a 180° pulse to neutral-
some protons before the 180° pulse differ- ize constant external inhomogeneities,
ently than after the 180° pulse. So some of the protons will experience larger differ-
the protons may still be behind or in front ences in magnetic field strength, when
of the majority of the protons that will the RF pulse is switched off. Due to this,
reach the starting line at the same time. they will be out of phase faster, the trans-
So from echo to echo, the intensity of the versal relaxation time will be shorter.
signal goes down due to so-called T2-ef- A curve describing the signal intensity
fects. A curve connecting the spin echo in that case is the T2*- (T2 star) curve. The
intensities is the T2-curve. star distinguishes this shorter transver-
May be we should illustrate this by an sal relaxation time from the T2 after the
example: imagine two buses full of peo- 180° pulse, which we have already talk-
ple, e.g. after a soccer or football game. ed about.
They are standing at a starting line (fig- The corresponding effects are named
ure 36). With two microphones, you re- T2*-effects. These T2*-effects are impor-
cord the signal (e.g. the singing from the tant with the so-called fast imaging se-
crowd) that is coming from each bus. The quences; we will hear about them later,
buses leave in the same direction. and cover them in more detail in the book
Listening to the singing of the crowds, on “MR Buzzology”.
i.e. recording the signal, you may recog- In our example with the buses this
nize that one signal disappears faster would mean that we just record the sig-
than the other. nals, as the buses drive away. The signals
This can have two different causes: The vanish due to extrinsic (bus speed) and
difference in signal intensities, the differ- intrinsic (exhaustion of the passengers)
ence in singing, may be due to differences properties under these circumstances
in inherent properties of the two groups (see figure 36).
(internal inhomogeneities); may be in The type of pulse sequence that we used
one bus, there are only the “party ani- in our experiment, is called a spin echo
mals”, who did not become tired as fast as sequence, consisting of a 90° pulse and a
the people in the other crowd. 180° pulse (causing the echo). This pulse
Or . . . may be one bus is driving faster sequence is very important in MR im-
than the other (loss of signal would thus aging, as it is the workhorse among the
be due to external influences, the external pulse sequences, which can be used for
magnetic field inhomogeneities). many things. It is important to realize
To figure out what is actually the reason that with a spin echo sequence, we can-
for the signal disappearing, you can make not only produce T2-, but also T1- and pro-
the buses turn around after a certain time ton density-weighted images. We will get
TE/2, and have them drive back with the to that a little later.
 The echo MRI made easy 53

Fig. 36: Without having the

buses come back (i.e. a
180° pulse), it is impossible
to say whether a decrease
in signal intensity is due to
inherent tissue properties
(the different shape of the
bus passengers), or due
to external influences, i.e.
different bus speeds.
54 MRI made easy The echo

a certain time TE/2 to send in the 180°

T2-weighted images – a closer look
pulse, transversal magnetization will
have become smaller. After waiting an-
Let us first look at such other time TE/2 (that is TE after the 90°
pulse is switched off), we will receive a
a T2-weighted sequence signal, the spin echo.
What did we do? First, we sent in a 90° The intensity of this echo is given by
pulse, resulting in some transversal the T2-curve at the time TE. This time
magnetization. Immediately after the between the 90° pulse and the spin
90° pulse, we have a maximum trans- echo is called TE = time to echo.
versal magnetization. However, this The time TE can be chosen by the op-
transversal magnetization disappears, erator. And as we can see from the T2-
due to T2-effects. How fast transversal curve, TE influences the resulting sig-
magnetization disappears, can be seen nal, and thus also the image: the short-
from a T2-curve; in figure 37, we have er the time TE, the stronger the signal
plotted T2-curves for two different tis- that we get from a tissue. To get the
sues, tissue A having a short T2 (e.g. best, strong signal, it may seem rea-
brain), tissue B having a long T2 (wa- sonable to use a short TE, because with
ter or CSF). The curves start at 0, which longer TEs, signal intensity decreases.
is the time immediately after the 90° With a short TE, however, there will be
pulse is switched off. When we wait for a problem (figure 37).

Fig. 37: T2-curves for two tissues with different transversal relaxation times; tissue A has a shorter T2
than tissue B, thus loses transversal magnetization faster. With a short TE (TEshort), the difference in signal
intensity is less pronounced than after a longer TE (TElong).
 The echo MRI made easy 55

Let us have a look at two different tis- heavily T2 -weighted. But (and there is
sues. Tissue B (water or CSF) has a longer always a “but”) if we wait longer, the
T2 than tissue A (brain). total signal intensity becomes small-
Both T2 -curves in this example start er and smaller. The signal-to-noise ra-
at the same point. If we only wait a tio becomes smaller, the image appears
short TE, TEshort, the difference in signal grainy.
intensity between tissue A and tissue B An example to illustrate this signal-
is very small, both tissues may hardly to-noise problem: when you receive a
be distinguished, as there is hardly any local radio station in your radio, this
contrast (which is the difference in sig- gives you a good signal, e.g. loud music
nal intensity of tissues). and only little static noise.
Consequence: with a short TE, differ- When you drive out of town, the sig-
ences in T2 do not influence tissue con- nal intensity of the radio station be-
trast very much. comes weaker, and you will hear more
As both T2 -curves diverge with a static noise; and when you drive even
longer TE, TElong, the difference in T2 - further away, you may not be able to
curves, and thus the difference in sig- discern the music from the background
nal intensity meaning contrast, is noise. And this is the same for the MR
more pronounced in our example. So it signal: there is always some noise in
might be reasonable to wait a very long the system, however, when the signal
TE; the resulting image should be very is strong, this does not matter much.
However, the smaller the signal, the
harder it is to differentiate it from the
background noise.

We’ll be able to
determine if you have a
brain in just a
moment…
56 MRI made easy The echo

Let us review some facts TR. A TR of less than 500 msec is con-
sidered to be short, a TR of more than
We have learned: 1,500 msec to be long (just to give you
7 The spin echo sequence a rough idea). A short TE is one that is
consists of a 90° and a as short as possible, a long TE also is
180° pulse. more than 3 times as long.
7 After the 90° pulse, protons de- A TE of less than 20 msec is con-
phase due to external and internal sidered to be short, a TE more than
magnetic field inhomogeneities. 80 msec to be long.
7 The 180° pulse rephases the de-

phasing protons (sometimes the

term spins is used interchangeably
for protons), and a stronger signal,
the spin echo, results.
7 The 180° pulse serves to “neutral-

ize” the external magnetic field in-

homogeneities. Fig. 38: Schematic
7 Signal decrease from one echo to illustration of a spin
echo pulse sequence.
the next, when using multiple 180°
pulses, is due to internal T2 -effects.
7 By choosing different TEs (differ-

ent times after the 90° pulse), the

signal can be T2 -weighted in varying
degrees – with very short TEs, T2 -ef-
fects have not yet had time to really
show up.
7 With longer TEs, the signal inten-

sity difference between tissues will

depend very much on their T2s, their
transversal relaxation time.
7 With very long TEs, there should

be even more T2 -weighting, however,

signal intensity as such would be so
small that at best it can just barely be
distinguished from the background
noise.

By the way: what is a short,

what is a long TR or TE?
A short TR is one that is about as
short as the smallest/shortest T1 that
we are interested in (remember: T1
was a time constant, not a time that
it takes for a tissue to regain its lon-
gitudinal magnetization!). A long TR
is about 3 times as long as a short
 Spin echo sequence MRI made easy 57

two or more times. The time to repeat a

The pattern of a spin echo sequence
pulse sequence was TR, time to repeat.
So what we get is the following scheme:

Let us go back to 1. 90° pulse – TE/2 – 180° pulse –

our spin echo pulse TE/2 – recording signal at TE.
sequence. After TR (time from the beginning of
one 90° pulse to the next 90° pulse)
This sequence can be illustrated sche-
follows another pulse cycle and signal
matically as in figure 38: 90° pulse –
measurement:
wait TE/2 – 180° pulse – wait TE/2 – re-
cord signal. For certain different rea- 2. 90° pulse – TE/2 – 180° pulse –
sons, such a pulse sequence is repeated TE/2 – recording signal at TE.
58 MRI made easy Spin echo sequence

Fig. 39: It is possible to determine signal intensity for a tissue using a spin echo
sequence by combining the T1- and the T2-curve for that tissue. The longitudinal mag-
netization after the time TR is equal to the amount of transversal magnetization we start
out with, as it is “tilted” 90 degrees. This transversal magnetization immediately starts
to disappear by a rate which is determined by the transversal relaxation time, and thus
by the T2-curve. The signal intensity of the tissue after a time TE can then be inferred
from the T2-curve at this time TE (which starts after TR!).

how much longitudinal magnetiza-

TR and TE
tion will be tilted 90° to the side (and
thus to figure out, with how much
To figure out how much signal you get transversal magnetization we start
from a certain tissue with certain pa- out with), we just look at the intensity
rameters of a spin echo sequence, you of the longitudinal magnetization at
actually have to do no more than com- the time TR.
bine its T1- and T2 -curves, as it is illus- The longitudinal magnetization
trated in figure 39. Here we have the at this point, “tilted” in the transver-
T1- and T2 -curve of a certain tissue. sal plane, is the starting point from
T1 represents the longitudinal relax- which transversal magnetization de-
ation and T2 represents the transver- cays. So we just attach the T2 -curve at
sal relaxation. Which parameter de- this point.
termined the amount of longitudinal How much signal we get with a spin
magnetization? That was TR. To see echo sequence to construct the image,
 Spin echo sequence MRI made easy 59

Fig. 40: By combining T1- and T2-curves, signal intensity of certain tissues can be determined for a
pulse sequence using TR and TE as illustrated, and as explained in figure 39. What happens, when we
choose a long TR, as illustrated? With a long TR, differences in T1, in longitudinal magnetization time
are not very important any more, as all tissues have regained their full longitudinal magnetization.
When we only wait a very short TE, then differences in signal intensity due to differences in T2 have
not yet had time to become pronounced. The resulting image is thus neither T1- nor T2-weighted, but
mostly determined by the proton density of the tissues (for this, ideally TE should be zero).

also depends on TE, the time that we as enough time has passed by to allow
wait after the 90° pulse. So we now even tissues with a long T1 to relax to-
only have to look for the signal inten- tally. So when we choose a long TR, as
sity at the time TE on the T2 -curve. we just said, then differences in T1 do
not really matter.
What image do we get, when we When we also use a short TE, differ-
choose a long TR and a short TE? ences in signal intensity due to differ-
This is illustrated in figure 40. Here ences in T2 have not had enough time
are the T1- and T2 -curves for two dif- to become pronounced yet.
ferent tissues. The signal that we get, is thus nei-
As we heard earlier, with a very long ther T1- nor T2 -weighted, but mainly
TR, all tissues will have totally recov- influenced by differences in proton or
ered their longitudinal magnetiza- spin density.
tion; differences in T1 of the tissues ex- The more protons, the more signal,
amined will not influence the signal, if you look at it simply (figure 40).
60 MRI made easy Spin echo sequence

Fig. 41: When we wait a long TR and a long TE, differences in T2 have had
time enough to become pronounced, the resulting picture is T2-weighted.

And what happens when we covered their longitudinal magnetiza-

use a long TR and a long TE? tion, thus differences in T1 (which de-
With a long TR, there are no prevail- termines how fast longitudinal mag-
ing differences in T1. With the long TE, netization is regained) will show up
however, differences in T2 become pro- in form of signal intensity differences
nounced. Thus the resulting image is (figure 42).
T2 -weighted (figure 41). When TE is short, differences in T2
cannot really manifest themselves, so
What if we use a shorter TR the resulting image is still T1-weight-
and a short TE? ed (there is a lower limit for TE, be-
With a short TR, tissues have not re- cause it takes some time for the 180°
 Spin echo sequence MRI made easy 61

Fig. 42: When we wait a shorter time TR, differences in T1 influence tissue contrast to a
larger extent, the picture is T1-weighted, especially when we also wait a short TE (when
signal differences due to differing T2s have not had time to become pronounced).

pulse to be “produced”, sent in and do transversal magnetization resulting to

properly take effect, see also in the disappear to a large extent. The result-
book “MR Buzzology”). ing signal will be so small, of so little
intensity that it cannot be used to make
What if we use a very short TR a reasonable image.
and a very long TE?
This is only a theoretical question. Why?
With a very short TR, there will only
be very little longitudinal magnetiza-
tion which is “tilted”. And with a long
TE, we even allow the small amount of
62 MRI made easy Spin echo sequence

If you have not been And what do you see in figure 44? The
same couple is having tea. Now, hav-
concentrating ing tea which is usually served hot, al-
for the last few minutes, you ways takes a long time. And in the illus-
are probably thinking about tration the long TEa makes two people
giving up right now. How to happy. This should remind you that a
remember this – even if you long TE gives a T2 -weighted image.
do not understand all of it (which hope-
fully is not the case)?
Try looking at figure 43. What can
you see? A man with short TRousers.
And considering the weather condi-
tions, this makes only one person in the
picture happy.
This should remind you that a short
TR (TRousers) gives a T1-weighted im-
age (only 1 is happy).

Fig. 43: What to choose

for a T1-weighted
image?
 Spin echo sequence MRI made easy 63

Fig. 44: What to choose

for a T2-weighted
image?
64 MRI made easy Image interpretation

Fig. 45: T1- (A), proton

(spin) density- (B), and
T2-weighted (C) images of
the same patient. The CSF
is black on the T1-weighted
image. However, it has
the strongest signal in the
T2-weighted image. On the
spin-density image, it is of
intermediate signal
intensity.

B
 Image interpretation MRI made easy 65

Some practical hints to

image interpretation
How can we tell from an image,
whether it is a T1- or a T2 -weighted im-
age, when imaging was done with a
normal pulse sequence, not one of the
fast sequences (which we will hear
about a little later)?
As a rule of thumb: if you see white
fluid, e.g. CSF or urine, you are dealing
with a T2 -weighted image. If the fluid
is darker than the solids, we have a T1-
or a proton-density image.
Look at the scan (figure 45): CSF
is dark, the grey matter is darker
(greyer) than the white matter; this is
a typical T1-weighted image.
In (B), CSF is still dark, even though
its signal intensity is slightly higher
than in the T1-weighted image; con-
C trast between the grey and white mat-
ter is becoming reversed. This is a pro-
ton or spin density-weighted image,
and as the grey matter has a higher
water content, i.e. contains more pro-
tons, its signal intensity is higher
than that of the white substance.
In (C), CSF has a higher signal inten-
sity than grey and white matter, the
image is T2 -weighted.
These are rules of thumb only. Actu-
ally, to be really sure, you would have
to look at two images taken with dif-
ferent imaging parameters. Why?
Look at figure 46. You can see that in
this example the T2 -curves start at dif-
ferent “heights”, and cross each other.
They do not have to run parallel, as
we depicted them in the previous il-
lustration, which was only done for
didactic reasons, as it is easier to un-
derstand at the beginning.
66 MRI made easy Image interpretation

Fig. 46: T2-curves of different tissues can intersect. The signal intensity
of the tissues is reversed choosing a TE beyond the crossing point (TEC ):
before this crossing point (e.g. at TE1), tissue A has a higher signal inten-
sity than tissue B. This means that image contrast is still determined by
differences in T1: the tissue A with the shorter T1 has the stronger signal
intensity. At TEC , both tissues have the same signal intensity, and thus
cannot be differentiated. After this crossing point (e.g. at TE2 ), the rela-
tive signal intensities are reversed, and tissue B has the stronger signal.

The fact that the curves intersect is 7 With a TE beyond the crossing point
very important: (TE2 ), tissue A will have a lower signal
7 With a TE before the crossing point than tissue B.
(TE1), tissue A will have a higher signal 7 Before this crossing point (which you

intensity. do not know, looking at an image nor-

7 With a TE right at that point (TEc), we mally), the relative signal intensities
cannot distinguish the tissues at all, as are still governed by differences in T1.
they have the same signal intensity. The tissue with the shorter T1 (or the
Thus, you might be unlucky, and higher proton density, if we have a long
choose a pulse sequence with just TR) still has the higher signal intensity.
those imaging parameters that do not Only with longer TEs does the T2-
allow tissue differentiation (which is weighting come up. Think about that
the reason for performing two differ- for a moment!
ent examinations with different T1-
and T2-weightings).
 Flow effects MRI made easy 67

How does flow influence The subject of how flow influences

the MR signal is rather complex and dif-
the signal? ficult, but let us at least get some idea
Now we have already heard about about it.
many parameters that influence the In figure 47, we have a body section
MR image, T1, T2 , proton density, pulse through which a vessel is crossing.
sequences, TR and TE – but there are When we send in our first 90° pulse,
more, e.g. contrast media, and flow. all the protons in the cross section are
The fact that flow influences the MR influenced by the radio wave. After we
signal has been known for a long time. turn the RF pulse off, we “listen” into
The first experiments on this subject the section and record a signal.
were carried out more than forty years At this time, all the original blood in
ago. Interestingly, this phenomenon our vessel may have left the slice being
was used to measure flow in the fuel examined. So there is no signal coming
pipes of satellite rockets, without hav- out of the vessel; it appears black in the
ing to put any obstruction into the flow image. This phenomenon is called flow-
lines. void phenomenon.

Fig. 47: Flow effects are responsible for the black appearance of flowing
blood, the signal void in blood vessels.
68 MRI made easy Flow effects

This is not the only way in which some relaxation, and there is some
f low may inf luence the image, there longitudinal magnetization again,
may be all kinds of things, e.g. also as shown by the arrows pointing
f low-related enhancement. back up. The protons in the blood
Illustration 48 shows a blood ves- vessel, however, have left the slice
sel going through a slice which is be- and been replaced by protons that
ing examined. (A) represents the sit- still have all of their longitudinal
uation before the 90° pulse and (B) magnetization.
immediately after the pulse, longitu- If we send in a second 90° pulse
dinal magnetization is “tilted” 90°. now, there will be more signal com-
If we wait some more time, before ing from the vessel than from its
we send in a second 90° pulse, like surroundings, because there is more
in (C), protons will have undergone longitudinal magnetization at this
 Flow effects MRI made easy 69

Fig. 48: Flow can have differing

effects on signal intensity, and
can also cause flow-related en-
hancement, which is explained
in detail in the text.

time. The whole subject of signal look it up in one of the comprehen-

strength and f low effects is actually sive standard text books (or before
much more complicated. that in the book “MR Buzzology”).
For example, when you do multi- They will also give you more infor-
slice imaging, i.e. taking images of mation on MRI angiography.
more than one slice at the same time In this technique, the fact that
(see page 83), the signal also depends f low inf luences the MRI signal is
on the direction of the flow. In addi- used positively by displaying the
tion, it differs over the cross section moving protons.
of a vessel, depending on the f low
profile, and whether there is lam-
inar or turbulent f low. If you want
to know more about this, you should
70 MRI made easy Contrast media

Fig. 49: Paramagnetic substances like gadolinium shorten the T1 and the T2
of their surroundings. The respective T1- (fig. 49A) and T2-curves (fig. 49B)
are shifted towards the left. In effect, this means that for a certain TR there is
more, for a certain TE, however, there is less signal.
 Contrast media MRI made easy 71

What about MR
contrast media?
Certain so-called paramagnetic sub-
stances have small local magnetic
fields, which cause a shortening of
the relaxation times of the protons
in their neighborhood. This effect is
named proton relaxation enhance-
ment.
The body contains such paramag-
netic substances under normal cir-
cumstances. Examples are degra-
dation products of hemoglobin, e.g.
deoxyglobin and methemoglobin,
which are found in hematomas, or al-
so molecular oxygen.
Gadolinium (Gd), a paramagnetic
substance, is used in MR contrast me-
dia, like in Magnevist® or Gadovist®.
Chemically, Gadolinium is a rare
Now don’t move earth, which, however, is toxic in its
for the next thirty free state. Because of this it is bound

minutes. If you
to “some other chemical” in a certain
way called chelation, which solves the
really have to, you may problem of toxicity. For example, in
breathe. Magnevist®, this chemical is DTPA.
The pharmacological properties of
many Gadolinium-containing contrast
media like in Gadopentetate Dimeglu-
mine or Gadolinium DTPA (Magne-
vist®) or Gadobutrol (Gadovist®) are
very similar to iodinated contrast me-
dia in conventional radiology: however,
these Gadolinium-containing contrast
media are even better tolerated.
The Gd-containing contrast media
have an effect on both the signal in-
tensity of T1- and T2 -weighted imag-
es, as they shorten the T1 and the T2 of
their surroundings (figure 49), mean-
ing that the respective curves are
shifted towards the left.
In effect, for a given TR, there is
more signal, for a given TE, there is
less signal.
72 MRI made easy Contrast media

Fig. 50: In (A), the T1-curves for tissue A and B are very close to each other,
resulting in only a small difference in signal intensity between the tissues
at TR. In (B), the T1-curve of tissue A is shifted to the left, as contrast agent
entered tissue A but not tissue B. At the same time TR, there now is a much
greater difference in signal intensity, i.e. tissue contrast.
 Contrast media MRI made easy 73

In figure 50, the signal intensity for two As loss of signal is often more dif-
tissues, A and B, is illustrated. The i.v. ficult to appreciate than a signal en-
administered Gd contrast medium en- hancement, T1-weighted images are
ters tissue A, but not tissue B. the predominant imaging technique
The T1 of tissue A becomes shorter used after contrast medium injection.
and the T1-curve is shifted to the left. As the contrast media are not distrib-
The result is that the signal from tis- uted evenly throughout the body, sig-
sue A at time TR is stronger than it was nals from different tissues will be influ-
before, and the two tissues can be bet- enced differently. Vascularized tumor
ter differentiated, because there is bet- tissues are enhanced, for example. This
ter contrast. may, for example, help with differen-
What happens, when we perform a tiation between tumor tissue and sur-
T2 -weighted examination after con- rounding edema, which might other-
trast medium application, we have seen wise be indistinguishable.
in figure 49: the T2 -curve is shifted to It is also important that the Gd com-
the left, reducing the signal coming at pounds do not go through the intact,
a given TE. but only through the disrupted blood-
brain barrier.
In general, it has been shown that
the use of contrast media increases le-

What magnetic sion detection and diagnostic accuracy

of MRI in very many cases.
field are you
hanging around
Ready for a review?
in? As we know by now, many
parameters, e.g. T1, T2, pro-
ton density, pulse sequence
parameters, influence the
appearance of tissues in an MR image.
7 With a short TR, we get a T1-weight-

ed image.
7 With long TE, the image is T2 -weight-

ed.
7 Flow effects can be variable, and

cover the spectrum from signal loss to

signal enhancement.
7 Paramagnetic substances, e.g. the con-

trast medium Gadolinium-DTPA,shorten

T1 and T2 of the surrounding protons.
This results in a signal increase in T1-
weighted images and a signal decrease
in T2-weighted images.
7 T1-weighted imaging is the preferred

technique after contrast medium injec-

tion.
74 MRI made easy Saturation recovery

Fig. 51: Schematic illustration of the partial

saturation/saturation recovery sequence.

Partial saturation/ concepts. So let us take a look at them.

Pulse sequences that use 90° pulses on-
saturation recovery ly, are the saturation recovery pulse

sequence sequence and the partial saturation

sequence (figure 51) (which we have
We already heard about the term pulse already discussed, but we did not give
sequence. Many different pulse se- them a name).
quences have been developed, and we Basically, the sequences are the same:
should be familiar with their basic they consist of two 90° pulses. The dif-
 Saturation recovery MRI made easy 75

Fig. 52: Signal intensity of tissues having a different

T1 depending on the choice of TR: With a long TR,
the saturation recovery sequence, image contrast
is determined mainly by proton (spin) density. With
a shorter TR, the partial saturation sequence, the
resulting image is T1-weighted.

ference is in the time interval between (the protons have relaxed, are saturat-
pulses, the TR (see page 45). ed), the signal is influenced by the pro-
You can see the effect in figure 52 ton density (Do you recall the stories
with the T1-curves (going uphill!) of with the short trousers and the long
two different tissues. If we send in the teas?). With a TR short, with the partial
second pulse after a long time, TR long, saturation (protons have not relaxed),
both tissues will have regained longi- the T1 becomes important for the signal
tudinal magnetization. With a TR long , intensity, so we get T1-weighted images
with the saturation recovery sequence (figure 52).
76 MRI made easy Inversion recovery sequence

Inversion recovery for the net magnetic moment pointing

up, now point down).
sequence This is illustrated in figure 54 for
In contrast to the spin echo sequence two tissues with different T1. The tis-
that we have mentioned before sue with the faster longitudinal relax-
(see page 57), the inversion recovery ation, i.e. the shorter T1, is in the bot-
sequence uses first a 180° pulse which tom row.
is then followed by a 90° pulse (figure If we do not do anything else, the
53). longitudinal magnetization will slow-
What happens? ly go back up, like a ball that is thrown
The 180° pulse turns the longitudinal into water. To get a measurable sig-
magnetization in the opposite direc- nal, however, we need some transver-
tion (all protons that were responsible sal magnetization. And for this, we use

Fig. 53: Schematic illustration of the inversion recovery sequence.

 Inversion recovery sequence MRI made easy 77

the 90° pulse. The 90° pulse “tilts” the original value. So we get a T1-weighted
magnetization into the transversal image – which is even more T1-weight-
(x-y-) plane, so it can be measured/re- ed than partial saturation recovery im-
ceived. ages.
The signal that we get depends on Interestingly, when the 90° pulse is
the time between the 180°- and the 90° sent in when the longitudinal mag-
pulse, the time after the inversion by netization goes from negative to posi-
the 180° pulse; this time is thus called tive, i.e. is zero, the tissue does not give
TI = inversion time. a signal! This may be useful when we
The signal intensity in an inversion do not want a tissue to show up in the
recovery image is dependent on T1, image, e.g. when we want to suppress
which determines how fast the longi- the fat signal – but let us not go into too
tudinal magnetization goes back to its much detail here.

Fig. 54: The inversion recovery sequence uses a 180° pulse which inverts the longitudinal magnetization, fol-
lowed by a 90° pulse after the time Tl. The 90° pulse “tilts” the magnetization into the transversal (x-y-) plane,
so it can be measured/received. The tissue in the bottom row goes back to its original longitudinal magnetization
faster, thus has the shorter T1. For the time Tl which is illustrated, this results in less transversal magnetization
after the 90° pulse.
78 MRI made easy Spin echo sequence

Spin echo sequence The disadvantage is, however that the

signal becomes weaker and weaker.
We have talked about the spin echo se- What were the imaging parameters
quence in detail already. that influenced the MR signal in the
It is composed of two pulses: a 90°- spin echo sequences?
and a 180° pulse (figure 55). These were: TE = the time between
You should be able to recall what hap- the 90° pulse and the echo and TR = the
pens by now: The 90° pulse establishes time between two pulse sequences, i.e.
transversal magnetization, which im- from one 90° pulse to the next.
mediately starts to decrease because What did the TE and the TR do?
the protons dephase. Some time (TE/2) They determined how the resulting
after the 90° pulse, we send in a 180° image was weighted: TE was responsi-
pulse, which rephases the protons. ble for the T2 -weighting, TR for the T1-
After the time TE, we get a strong sig- weighting.
nal, the spin echo. As we have heard, we If you cannot remember this or still
can produce not only one, but several are feeling unsure, please go back to
echoes using more than one 180° pulse. pages 48 – 63 again.

Fig. 55: Schematic illustration of a spin echo pulse sequence. This is repeatedly
illustrated, as the spin echo sequence is so important.
 Fast imaging MRI made easy 79

(see pages 58-61), as we have already

Fast imaging sequences
seen; so there is only very little longi-
tudinal magnetization to be tilted by
What about those fast the next pulse, yielding very little sig-
nal.
imaging sequences? These problems are solved as follows:
Normal imaging sequences take quite 7 We use a different way to refocus

some time, during which it may be dif- the dephasing spins: instead of a 180°
ficult for the patient to lay completely pulse, we apply a magnetic field gradi-
still. In addition, there is always some ent. This means that an uneven mag-
unavoidable motion, like respiration netic field, a gradient field, is added/su-
and heart beat. All these movements perimposed on the existing magnetic
unfortunately decrease image quality. field.
To help with these problems, fast The magnetic field gradient is
pulse sequences were developed, switched on for a short time. This re-
which take less time. Most of these sults in even larger magnetic field in-
have strange names such as FLASH homogeneities in the examined slice.
(Fast – Low – Angle – Shot), or GRASS (The magnetic field inhomogeneities
(Gradient – Recalled – Acquisition – at that already exist at that time are due
Steady – State). These sequences are to inhomogeneities of the external
very important nowadays in daily magnetic field, and the internal mag-
practice. netic field inhomogeneities inside of
Here is just a rough outline – you the tissues, which we talked about ear-
can find more details in textbooks or lier – if you do not remember this, go
the book “MR Buzzology”. back to page 27 for a short recap).
As you may already have noticed, Due to these larger magnetic field in-
the TR is the most time consuming pa- homogeneities, transversal magneti-
rameter of an imaging sequence (see zation, and thus the signal, disappears
also pages 57 and 83). It makes sense to faster (protons dephase faster!). Then
shorten TR if we want to make imag- the magnetic gradient is switched off,
ing faster. And this is done in the fast and after a short time turned back on
imaging sequences. with the same strength, but in the op-
But with a decreasing TR, there are posite direction.
some problems: The faster moving protons now be-
7 Firstly, with a spin echo sequence come the ones that move slowly, and
we used a 180° pulse to refocus the de- vice versa (similar to what happens af-
phasing spins. Unfortunately, we can- ter a 180° pulse).
not use a 180° pulse for this purpose, This results in some rephasing, and
when we do imaging with a very short thus the signal increases again to a
TR: it requires some time to “produce” certain maximum, which is called a
and to deliver a 180° pulse, and with a gradient echo. After this echo, the sig-
very short TR, there will not be enough nal decreases again.
time for that between the 90° pulses. What to do about the second prob-
7 Secondly, with decreasing TR, lon- lem, the small amount of longitudinal
gitudinal magnetization will have re- magnetization with a short TR? The
covered less and less between pulses 90° pulse, e.g. in a spin echo sequence,
80 MRI made easy Fast imaging

abolishes longitudinal magnetization; We save imaging time because

longitudinal magnetization, however, – with small flip angles we only need
starts to recover immediately after an RF pulse of short duration;
the 90° pulse, depending on the T1 of – we do not use a 180° refocussing
the tissue examined (if you have for- pulse (which takes time to be generat-
gotten, see page 38). The trick with the ed and to take effect);
fast imaging sequences is not to use a – we do not have to wait long TRs for
90° pulse, but pulses that cause small- enough longitudinal magnetization
er “flip angles” (mostly in the range of to reappear, as with small flip angles
10°-35°). there is always a reasonable amount
With these flip angles smaller than of longitudinal magnetization left af-
90 degrees, longitudinal magnetiza- ter the initial pulse.
tion is not totally abolished. Instead, With these fast scans, it is possible
there is always a substantial amount to do imaging in a second or even less.
of longitudinal magnetization left,
which can be “tilted” by the next
pulse; this gives a reasonable signal
Well, time to repeat
even if the next pulse comes in after a and take a break.
very short TR. Partial saturation and
As we have heard, a 180° pulse nor- saturation recovery se-
mally “neutralizes” the effects of ex- quences use 90° puls-
ternal magnetic field inhomogeneities. es. TR is relatively short
The decay of transversal magnetiza- with partial saturation and relative-
tion is then due to so-called T2-effects ly long with saturation recovery.
(see figure 35).
When we do not use such a 180°
pulse, the protons experience larger
magnetic field inhomogeneities and
get out of phase faster. Signal intensi- I can explain
ty decays faster, and in this case is due
everything, your
honor. The intake
to so-called T2*-effects (pronounced: T2
star-effects), which has already been
illustrated in figure 35. of alcohol
Besides these T2*-effects, other fac-
resulted in my
client’s...
tors, e.g. the flip angle, influence signal
intensity in the fast imaging sequenc-
es, which are also called gradient echo
sequences for obvious reasons.
Here are some guidelines about gra-
dient echo imaging:
7 Larger flip angles produce more T1-

weighting.
7 Longer TEs produce more T2*-weight-

ing.
7 With fast scans, intense signals of-

ten come out of the vessels.

 Imaging time MRI made easy 81

While saturation recovery yields

proton (spin) density images, the
About imaging time
As we have just seen, fast imaging se-
images are T1-weighted with partial
quences decrease imaging time.
saturation.
Is there any other way to decrease
7 A spin echo sequence has a 90°
this time? What does actually deter-
pulse, which is followed by one – or
mine the imaging time?
more – 180° pulses, to rephase the
For MR imaging with normal pulse
dephasing protons resulting in one –
sequences, this can be easily calculated;
or more – spin echoes. This sequence
the acquisition time (a.t.) is:
can give proton density-weighted,
a.t. = TR x N x Nex
T1-weighted, or T2 -weighted images.
This looks a little complicated but it
This is determined by the imaging
isn’t really. Let us start at the back. Nex
parameters which are chosen (TR,
is the number of excitations. What does
TE).
that mean?
7 In the inversion recovery sequence,
For certain reasons, it is necessary to
a 180° pulse is followed by a 90°
use not only one signal measurement,
pulse, resulting in T1-weighted im-
but to repeat the measurement several
ages.
times. As the MR signal coming out of
7 Fast imaging sequences use flip
the patient is very weak, it may be good
angles that are smaller than 90°,
to add up signals from several meas-
and so-called gradient echoes. Image
urements, to take several “averages”,
weighting is also determined by the
to get a good quality image.
specific type of sequence and the im-
Actually, what you get is an image
aging parameters chosen.
with a better signal-to-noise ratio.
Naturally, imaging time increases
with every additional measurement.

...increased spin, ...that caused the

which in turn coins in the till to
initiated a become magically
magnetization... attracted to him.
 Imaging time MRI made easy 81

While saturation recovery yields

proton (spin) density images, the
About imaging time
As we have just seen, fast imaging se-
images are T1-weighted with partial
quences decrease imaging time.
saturation.
Is there any other way to decrease
7 A spin echo sequence has a 90°
this time? What does actually deter-
pulse, which is followed by one – or
mine the imaging time?
more – 180° pulses, to rephase the
For MR imaging with normal pulse
dephasing protons resulting in one –
sequences, this can be easily calculated;
or more – spin echoes. This sequence
the acquisition time (a.t.) is:
can give proton density-weighted,
a.t. = TR x N x Nex
T1-weighted, or T2 -weighted images.
This looks a little complicated but it
This is determined by the imaging
isn’t really. Let us start at the back. Nex
parameters which are chosen (TR,
is the number of excitations. What does
TE).
that mean?
7 In the inversion recovery sequence,
For certain reasons, it is necessary to
a 180° pulse is followed by a 90°
use not only one signal measurement,
pulse, resulting in T1-weighted im-
but to repeat the measurement several
ages.
times. As the MR signal coming out of
7 Fast imaging sequences use flip
the patient is very weak, it may be good
angles that are smaller than 90°,
to add up signals from several meas-
and so-called gradient echoes. Image
urements, to take several “averages”,
weighting is also determined by the
to get a good quality image.
specific type of sequence and the im-
Actually, what you get is an image
aging parameters chosen.
with a better signal-to-noise ratio.
Naturally, imaging time increases
with every additional measurement.

...increased spin, ...that caused the

which in turn coins in the till to
initiated a become magically
magnetization... attracted to him.
82 MRI made easy Signal to noise

Repeated measurements result in

a better signal-to-noise ratio.

To illustrate this: Just imagine that you our formula: What is “N”? As you know
are sitting in a large audience, where peo- from other imaging methods (or your
ple are making a lot of noise. Someone sit- PC), pictures are made of picture ele-
ting next to you whispers something in ments, which all together make up the
your ear, but you cannot really under- image matrix, e.g. a 1,024 x 768 matrix
stand him, because there is so much back- has 768 rows of 1,024 picture elements
ground noise. What you will probably do, (pixels).
is ask him to repeat what he said once or In our equation, N is the number of
several times. You mentally add up the rows in a matrix, like rows in a letter.
information which you receive each time. The more rows you have, the more time
As this signal is always the same, it will it takes for the image.
increase by adding it up. The background Just think about this as if you were
noise, however, is not always the same. writing a letter: if you have paper with
Instead, it is random and fluctuates and 5 rows on a page, you will finish a page
does not add up the way the signal does. faster than if you have 25 rows to write.
So altogether you will have a better sig- However, you have more content, more
nal-to-noise ratio (which you would also detail on a page/picture, when you work
have if the person spoke louder). Back to with more rows.
 Multislice imaging MRI made easy 83

And why does TR influ- The longer the TR, the more slices we
can excite in the meantime.
ence acquisition time? So by just adding a little extra time,
If you choose a long time TR to repeat we will examine many slices instead
your pulse sequence, to perform ad- of one, and imaging time per slice de-
ditional signal measurements, imag- creases substantially.
ing takes longer than with a short TR. We perform so-called multislice im-
However, there is a trick that can short- aging. Another way to possibly reduce
en imaging time a bit. TR, and thus imaging time, is the use
While we are waiting to repeat our of a contrast medium: as we have seen,
imaging sequence in one slice, i.e. Gadolinium-DTPA shortens T1.
while we wait for TR to go by (slice A And when T1 is shorter, the TR can al-
in figure 56), we might as well make so be shorter, without a loss in signal
measurements in one or more differ- intensity of the tissue in question (see
ent slices (slices B, C and D in figure 56). figure 49).

Fig. 56: Multislice

imaging: while we wait
for the time TR to pass
by for another signal
measurement in slice A,
we perform signal meas-
urements in additional
slices. So during time
TR, we actually recorded
signals for more than
one slice.
84 MRI made easy MR signal intensity

Let us review called gradient coils. This gradient field

modifies the strength of the original
important factors that magnetic field.
influence signal In figure 57, magnetic field strength
increases for different cross sections
intensity in MR. from the feet towards the head.
These are: Consequently, the protons in the dif-
7 proton density (page 45) ferent slices experience different mag-
7 T1 (page 24) netic fields, and thus have different pre-
7 T2 (page 28) cession frequencies.
7 flow (page 67) So the RF pulses which disturb the
7 the pulse sequence (page 74-80) protons in the different slices, must
7 TR (page 45) have different frequencies as well −
7 TE (page 54) otherwise there would be no resonance.
7 TI (page 77) As gradient fields can be superim-
7 flip angle (page 80) posed in any direction, it is possible to
7 use of contrast medium (page 71) define not only transversal slices, but
all kinds of different imaging planes
If you are not sure about any of these, without moving the patient. The gradi-
go back to the corresponding page. ent field that enables us to examine a
If you feel familiar with these facts, specific slice, is also called slice select-
continue with the next section, and ing gradient.
learn about some important things in
MR imaging that we have not talked
Hmmm.
Sounds
about yet.
My wife
always says, familiar.
How can we select a They’re
slice which we want to I’m easy to see
through … better than
examine? an MR.
When we put a patient into an MR scan-
ner, he or she is in a rather homoge-
neous magnetic field.
So all the protons in the whole body
have the same Larmor frequency, and
will be excited/disturbed by the same
RF pulse. To examine a specific slice
only, a second magnetic field is super-
imposed on the external field, which
has different strengths in varying lo-
cations. The magnetic field is therefore
stronger or weaker in some places than
in others (figure 57).
This additional field is called a gra-
dient field, and is produced by the so-
 Spatial encoding MRI made easy 85

Fig. 57: Magnetic gradient fields are superimposed on the field of the MR magnet,
so that different cross sections of the body experience magnetic fields of differing
strength. In the illustration, the resulting magnetic field strength is increasing from 1.4
Tesla at the feet to 1.6 Tesla at the head. As magnetic field strength and precessing/
resonant frequency are directly correlated (Larmor equation), the resonant frequency
at the feet is about 60 MHz, while it is about 68 MHz at the top of the head in our
example. By selecting a certain RF pulse frequency, we determine the location of the
slice which we examine.
86 MRI made easy Slice thickness

How can we determine head. If we, however, have a steeper gra-

dient field, i.e. one that has more differ-
or select a certain slice ence in field strength over a specific dis-
thickness? tance, the precession frequencies will
also vary to a larger degree, let us say
We can select a different slice thickness from 56 MHz to 72 MHz.
in two ways (figure 58): If we now use an RF pulse of the same
One solution is to send in not on- bandwidth as in A, containing frequen-
ly one specific frequency (which is not cies between 64 and 65 MHz, the slice
done in practice), but an RF pulse that thickness in our example C with the
has a range of frequencies, which is of- steeper gradient field is, however, small-
ten referred to as bandwidth; the wid- er than in our example A with the more
er the range of frequencies, the wider shallow gradient field.
the bandwidth, the thicker the slice in So using the same range of radio fre-
which protons will be excited. This is il- quencies, the same bandwidth as it is
lustrated in figure 58. called, slice thickness can be modified
If we use an RF pulse with frequen- by the slope of the gradient field.
cies from 64 to 65 MHz, we will get the
thickness of slice 1 (figure 58A). If, how-
ever, we only use frequencies from 64
to 64.5 MHz, the protons in the thinner I have this
slice 2, will show resonance (figure 58B).
There is another way to select a differ-
feeling that it
ent slice thickness: isn’t really an
In our example, we used a gradient MRI…
field that “produced” precessing or reso-
nant frequencies starting at 60 MHz at
the feet, up to 68 MHz at the top of the
 Slice thickness MRI made easy 87

Fig. 58: There are two ways to determine slice thickness. The first is to use an RF pulse that has not only one spe-
cific frequency, but a certain range of frequencies, a so-called bandwidth. If, for example, we send in an RF pulse,
which contains frequencies between 64 and 65 MHz, protons in slice 1 will be influenced by the RF pulse.
When the RF pulse only contains frequencies between 64 MHz and 64.5 MHz, thus has a smaller bandwidth, slice 2,
which is half as thick as slice 1, will be imaged.
When there is more difference in magnetic field strength between the level of the feet and the head, i.e. the
magnetic gradient is steeper, the resulting slice will be thinner, even though the RF pulse bandwidth is the same.
This is illustrated in (C), where the magnetic field strength varies more between the feet and the head than in (A);
the corresponding resonant frequencies are 56 to 72 MHz in (C) vs. 60 to 68 MHz in (A). Using the same RF pulse
containing frequencies from 64 to 65 MHz results in imaging of a thinner slice 3 in (C) than in (A).
88 MRI made easy Frequency encoding

Where does the signal image? The trick is similar to the slice select-
ing gradient, which is turned on only during
come from? application of the RF pulse.
Now we have selected position and After the RF pulse is sent in, all protons in
thickness of our slice. But how can we the slice precess with the same frequency.
find out, from what point of our slice a We now apply another gradient field which
certain signal is coming from – informa- – in our example – decreases from left to right.
tion that we must have to construct an So the precession frequency of the protons

Fig. 59: To determine where

in a certain slice a sig-
nal comes from, we use a
magnetic gradient field. In (A),
nine protons in the same slice
are depicted. They precess in
phase with the same frequency,
after the RF pulse is sent in. A
magnetic gradient field is then
superimposed on the external
field, which in (B) decreases in
strength from left to right. The
protons in the three columns
now experience different
magnetic fields, and thus give
off their signals with different
frequencies (e.g. 65, 64 and
63 MHz). The corresponding
magnetic gradient is called the
frequency encoding gradient.
We now can tell from which
column a signal comes from,
but still cannot pinpoint the
exact place of origin.
 Frequency encoding MRI made easy 89

will also decrease from left to right (in protons in one column will still have sig-
our example, the precession frequencies nals with the same frequency. We now
are 65, 64 and 63 MHz, respectively). can tell by the frequency from which col-
The result is that the protons in the umn a signal comes from, but still can-
different columns emit their signals not pinpoint the exact place of origin in a
with these different frequencies. The particular column. As this is not enough
gradient applied is thus called the fre- spatial information, we have to do some-
quency encoding gradient. However, all thing else.
90 MRI made easy Phase encoding

Fig. 60: To find out where in a column with the same frequency a certain signal comes from, we use an ad-
ditional gradient. In (A), the column with the precession frequency of 65 MHz from figure 59 is depicted. We now
switch on a gradient field, which is stronger at the top than at the bottom of the column for a very short time (B).
The proton at the top thus precesses faster than the one in the middle, which in turn precesses faster than the
proton at the bottom. This difference in precessing frequency only lasts for a very short time; when the gradient
is switched off, all protons experience the same magnetic field again, thus have the same 65 MHz precession
 Phase encoding MRI made easy 91

Theoretically, we could use the same trick

with the magnetic gradient again. This, how-
ever, causes some practical difficulties, e.g.
this may result in two points at different lo-
cations having the same frequency. To solve
the problem, something different is done,
which is illustrated in figure 60.
In figure 60, we just look at the protons of
one column from figure 59, the 65 MHz col-
umn. The protons are in phase after the RF
pulse “whipping”. Now we apply a magnetic
gradient along this column for a short time.
This causes the protons to speed up their pre-
cession, according to the strength of the mag-
netic field to which they are being exposed.
In our example (figure 60B), the proton at the
top thus precesses faster than the one in the
middle, which in turn precesses faster than
the proton at the bottom.
When this short gradient is switched off,
all the protons of the column experience the
same magnetic field again, and thus have the
same precession frequency. However, there is
an important difference: Formerly, the pro-
tons (and their signals) were in phase. Now
the protons and their signals still have the
same frequency, but they are out of phase.
(This can be viewed as if their magnetic vec-
tors come by the antenna at different times.)
As the gradient which we used causes pro-
tons to precess in different phases, it is called
the phase encoding gradient.
How this phase encoding actually works
is too complicated for a basic introduction; if
you are interested in more details, grab a copy
of “MR Buzzology” for further reading.
For here it is enough to know that all of
these different signals now can be assigned
to a certain location in the slice, so we now
can reconstruct our image.
frequency again (C). However, now we have a little difference among
these protons: even though they precess with the same frequency
again, they are a little out of phase, and consequently give off signals
of the same frequency, which are different in phase and, because of
this, can be differentiated. The corresponding gradient is called the
phase encoding gradient.
92 MRI made easy Spatial encoding

Let us repeat: With all this frequency and phase in-

formation we can now assign a cer-
7 We can select a slice to tain signal to a specific location which
be examined by using a results in our MR image (... finally!).
gradient field, which is su-
perimposed on the exter-
nal magnetic field. Protons along this A few more basics
gradient field are exposed to different So far we have discussed just about
magnetic field strengths, and thus every important aspect of MR basics.
have different precessing frequen- But: why have we always talked on-
cies. As they have different preces- ly about the proton, the hydrogen nu-
sion frequencies, we can send in an cleus? What about other nuclei?
RF pulse that contains only those fre- As you recall, atoms have a nucle-
quencies, which excite the protons in us made up by protons and neutrons.
the slice we want to image. An exception is the hydrogen nucle-
7 Slice thickness can be altered in two us, which only consists of one proton.
ways: by changing the bandwidth And when we talk about the proton,
of the RF pulse, or by modifying the we talk about the hydrogen nucleus,
steepness of the gradient field. as both are the same (the terms pro-
7 The slice selecting gradient is only ton and hydrogen nucleus can thus
turned on during the RF pulse. be used interchangeably). The hydro-
7 To determine the point in a slice gen nucleus is best for MR imaging, as
from which a certain signal is com- hydrogen occurs in large abundance
ing, we use two other gradients, the throughout the body. Hydrogen also
frequency encoding gradient and the gives the best signal among the nu-
phase encoding gradient. clei: from an equal number of differ-
7 The frequency encoding gradient is ent nuclei in the same magnetic field,
sent in after the slice selection gradi- hydrogen gives the most intense sig-
ent. In our example, it is applied in the nal. All of the routine MR imaging is
direction of the x-axis. This results in proton/hydrogen imaging nowadays.
different precession frequencies along However, research is being done on
the x-axis, and thus different frequen- the use of other nuclei, like sodium.
cies of the corresponding signals.
7 The phase encoding gradient is

turned on for a short time after the

RF pulse, along the y-axis in our ex-
ample. During this short time, the
protons along the y-axis precess with
different frequencies. When this gra-
dient is switched off, they go back to
their former precession frequency,
which was the same for all of them.
Due to this phase encoding gradient,
however, the protons and their sig-
nals are now out of phase, which can
be detected.
 Nuclei for MR MRI made easy 93

Can we use all other ics, so we will only talk about the pro-
tons). Why an odd number? Just think
nuclei for imaging? about the proton as a little bar mag-
The answer is no. There are two im- net. If you have a nucleus with two
portant pre-conditions both of which (or any other even number) protons,
must be fulfilled. these little bar magnets would cling
7 Firstly, we can only use nuclei that together like any other magnets (op-
have a spin. posite poles attract).
This can be easily explained: as we The result: their magnetic moments
saw at the beginning, the protons were would cancel each other out. If we
spinning around, and thus their elec- have a nucleus with an odd number
trical charge was also spinning, mov- of protons, e.g. three, pairs of protons
ing. And the moving electrical charge will still cling together and neutralize
was the current that caused the mag- each other. However, there will always
netic field of the proton, which was be one proton left that still has a mag-
the basis for everything. If it weren’t netic moment. Nuclei with odd num-
for the spin, there would be no mag- bers of protons thus have a magnetic
netic field. moment, and can principally be used
7 Secondly, the nucleus must have an for MRI.
odd number of protons (and neutrons, Examples are: carbon-13, fluorine-19,
but this will go into too much phys- sodium-23, phosphorus-31.

What a day!
Give me a
180°-rephasing
high-frequency
impulse!
94 MRI made easy Hardware

MR Hardware – precession frequency can be, has already

been illustrated on page 27. Homogenei-
an overview ty of the magnetic field can be improved
Let us have a look at some hardware. The by making some electrical or mechani-
most important part of the MR machine cal adjustments, a process called shim-
is the main magnet, which has to be ming.
pretty strong to allow MR imaging. The
strength of a magnet is given in Tesla
or Gauss, where 1 Tesla = 10,000 Gauss. Types of magnets
Gauss was a German mathematician, In MRI, different types of magnets have
who was the first to measure the geo- been used – here a short description.
magnetic field of the earth. Tesla is con-
sidered to be the “father” of the alternat- Permanent magnets:
ing current. He was a peculiar fellow, Everybody is probably familiar with
having refused to share the Nobel prize a permanent magnet. It is that type of
with the inventor Thomas Edison in the magnet that fascinates little kids. This
early 1900s. kind of magnet is always magnetic and
Magnets used for imaging most- does not use any energy for work, which
ly have field strengths up to 1.5 Tesla, are its advantages.
meanwhile also 3 Tesla magnets have Possible disadvantages are thermal
become popular (they are referred to as instability, its limited field strength, and
“high field”, with the term “ultra-high its weight (a magnet of 0.3 T may weight
field” used for even stronger magnets). about 100 tons!).
Basically, the stronger the magnet, the This magnet is only used in low field
better the MR signal. Unfortunate- systems today.
ly, technical problems and image arti-
facts also increase with magnetic field Resistive magnets:
strength. Because of that, only magnets In a resistive magnet, an electrical cur-
up to 3 Tesla are useful for general clini- rent is passed through a loop of wire
cal work at present. and generates a magnetic field. Resis-
Just to get an idea about the strength tive magnets are therefore also called
of the MR magnets: the earth’s magnet- electromagnets. They are only magnet-
ic field is between 0.3 and 0.7 G, the mag- ic as long as there is an electrical cur-
net of a refrigerator door has about 100 rent flowing through them. Thus, they
G = 0.01 T). use electrical energy.
The magnetic field of MR magnets has As there is a resistance to the flow of
to be very homogeneous, as it directly the electricity through the wire, these
determines the precession frequency. magnets get warm when in operation,
The homogeneity is quoted in terms as and have to be cooled.
ppm, part per million, in a defined vol- Compared with permanent magnets,
ume. To calculate this, the difference be- they achieve a higher field strength.
tween maximum and minimum field However, resistive magnets are not very
strength is divided by the average field practical with high field strengths, be-
strength and multiplied by one million. cause they create lots of heat that must
How detrimental even rather small in- be dissipated. In general, resistive mag-
homogeneities and thus differences in nets are no longer of central interest.
 Hardware MRI made easy 95

Superconducting magnets: will be a loss of superconductivity (so-

Superconducting magnets are the ones called quench), and sudden resistance
most widely used in MR machines. to the flow of electricity. This results in
They also make use of electricity, but rapid heat production, which causes
they have a special current-carrying cryogens to boil off rapidly (these leave
conductor. This is cooled down to su- the system via the so-called quench
perconducting temperature (about 4 °K lines). Advantages of superconduct-
or -269 °C). At this temperature, the cur- ing magnets are high magnetic field
rent conducting material loses its resis- strength and excellent magnetic field
tance for electricity. So if you send in an homogeneity. (This is in the order of
electrical current once, it flows in there 10-50 ppm over a region 45 cm in di-
permanently, creating a constant mag- ameter). High field strength and field
netic field. So-called cryogens (helium, homogeneity facilitate very detailed
nitrogen) are used for cooling of these and fast studies, and allow for spec-
magnets, and have to be refilled once in troscopy.
a while. Disadvantages of the superconduct-
When for some reason the tempera- ing magnets are their relatively high
ture rises above the superconducting costs, and use of rather expensive
temperature in these magnets, there cryogens.

Your insurance card, please. And do you have

your protons with you?
96 MRI made easy Hardware

Which is the ideal field strength? Gradient coils

This question is as easy to answer as Gradient coils are used to systemati-
the question about the ideal horsepow- cally vary the magnetic field by pro-
er for a car. Here are some of the pros ducing additional linear electromag-
and cons: netic fields, thus making slice selec-
7 higher field strength allows for a bet- tion and spatial information possible.
ter spatial resolution and faster exami- As we have three dimensions in space,
nations, and may be used for spectros- there are three sets of gradient coils. As
copy; these coils bang against their anchor-
7 low field systems on the other hand ing devices, they are the cause of noise
offer better tissue contrast, are cheaper that you can hear during an MR exam-
in price and in operating costs. ination.
Most MR units today are 1.5 Tes- Surface coils
la systems, with 3.0 Tesla systems be- Surface coils are placed directly on
coming increasingly popular. the area of interest, and have different
shapes corresponding to the part to be
examined.
Another piece of They are usually receiver coils only,
most of the received signal coming from
hardware: the coils tissues nearby; deeper structures cannot
In MRI, radio frequency coils are nec- be examined with these coils. As with
essary to send in the RF pulse to excite the head coils, the RF pulse is transmit-
the protons, and to receive the result- ted by the body coil in these cases.
ing signal. Coil technology is extreme- Shim coils
ly important. The same or different As we have already mentioned in con-
coils can be used for transmission of nection with the magnets, magnetic
the RF pulse and receiving the signal. fields have inhomogeneities. Better ho-
A variety of coils are in use − here just mogeneity can be achieved by electri-
a few comments. cal and mechanical adjustments. For
Volume coils this process, which is called shimming,
Volume coils are used in all MR units. the shim coils are used.
These completely surround the part of
the body that is to be imaged. These
volume coils should be close to the size
of the subject.
Why do MR units require
The body coil is a permanent part of special facilities?
the scanner, and surrounds the patient. As the systems usually weigh a lot,
It is important, as it is the transmitter there are certain static requirements to
for all types of examinations. It also re- be met. However, there are additional
ceives the signal when larger parts of important factors.
the body are imaged. Head coils, the The strong magnetic field of the MR
most frequently used dedicated coil system extends beyond the magnet.
type, may act as receiver coil (with the Naturally, the magnetic field can at-
body coil transmitting the RF pulses), tract (even very heavy!) metallic ob-
or may transmit the RF pulses as well jects and transform them into projec-
(so-called transmit-receive coils). tiles! So these have to stay outside the
 Hardware MRI made easy 97

examination room. Also the magnetic

field influences mechanical and electri-
cal devices, like computers, monitors,
pacemakers and X-ray units – so such
devices must be kept at a certain dis-
tance away from the MR unit.
On the other hand, there are also ex-
ternal influences. The whole air is full

If you listen to the

of radio waves – just think about all the
stations which you can receive on your
music you say you do, radio. To prevent interference between
your protons are outside radio waves and those sent

already accustomed to
from the MR unit, the whole system
has to be shielded by a Faraday cage.
quite a bit! In addition, it has to be taken into ac-
count that larger metallic objects, es-
pecially when moving (like elevators,
cars), can influence the magnetic field
of the scanner, and should also be kept
away from the MR unit.

MR spectroscopy
MR spectroscopy has been in use for a
long time, long before MR was used for
imaging. The procedure is used as an
analytical tool, as it can identify vari-
ous chemical states of certain elements
without destruction of the sample.
Meanwhile, spectroscopy and imaging
may be combined (spectroscopic imag-
ing). This enables us to obtain in vivo
information about the chemistry and
metabolism in specific locations, like in
the brain, the liver, or even the heart.
As these measurements can be re-
peated without harm, follow-up stud-
ies of cell physiology are possible. This,
for example, can be useful in the evalu-
ation of certain diseases and the effects
of therapy.
As spectroscopy requires very ho-
mogeneous magnets with higher field
strengths, it can only be performed
with the use of MR units which have
superconducting magnets.
98 MRI made easy The final

The final review

Now that you have made it
up to here, it is our sincere
hope that you know a little
bit (more?) about MRI. A fi-
nal review?
Yes, but let us try a different approach
this time: take a look at the index on the
following pages. Check and see if you
understand all of the terms mentioned.
If not, refer back to the page numbers
listed for a short review.
If you understand all or at least most
of it, be happy about it!

If you’ve made it this far, a second book

shouldn’t be any problem at all.