Latest / European Society for Vascular Surgery / The Secret Life Of MRI/MRA

Transcript

The Secret Life Of MRI/MRA · European Society for Vascular Surgery

0:12
Welcome to the ESVS Podcasts. I'm Justin Woolgar
0:16
and today you join me for another exciting episode
0:20
in our Secret Life series. where I'll be looking
0:23
at the engineering behind some of our complex
0:27
pieces of kit that make our job as surgeons easier.
0:31
This series of podcasts was inspired by an old
0:35
TV series called The Secret Life of Machines,
0:39
where everyday objects that we surround ourselves
0:42
with, such as telephones, televisions, radios,
0:46
etc., were broken down and explained in simple
0:50
language. In this episode, I'll be looking at
0:54
possibly one of the most complicated machines
0:56
out there. Yes, the Magnetic Resonance Imaging
1:00
or MRI scanner. And specifically, how we use
1:05
it for MR and geography. Unlike the TV series,
1:10
here at the ESVS, we take things to a different
1:13
level. Not only do we take things apart, but
1:17
we build machines from scratch. and I will tell
1:20
you exactly how to do it. In our previous episodes,
1:25
I have demonstrated a homemade Doppler machine
1:28
and a stent. And in this episode, as promised,
1:33
I will be showing you how to build your own MRI
1:36
machine. No more hounding the radiologists with
1:40
those request forms when you can simply do the
1:43
job yourself. In order to understand how an MRI
1:46
scanner works, we have to go back to basics.
1:50
In a nutshell, MRI creates detailed images inside
1:55
the body using strong magnetic fields, radio
1:59
waves and signals from hydrogen atoms. The scanner
2:03
produces an extremely powerful magnetic field
2:06
which causes the hydrogen atoms in your body
2:09
to align either with or against the magnetic
2:12
field. This happens because the hydrogen atoms,
2:16
of which there are many, due to the high water
2:19
and fat content, behave like tiny bar magnets
2:23
when exposed to this intense field. The scanner
2:27
then emits a serious pulse of RF, or radio frequency,
2:32
which knocks the hydrogen atoms out of alignment.
2:36
When the pulse stops, the nuclei try to return
2:40
to the original alignment, emitting a small radio
2:44
signal as they relax back. Relax. All you need
2:49
to do is detect these signals, make sense of
2:52
it with a computer and hey presto, an MRI image.
2:57
Believe it or not, a DIY MRI scanner is actually
3:01
quite doable when you break down the main components.
3:05
If you can actually imagine yourself building
3:08
one, then you'll certainly get a better understanding
3:11
of how it actually works. The basic parts of
3:16
a scanner consist of a super vicious magnet,
3:19
which we will talk about in a moment, a load
3:22
of copper wire known as the gradient coils, which
3:25
vary the magnetic field and allow precise 3D
3:29
localization of the signals, another set of coils
3:32
to provide the RF input, and some form of detector
3:36
coils. I think I'm going to need shares in a
3:39
copper mine. The biggest challenge when building
3:42
your own MRI scanner is the magnet. The typical
3:45
magnet in your hospital MRI is truly a marvel
3:49
of engineering and physics. It is essentially
3:52
a large superconducting electromagnet made from
3:57
niobium titanium alloy wire embedded in a copper
4:01
matrix. To get it into a superconducting state,
4:06
The whole thing has to be cryogenically cooled
4:09
to minus 269 Celsius with liquid helium. It's
4:16
at this temperature the magic starts. To get
4:20
it going, simply apply 500 amps and close the
4:24
switch. Because the resistance of the system
4:28
is virtually zero, the current in the electromagnet
4:32
circulates continuously. and never dies out.
4:37
Effectively, the magnet is on all the time. So
4:41
if you're thinking of having a party in the hospital's
4:45
MRI suite late at night, you might have to rethink
4:49
that idea, especially if you have any metal in
4:52
your pockets. To give you an idea of how strong
4:57
these magnets are, let's compare it with a typical
5:01
fridge magnet. You know, one of those tacky souvenirs
5:05
you buy when you go to Majorca or Benidorm. They
5:09
typically have a strength of about 1 milliTesla
5:12
compared to the 1 .5 Tesla of your typical MRI
5:17
scanner. In other words, an MRI scanner is 1500
5:23
times more powerful than a fridge magnet. This
5:27
becomes even greater when you think about the
5:29
three Tesla machines that are out there. Well,
5:33
now that we know what we need, I'm off out to
5:37
my garage and see if we can knock up a suitable
5:39
magnet. I think I might have to improvise somewhat
5:43
as I don't have any niobium titanium alloy wire
5:48
or liquid helium to hand, but I do have a few
5:52
neodymium speaker magnets and an old carbon dioxide
5:57
fire extinguisher to cool things down. this might
6:02
just work. Welcome back to my rather cold workshop
6:06
and today we are going to see if we can build
6:10
an MRI magnet. Now this is not going to be a
6:15
huge machine so I'm figuring that I'm going to
6:18
probably need a mounting frame of about 60 to
6:21
70 centimeters in diameter possibly about half
6:25
a meter long and In order to make the magnet
6:29
we're going to use some of these and in my tin
6:34
I've got a couple of hundred neodymium magnets
6:38
which we're going to stick all the way around
6:42
the circumference of the mounting frame and I
6:47
think we might need a few more than a couple
6:49
of hundred possibly in the order of two to three
6:52
thousand so I'm looking at my tin. I'm a bit
6:55
short but I'm sure we can find some more to do
6:58
the job. Now the other thing which we have to
7:01
do is cool the whole lot down. Now what I've
7:04
got in front of me here is a rather vicious carbon
7:10
dioxide fire extinguisher which we're going to
7:14
cool the whole whole lot down with to make it
7:17
well it won't be superconducting completely but
7:20
it might just help the magnets somewhat. You
7:23
get my drift. The next thing we are going to
7:26
need for our MRI scanner is a set of gradient
7:29
coils. These are a number of copper coils arranged
7:33
in separate x y and z axes and allow the MRI
7:39
to know where a signal is coming from. The gradient
7:43
coils have switched DC current flowing through
7:45
them and create a slight change in the magnetic
7:48
field strength across space. The change in magnetic
7:53
field that occurs when the RF is applied allows
7:57
the source to be localised using either frequency
8:00
or phase encoding, which I'm not going to get
8:03
into. And providing my computer is powerful enough,
8:08
allows full 2D and 3D localisation. Have you
8:13
ever wondered why MRI scanners are so noisy?
8:17
Well, it's all due to the current in the gradient
8:20
coils interacting with the magnetic field resulting
8:24
in vibrations. In powerful scanners or when there
8:29
is fast gradient switching, as in diffusion -weighted
8:33
MRI, this can actually cause the patient to feel...
8:37
a tingling sensation in their fingers or toes
8:39
due to induced voltages in the peripheral nerves.
8:43
It really gives an idea of just how powerful
8:45
these magnetic fields are. I'm going to find
8:48
a large diameter section of plastic pipe about
8:53
a foot across and wind each of these coils by
8:57
hand to cover all the directions and glue them
9:00
into position. Easy peasy. This will then sit
9:05
inside my magnet array. As you might have guessed,
9:11
this is not going to be a full -size machine.
9:13
I can't afford the copper. Right, what else do
9:17
we need? Ah yes, another set of coils this time
9:20
for the RF transmit and receive. The RF coils
9:24
are essentially multiple loop antennas matched
9:27
to 50 ohm impedance, which happens to be the
9:31
standard in RF engineering. I'll lash up another...
9:35
former with a load of coils and hopefully that
9:39
will do the job that will then sit inside my
9:41
gradient coil former so so far so good and i
9:46
think you might be able to get the gist of how
9:48
this thing is being constructed the rf will be
9:53
relatively easy to get the ham radio transmitter
9:57
plus my homemade amp should give it about 60
10:00
odd megahertz at a couple of kilowatts with a
10:03
bit of a push. So after putting all that lot
10:05
together, we need to have a quick tea break and
10:08
talk about how exactly we are going to interpret
10:11
our MRI or MRA image if this whole setup actually
10:16
works. I'll try and keep this explanation as
10:19
simple as possible as it is somewhat difficult
10:22
to get your head around it. After the radio frequency
10:25
pulse, the decay of the magnetic resonance signal
10:28
is described by two relaxation processes, T1
10:33
and T2. Essentially, T1 describes the rate at
10:37
which magnetization realigns itself along the
10:41
external magnetic field direction, and T2 describes
10:46
the rate of decay of the magnetization component
10:50
along the transverse axis. MRA sequences typically
10:55
use either time of flight techniques to encode
10:59
flow as an apparent T1 shortening through the
11:04
washing of fully relaxed blood from outside the
11:08
image volume. This results in an enhancement
11:12
of vascular structures relative to stationary
11:16
tissues by inflow enhancement or flow -related
11:21
dephasing, which I'm not going to get into. To
11:24
keep it simple, T1 weighted MRA uses time of
11:30
flight to highlight arteries with fast flow without
11:34
contrast agent. This makes them appear very bright.
11:39
This type of sequence is often used to visualize
11:42
the intracranial arteries. If we compare this
11:47
with T2 imaging, the blood appears black, vessel
11:51
walls are visible and thrombus is bright and
11:55
easily visible. However, just to confuse you
11:58
even more, blood can appear bright on T2 -weighted
12:03
images when it is slow or non -moving. Let's
12:07
explain how this works. The classic T2 appearance
12:11
is for blood to appear black. In standard spin
12:16
echo T2 sequences, a 90 degree pulse excites
12:21
spins followed by a 180 degree pulse to refocus
12:27
them. The signal is measured during the echo
12:30
time. In order for the signal to appear, the
12:34
spins must still be in the slice for both pulses.
12:39
As blood flows into the slice, it receives a
12:43
90 degree pulse, but moves out of the slice before
12:47
the 180 degree pulse. Hence, there is no refocusing
12:52
and no signal, causing the fast flowing blood
12:56
to appear black. This is called a flow void.
13:01
Clinically, fast arterial flow appears black
13:05
as do normal veins. A stenosis could also appear
13:10
very black. if there is fast turbulent flow.
13:14
When blood flow is reduced or stagnant, it stays
13:19
in the slice long enough to get both RF pulses
13:22
and a signal appears making blood appear bright.
13:27
Radiologists deliberately design sequences to
13:31
keep the T2 blood signal and a good example of
13:35
this is QISS or quiescent interval single shot
13:40
MRA. In this technique the fresh blood flow into
13:45
the slice is timed using cardiac gating to gain
13:49
a quiescent interval combined with a fast imaging
13:53
sequence allowing detailed images of the blood
13:56
vessels. This technique is particularly useful
13:59
in diabetics with peripheral vascular disease,
14:03
where you need to visualize the tibial vessels,
14:06
which may be quite difficult to image on other
14:10
techniques. So you might be asking yourself,
14:14
if T1 or T2 image sequences show blood as bright,
14:19
why do we use gadolinium contrast? The answer
14:24
to this question is that even though blood can
14:27
look bright on T1 or T2, that so -called brightness
14:32
can be a bit unreliable and non -specific. Gadolinium,
14:37
on the other hand, makes blood predictably, selectively
14:40
and controllably bright, which is what you need
14:43
for looking at blood vessels. Gadolinium works.
14:48
by shortening T1 relaxation, therefore making
14:51
blood consistently bright on T1 weighted images
14:55
and is independent of flow speed. So instead
15:00
of asking, is this bright because blood is moving
15:04
slowly? The answer you get is this is bright
15:08
because contrast is in the vessel. A good analogy
15:12
is to compare non -contrast MRA to a road. Sometimes
15:17
it reflects sunlight, sometimes it does not,
15:20
compared to gadolinium MRA where every road is
15:24
painted with high -vis paint. One of the other
15:28
contrast protocols you might see when you look
15:31
at MRA images in patients with tibial vessel
15:34
disease is TRIKS, T -R -I -C -K -S. And I'm not
15:39
talking about performing pranks or practical
15:41
jokes. TRIX stands for time -resolved imaging
15:45
of contrast kinetics and is basically a rapid
15:49
sequence of 3D angiograms showing arteries filling
15:53
first and then veins. The problem with standard
15:57
contrast MRA is that you have to image at the
16:00
arterial phase. Timing is critical. Inject too
16:03
early, the arteries will appear weak and if you
16:06
inject too late, you get venous contamination.
16:10
So how does this differ from standard T1 or T2
16:13
imaging? Well, instead of trying to get one perfect
16:17
image, TRIX continuously images while contrast
16:21
passes through, so you get multiple time points,
16:24
each showing a snapshot of contrast distribution,
16:27
a bit like a movie of contrast flow. TRIX is
16:31
especially useful for small distal vessels where
16:34
there might be delayed arrival of contrast, which
16:36
standard MRA may miss. The Trix sequences will
16:41
be able to pick them up. You could ask why not
16:43
just use Trix alone? You could do, I guess, but
16:48
Trix does not show wall pathology particularly
16:51
well. It's contrast dependent and its resolution
16:54
is somewhat limited. The images can be really
16:57
grainy. Anyhow, enough of that radiological gobbledygook.
17:01
Let's get back to our MRI scanner, as I think
17:05
we are now at a point where we can test it. Don't
17:08
forget, kids, don't try this at home. Leave it
17:11
to the professionals like me. OK, folks, it's
17:13
back in the garage and we've got everything set
17:15
up here. A core of vicious magnets, which we
17:19
need to cool down a bit with the fire extinguisher.
17:22
So here goes. Let's give them a bit of a chill.
17:28
I think that will probably do it. Right, we have
17:32
to switch on everything now, so let's give it
17:36
a go. Right, we'll switch on the gradient coil
17:40
amplifier and apply 3 ,000 volts from the power
17:44
supply to the RF amplifier. We're going to place
17:52
a plastic bottle filled with water as our specimen
17:54
inside the core of magnets and coils. Just need
17:58
to wait a few seconds and three, two, one, energize.
18:22
And just need to wait a few seconds to see if
18:27
we can get an image acquired. Bit of fast furrier
18:30
analysis of the data and let's see what we've
18:34
got. Folks, we have an image and I'm going to
18:38
try and describe it to everyone listening. It's
18:41
grey and blotchy, rather like someone has poured
18:45
a bottle of ink on newspaper. Essentially...
18:49
It looks like your standard typical MRI picture.
18:52
Can't make head nor tail of it. If you ask me,
18:56
I think that's a result. I think my scanner could
19:00
do with a few more tweaks to make the image even
19:03
more unintelligible and on par with the usual
19:06
MRIs. So with that, it's goodbye from me and
19:10
see you in the next episode. If you want to listen
19:13
to this again when you're building your own MRI
19:16
scanner, remember... All ESVS podcasts are available
19:20
on Apple Podcasts, Spotify and the ESVS library.
19:24
Bye for now.