Live 1057

Free Internet Radio Stations
Diffusion Weighted Imaging – radiology video tutorial (MRI)

Diffusion Weighted Imaging – radiology video tutorial (MRI)


Hello and welcome to this second
pre-course video as part of our adult brain MRI review course. In this video
we’re going to discuss diffusion-weighted imaging which is a crucial part of modern MRI imaging and is really an integral
part of imaging almost all brain pathologies. One of the challenges with
diffusion-weighted imaging is that the understanding of the underlying
processes and the nomenclature is a little confusing. At a fundamental level,
diffusion-weighted imaging seeks to measure the ease with which water
molecules are able to diffuse in any particular voxel and therefore it gives us an insight into essentially the histology of that tissue, how cellular it is, what the extracellular space is, what the intracellular space is. This can be very useful in
distinguishing various entities as well as tumour grading for example. Although the
exact mechanisms and the proportion of
diffusion-weighted imaging signal generated by water in different cellular and extracellular
compartments remain controversial and remain a focus of study, as a general
principle one can think about the majority of
diffusion-weighted imaging looking essentially only at
extracellular fluid. And the terminology most commonly used is whether or not a mass or a region of the
brain demonstrates restricted diffusion, which actually implies whether there is
abnormally decreased diffusivity, in other words
water molecules are having a greater degree of difficulty
diffusing long distances compared to what that tissue
should exhibit were it normal. A somewhat less frequently used term is facilitated diffusion which is the
opposite and that’s used to denote when water molecules can diffuse greater
distances than would be expected and so generally restricted diffusion
is due to a reduction in the size of the
extracellular fluid compartment between cells, whereas facilitated
diffusion results from widening of the extracellular fluid compartment, such as the addition of additional fluid
as is seen in vasogenic oedema. In most centres we are supplied not only
with what we usually refer to as the DWI, which is
usually a B1000 image, and we’ll see what that means in a second, but also with the ADC map which is Apparent Diffusion Coefficient.
The diffusion coefficient is a real physical value of how easy it is for water molecules to diffuse and it’s measured in millimeters squared per second, and these are true values, and on scanners that are appropriately set up you can put a ROI on any the
part of the brain and read off the mean value and this is very
reproducible both within the magnet and across magnets. Typical values, and I’m going to use very
rounded numbers here just to make it easy to
remember, is that CSF is typically 3200 x 10⁻⁶ whereas white matter is around 800 x 10⁻⁶ now putting aside normal images we should, at this point,
talk about the concept of T2 shine through. Again there’s a lot of confusion around
what exactly this means, but most of us know that trying to
interpret DWI images in isolation, or the B1000 images in isolation, can lead to difficulty distinguishing true restricted diffusion from what we
call T2 shine through. On this side we have an acute ischaemic stroke with very high DWI signal and here a
low-grade glioma demonstrates some elevation of signal. Now on these images you can probably assume
that this is real and in this one it’s hard to know,
because gliomas can demonstrate variable diffusion restriction
depending on the cellularity and the grade, therefore we would always
want to review ADC maps and on this side
we can confirm that the restricted diffusion is abnormally high, so the ADC values are lower than normal, it is blacker than
what that tissue should be. Whereas in the low-grade glioma that
tissue demonstrates higher signal intensity
on ADC, so higher ADC values. This is not
restricted diffusion, or not abnormally restricted diffusion,
but rather facilitated diffusion. So we would usually say that the
high signal on diffusion-weighted imaging is due to T2 shine through. One of the problems with this
terminology is that makes it sound as if you either have true restricted diffusion or
true T2 shine through, when in fact all diffusion-weighted
images have a combination of both T2 signal and diffusion depending on the
tissue. So going back to our normal examples, most of us will have encountered at some point this third sequence that forms the
trifecta of diffusion weighted imaging which the B0, B1000 and apparent diffusion coefficient maps or ADC maps. And one the challenges
is that at first glance a B0 and an ADC look quite similar and one can mistakenly look at the B0 value to try and
establish whether something has a T2 shine through component or not. Now the relationship between B0 and B1000 is that the combination of the two
allows you to calculate an ADC and this is in fact why we obtain the two, and this is how we generate the ADC
maps. The relationship between these three
sequences is actually this. And trust me we’re not going to
go into the too much detail, but it is worth looking
at because it actually helps you understand and conceptualise what is
going on with these sequences. So we have three components: we have the
signal intensity of diffusion-weighted imaging, this is
your DWI or B1000 image, is equal to the signal intensity
on your B0 which is just a T2-weighted EPI sequence, times the negative exponential of B x ADC. And so what this results in is: if you have a very high T2 signal, such as CSF and very high ADC
values, say 3200, then you’re going to end up with a very small number. So this value is
attenuated very significantly and you end up with
black CSF. In contrast, if you have an intermediate
T2 signal and intermediate ADC you will end up with intermediate signal on DWI. And if that sounds a little bit
confusing or not terribly intuitive, that’s true and that’s the reason why just looking at a DWI image makes it difficult to establish
how much of a true restricted diffusion component there is and how much T2 shine through there is. Another term that causes a great deal of
confusion to residents is the term “the B value” and what
exactly does this denote. It appears twice in this, once as a B0 and once within the
exponential component. And the B is actually made up of
a number of terms and we really don’t need to know this, but it
suffices to say one of these is the gyromagnetic ratio which is fixed according to the
strength of your magnet. A couple of them are part of the sequence
parameters and the one that I think makes it
easiest to understand what is going on in diffusion, and the one that will be using
conceptually to understand the effect of changing your B value is the time between your gradient pulses. In
other words how long you wait between the start of the
sequence and when you read out the echo having given enough time for the water
molecules to diffuse through the tissue. So we’re going to look and pretend that
that’s the only thing that we’re changing. I took the liberty of getting one of my
long-suffering technologists to be scanned at multiple B values and here we can see
a normal brain ranging from the B0 to the B2000 and most centers would perform B1000 in the brain and this is the DWI image that we’re mostly
familiar with. And as you can see, there is a
smooth gradation across the signal intensity and if we
concentrate on CSF for example it remains bright until we get to about a B value of 500 at which point it’s isointense, and then it becomes progressively darker. In contrast, if we have a look at
white matter across different B values, it remains essentially the same signal
intensity regardless of the B value. So let’s have a look at
these two voxels and try and understand what’s
going on to account for these differences. So here we have at the bottom we have our B
values which are really let’s just say time, and we’ve denoted water molecules as blue dots. Here we have only water molecules
and we have a lot of them therefore we have a lot of T2 signal on
our B0, whereas in the white matter we have fewer water molecules, and they
are trapped or constrained by axons. So as we move forward in time and go to a B value of 100 we’ve only waited a little bit, water molecules have only moved slightly; nothing much has changed. as we start to move further forward and
double the time out to a B of 200 you’ll see now that some of these water molecules have left the voxel. That means that they are no longer
present to return signal and as expected the CSF signal has dropped somewhat. Now because these water molecules are stuck
between these axons, their ability to diffuse long distances is much reduced. So as we go forward in time we can see that more and more of the water
molecules that are in the CSF within the ventricle have left the voxel, whereas
only a couple of water molecules within the white matter are able to leave, accounting for why the
signal intensity really doesn’t change much. By the time we get to a B1000 there are very few water molecules left in the original voxel
and therefore there is very little signal, whereas most of the water
molecules remain in the white matter, and therefore the
signal has not really changed. Let’s have a look at a few examples to see the
application of diffusion-weighted imaging and what the underlying processes
that account for this restricted diffusion are. The one that we are most familiar with is acute ischaemic stroke, and really the introduction of
diffusion-weighted imaging was revolutionary in the assessment of acute stroke, because diffusion-weighted imaging
demonstrates stroke within a few minutes, in fact probably a
few seconds, of secession of normal blood flow. So in this pretend voxel here we’ve denoted
cells as orange and water molecules as blue. And as soon as the blood flow to a part of the brain is
ceased, ATP is depleted very quickly and therefore
sodium potassium pumps cease to work the result of this is that sodium floods
into the cell taking with it water and resulting in cellular swelling. This cellular swelling
results in turn in narrowing of the extracellular
compartment such that the distance between adjacent cells is much smaller than before and
therefore the distance that these water molecules can bounce around in is reduced. As a result, ADC values drop and there is true restricted diffusion. As I mentioned before, the exact physics behind exactly what is happening and
what compartments are contributing to restricted diffusion is still an ongoing area of research, as water molecules not only are trapped in the extracellular space, but do move into the cell and do adhere via various forces to cell membrane and organelle
membranes. But the concept of extracellular space
being a determinant for ADC values is one that’s very helpful so
that’s what we are going to use for this discussion. And so as we mentioned before normal
white matter would be in the order of about 800 whereas an acute infarct will be
much lower than that, typically let’s say half that, about 400.
In contrast, if we look at a high-grade glioma, a glioblastoma, we
have this typical appearance of a heterogeneously enhancing mass
with areas of necrosis which gives this sort of ADC, where the enhancing components typically
are the ones that show the reduced ADC value, and by reduced ADC values in gliomas we’re actually talking about
areas that return to approximately normal white matter
diffusion, so about 800. When we look at these
regions of lower ADC values these correlate with areas on histology
of high cellularity, the cells are larger, they’re dysplastic and the extracellular space is compressed.
Note that in areas of necrosis, as demonstrated here, there’s
not much cellular structure and therefore the ease with which
water molecules can diffuse is increased and here we have very high ADC values. If we look at this diagrammatically, we
see that the extracellular space is narrowed due to these large malignant
cells. One of the very useful features of ADC values in imaging gliomas is that they’re fairly tightly correlated with tumour grade. These are very rough figures again, but
grade 2 have facilitated diffusion compared to
normal white matter of say 1200, whereas grade 3 are around
1000 and grade 4 tumours, so glioblastomas, return to
approximately the values of normal white matter. One final example,
that of lymphoma. We can see on these T1 post-contrast and ADC maps, a patient with a cortical lymphoma, very
vividly enhancing tumour involving cortex, which have strikingly low ADC
values. And this is due to high cellularity. These
are small round blue cell type tumours and exhibit
similar diffusion restriction to other tumours of that sort of histology, namely medulloblastoma, and PNETs for example, because of lots of
small blue cells tightly packed together. Diagrammatically it would have this
appearance. The cells themselves are not terribly large, but they are tightly packed, reducing the extracellular component. It’s
worth noting on this image that there is vasogenic oedema and note that where there is the influx of
additional water molecules spreading out normal axons, then we have facilitated diffusion. One final point to cover is diffusion tensor imaging or tractography, because this is just an extension of what we’ve seen before. If we go back to our diagram of normal white matter we will recall that as time passes, and as you go from a B0 to a B1000, water molecules move, but are largely restricted in their motion by the
surrounding axons such that the direction and distance that they move is aligned with axons rather than across axons. We have the ability to measure the
diffusion in different directions. If we measure enough directions we are
able to calculate for each voxel what the dominant direction of fibers in that
voxel is and therefore they can generate maps of white
matter tracts as indicated. So I hope that has been of some use, and I
look forward to your company at the end of August

18 comments on “Diffusion Weighted Imaging – radiology video tutorial (MRI)

  1. Dr. Gaillard, I'm afraid that your explanation of the physics of diffusion is not entirely accurate. If you would like to know where you went astray, I would be happy to advise you.

  2. Thanks Eric Diaz – there are many simplifications here, but I would love your comments – happy to email me or post them here for everyone's benefit. ๐Ÿ™‚

  3. As Eric was kind enough to point out, there are a number of simplifications in this video to try and convey the general 'idea' of how DWI works. For example when water molecules are shown leaving the 'voxel' resulting in loss of signal, what is not shown is a statistically equivalent number of water molecules from adjacent regions moving into the area, not carrying any signal.ย 

    Also, I regret using the word 'voxel' as the loss of signal due to water motion is independent of voxel size. So 'region' would be better. Additionally there is no boundary over which a molecule moves going from 100% to 0% signal, but rather a gradual decrease with increasing distance.ย 

    Thanks for the comments.

  4. Great presentation ยก … thank you Dr. Gaillard. I am a medical resident in Ecuador and this explanation has been really helphfull. Thanks a lot again

  5. Dr Gaillard, I am a neuroradiologist, big fan of yours, but this explanation of the water molecules leaving the voxel is wrong. I understand sometimes we need to simplify things for students but this is too inaccurate to be considered just a simplification.

  6. I understand that DWI imaging always contains T2 signal. What I don't understand, is why we don't just always look at the ADC image to evaluate diffusion, if that is the "true value". Any explanation would be appreciated. Oh and btw 5th year med student here, reviewing neuroradiology.

  7. i am very new in the field of radiology just started residency …..how decrease water content cause bright DWI imaging ??

Leave a Reply

Your email address will not be published. Required fields are marked *