Create an account and get 3 free clips per day.
Chapters
Outline | Physics of MRI 7: Artifacts - Part 1
Outline | Physics of MRI 7: Artifacts - Part 1
2012artifactschapterfull videolectureMRIseriesUHN
What Are Artifacts : Wrap Around Artifact - Remedies | Physics of MRI 7: Artifacts - Part 1
What Are Artifacts : Wrap Around Artifact - Remedies | Physics of MRI 7: Artifacts - Part 1
2012acquireanatomyartifactbottombrainchapterdirectiondoubledextentfieldfrequencyfull videoillustrateimageincreasematrixMRIoccursphasescansimplyspatialUHNviewwrap
Truncation Artifact : Gibbs Ringing - Remedies | Physics of MRI 7: Artifacts - Part 1
Truncation Artifact : Gibbs Ringing - Remedies | Physics of MRI 7: Artifacts - Part 1
2012accuratelyacquireacquiringartifactchapterdatadecreaseedgesessentiallyfieldfull videoimageincreasematrixresolutionscansizespatialspineUHN
Chemical Shift Artifact - Causes and Remedies | Physics of MRI 7: Artifacts - Part 1
Chemical Shift Artifact - Causes and Remedies | Physics of MRI 7: Artifacts - Part 1
2012acquiredartifactbandwidthchapterchemicaldecreasedifferencedirectionevolvefatfrequencyfull videoimageincreasemagneticmentionminimizeminimizingoccursphasepositionpulsereconstructiverotationalshiftsimplysusceptibilityteslaUHNvialswater
Susceptibility Artifact - Remedies | Physics of MRI 7: Artifacts - Part 1
Susceptibility Artifact - Remedies | Physics of MRI 7: Artifacts - Part 1
2012artifactartifactsbloodbrainchapterchemicalclotdecreasesdeterminesdifferencedifferencesechofieldfrequencyfull videoholeillustrateimageinsidemagneticmanifestationrotationalscannershiftsignalspinsubstancesurroundingsusceptibilitytissuetissuesUHNvectorvoxelwater
Transcript

Hi, my name is Marshall Sussman I'm an MRI Physicist at University Health Network and University of Toronto. I'm giving a series of lectures on basic MRI physics and the topic of today's lecture is going to be Magnetic Resonance Imaging Artifacts. This lecture is divided into two parts, first part is gonna begin right now. To give a bit of an outline

of what I'm gonna speak about today. I'm gonna run through a series of artifacts that are found in MRI. I'm going to give a description of what each of these artifacts are and then discuss what causes them. And finally I'm gonna talk about what remedies are possible to minimize those artifacts if a remedy is possible. So, first of all what are artifacts? So artifacts are simply any feature of

in MRI we call that, here we have, let's say we acquire an image of the head, and in MRI as we know we acquire the data in k-space. You then perform a fourier transform and that generates the resulting image. Now because our data is acquired in k-space in order to understand how these other facts actually

appear and how they occur, we really have to examine what's going on in k-space. Let's begin with the first type of artifact which is wrap round or aliasing artifact. So here we have an example of an image that has wrap around

artifact. So you can see an image of the brain and on top and the bottom of the brain where the two arrows are pointing, you can see that the anatom is essentially wrapped around into the either side of the image. So the top of the brain is wrapped around into the bottom side and the bottom side is wrapped around into the top side. Here we have another example, this is an image of the knee

and you can see we see something similar here. So we have the anatomy wrapping around in left-to-right direction where you see the arrows. So what is it that cause wrap-around artifact? Well, the wrap-around artifact results if your field of view is smaller than the physical extent of the anatomy. So just to illustrate an example of that. Here we have an image of the knee

and you can see that the field of view is indicated by the arrow in that direction. You can see that the field of view that we've prescribed for a scan is smaller than the physical extent of the anatomy. So as a result, the anatomy wraps around from one side of the image around to the other. Now you may look at this image and ask the question, okay, why do we just get the imaging? Why

do we just get aliasing in left-to-right direction and not in the top-to-bottom direction, because as you can see the anatomy is also bigger physically than the field of view in the top to bottom direction? The reason for that is because wrap-around artifact only occurs in the phase encode direction. So in this particular image the phase encode direction is the left-to-right direction.

The Vertical direction is the frequency encode direction, aliasing does not occur, and that's why we don't have artifact in this case. One trick that is often used in MRI as a result of this fact is that we often like to set the largest direction of the anatomy to be the frequency encode direction. So in this example, you have the

largest extent of the anatomy is obviously in the top to bottom direction because that is the full extent of the leg. So in this case to minimize wrap-around artifact we would set the frequency encode direction to be in the top to bottom direction. Now I should also mention that you can also get aliasing in 3D scans. So you can also get wrap-around artifact in three-dimensional scans. So

here we just have an example of that this is an image of the leg and you can see circled in the center of the image is that sort of wided out region. And that's from wrap-around direction in the through plane direction of this image, because this was a 3D scan. So one thing I should mention is that this occurs if you acquire a true 3D acquisition. Through plain wrap-around artifact doesn't

occur if you're doing a 2D multi-slice acquisition. So the reason for this is because the 3D scan has phase encoding in both the y and z direction. So you only get wrap-around artifact in phase encode directions. So if you're doing a 2D multi-slice scan there is no phase encoding in through-plane direction. So what are some of the remedies that we can use to deal with wrap-around artifact?

So as I mentioned wrap-around artifact results when the field of view is smaller than the extent of the anatomy. So the obvious thing to do would be to simply increase the field of view, to be larger than the extent of the anatomy. Now that would get rid of the wrap-around artifact but unfortunately doing that is not free, there's a couple of costs. Because in particular if we just

simply increase our field of view without changing anything else that would cause a decrease or a worsening of our spatial resolution. So for example if we double our field of view, our spatial resolution goes down by a factor of two. Now you could compensate for that by increasing the matrix size so you could say if you doubled your field of view, if you correspondingly doubled your matrix

size, you'd maintain the same spatial resolution. But of course the cost of doing that is we now increase our scan time, so we've increased it by a factor of two. So increase the field view you will get rid of aliasing artifact or wrap-around artifact but it comes with a cost and depending on the situation you have to decide if that's a cost you're willing

to pay. Another approach that can be used is to swap the phase and frequency encode directions. So I mentioned earlier that aliasing only occurs in the phasing encode direction. So if the anatomy we're interested in, if the aliasing or the wrap around artifact lies over the top of the anatomy that we're interested in interrogating, then we could simply swap the phase and frequency encode direction

to rotate the wrap-around artifact. So this won't get rid of the wrap-around artifact but it will shift it by 90 degrees. So if there's some anatomy you're interested where the wrap-around artifact is we can swap phasing frequency and we can uncover what's lined underneath that artifact. The last solution to dealing with wrap-around artifact is just simply to live with it. And that's probably by

far the most common solution to the problem. So just for an example here we have an image of the abdomen and you can see we have wrap-around artifact in the images pointed by the arrow. But because the wrap around artifact really doesn't overly any anatomy we're really interested in, it really doesn't matter it just simply occurs and it's there

but it doesn't bother us, so we just simply live with it. And it's not that uncommon to actually encounter wrap-around artifact in images but in most cases it's inconsequential. Second type of artifact that we're gonna discuss is truncation artifact or the technical name for it is Gibbs-ringing. So to illustrate an image of that

here we have an image of the brain and if you look closely to where

the arrows are pointing, you can see that we essentially have a repetition of the edges, so essentially a ringing of the edges as indicted by the arrows. So what causes that? Well because the problem is that if you have an image with a very sharp border so for example a very sharp border from dark to light which occurs let's say the edge of the skull.

This can't be accurately represented by the spatial resolution that we've prescribed. And in that case the resulting artifact is what you saw which is a ringing or essential a replication of the edges in the image. So why is it called truncation artifact? Well because remember when we're acquiring data in MRI, we again as I mentioned earlier, acquire the data in K-space. And let's say to fully represent

the image, or to accurately represent the image we have to cover the extent of K-space as indicated by the grid as you see here. However, if instead of acquiring that amount of K-space, if we essentially truncated the amount of data we're acquiring, so let's say we only acquire a portion of K-space then as we saw on our earlier lectures the effect of truncating the K-space acquisition

like that is to decrease or make the spatial resolution worse. And when we don't have sufficient spatial resolution, that's when we can get that truncation artifact on Gibbs ringing. So what are the remedies for a truncation artifact? Well, as we mentioned the cause of truncation artifact, is that we have insufficient resolution. So obviously the solution is going to be simply to increase the

spatial resolution or improve the spatial resolution. And we can do that in two ways, we can either increase the matrix size or decrease the field of view. But as we saw in the wrap-around artifact slide, this leads to some cost if we increase the matrix size that causes an increase in our scan time. If we decrease the field of view that will maintain our same scan time but it runs the risk of introducing

aliasing artifacts if we make our field of view too small. So as you see the solution for getting rid of truncation artifact is relatively straightforward. But we have to decide whether or not the penalties that we have to pay in order to do that are worth it and that obviously depends on the individual situation. Just to give an example of this remedy, here we have an image of the spine

that was acquired with an image matrix size of 128 by 128. And you can see that in this image we have the truncation artifact as indicated by the arrows. So we have essentially the ringing off of the edges. And in this example this can actually mimic pathology. So to get rid of that we simply increase our matrix size so the image on the right has a matrix size 256 by 256. And now you can

see we've completely eliminated the truncation artifact or the ringing artifact. Now I should mention that this type of artifact is relatively rare in comparison with the past. And the reason for that is the MR scanners nowadays are much better than what we had previously. So as a result we typically acquire our images with much higher spatial resolution than we used to. So the appearance of truncation artifact

is much more of a rare event. Just as an example. We would never acquire a spine image with matrix size 120 by 128 nowadays. So the next artifact we're gonna talk about is chemical shift artifact. So just to illustrate an example of that here we have an image

of the abdomen. And the classic appearance of the chemical shift artifact is alternating bands of white and

dark as indicated by the arrows on either side of the fat water interface. So what causes chemical shift artifact. Well chemical shift artifact is as a result of the, as I said if you have both fat and water interface, the reconstructive position of fat is shifted relative to the true position of the fat. So just to illustrate an example that, let's say we have two vials, one needs

fat and one contains water and let's say this is their true position if we acquire an image of those vials then the position of the fat bowel will be shifted to its true position in reconstructed image. And that's exactly what you see in this case here. So the fat water interface, the whole anatomy is shifted over. So on one side where the signals pile on top of each other you get a

constructive interference, and a bright line. On the other side where the signal shifted away there is essentially a void where you get the dark line. I should mention that chemical shift artifact only occurs in the frequency encoding direction. So in most pulse sequences, there is a couple exceptions to that rule but in most pulse sequences it only occurs in a frequency

encode direction. So that's often a hand that can tell you which direction of the phase and frequency encode direction in an image. So for example in the image I've showed you here, the frequency encode direction is in the right left direction. And another give away that will tell you that, that's also gonna be the largest direction of the anatomy. So typically do minimize wrap

around effect we also wanna make the right left direction our frequency whole/g direction. So what causes chemical shift artifact? Well it's really different due to the difference in rotational frequency between fat and water and the strength of this shift, the magnitude of the shift is directly proportional to how far apart those rotational frequencies are separated. The image on the bottom shows you the

frequency spectrum of fat and water at 1.5 Tesla and you can see that the rotational frequencies are separated by a size of 225 Hertz. So they rotate about 225 cycles per second differently between fat and water. I should also mention here that the difference in the frequency varies with the magnetic field strength. So for example if we are to double our magnetic field strength to 3

Tesla, then we shift between fat and water Will increase and so to with our chemical shift artifact. So what are the solutions for minimizing chemical shift artifact? The first one will be to use fat saturation. So if just we simply eliminated this source of the fat signal, then there would be no chemical shift artifact cause there is no fat water interface. Another approach is

to increase the bandwidth. So why does that work? Well because recall from my earlier lectures, the effect of increase in the bandwidth if we maintain the same number of points we are acquiring is to decrease the readout duration. Now if you recall that the chemical shift artifact arises due to the difference in frequency between fat and water.

And that's because over time a phase evolve between fat and water due to the difference in frequency. Well if we simply minimize the readout duration, there is just simply less time for that phase to evolve and therefore we minimize the amount of chemical shift artifact. Here we just have an example of that, we have an image on the left acquired of the brain with a low bandwidth and on the

right we have an image of the brain acquired with a high bandwidth. And if you look to the region highlighted by the arrows, you can see that by increasing the bandwidth we can decrease the amount of chemical shift artifact. So as before I should mention that doing that is not free, because as we know if we increase the bandwidth that decreases the signal to noise ratio. So again there's

a cost to minimizing chemical shift artifact. The next artifact we're gonna talk about is susceptibility artifact. So first of all what is magnetic susceptibility? Magnetic susceptibility is simply

a characteristic property of a substance that determines essentially the magnetic field that that substance sees. So just to illustrate an example of that, let's say we put a person inside

an MR scanner and let's say the magnetic field inside the MR scanner is B0. However that's not the field that is seen inside the body. So let's look at a region of the brain here. So the actual magnetic field inside the tissue is gonna be equal to the main magnetic field B0 times one plus the susceptibility of that tissue. Now this is important because different tissues have different susceptibilities.

So for example if we looked at a different type of tissue it would have a different susceptibility let's say kie/g 2 and therefore the magnetic field that that tissue sees would be different than the first one. Just to give an example of some common, there are some substance with different susceptibilities. Here we have a table, we can see that water has a susceptibility of -9 * 10 ^

-6. Deoxyhemoglobin has a value of 3 * 10 ^ -7 and this obviously gonna be present in blood so you can see that depending on the oxygenation state of blood. The susceptibility of blood may change and that's responsible for the bold effect which is using FMRI. Air you can see has a susceptibility of 4 * 10 ^-7. Hemosiderin

which is often occurs when you have blood clots. Has significantly the largest susceptibility 4 * 10 ^-5. And iron which is a fair magnetic object has a very susceptibility much greater than a 1,000. So any time you have the presence of tissues with different susceptibilities that can give rise to susceptibility artifact. So why is that? Well, again recall that susceptibility

determines what magnetic field each tissue sees. So if we look at a particular tissue, the rotational speed is gonna be proportional to the strength of the magnetic field. And the rotational frequent in particular is equal to the gyromagnetic ratio gamma times the magnetic field in the tissue. However, because the strength of the magnetic field in the tissue depends on the susceptibility,

that means the rotational frequency also will depend on the susceptibility. So as in the case with chemical shift artifact if you have different tissues with different susceptibilities, it'll have different rotational speeds and that will lead to artifacts. So, the manifestation of chemical shift artifacts has some characteristics similar to that of chemical shift artifact, because again it's

due to differences in rotational frequencies. So just to illustrate an example of that, here we have a susceptibility artifact in this image of the brain. And this is due to the difference in susceptibility between the tissue and the air. And again you can see we have this sort of bright-dark region which is again reminiscent of what we had with chemical shift artifact. So in the case of susceptibility

artifact there is initial manifestation of how this appears in images. Consider the figure below. So let's examine a particular voxel withing the body. Let's consider as in our previous examples that the magnetization with in that voxel consists of three separate water molecules. So as before the net signal we get is going to be the vector sum of those individual

water molecules. Now if we have within that voxel if we have tissues with a range of different susceptibilities that means that the magnetic field inside that voxel will also have a range. In other words the magnetic field is gonna be inhomogeneous inside that voxel. So as we saw before water molecules that see different magnetic fields will rotate at different frequencies. So over time what's

gonna happen is magnetization within that voxel is going to dephase relative to each other, and because the signal is a vector sum of the individual water molecules, that's gonna lead to a signal loss because of a destructive interference. So if you have susceptibility differences particularly at interfaces where you have tissues of different susceptibility that can lead to signal losses in your

image. So here we have an example of that, this is a gradient echo image of the brain and this is a case where this person had a blood clot in their brains. So blood clots as we saw before I think contain hemosiderin which has a significantly different susceptibility than the surrounding tissue, and that's one of the reasons why you see large black hole in the brain. We

have a second example this is the case where the person has a metallic clip in their brain as you can see on the X-ray image on the left. If we look at the corresponding MR image you can see again we have a large black hole because there's a large difference in susceptibility between the metal and the surrounding tissue. So what are the remedies for susceptibility artifact? Well, as we

saw that the cause of susceptibility artifact is similar to that of chemical shift artifact, and that because of a difference in frequency between different tissues. So as with the case of chemical shift artifact, the solution is to increase the bandwidth which again as I discussed earlier decreases the duration of the readout which then decreases the amount of time

of which phase can evolve and again leading to artifact. Another solution is to use a spin echo base sequence cause this will minimize the phase distribution because as we saw in earlier lectures spin echo imaging corrects for the effects of inhomogeneities. So here we just have an example of the image on the left has occurred with a gradient echo image so again you see the large hole, black

hole, due to the differences in susceptibility between hemosiderin and the surrounding tissue, when we use spin echo image you can see that that black hole is largely cleaned up. Again because the spin echo image corrects for the inhomogeneities caused by the difference in susceptibility between the blood clot hemosiderin and the surrounding

tissue. So that brings us to the end of the first part of the MRI artifacts lecture and in this lecture we covered a number of different

Disclaimer: Content and materials on Medlantis are provided for educational purposes only, and are intended for use by medical professionals, not to be used self-diagnosis or self-treatment. It is not intended as, nor should it be, a substitute for independent professional medical care. Medical practitioners must make their own independent assessment before suggesting a diagnosis or recommending or instituting a course of treatment. The content and materials on Medlantis should not in any way be seen as a replacement for consultation with colleagues or other sources, or as a substitute for conventional training and study.