Introducing MRI: Diffusion Imaging (49 of 56)

we're going to talk first today about diffusion so when we talk about diffusion measurements that are made with em are so first we should understand exactly what it is that we're we're talking about before we talk about how to measure it what's the phenomenon we're talking about before we discuss how we're going to measure it with imaging so if we think of a simple example we have a right so like a bucket of water sitting here on the table or like my little glass of water and if it's sat here all night so it was perfectly still you would walk in and you would look at it and it would look like it was absolutely perfectly still but the reality is right everyone probably knows is that that water is composed of many individual particles molecules and it's very far from still there is continuous molecular motion going on in there and in this environment where all we have is water molecules let's assume it's distilled water there's no impurities nothing else in there those molecules are freely able to move in any direction the only potential impediment is that they're going to run into other water molecules and this is what is generally referred to as a freely diffuse Abul some solvent we can do a little experiment where you drop a dropper full of ink or food coloring in there you can watch it randomly right defuse through the solution to give it an even color over time the rate at which that occurs can actually be measured and that tells us something about the rate of which those water molecules are moving around so we're obviously not very interested in imaging tissue right in animals or humans about this scenario because this doesn't pertain in our example of a patient in the mr scanner so if we look at some tissue sample whether it's doesn't matter whether it's the liver or the prostate or the brain or wherever it happens to be if we look at some region of tissue it's a much more complicated scenario than this one over here because in this tissue while there is a lot of water in there and that water of course is where our mr signal derives from if you think about what's happening in this region in addition to having many water molecules there is all kinds of other stuff which could be things like macromolecules membranes right organelles whatever whatever tissue components are in there which we can think of in a crude sense as setting up all kinds of roadblocks to the free movement or diffusion of water molecules so if we could somehow make a measurement of how rapidly water molecules are able to move through this sample of distilled water and compare that to how rapidly they move through this sample of tissue with all of this other stuff in it right the rate of diffusion which is measured as D right so this D and our liquid is going to be much greater right than the diffusion over here in tissue and what we want to do with diffusion imaging is be able to detect how rapidly these water molecules are moving around now we already have a way to do that because as we discussed previously in talking about flow that if we are acquiring some kind of image and I think the example that we gave was an image of the abdomen let's say we talked about flow in the splenic artery that when we spatially encode this image such that we have a frequency encoding gradient on that in the presence of that frequency encoding gradient the spins that move along the direction of this gradient magnetic field after a period of time will accumulate more phase than other spins which might be stationary okay and we don't have to go back into all that detail about whether it's a linear or quadratic or whatever it happens to be but suffice it to say that anytime we have a gradient magnetic field when we put our spins generate our transverse magnetization put those spins in the transverse plane since they're in different physical locations along the gradient they see different strengths different beam at different net magnetic field strength and therefore process at different frequencies and during the period of time that gradient is on there's some amount of phase change which leads to less net transfers magnetization that's true even if the spins are stationary if they're moving along this gradient magnetic field then there's going to be an even greater phase shift and a greater loss of signal so if we made a comparison between the stationary spins and the moving spins at the end of the day we would detect less signal from the region where the spins were moving and more signal from the region with the spins were stationary okay and this is exactly the mechanism that we use to detect diffusion and an mr image we turn on a gradient magnetic field and look at the amount of signal loss that occurs while that gradient magnetic field is on and that amount of signal loss is going to be proportional to the rate of diffusion or we could just say rate of motion at this point in time diffusion is just a form of motion now keep in mind that given what I've just said if we look at any mr pulse sequence so let's just make up some example mr pulse sequence we'll start out with a 90-degree RF pulse that's slice selective okay and we know that we have to phase in code we have to frequency in code and in the middle of that frequency encoding period is when we're going to have our te so if this is all we do and I just left out for clarity all those D phasing other components the gradients but they're of course all happening if this is what we do notice that during the time between when we generate our signal and when we sample it of course we're sampling throughout this period of time our tissue has been exposed many times over to gradient magnetic fields okay so when the sly select gradient is on or when the phasing code ingredient is on or when the frequency encode ingredient is on we know that our spins that have been some of them are let's say stationary some of them are moving through the tissue and of course there's movement in all different directions if we talk about diffusion in tissue it's random can go in any direction so there's all kinds of motion going on along these gradient magnetic fields when they're on whenever we have movement along a gradient magnetic field there's going to be an accumulation of phase proportional to the amount of movement along that gradient magnetic field which of course depends on the velocity or speed of movement and they'll be resulting loss of signal so in this image what type of image is this by the way based on the information I've given you here this is just a simple gradient echo image that's all so when we look at any voxel signal from any location in this gradient echo image that signal is modulated by the presence of diffusion and voxels that are let's say or I shouldn't even say diffusion modulated by the presence of motion right if we have a voxel that is in the splenic artery and we compare it to a voxel that's let's say in the spleen in solid tissue there is going to be a greater amount of signal loss for those arterial spins which are moving along than the ones that are in the soft tissue similarly if we make a comparison between spins let's say in the bladder and fluid or in cerebrospinal fluid or wherever and compare that to tissue spins there are going to be different amounts of signal loss which are inherent in this signal so at this point I think hopefully it's clear that anytime you look at any mr image that that signal is influenced by diffusion long before anyone ever thought of diffusion-weighted imaging per se all mr images had their signal modified based on the presence of diffusion so we're interested in actually detecting that diffusion specifically and all we have to do then is take this pulse sequence and make it particularly sensitive to the presence of diffusion well what determines the amount of signal loss in any location that will occur based on that diffusion we said yesterday when we talked about accumulation of phase due to flow along a gradient magnetic field we said that it changes as opposed to stationary spins which undergo a linear change with time we said that this was a quadratic we're not going to deal with all those putting those gradient loads together don't worry all I want to point out is that this constant term here contains a lot of important information while it's true that with motion we have a more rapid rate of change right amount of phase change what else determines that amount of phase change well ultimately what it is is during the period of time that we observe the sample before we measure the signal what has been the total range of gradient exposure to those spins so if we set up our image and this is our gradient magnetic field how much ground has been covered along that gradient magnetic field what's going to determine that well it's going to be a combination of the rate of speed how fast the spins are moving and how long we observe them for so the longer period of time that we watch the greater the amount of phase change it will occur and the faster they move the more different ranges of field strength they'll experience and the more the phase change will be so the signal intensity that or this the change in signal intensity that occurs as a result of diffusion is dependent on the rate of diffusion yes but it's also dependent on the time of observation how long we let this occur the other thing that it's dependent on is the strength of this gradient magnetic field if that gradient magnetic field is stronger then over the same period of time at the same speed or velocity we will have a greater change in phase so these three factors all determine the change in signal amplitude as a result of diffusion now this one D this is the rate of diffusion how fast the spins are moving that's not a parameter we control that's the one we're trying to figure out but we have direct control over these two and if we go back to our pulse sequence and we want to make this pulse sequence particularly sensitive to diffusion so what we can do is turn on a gradient magnetic field of very high amplitude and we turn it on as a bipolar gradient magnetic field equal and opposite Globes here so that whatever is not moving of course there's really nothing that isn't moving at all but whatever is truly stationary would have no net effect but using a very large amplitude gradient magnetic field and yes the amount of time that we leave it on also matters this will make the signal that we detect and put into our image also right right or much more strongly a function of diffusion along this one linear dimension so this image as we've edited it here in blue is going to be sensitive only to diffusion along the phasing code in gradient or only extra sensitive to diffusion along the direction of phase encoding so if this is the phase encoding direction in our image any movement that is moving top-to-bottom in the image there will be much more attenuation of the signal coming from those moving spins in proportion to how fast they're moving any questions about this yes and number three there this number three they bring me the slide yes um good graph of I guess that's they defaced yes where's was blessing on a component in there in here hey it sure is Liz with which it's all incorporating in this constant term I mean this is a much more complicated equation than just KT squared it's saying that it's proportional to the square of time but also dependent on other things including the gradient strength right including the velocity so it's all it's all embedded in there in other words if you if we're doing phase contrast MRA trying to quantify the speed of flow with people to install Q flow imaging member I showed you the CSF flow study yesterday with that graph where we showed at different points during the cardiac cycle how fast was CSF flowing the cerebral aqueduct that would be in order to quantify that you would need to know everything else so you would need to know the amount of time that you observed it for you would need to know what the gradient strength was etc etc and then you would solve for velocity which is also in this and and this graph is also related to the the relationship shown in the bottom of the defacing they were related so these parameters are right here that's all so basically what I was telling you is that the rate of diffusion which is the velocity of movement and or is roughly equivalent to the velocity of movement diffusion is a little bit different than than gross you know bulk flow and the gradient strengths are all part of this term okay any other questions about this so far so we are able now to take our image and make it sensitive to diffusion along one direction and how sensitive it will be meaning how much signal change will occur for a given amount of a given rate of diffusion will be proportional to both the strength of this gradient and the amount of time we leave it on those two factors right the gradient amplitude and the time that we leave the gradient on are expressed as something called the B value anyone heard of that if you look at your diffusion-weighted images they'll say the B value was a thousand or 800 or whatever it happened to be so that B value is an expression of how much gradient we're exposing our sample to and therefore how much we are sensitizing it to diffusion so a B value of a thousand as opposed to 500 means that the gradient was either stronger had a steeper slope or was left on for a longer period of time or a combination of the two now this image is only going to be sensitive to diffusion in a single direction when we talk about diffusion let's say think of a clinical example where we commonly use diffusion imaging if you're looking to detect an acute infarct so if we're looking at let's say an image of the brain some axial slice of the brain and we're looking to see whether this patient has a middle cerebral artery territory infarction so one of the things that we know about infarct although it's still many years later not totally clear exactly why this happens is that in the setting of acute infarction in the brain there is a dramatic decrease in the rate of diffusion and we'll talk in a little bit about why that allows us to detect it but suffice it to say that we're looking for a change in diffusion in this tissue as opposed to the unaffected tissue now which direction are we interested in measuring that change in diffusion along what's the typical direction of diffusion which becomes restricted or declines in an MCA in fart anyone know there there is no specific direction that's my point is that as we're talking about this now we're interested in knowing about in a specific location as opposed to some other specific location what's the total magnitude of diffusion I'll care which direction it's going in I just want to know how fast those spins are moving so this approach in blue is only giving us information about movement along a single direction it's really only part of the answer to the question that we're trying to address so we can however make this image sensitive to see if we can do this just going to decrease the size to make it a little bit more bearable okay so we can make this image sensitive to diffusion in any direction simply by applying those gradient magnetic fields in all directions so in this scenario we have the same strength of this extra application of a gradient magnetic field applied in both XY and Z so if I am a population of proton sitting in some voxel in the brain no matter which direction I move there is going to be a change in B net and proportional to how rapidly I move there will be a greater amount of phase change yes why do you need all three directions wouldn't a minute piece if you isn't the point of the fusion then it's just randomly in any direction and then if you can just detect the diffusion in one with one gradient Y which is made all through okay so let's blow this up and if we look at what's going on within this single voxel because our signal is always the aggregate signal from everything happening in a voxel proton density modulates our total signal in the image based on the fact that the total number of protons within that volume of tissue gives us an aggregate signal that we we write into our image so let's say in this in this example that we have a whole bunch of protons and some of them are moving this way some are moving this way and some are moving in and out and let's just make things simple and make an assumption that we have these spins in equal proportions so within this voxel you know there's a million protons one-third are moving in and out of the board one-third are moving left to right and one-third are moving up and down and let's also just keep it simple let's assume that those different populations are moving only in that one direction and reality of course is a whole spectrum of all different directions and there they should be you know depending if it were a if we're a freely diffuse Abal solvent like this they would be all randomly aligned in tissue there may be some other alignment but let's just keep things simple and assume that we have this equal distribution along only three directions so if we do our what's a diffusion weighted imaging sequence here it's a gradient echo image where we added this extra gradient magnetic field to make it sensitive to diffusion if all we did was the blue example well that's phase which let's just assume that that's up and down so we would detect a change in signal proportional to the diffusion of one third of the spins okay but if we're sensitive to diffusion in all directions we would have three times that signal change so it basically improves the contrast to noise of our diffusion measurement there's going to be a greater change in signal with diffusion if we're sensitive in all directions now in a very short while we're going to talk about reasons why we would want to be sensitive in only one direction but for now we're our question in this case of let's say looking for a brain infarct is just what is the magnitude of diffusion the direction of diffusion in this voxel really doesn't interest us it's not part of the clinical question that we're asking it's not relevant so we want to be maximally sensitive to all the potential information that's there so let's sensitize in all directions is everyone clear about that any questions yes how can we have all three ingredients on the same time if it was going to ask the same thing okay you're a team so right that's a question right which I was going to bring up even if you didn't ask it well what happens we have all three gradients on at the same time our gradient magnetic fields are expressed as vectors and they sum so in the presence of all three of those gradient magnetic fields we actually have a single linear gradient magnetic field which is on a compound oblique direction through the tissue so the answer is it doesn't matter because the only thing that matters in this case it's true you're absolutely right to correct observation but what matters to us in this case is that no matter which direction those spins diffuse that they will see a gradient of magnetic field strength we're not using this for spatial encoding when we talked about spatially encoding like in the phase encoding direction or the frequency encoding direction if you had two gradients on at the same time it meant that your spatial encoding was on that oblique direction but here this is not about spatial encoding this is about encoding signal change due to diffusion okay so we can determine how much of a signal change is going to occur based on the strength and duration of the gradients that we put on and very commonly by the way these gradient magnetic fields are actually will look like this they are put on this is kind of a detail that we don't need to worry about but they are pulsed in this way once before and once after the phase encoding gradient or very commonly this is actually done with a spin echo pulse sequence this is not to scale but typically the phase encoding green diffusion sensitizing gradients are placed both before and after the 180-degree RF pulse but keep in mind that this idea of placing additional gradient magnetic fields to make your signal sensitive to diffusion is totally independent of the pulse sequence that you choose you could do this with gradient echo imaging spin echo imaging fast spin echo imaging echo plane our imager you can do with whatever you want inversion recovery it really doesn't matter it's an extra that we can throw in there to modulate this signal okay now that being said that you could choose any pulse sequence that you're interested in does it matter is there an optimal choice does anyone know what we usually use we talked about this yesterday with the safety discussion right impatience with retain pacemaker wires we typically don't do diffusion-weighted imaging we try not to because we're concerned about the gradient magnetic field inducing currents in those retained wires and affecting myocardium causing arrhythmias that's because those gradients are very rapidly switching and those are echo planar acquisitions so epi is the typical pulse sequence that is used for diffusion-weighted imaging it doesn't have to be why choose echo planar imaging okay because of the speed why because it is fast you want to I'm sorry detect what defect occasions and he'll be fine clean why is the diffusion signal going away I don't understand auntie Xiu % in that petition that's been fun I'm not sure I understand what you're saying I thought that like that you only have a small window to do the diffusion imaging and then you reasonable not be detectable oh you mean physiologically that after a period of time the effect isn't well no not that short of window I mean words we're talking about you know seconds to minutes here these so it's true right that these acute changes in an infarct in diffusion-weighted imaging and that that's really not the matter of our discussion today are present very early on and they persist for several days to a couple of weeks depending on what we're talking about but they persist orders of magnitude longer than the time it takes to do any kind of imaging no actually that's not true what we'll talk about how we quantify it because it will see that it's a comparison between images that are both with and without diffusion sensitization that we use to to quantify the rate of diffusion or speed so consider this that if we're looking at this part of the brain so again when we look at a brain image you sit down at your pax workstation and you're looking at this t2-weighted image of the brain it looks like a nice static picture and we've already discussed how the tissue in fact when the patient was lying in the scanner had all this diffusion going on the diffusion is very small-scale motion I mean during the period of time that we acquire an image or a line of k space for that matter how far do you think a water molecule moves not very far right it's a very small scale movement at the same time there is other motion which we call bulk motion which is going on in the brain simply because the patient's heart is hopefully continuing to beat during the scan and with every cardiac cycle there is expansion and contraction of tissue right at the vascular intravascular volume changes this happens with respiration as well as maybe gross patient motion but even that physiologic motion is substantially greater than the degree of motion that we are trying to detect as diffusion and that type of motion also causes movement of tissue along these gradient magnetic fields so if we want to be maximally sensitive to this small scale molecular motion it's important that we eliminate other types of motion from the equation and that's the reason why this is done with very fast imaging techniques and pretty much always done using echo planar imaging where we can read out not just a line of case base but an entire image right in the order of you know a few milliseconds faster or fast enough to freeze any motion due to physiology and certainly due to the patient's actual gross motion now in terms of actually measuring diffusion so if we look at two images and we pick some location in those two images and in our first image this is an echo planar image and the second one is the same echo planar image with some amount of diffusion sensitization so essentially we are doing the same thing twice we scan the patient's brain using whatever pulse sequence we've chosen as I've said generally epi and we come back and we do it again making a single change and that single change is turning on all three of those diffusion sensitizing gradients if we now compare the signal intensity in our echo planar image to the signal intensity in our diffusion-weighted image this is sometimes called a b0 image the amount of diffusion sensitization is zero that means that those gradients are off and this would be the B whatever it happens to be 1000 is a typical B value so if we compare those two signal amplitudes what goes in between equal greater than or less than remember the signal is attenuated in proportion to diffusion and this is not just qualitative this is quantitative so the amount of signal change that occurs is directly proportional to well since two of our factors the gradient strength and how long we left the gradient on are known right this is this ratio is proportional to the only variable that's left which is how fast the water molecules are percolating or diffusing through the tissue during that period of time so from this information we can actually compute something that we call the apparent diffusion coefficient which is a quantitative measurement of the rate or the speed of diffusion it's not a velocity by the way why why would it be incorrect to say that this is the velocity of diffusion ballast is a vector velocity implies direction there's no directional information contained in the ADC right that's a function of the fact that we turned on all of these gradients at the same time we have no way of knowing the direction that this diffusion is moving just that it's moving and how fast okay so when we talk about the apparent diffusion coefficient we are essentially talking about a measure a scalar measure of the speed of diffusion how fast the protons are going and that of course is an aggregate measure from this voxel of tissue all of the protons within that voxel it's the aggregate measure of their rate of diffusion and if we want we can make another image we're in this voxel right we take the ADC and we represent it as a visual scale a grayscale a color scale you can do whatever you want and put those values in this image and we do it to a typical what we call a diffusion weighted imaging study you will have all three or you better have all three of these images 1e 0 image which is just the baseline echo plane or image what does it look like by the way to single-shot EPI image what does it look like contrast wise hmmm t2-weighted image it looks like t2-weighted and typically by the way we are going to be using spin echo echo planar imaging no reason you couldn't use gradient echo echo planar imaging but remember that the presence of these gradient magnetic fields is causing signal attenuation so we use the spin echo approach just to get as much signal-to-noise as we possibly can so this is our spin echo EPI image which looks like a t2-weighted image if we compare if we predict we'll look at them in just a second what that image looks like relative to the diffusion-weighted image you guys have all seen these but if you had to make one overall statement about the difference between these two images it would be what we're going to have much lower signal-to-noise in the diffusion weighted image and now it should be clear that these images would show up on your pax work list are not images that contain any kind of signal that we detected from the patient they are computed or quantitative images derived from these two now one more point that I just want to mention here is that if we think about pathology so one of the things that we tend to use diffusion-weighted imaging for it the most is to look for evidence of tissue injury right you guys are most commonly used to using this to look at stroke that acute infarct demonstrate very obvious abnormalities on diffusion-weighted imaging that's actually true for other forms of tissue injury you can see this in infection you can see it in contusion but stroke is a good example and what we expect to see when you look at an image of a patient with an infarct is that there will be a region which has a higher signal amplitude relative to the normal unaffected tissue that's the typical finding in acute infarction why is that the case why does the stroke light up like a light bulb so to speak of a diffusion-weighted image so on the echo planar image this t2-weighted image you may or may not write to see much of anything going on in this area but the pathophysiology in terms of diffusion that is happening in this infarct is that there is as we already said there's a dramatic decrease in diffusion or what is sometimes called restriction of diffusion but the water molecules are constrained and they're not able to move as much as a result of that decrease in diffusion that means that when we perform our diffusion-weighted image that in the presence of these gradient magnetic fields the region of the infarct spins are undergoing relatively little movement and there is therefore relatively little attenuation of the signal whereas in normal unaffected healthy tissue there is a much larger amount of movement the diffusion is not restricted and as a result there is a large amount of signal attenuation so when we compare what things look like on our B zero and our diffusion-weighted images there is going to be a large degree of davit nua ssin of signal in the normal tissue and much less attenuation of the abnormal signal so the contrast between these two areas is such that we have a much higher signal amplitude in our diffusion-weighted image relative to the normal tissue but if you actually go in and measure the signal in the area of the infarct on your b0 image and compare that to the measure of signal on the diffusion-weighted image where this thing lights up like a light bulb how will they compare well they won't be the same it'll probably be a little bit lower on the diffusion-weighted image there's still some degree of signal attenuation so I just want to be clear that we tend to look at these diffusion-weighted image and you see this huge bright area that represents the infarct it does not represent that the signal ant was augmented or increased relative to the b0 image simply that it didn't undergo the same change as normal tissue and that's what's reflected in the ADC image now let's say we have some other type of pathology which for whatever reason is causing us to have very high signal amplitude on the b0 image maybe it's a tumor so that tumor has very high signal why because perhaps it has a very long t2 and if we look at what happens during our relaxation the tumor is doing this the normal tissue is doing that and we have this contrast where there's a much higher signal amplitude in the tumor well we've acquired this t2-weighted image t2-weighted echo planar image and now we're going to do it all over again adding this diffusion sensitization in that tumor there is diffusion it's tissue like any other tissue now of course right tumors may have different right diffusion coefficients the rate of diffusion may be slightly different but you know in a gross sense there is diffusion going on here just like there is in the other tissue and therefore if we look at this area on the diffusion-weighted image and we compare what happens between image number one no diffusion sensitization and image number two with diffusion sensitization we should expect to see a decrease in signal however if we compare on the diffusion-weighted imaging what this area looks like relative to our normal unaffected tissue it actually might look much higher signal even though if we go and make a comparison between the normal tissue before and after that there was a similar degree of decline so the reason for that is simply that this had so much more signal on our b0 image that even after that relly Creason signal amplitude it still has a large amount of signal right and this is called what right this is the famous T - right shine through artifact now if only if the only thing you're looking at is this diffusion-weighted image you will see this area of very high signal relative to the rest of the brain how do you know whether it is high signal in the sense of this infarct or high signal due to this t2 shine through artifact well the way you do that is you actually make a measurement and you compare before and after in the infarct because there is so little signal change from image number one to image number two the ADC that we measure is going to be very small if this is a grayscale ADC image it's going to be black because it's at the low end of our ADC scale whereas this area will have a much more substantial change in signal and will be somewhere in the mid to upper range of ADC so it's looking at a quantitative diffusion-weighted image that allows us to make that differentiation between whether something is truly restricted or slow diffusion or if it is simply an artifact due to high signal amplitude that was present on the b0 image looking at the b0 image enough isn't good either because I'll show you in a minute right these infarcts often show up on those b0 images as well write that pathology can show up due to differences in t2 so to make this assessment you have to look at a quantitative image right and you really should never be reviewing diffusion-weighted imaging in the absence of a quantitative apparent diffusion coefficient image all right any questions about this yes I will diffusion-weighted imaging even though there's a signal loss sir signal attenuation and then and and but it still looks bright so I know this is really basically why does it look so it's relative to the other tissue one of the things that you will notice if you take in a global look at image number one be zero and image number two be 1000 and you compare the signal in the brain to the signal that you can detect in the background where there should only be noise there's a whole lot more signal visible out here these are dramatically lower signal-to-noise images so the reason why it looks like it is high signal is simply because it is so much higher signal than the remainder of the brain right that when we window and level this image for display we're doing that based on the overall image if you actually displayed these at the exact same window and level setting right this the remainder of the brain would be not even visible and the area of the infarct would look a little bit lower signal amplitude than it did on the b0 image so it's just a display thing okay so what we're detecting here is contrast it's not an absolute change in signal of the infarct it's the contrast between the infarct t'me and the normal tissue okay anything else so let's look at some examples of this okay so this is a normal person I don't know if they're normal but the images are normal so the top row right I've indicated on the left side these are your t2-weighted echo planar images the B value is zero the second set of images are exactly the same thing repeated with a B value of a thousand that just means we have those sets of bipolar gradients at a gradient amplitude and duration that corresponds to a B value of a thousand the images in the bottom are the apparent diffusion coefficient images so these are computed quantitative images based on the change in signal amplitude from t2-weighted image to diffusion-weighted image so if you notice if you look at these images and for example take a look at the cerebrospinal fluid in the ventricles so these are t2-weighted images as we would expect like on any other t2-weighted image there's high signal in the CSF there is dramatically low signal in that same cerebrospinal fluid on the diffusion-weighted image why because this is like the glass of water there is free diffusion of those water molecules during the period that we're sensitizing our image to diffusion they cover a tremendous amount of ground and undergo a huge attenuation of their signal and when we quantify that signal change it comes out at the top end of the ADC scale the ADC is very high and vice versa if you look at tissue if you look and make a comparison between let's say white matter in the internal capsule and gray matter or even the subcortical white matter you notice that the white matter seems to be a little darker on the diffusion-weighted image right that's also due to the rate of diffusion and lastly to make a comparison between the b0 image and the diffusion-weighted image by looking at the background this is a much lower signal-to-noise image and that's all a function of the fact that we destroyed all the signal deliberately by turning on those diffusion sensitizing gradients okay oh sorry this artifact over here okay sure which you see all the time and you have to always you're on call and you have to know what to do with that right so you see it commonly along the temporal bone we don't have it here but you'll often see it around the orbits these are echo plane our images they are constructed right by applying an RF and then having a long series of gradient echoes all of those gradient echoes a gradient echo image in general is sensitive to magnetic susceptibility gradients remember the artifacts that we looked at yesterday where when you have something in your subject that you're imaging like the metal hairpin right because it causes significant disturbances in the magnetic field strength it can cause signal loss and it can also cause signal to be displaced in the image to be spatially miss registered so here we're along the temporal bone where there's a lot of bone and air there's a gradient of magnetic susceptibility and the signal is just simply being displaced why do you see it on the diffusion-weighted image but you don't see it on the b0 image that's also an echo planar image hmm any ideas how can we see that here if the same effect going on here all right this part of the looks like brain actually on this image doesn't it of course all of us anatomic experts know that's sitting right on top of the temporal bone so how come this signal is showing up in the diffusion-weighted image it's basically being displaced from the next slice down but we don't see it on the b0 image what's different between these two images extra gradient magnetic fields those gradient magnetic fields increase the sensitivity of the image to these magnetic susceptibility effects so your diffusion-weighted images are even more sensitive than your b0 images to those effects you'll see this around the sinuses if you have a patient that has I don't know you know some metallic fragment and their skin or air anything like that you'll have these large gradients of magnetic susceptibility and see these types of effects so they're the kinds of things you just have to get used to looking at and know when to expect them okay this is just an example showing you how we quantify the ADC right so two images this is the diffusion weighted image on the top this is the apparent diffusion coefficient image on the bottom and what I'm showing you on the right is a graph where we took a region of interest alright this one is number two and we simply plotted the signal intensity on the b0 image which would be on the right and on the B 1000 image on the left the slope of that line is extremely steep when we look at CSF and is much flatter when we look at tissue the diffusion coefficient as we are describing it is being measured based on these two measurements so essentially it is a met a two-point approximation we're really looking at the slope or the magnitude of change between these two images one thing I just want to mention briefly for those who may be interested is that what would happen if we looked at signal intensity relative to B value well if the B value is very small like zero we'll have a large amount of signal intensity if the B value is extremely high we know we will have a small amount of signal intensity if we just make those two measurements these two points show us a line and the slope of that line is proportional to the ADC if on the other hand we acquire many many images each one with a progressively increasing B value we do not get points that line up on this line we get an exponential and it's the time constant of that exponential that represents the ADC so if you want to accurately or with as much accuracy as possible quantify the ADC you need to image with multiple B values which of course is time consuming and requires more complicated computations in clinical diffusion-weighted imaging we are making a two-point approximation of this exponential decay which does a reasonably good job of answering the questions that we have in clinical imaging like in brain imaging of stroke but one thing to be aware of is that if there is an area in the brain where there is let's say only a subtle disturb of diffusion there's only a small decrease in diffusion it might be something that is not detectable by this two-point approximation so when you look just as sort of a little clinical purl when you look at your diffusion-weighted images and you see an area where it looks like maybe there's some high signal on that diffusion-weighted image and what do we do right away as we rush to look at the ADC image and see whether it looks black on the ADC image and you look at the ADC image and it looks okay so therefore this must be some artifact T to shine through something like that it's also possible that we are simply not able to detect that alteration in the ADC with this two-point approximation and whether and this is a particular problem with relatively subtle changes in ADC so you have to kind of qualify your interpretation by saying that we may not be able to fully characterize this based on the way that we've measured it okay so the diffusion-weighted images will be sensitive to these relatively small changes in ADC right by showing us a signal right a higher signal than the surrounding normal brain tissue but the specificity or I guess maybe it's the sensitivity of the ADC image for picking that up may be insufficient so we can have scenarios where we see something on the diffusion-weighted image and are not able to detect it on the ADC image now for you know large acute infarct this is not a problem but when you're looking at more subtle abnormalities it may be an issue okay so here is an example again b0 is at the top all right this is B 1000 and this is the ADC image so what's wrong with this person right there's a large area a very high signal in the diffusion-weighted image corresponding to the right ACA and mca territories this is an acute infarct there's actually some high signal on the t2-weighted image but this is such a large abnormality with such a dramatic effect on diffusion that even that it shows up on the diffusion-weighted image and is still clearly dark on the ADC image so there's no issue of T to shine through over here because there's such a dramatic change in signal from B 0 to B 1000 but if you look carefully and you take a look in the left frontal lobe you notice that there is an area where there's high signal on the t2-weighted image and there's also some high signal on the diffusion-weighted image but when you look at the ADC image if anything it's higher signal than the normal tissue right so this is a t2 shine through artifact in this case this was actually a fungal infection in the sinuses and this is cerebritis and edema in the left frontal lobe and because of invasion of the carotid there was infarct of the right AC AMCA territories okay all right so any questions about diffusion-weighted imaging sometimes we see a be 500 or like smile is okay in body imaging perhaps right so in some settings so I mean I can give you an example that's more near and dear to my my heart the delivery so we also sometimes do diffusion-weighted imaging in the spinal cord okay or something called diffusion tensor imaging which we'll talk about next which is really just a variant of diffusion weighted imaging in the spine the problem that we have is that we have all of these bones right this is our sagittal cartoon and we're interested not in the bone we're interested in the spinal cord because of all of this bone surrounding the spinal canal there are some pretty significant gradients of magnetic susceptibility and in doing a diffusion-weighted imaging exam where you're using echo planar imaging even though it might be spin echo echo planar imaging with that long string of gradient echoes there's already going to be problems with distortion and signal loss which will only be amplified as you turn on the diffusion sensitizing gradients using the same magnitude diffusion sensitized ingredients we use in the brain be of a thousand can create significant problems with image quality by reducing that B factor that B value we can minimize some of those artifacts so that's one of the considerations the other consideration is just that by experiment and evaluating a new application of diffusion weighted imaging maybe in the liver per se you need to determine what B value gives you the optimal signal-to-noise there's a trade-off and B value that as you increase your B value you are increasing the signal change that occurs due to diffusion but at the same time you're causing your diffusion weighted image to have lower signal-to-noise so there is a trade-off between signal-to-noise and contrast due to diffusion that you have to take into consideration okay the the B value is a product of the time and the and the strength yes does that only apply to diffusion-weighted imaging does that as opposed to what women anytime a turn-on we didn't so any of our images or diffusion-weighted images in a sense because we have gradient magnetic fields and you're absolutely right if the gradient magnetic field is stronger or on for a longer period of time we're going to see more effects due to diffusion what is the what's the relative I mean you do a much better line but how much bigger we actually pack in the magnitude of these gradients they're huge right in order to do this right and this is something I don't really emphasize that much anymore because diffusion-weighted imaging has sort of become a standard such a standard clinical tool that if you went into private practice tomorrow and went shopping to buy a new mr scanner you probably wouldn't be able to find one that you could buy new that didn't have the ability to do diffusion weighted imaging so not that many years ago that wasn't the case and you could have gone out and bought a scanner that actually wouldn't be capable of doing diffusion weighted imaging simply because it couldn't drive the gradients strong enough and fast enough to accomplish the diffusion sensitization in the earliest days of diffusion weighted imaging you know when you when you know the GE 1.5 tesla signal scanner was like the state of the art so the few places that were doing diffusion-weighted imaging very early on this is like I think I was an intern at this time they what they did was they used to have to have this huge thing that they bought and attached to the back of the scanner which was an extra powerful gradient set they had to retrofit the equipment so it definitely requires a significant increment of gradient performance to be able to do it it's not what you're using for your typical imaging and I don't want to mislead you when I looked at this pulse sequence initially before we added those gradients and said this image is sensitive to diffusion those aren't effects that we can really detect but it is there I just want you to realize conceptually that anytime you turn on a gradient magnetic field your attenuating the signal due to diffusion okay I don't mean to say that there's usable or visually detectable diffusion information

Channel: Albert Einstein College of Medicine

Views: 79,953

Keywords: Magnetic Resonance Imaging (Diagnostic Test), Medical Imaging (Field Of Study), Diffusion MRI, MRI, diffusion imaging, MRI lecture, MRI course, michael lipton, gruss magnetic resonance research center, Albert Einstein College Of Medicine (College/University)

Id: dW8Yh-c2xVY

Length: 67min 12sec (4032 seconds)

Published: Tue Sep 23 2014