Transcription
Hello everybody, and welcome back. Today, we're going to be looking at fast or rapid pulse sequences.
Now, up until now, we've been looking at two separate types of pulse sequences: both spin echo and gradient echo sequences. And for both of these sequences, we acquire a single line of k-space per pulse repetition. We apply a specific degree of phase encoding and then we apply a frequency encoding or readout gradient. When we're converting that analog signal into a digital signal that we apply in a line of k-space, we then wait until the TR before we repeat the process with a different degree of phase encoding. And it's the degree of phase encoding that determines where we place that data within k-space. So, the rate-limiting step here is how long it takes to get to our TR for each line of k-space. It takes a full pulse repetition to fill that line of k-space. Only then can we enter a new line of k-space. And to get enough data for the entire slice, we need to fill the whole of k-space, and that takes an incredibly long time.
Now, there are very few pulse sequences that you'll use clinically that only fill a single line of k-space per pulse repetition. And we can use multiple different techniques to fill multiple lines of k-space at one time. And when we do that, that's what's known as fast pulse sequences or rapid pulse sequences. Now, we've actually covered two of these previously, and I'm going to touch on them now and show you how they differ from the third type. So, today we're going to be looking at fast spin echo sequences, short TR gradient echo sequences or rapid gradient echo sequences, and echo planar imaging. Now, we haven't covered echo planar imaging. We have covered the first two, so let's cover those first two first.
Now, fast spin echo imaging is part of our spin echo pulse sequences. And if you've cast your mind back to that talk, we looked at multi-echo sequences, multi-slice sequences, and fast spin echo imaging. Now, what exactly are we doing during fast spin echo imaging? Well, we apply a 90° radio frequency pulse, flipping our spins into the transverse plane, or flipping our net magnetization vector into the transverse plane. That vector then loses transverse magnetization at a rate of T2 star or free induction decay because of both spin-spin interactions as well as the local magnetic field inhomogeneities. We then apply a 180° radio frequency pulse that will allow for those dephased spins to start to rephase in the transverse plane, reaccumulating signal back up to the levels of the T2 decay constant, before then losing signal at free induction decay or T2 star during this reaccumulation and loss of phase again. We measure that signal at TE and we measure that signal for a specific phase encoding gradient. This signal can then fill one line of k-space.
Now, instead of then waiting for all those spins to fully regain their longitudinal magnetization before then flipping 90°, what we can do is these dephased spins can again be flipped with 180° RF pulses and be allowed to rephase and reaccumulate signal. That reaccumulation or rephasing can then be read out again and fill a separate line of k-space. Now, why can it fill a separate line of k-space? Well, we apply what's known as a rewinder phasing encoding gradient to cancel out the effects of this initial phase encoding gradient. So, during this next echo, we can apply a different phase encoding gradient that corresponds to a different line of k-space. We can then repeat these echoes multiple times prior to then repeating the sequence itself at the next TR. And this is why this is called fast spin echo imaging. We are creating multiple echoes.
Now, a couple of important points here. One, when we acquire signal, where we put it into k-space is determined by the amount of phase encoding gradient that we apply in the y-axis direction. The second is that these echoes are being created by this 180° RF pulse. These are spin echoes that are being created. Now, if we were to look at k-space and see how we go about filling these lines, well, during this first spin echo, we would then fill that line of k-space from left to right based on the degree of phase encoding that we have applied. During our sequence, within the same TR, we will then apply the second line of k-space and then acquire the third line of k-space, each corresponding to the degree of phase encoding that we have applied during these echoes. Obviously, the initial echoes are going to have more transverse magnetization because they are earlier on in the sequence. We are losing signal at a rate of T2 here. The spin echo accounts for those local magnetic field inhomogeneities, that accounts for the free induction decay here, and brings the signal back up to levels of T2. What it can't account for is the spin-spin interactions that cause this T2 decay. As we head longer and longer, further down more and more echoes, the actual amount of transverse magnetization is going to be reduced.
Now, when we fill k-space, we know that the middle of k-space provides us our contrast information. It provides us with the maximum signal differences within our tissue, the differences in either the T1 recovery or the T2 decay constants between tissues. Now, why does the center of k-space give us our contrast information? Well, the center of k-space has very little phase encoding gradient. We haven't dephased those spins with a phase encoding gradient to a large degree. And because we haven't applied much phase encoding gradient, we keep most of the signal from these pulse sequences that are required to make the middle of k-space. Here, as we head further and further out into k-space, we are applying a larger and larger phase encoding gradient. The spins are more dephased based on the phase encoding gradient that we have applied. So, these peripheries of k-space don't give us a great deal of signal themselves, but what they do give us is still the ability to determine the differences between the edges of tissues. The more phase encoding gradient we apply along the y-axis, the more out of phase the spins will be based on their y-axis location, the better we're able to delineate edges within our image. So, the edges of k-space give us edge definition, and the center of k-space gives us contrast.
Now, you'll see why this becomes important as we go through these fast pulse sequences, but just bear that in mind. The order at which we fill k-space will determine what contributes to contrast and what contributes to the edge definition within our image. Now, here we can't actually speed up the rate at which we do the 180° RF pulses. That takes a set amount of time. What we can do is bring this 180° pulse closer to our initial 90° pulse, and that itself will also compress the pulse sequence. You can see now how we can acquire multiple different lines of k-space based on the number of echoes that we generate. The number of echoes that we generate is what's known as the echo train length, and our acquisition time is reduced by a factor of the echo train length. If we have 10 echoes within our pulse sequence, we will reduce that spin echo time, the full acquisition time for that individual slice, by a factor of 10. You may also hear these being called turbo echo or turbo spin echo imaging. They're synonyms. They describe the exact same thing.
Now, what this allows us to do is fill multiple lines of k-space but still keep our TR relatively long. We want to keep our TR long because our TR determines how much magnetization is going to be flipped into the transverse plane at our next pulse sequence. If we don't allow spins to fully recover their longitudinal magnetization before we then flip them 90°, we are going to have reduced signal. With shorter TRs, this allows us to keep long TRs but still fill multiple lines of k-space, still get a lot of data while we're waiting for that TR. If we were just acquiring one line of k-space and we had a really long TR, our scan is going to take hours to complete because we need to individually fill lines of k-space, we need to take multiple signal averages of k-space, and then we need to take multiple slices of the image that we're trying to acquire. This allows us to acquire images much quicker.
The next type of fast imaging that we've looked at is what's known as short TR gradient echo imaging or rapid gradient echo imaging. We apply a radio frequency pulse that flips spins a certain degree into the transverse plane, or flips the net magnetization a certain degree into the transverse plane. We apply a specific degree of phase encoding gradient and then we apply our frequency encoding gradient. Initially, a dephasing frequency encoding gradient that allows for loss of signal or causes a loss of signal, and then a rephasing frequency encoding gradient when we actually read out our signal at TE. That rephasing frequency encoding gradient creates what's known as a gradient echo. It's not the same as a spin echo, which is allowing spins to rephase, accounting for those local magnetic fields. This rephasing frequency encoding gradient allows for rephasing of the dephasing that we've caused with our frequency encoding gradients.
Now, remember a really important point when it comes to reading out signal from a slice, we can only read out signal during a frequency encoding gradient. That's why the frequency encoding gradient is often known as the readout gradient. If we don't have a frequency encoding gradient applied during readout, we won't be able to do a transform that information and figure out where the signal is coming from along the x-axis. Now, we're able to have very short TR times during this gradient echo because we use small flip angles. The use of small flip angles allows us to have short TR times while still getting T2-weighted images. The short flip angle, if we were to have spins in the longitudinal plane, a short flip angle will allow for full longitudinal recovery in this very short period of time, allowing for the negation of T1 differences between tissues. Being able to do this, being able to generate a single line of k-space in a very short amount of time, means that we can repeat this sequence multiple times, filling multiple parts of k-space. So, we can then fill k-space based on the phase encoding that we've applied to each individual pulse sequence. Here, when we flip the spins here at the next TR, in this case 50 milliseconds, we create another free induction decay curve. We are not sampling this initial free induction decay curve, and each single flip here will give us a separate line of k-space based on the degree of phase encoding. What this allows us to do is because we are generating a new free induction decay for each RF pulse, our sampling time, our TE from the free induction decay curve, is the same, and therefore our contrast contribution is going to be the same for every single line of k-space that we do. When we looked at our fast spin echo imaging, here you can see that our TE is getting longer and longer. Our level of contrast or degree of contrast contribution based on the T2 decay curves of the tissues is going to be different for different parts of k-space. And depending on what we fill the middle of k-space, that's going to be the predominant contrast weighting within the image. Our gradient echo allows us to keep a consistent contrast throughout k-space. Again, with this fast sequence, we still can't compress the sequence much smaller than this one because we need to allow for full longitudinal recovery of the spins, and two, because we still are applying multiple RF pulses.
This brings us to our final rapid imaging sequence, which is known as echo planar imaging. Take this pulse sequence here. We've got a basic gradient echo sequence here. We apply a 90° radio frequency pulse, and the spins start to dephase at a rate of T2 star. We then apply a dephasing frequency encoding gradient, which causes a rapid loss of transverse signal because we're applying that frequency encoding gradient along the x-axis, and apply a rephasing frequency encoding gradient, which we read out our signal. Now, during the frequency encoding gradient, we can actually read out signal both in the dephasing and the rephasing frequency encoding gradient. What would happen then if we were to repeat that process? What we would generate is another gradient echo. We are generating an echo based on the dephasing and rephasing frequency encoding gradients that we ourselves have applied to the slice. This rephasing and dephasing isn't occurring as a natural process like we saw in this spin echo where the laggers start to get caught up by the leaders after our 180° RF pulse. Here, we are forcing dephasing in the frequency encoding direction and then forcing rephasing in the frequency encoding direction, and that's what's generating these gradient echoes. It turns out we can repeat this process multiple times during our free induction decay curve, and what we are generating here are multiple gradient echoes.
Now, the limitation here is a hardware limitation. How quickly is our machine able to apply these alternating frequency encoding gradients, and how rapidly is a machine able to convert that analog signal into a digital signal that we place into k-space? If we can sample each one of these frequency encoding gradients rapidly enough to allow us to get enough x-axis resolution in the type of image that we're trying to create, then we are able to do what's known as an echo planar image. Now, where does the phase come into this? Well, we can apply what's known as a phase blip. Initially, we have no phase encoding gradient, and we apply a very small amount of phase encoding gradient that allows us to place the data that we've acquired during the first dephasing frequency encoding gradient into a line of k-space. We can then add one phase encoding gradient blip for each frequency encoding gradient that we apply here. Now, we are not rewinding the initial phase encoding gradient, so each phase encoding gradient blip that we add on, we are adding more and more phase encoding gradient to the slice. So, when we go about filling k-space now, we are going to add a little bit of phase for each phase encoding blip here, and for each line of k-space that we acquire during each frequency encoding gradient, we are going to fill a separate line of k-space.
Now, our first frequency encoding gradient is actually in the opposite direction to where we normally read out that analog signal. So, when we place it into k-space, we're actually going to place it from right to left, and it's going to go into the center of k-space because it's only had a very small phase encoding gradient contribution here. We add an extra bit of phase encoding and we apply the frequency encoding gradient in the opposite direction along our x-axis. Now, we fill k-space from left to right. We repeat this process, adding incrementally more phase encoding gradient as we acquire each line of k-space during these alternating frequency encoding gradients. This allows us to rapidly fill k-space with this data that we're acquiring during these frequency encoding gradients and allows us to fill a large portion of k-space in one single free induction decay. We will get to a point where the level of transverse magnetization is so small and it's not more than background noise, we're not going to get any useful information, and we will need to repeat this process. We will need to repeat another 90° RF pulse and fill a separate section of k-space.
Now, if we fill multiple chunks of k-space at a time, that's what's known as multi-shot echo imaging, multi-shot echo planar imaging. Now, the example we're looking at here, we are sampling signal coming from a gradient echo. Each one of these is causing a repeat gradient echo to occur, and we are sampling free induction decay. What happens if we were to precede this by a spin echo? We flipped our spins 90°, allowed them to lose signal, a free induction decay, and applied a 180° RF pulse. This is going to cause reaccumulation of signal, as we've seen, and then loss of signal at the rate of free induction decay. We have generated a spin echo. We can sample this spin echo with the echo planar imaging technique that we looked at previously. This allows us to fill k-space in a slightly different way to how we filled the gradient echo echo planar imaging. Remember, the echoes that we are generating here are still gradient echoes. We are still generating those echoes by rapidly shifting the frequency encoding direction. The degree of signal that we get is higher in this case because of the spin echo that we've generated. We're not purely sampling free induction decay, we're sampling this echo that we've generated. How then do we go about filling k-space here? Well, you'll see that initially our signal is going to be weaker. It's going to build to a high signal and then we're going to lose signal again. When we sample free induction decay alone, we went from high signal to low signal, and we wanted to put our high signal areas into the middle of k-space to allow for our high signal to contribute to the contrast within our image, and we place this high signal with low phase encoding gradients. In this case, we also want our high signal to fill the middle of k-space here. So, we want very minimal phase change whilst the largest part of our echo is being generated. How we do that is apply a large phase encoding gradient prior to our 180° RF pulse that dephases the spins a set amount. We can then apply incrementally more phase encoding gradients like we did in the previous sequence, rephasing those phase encoding gradients ever so slightly until we reach a midpoint here, until we have accumulated enough phase that cancels out this initial phase encoding gradient before we then deaccumulate those phase encoding gradients with the additional phase encoding blips. That allows us to fill k-space again from right to left and left to right, but start at a different area. We start at the degree of phase encoding that we have initially encoded into our sequence. We can then sequentially fill k-space as we add each individual blip here and allow for the large signal to be at a point where we are not applying a great deal of net phase again. In this sequence, we have filled a part of k-space here, and that's what's known as multi-shot EPI.
Now, we can actually generate echo planar or EPI imaging that allows us to fill the entirety of k-space in a single shot, and that's what's known as single-shot EPI imaging. And single-shot EPI imaging allows us to take a full image of a complete slice within the patient within a very short period of time, within a period of time of 20 to 100 milliseconds. You can see the advantage to this. You can rapidly reduce motion artifact. If you're trying to image say the heart and you want to get a quick snapshot without getting all of that motion in here, we can do EPI imaging to rapidly fill k-space. There's obviously going to be disadvantages at the peripheries here, especially when we're doing a gradient echo EPI, we're going to lose signal rapidly. The peripheries of k-space are going to have very little signal amplitude, so we may get a slight loss of spatial resolution. Often, when we do single-shot EPIs, we reduce the matrix size, the resolution that we want to acquire within our image, to allow us to fill k-space rapidly. And as you'll see as we go to DWI imaging, we go to tensor imaging, look at functional imaging, there are certain image techniques that require rapid detection of signal at a set point in time, especially when it comes to DWI imaging, which we're going to be looking at in our next talk. We want to see what's happening right now at this given point of time and compare it to a different point of time. And if we can take a snapshot of a single slice rapidly, that allows for those comparisons to be more accurate. And that's going to become very necessary when we go into DWI imaging in the next talk.
So, hopefully, you can see the subtle nuances, the differences between fast spin echo imaging, short TR gradient echo imaging, as well as EPI or echo planar imaging. All of these techniques fill k-space rapidly, allow us to take images much faster, but they have drawbacks with that loss of signal. And depending on how we fill k-space, we can have contribution to the signal based on whether we fill the peripheries of k-space with an early echo or whether we fill the center of k-space with our higher signal echoes. And for now, I think that is a good place to leave it there. Obviously, there are differences between the different manufacturers, and there are subtleties as to how we can go about filling k-space, but I want you to understand these three major sections when it comes to rapid imaging. In our next talk, we're going to move on to DWI and perfusion imaging. I'll see you all in that talk. Goodbye everybody.