Transcription
Hello everybody, and welcome back. In the previous talk, we looked at spin echo pulse sequences. We saw that two successive radio frequency pulses could lead to the generation of what's known as a spin echo. And if we used a 90-degree followed by a 180-degree radio frequency pulse, we could account for the local magnetic field inhomogeneities that led to the rapid loss of transverse magnetization within our sample. And that rapid loss was known as free induction decay, or T2 star. And the spin echo that was generated created a signal that was much closer to the true T2 signal from within our sample.
Now, today I want to look at gradient echo pulse sequences. And I want to separate the topic of gradient echo into three separate talks. Our talk today is going to focus mainly on what exactly a gradient echo is, and how it differs from a spin echo, and why we need to go about actually generating a gradient echo.
Then, in the next talk, we're going to be looking at flip angles and how we can utilize different flip angles to actually modulate contrast within our image, as well as touching on what's known as the Ernst angle, which is the angle, the flip angle that gives us the best signal based on a T1 characteristics of our sample, as well as the TR of our pulse sequence.
The final talk in this gradient echo module is going to be looking at different specific gradient echo sequences: coherent sequences, incoherent sequences, as well as steady state free procession sequences. Now, my hope is that by the end of these three talks, you're going to have a good understanding of how exactly gradient echo pulse sequences work. For me, they're much more complicated to cross conceptually than spin echo pulse sequences. After that, we can move on to our last broad group, the inversion recovery sequences.
Now, you can see I've drawn two separate pulse sequences: a spin echo sequence that you should be familiar with by now, and a gradient echo sequence. And we're going to be looking at the differences in the next three talks. You'll notice that a gradient echo doesn't have this 180-degree pulse here, and the TR and TE in gradient echoes are much shorter than they are in spin echo sequences. And we're going to see why we are able to do that in the next three talks.
So let's go to a pulse sequence here. This is a gradient echo pulse sequence. We've got no second RF pulse, just one single RF pulse. And we flip our spins a certain degree, let's say 90 degrees for example. And as you'll see, especially in gradient echo, we often don't use 90 degrees. We often use smaller flip angles. And the reason for that will become apparent. So we flip our spins at certain degrees, let's say 90 degrees in this example, and it's slice selected. Only the specific slice is going to flip into the transverse plane, giving us transverse signal.
Now, if we were to apply a frequency encoding gradient and we would just to measure the signal coming from this slice, that signal is going to decay at free induction decay, or T2 star. There are two mechanisms for that loss of transverse signal: spin-spin interaction, as well as loss of phase due to the local magnetic field inhomogeneities. So our spins here are going to precess at the Larmor frequency. They're experiencing no gradient except the main magnetic field. But we are losing signal here. The vectors themselves have still got transverse magnetization, but they have dephased from one another through both T2 relaxation, spin-spin interaction, as well as the T2 star free induction decay because of those local magnetic field inhomogeneities.
Now, this loss of signal, we've said, is T2 star decay. Now, when we looked at spin echo, we could then rephase these spins that have dephased like that because of the local magnetic field inhomogeneities. We could use a 180-degree pulse to flip those spins, change their orientation, and allow the spins at dephased faster to now lag behind the spins at dephased slower. That change in orientation allowed those spins to then rephase and create a transverse signal that was much closer to T2, or true T2 signal from within the slice.
Now, in gradient echo imaging, we don't use that 180-degree RF pulse. And when I think about the 180-degree RF pulse, the way I like to think about it is two separate spins here, and they're running a race. One is a fast spin, and one is a slow spin. In spin echo, what happens is one goes faster than the other spin based on the local magnetic field inhomogeneities. Now, we have no control over those local magnetic fields. Those inhomogeneities are constant within our sample. So what happens is one will precess faster than the other. They will dephase because of those differences in precession. Now, the differences in speed is because of those local magnetic field inhomogeneities. What happens with a 180-degree RF pulse is the direction of those spins change. The orientation changes. Now, the faster spin is still traveling faster, but the direction it's going is different. If these two spins were in a race, now this fast spin needs to catch up with the slow spin. The direction of the race has changed. So remember that analogy of the direction of the race changing based on that 180-degree flip in spin echo imaging.
Now, if we're to look at gradient echo imaging, we can see that when we are actually taking the sample, we know that we need to apply a frequency encoding gradient. We can't just measure that sample getting lost at free induction decay. We need the frequency encoding gradient in order to do that one-dimensional Fourier transformation and figure out where that signal is coming from along the x-axis of our slice.
Now, there's a problem. When we apply a gradient along the x-axis, not only are we losing phase now because of spin-spin interaction, as well as losing phase because of local magnetic field inhomogeneities, but we're also going to lose phase if we apply a gradient along the x-axis here because these spins are now going to precess faster than these spins here. The precessional frequency of these spins being faster contributes to further dephasing. Now, there's three mechanisms for dephasing. Notice now what is happening. As these spins spin faster, the whole sample goes out of phase even quicker than it did when we just let that sample go.
Now, unfortunately, this is something that we can't really avoid. We have to apply a frequency encoding gradient to get spatial localization of signal within our sample. Now, because we have to apply this frequency encoding gradient, we get an even more rapid loss of transverse magnetization here. So our frequency encoding gradient is a double-edged sword. We need it to delineate those signals, but it leads to even sharper decline in our transverse magnetization. And we can only measure transverse magnetization in MRI. We can't measure longitudinal magnetization. This is the only signal we have to generate our image. So how do we go about surpassing this? Well, you may have guessed it. We need some sort of mechanism to regenerate that signal here, to get signal that more closely matches our T2 star decay.
Now, gradient echo can't account for the local magnetic field inhomogeneities. We don't have that 180-degree flip that allows our leaders to catch up with our laggers. So let's look at how exactly we go about generating a gradient echo. Now, I mentioned this when we looked at frequency encoding gradients. What we can do is apply an equal and opposite frequency encoding gradient along that x-axis of our slice. Now, remember, this rapid loss of signal is happening because of a gradient that we are applying, a known gradient at a known strength across our slice here. It's not local magnetic field inhomogeneities that we can't change. So we apply what's known as a dephasing gradient before we apply our readout gradient here.
Now, we are adding to the magnetic field on the left-hand side of our slice, and we're taking away from the magnetic field on the right-hand side of our slice. These spins are still going to be precessing in the same way. They're still going to be precessing in this example in a clockwise fashion. However, these spins are going to precess or resonate slower than these spins here. And we're going to get dephasing. The same dephasing that we saw before. This gradient is the other way around, but the dephasing is still the same. The magnitude of our gradient field strength is the same for the whole slice here. The orientation is just slightly different.
Now, if we were looking at our analogy of spins precessing faster or slower, we saw that in spin echo, one precessed faster because of a local magnetic field inhomogeneity, and one precessed slower. Then we changed the direction of the race, and the faster ones caught up with the slower ones. Here, now we've got faster spins because of this extra magnetic field, and slower spins because of this gradient magnetic field here. We've got some spins going faster, some spins going slower. These faster ones, think of them as having a tailwind. They've got wind pushing them along. They're running at a set rate, but they're going slightly faster because they've got this gradient field behind them. The slower ones have got a headwind. There's wind coming into them. They've got a gradient field against them. They're precessing slightly slower than the Larmor frequency here.
What we can do then is allow that to play out. We've got our dephasing gradient causing that loss of signal that we've seen before. And that sharp loss of signal here is because of this dephasing gradient. Now, what we can do is apply an equal but opposite frequency encoding gradient. Our spin that had the tailwind now has a headwind. It's going to precess slower. Our spin that had the headwind now has a tailwind. We haven't changed the orientation of these spins. They're still processing in the clockwise direction. However, we've changed the rate of resonance here. Our fast spin is now going to be our slow spin, and our slow spin is now going to be our fast spin.
If we apply this gradient for the exact amount of time that we applied the dephasing gradient, and the strength of that gradient represented by the area under this curve here is the same, but in the opposite direction, at TE here, our initial fast spin would then have been caught up by our initial slow spin because of the change in this gradient. We haven't changed orientation, but now they've traveled the same distance. We will get re-accumulation of phase up until TE, and then we will get loss of phase again because we are applying this frequency encoding gradient, as you can see from this diagram, for twice the amount of time as we applied our dephasing gradient.
So watch now as I play this. Watch how these spins go faster than these spins, and there's an eventual re-accumulation of phase before we get that de-accumulation of phase, or dephasing again. Watch how these spins start to generate some signal. They start becoming more and more in phase at TE. They're all in phase, and then we get loss of that signal, rapid loss of that signal again because of this frequency encoding gradient that we're applying. Now, that is the basis of a gradient echo.
Now, we're only accounting for the loss of phase because of the gradient that we've applied along the x-axis of our slice. This gradient that we're applying, it's not a radio frequency gradient that is allowing spins to flip from the Z axis like this. When we apply a 180-degree radio frequency pulse, we're changing the orientation of those spins. That's not what we're doing along the frequency encoding gradient here. We are just changing the rate at which those spins precess. And that change in rate of precession is what's leading to dephasing. So we allow some spins to dephase faster initially without dephasing gradient, and then we apply the rephasing, or the readout gradient, allowing those spins to catch up and rephase at TE, which gives a signal that is similar to T2 star, or our free induction decay.
So our gradient echo is only going to give a signal that is equal to our free induction decay. We can't get up to that T2 signal that we got in our spin echo pulse sequences. That's one of the main differences between spin echo and gradient echo.
Now, you may notice from this timeline here that TE is quite far away from our RF. And if we're losing signal at free induction decay, we know that happens quite quickly. We don't have any 180-degree RF pulse here to get in our way. An RF pulse is not like a gradient pulse. Gradient pulses, we can change the magnitude of those gradient pulses. Remember back to the bandwidth talk, the magnitude of that gradient magnetic field strength determined our bandwidth. The stronger that magnetic field strength, the wider or broader our bandwidth, the quicker this readout happens. A radio frequency pulse, we can't change the strength. The time it takes to flip spins 90 degrees is a set amount of time. It occurs at a set radio frequency, and that radio frequency pulse needs to match those Larmor frequencies in order to flip those spins. So both the 90-degree radio frequency pulse and the 180-degree radio frequency pulse take a finite set period of time that we have to wait for. We can't make those shorter.
Now, because we don't have that 180-degree radio frequency pulse in gradient echo, we can move that TE much closer to our RF here. And you can see this process occurring, allowing us to get more signal from this T2 star decay. We don't have that 180-degree pulse to get in our way. And this is one of the benefits of gradient echoes: is we can get quite rapid imaging because our TE can be really close to our RF.
Now, we've seen that in T1 imaging, we want a short TE to negate those T2 differences within our tissue. The same in proton density. We want very short TEs. And that's why gradient echo is great for proton density and T1 weighted imaging. And when we're measuring that signal that's occurring during this frequency encoding gradient, that is converting the analog signal into a digital signal that we apply in k-space. And this sequence here is going to fill one line of k-space.
Now, again, where do we place our TR? That's always the question in these sequences. Our spins dephase, they lose transverse magnetization, but they are still in the transverse plane. We get that rapid T2 loss. Then we have to wait a long time for those spins to go back into the longitudinal plane. They're out of phase, so there's no transverse signal, but they still have magnitude in the transverse plane. Our spin echo allowed us to rephase those, and we could do multiple spin echoes while we're waiting for the next TR. Gradient echo is not the same because this T2 star is quite rapid. We can, as we'll see later, flip these gradient pulses, but we're not really going to get more than three or four gradient echoes before we completely lose our signal because of those local magnetic field inhomogeneities.
Now, we're faced with a problem here. Although we can do this really rapidly in gradient echo imaging, where do we place our next RF pulse? If we do it too quickly, and we've flipped our spins 90 degrees, they are not going to have gained much transverse magnetization. And if we then flip our spins at our next RF pulse, remember the magnitude of the transverse magnetization is only the amount of longitudinal magnetization that we have recovered. That's why we normally have to wait so long for the TR for our next radio frequency pulse.
Now, it turns out there's a way around it in gradient echo imaging, and it all has to do with the flip angles that we use. We don't often use 90-degree flip angles in gradient echo imaging. And I'm going to show you in the next talk how we can use much smaller flip angles, which allows us to bring our TR much closer. It allows us to keep the longitudinal magnetization within our sample, and we can make both T1 and T2 weighted images with very short TRs. Our flip angle allows us to modulate contrast within our image. That's all for the next talk.
Again, if you're studying for an MRI exam, these are the kind of concepts that come up. And I've included them in a question bank that I will link below in the description. Go and check it out if you are preparing for a specific exam. I hope you understand how a gradient echo is formed. Don't get too caught up on TRs and TEs and flip angles. That's all going to be covered in the next talk. So until then, goodbye everybody.