Why Don't Electrons Replace Protons in MRI Imaging?

[dangus]

MRI imaging uses primarily the magnetic fields of spinning hydrogen protons to generate images. I read somewhere that the magnetic fields generated by spinning electrons are stronger then those created by spinning protons. Why then are electrons not used in MRI imaging?

Thanks

 

[Drakkith]

This article explains the physics behind how it works. Put simply, protons in hydrogen atoms take the right amount of energy for their spins to be flipped against the applied magnetic field with a radio frequency. The RF is then removed and the protons align themselves with the magnetic field again, releasing a radio frequency as they do, which is then detected by the machine. I'm guessing that the frequency required to flip the spin of an electron is probably much to low to be useful in this manner.

http://en.wikipedia.org/wiki/Magnetic_resonance_imaging

 

[K^2]

Other way around. EPR frequencies are too high. In 1 Tesla field, you get 28GHz for free electron resonance. For protons, it's closer to 40MHz.

 

[Drakkith]

Other way around. EPR frequencies are too high. In 1 Tesla field, you get 28GHz for free electron resonance. For protons, it's closer to 40MHz.

Hmm. How do you calculate that? I figured electrons would be easier to flip since they are less massive. Obviously I was wrong!

 

[Dale]

Yes, the electron resonance is in the microwave range for typical field strengths.

 

[K^2]

Hmm. How do you calculate that? I figured electrons would be easier to flip since they are less massive. Obviously I was wrong!

Resonance frequency is proportional to energy splitting in magnetic field. That energy is proportional to magnetic moment and strength of the field. Because angular momentum of proton and electron are the same, the magnetic field of an electron is a lot stronger. Classically, you expect electron's moment to be roughly 2,000 times greater due to being 2,000 times lighter, and if you look at the numbers I gave, it's not far off.

Here is a good article to put a bit more theory with this. Wikipedia: EPR

 

[Drakkith]

Thanks. I'll have to look more into magnetic moments and such.

 

[dangus]

Thanks guys. I don't know enough about how the RF is actually generated though. Is it not feasible to generate RF pulses at 28GHz?

 

[K^2]

Thanks guys. I don't know enough about how the RF is actually generated though. Is it not feasible to generate RF pulses at 28GHz?

It's tricky enough by itself, but you also have to do some signal processing to do anything complex. EPR is done to study materials. It's just that pulse sequences for MRI get complex, and so doing them at these frequencies becomes complicated. Receiving the signal is even more complicated. For basic EPR, you don't need much. But for MRI, you need to get the whole spectrum and Fourier transform it.

 

[Dale]

Thanks guys. I don't know enough about how the RF is actually generated though. Is it not feasible to generate RF pulses at 28GHz?

Besides what was mentioned above, the other problem is that these frequencies are good at cooking meat.

 

[Vanadium 50]

But that would only be a problem if we were made out of meat...oh...wait...

Since this already works with protons, what would be the point of making it work with electrons?

 

[Drakkith]

But that would only be a problem if we were made out of meat...oh...wait...

Since this already works with protons, what would be the point of making it work with electrons?

I'm sure the OP is simply curious as to why we don't use the electrons, not suggesting we do.

 

[K^2]

Since this already works with protons, what would be the point of making it work with electrons?

Actually, there would be. Stronger magnetic moment would mean that you can reduce magnetic field for the same signal/noise ratio. One of the problems with human body is that it is filled with water, which is diamagnetic. That has tendency to distort magnetic field, and without a uniform field, it's very difficult to get a good MRI scan. The stronger the field, the more distorted the field becomes. It's not a huge obstacle in MRI, but it is in fMRI.

On the flip side, and thanks to DaleSpam of reminding me about this, microwaves have trouble penetrating human flesh due to conductivity. So even if you could reduce the RF power to the point where it won't cook the flesh, it won't penetrate very well. That might actually be the bigger problem than everything else mentioned above.

 

[marcusl]

I don't think this answers Vanadium's, and the OP's, question. First of all, fMRI ("functional MRI" or brain function mapping) is just a flavor of MRI, which itself is a simple flavor of NMR. It really doesn't address electrons at all. Second and more importantly, NMR is a far richer phenomenon than EPR. Because of the couplings that spins in one nucleus have to its neighbors, and to the matrix (bone, interstitial fluid, or in the case of a solid, the crystal lattice), the nuclear resonance spectrum is extremely complex. It reflects chemical structure, coupling strengths, the distances to other nuclei, molecular conformation (shape and folding of a protein, e.g.), and the solidity or fluidity of the matrix, among other things. By contrast, the outer electron in an atom (which contributes to EPR) participate in chemical bonds with neighbors but otherwise do not reflect the same diversity seen in NMR. Only atoms with an unpaired electron can resonate, dramatically limiting the cases where EPR exists. These are the main reasons that EPR is a sideshow in the research world while NMR and MRI are ubiquitous.

BTW, EPR stands for electron paramagnetic resonance. It used to go by the name ESR or electron spin resonance.

 

[Mike H]

My understanding of the in vivo EPR imaging that does go on is primarily focused on small animals (mice, in particular) and the occasional mention of (for example) human limbs that are placed in a resonator. Of course, tissue cultures and ex vivo samples are also possible with this sort of setup. They do, however, seem to be devised to work at lower field strengths. Clearly, if your interests involve the role of free radicals in biological systems, EPR imaging offers the ability to directly interrogate those species. (See, for example, here.) Insofar as to reasons for wanting to do EPR imaging, it's a lot like wanting to develop alternatives/complements to ¹H MRI - you could work with something with a great deal less natural background signal (example - ¹⁹F MRI), or with a more biochemically informative nucleus (example - ²³Na MRI).

I'm not sure I'd entirely agree with EPR being a sideshow - it really depends on what one's show is in the first place, after all. It's something of a specialized niche, I will agree, although I get the impression more and more people are becoming intrigued by its benefits. It's been invaluable in inorganic & solid state chemistry, and as I've alluded to earlier in this post, for understanding the nature and role of free radicals in a number of contexts. With the development of pulsed EPR methods, one can obtain longer distance constraints for materials and biological assemblies, typically between 20 to 80 Angstroms. Also, it's a natural choice for examining dynamics at the nanosecond time scale.

 

[K^2]

Second and more importantly, NMR is a far richer phenomenon than EPR. Because of the couplings that spins in one nucleus have to its neighbors, and to the matrix (bone, interstitial fluid, or in the case of a solid, the crystal lattice), the nuclear resonance spectrum is extremely complex.

And yet, very little of it is used in actual MRI. For the most part, what MRI systems are built to measure are water concentrations and water diffusion. A lot of the things you mention, like NMR on a crystal lattice, isn't even possible with MRI. If you've ever seen how solid NMR is done, you know why. If not, look it up.

EPR would be quite sufficient for medical uses, because everything a medical MRIs are built to measure can, in fact, be measured with EPR, and all the same imaging techniques apply to both.

 

[dangus]

I have another question about MRI physics. I was going to make a new thread, but it seemed easier to just post it here since you all seem to know a lot about the topic.

When the perpendicular RF pulse is applied in a MRI machine the net magnetic moment tilts from the Z axis into the X-Y plane. What happens to the individual magnetic moments of all the protons though? Do the individual magnetic moments tip as well?

My previous understanding was that the individual magnetic moments don't change orientation, instead the tipping of the net magnetic moment is caused by the precession of all the protons in phase (which creates the X-Y component) and the elevation of some protons from the low energy state to the high energy state (thus shrinking the net Z component). Is this true?

 

[K^2]

No. If you apply a 90° pulse, individual moments align with XY plane, and precess in that plane with resonant frequency.

 

[marcusl]

And yet, very little of it is used in actual MRI. For the most part, what MRI systems are built to measure are water concentrations and water diffusion.

Not so. Virtually all clinical diagnosis is based on T1- and T2-weighted contrast enhancement, arising from spin-lattice and spin-spin couplings, respectively. To oversimplify and exemplify: T1-weighted images particularly highlight differences in "solidity" between tumor and tissue, meniscus and bone, etc. while T2 weighting emphasizes fluid due to inflammation and is used in spinal imaging because it highlights cerebral spinal fluid so that disks, impingements and bulges are visible.

 

[marcusl]

I have another question about MRI physics. I was going to make a new thread, but it seemed easier to just post it here since you all seem to know a lot about the topic.

When the perpendicular RF pulse is applied in a MRI machine the net magnetic moment tilts from the Z axis into the X-Y plane. What happens to the individual magnetic moments of all the protons though? Do the individual magnetic moments tip as well?

My previous understanding was that the individual magnetic moments don't change orientation, instead the tipping of the net magnetic moment is caused by the precession of all the protons in phase (which creates the X-Y component) and the elevation of some protons from the low energy state to the high energy state (thus shrinking the net Z component). Is this true?

The classical and quantum mechanical pictures of NMR are sufficiently different to cause confusion if you mix and match the concepts. In the classical view, the spins are tipped away from the z axis by the RF field, whereupon they precess. The longer the field is applied, the greater the polar angle from the z axis. A CW field will send the spins from z to the xy plane (90) to -z (180) and back up to 360 or 0, and then it continues on. To tip them into the xy plane, the RF is turned off.

In the quantum picture, the spins are a two-state quantum system (spin up or down) with an energy gap (the Zeeman splitting). Transitions accompany the absorption or emission of a photon of correct energy. The spins in this view are always aligned or anti-aligned with B0.

 

[dangus]

The classical and quantum mechanical pictures of NMR are sufficiently different to cause confusion if you mix and match the concepts. In the classical view, the spins are tipped away from the z axis by the RF field, whereupon they precess. The longer the field is applied, the greater the polar angle from the z axis. A CW field will send the spins from z to the xy plane (90) to -z (180) and back up to 360 or 0, and then it continues on. To tip them into the xy plane, the RF is turned off.

In the quantum picture, the spins are a two-state quantum system (spin up or down) with an energy gap (the Zeeman splitting). Transitions accompany the absorption or emission of a photon of correct energy. The spins in this view are always aligned or anti-aligned with B0.

It would seem then that my previous description of the tipping net magnetic moment was the quantum mechanics description of the system then. Correct?

Assuming my description is correct, I am having trouble rationalizing something. In the quantum mechanics view T1 relaxation occurs as the high energy protons drop back down to their low energy state, this increases the number of low energy protons relative to high energy protons and thus regrows the Z component of the net magnetic moment. T2 relaxation occurs as the protons stop precessing in phase and thus their horizontal components are no longer additive.

In this quantum mechanics description the vertical and horizontal components of the magnetic moment are completely independent. Therefore it would seem that T1 relaxation can have no impact on the signal. Even if we wait 0 time between excitation cycles such that no T1 relaxation occurs, the next RF pulse will still bring all the protons into phase (both the high and low energy protons) and the horizontal component will be the same as if we waited a long time between cycles. You see what I am saying? Does it make sense?

 

[dangus]

Im having another problem with the quantum mechanics view. The classical view says that we can perform a 90° pulse, a 180° pulse a 270° pulse etc with increasingly strong or long RF pulses. But in the quantum mechanics view this doesn't seem feasible. Once the protons are all elevated into the high energy state by a 180° pulse, it is impossible to continue the rotation. All the protons are already elevated to the high energy state, so it doesn't seem like further RF pulsing would do anything.

Furthermore, by the quantum mechanics view, the 180° pulse would also bring all the protons precession into phase and this 180° pulse should have a magnetic moment with a horizontal component, but it doesnt.

 

[K^2]

In the quantum picture, the spins are a two-state quantum system (spin up or down) with an energy gap (the Zeeman splitting). Transitions accompany the absorption or emission of a photon of correct energy. The spins in this view are always aligned or anti-aligned with B0.

No, they are not. Take a spin in m=-1/2 in B₀, apply a 90° pulse. If you now measure <S_x>, you'll get (1/2)Sin(ωt), while <S_z> is 0.

The orientation of spin in quantum picture follows perfectly the classical picture.

Not so. Virtually all clinical diagnosis is based on T1- and T2-weighted contrast enhancement

In other words, you do proton density measurement with and without spin-echo. You're still not measuring any of the actual spin-spin interactions you've been talking about.

Though, yes, I'll grant you, the difference between T1 and T2 does depend on tissue, so you do get different contrasts because of it. Thing is, you also have T1 and T2 in EPR, and you'd get roughly the same difference between contrasts with and without the echo.

 

[dangus]

No, they are not. Take a spin in m=-1/2 in B₀, apply a 90° pulse. If you now measure <S_x>, you'll get (1/2)Sin(ωt), while <S_z> is 0.

The orientation of spin in quantum picture follows perfectly the classical picture.

I was also under the impression that in the quantum view the spins are only either aligned or antialigned. I obviously don't know enough to argue it one way or the other though, but I am curious to hear what marcusi says. I will share a link though, which is one of the sources I got my information about MRI from and it seems to imply that the spins are only aligned or antialigned. That being said this source could very easily be wrong.

http://en.wikibooks.org/wiki/Basic_Physics_of_Nuclear_Medicine/MRI_&_Nuclear_Medicine#Nuclear_Spin

 

[K^2]

I was also under the impression that in the quantum view the spins are only either aligned or antialigned.

Not at all. This only happens after a measurement. If I measure the energy of a spin in the magnetic field, I'll get results corresponding to either +1/2 or -1/2, with spin state collapsing to one or the other. But if the magnetization was only aligned or anti-aligned with magnetic field, you'd never get transverse magnetic field, which is what you actually measure in resonance experiment.

 

[marcusl]

Not so. Virtually all clinical diagnosis is based on T1- and T2-weighted contrast enhancement, arising from spin-lattice and spin-spin couplings, respectively.

In other words, you do proton density measurement with and without spin-echo. You're still not measuring any of the actual spin-spin interactions you've been talking about.

I don't follow you here. Both T1- and T2-weighted images are typically taken with spin echoes. The interplay of the echo time TE and the repetition time TR determine the contrast.

Though, yes, I'll grant you, the difference between T1 and T2 does depend on tissue, so you do get different contrasts because of it.

Thank you, I'll take your grant--especially since it lies behind most of the clinical utility of MRI.

Thing is, you also have T1 and T2 in EPR, and you'd get roughly the same difference between contrasts with and without the echo.

Same comment as above regarding spin echo. You'd have to look at something special like free radicals to get an EPR signal in the body as Mike H mentioned, giving a very differently slanted image. I've heard of trying to image free radicals created from radiation exposure in cancer therapy using EPR at very low field strength so that RF penetrates. It's not clear to me, though, why you'd get the same tissue contrast in EPR as in MRI.

 

[marcusl]

I'm not sure I'd entirely agree with EPR being a sideshow - it really depends on what one's show is in the first place, after all. It's something of a specialized niche, I will agree, although I get the impression more and more people are becoming intrigued by its benefits. It's been invaluable in inorganic & solid state chemistry, and as I've alluded to earlier in this post, for understanding the nature and role of free radicals in a number of contexts. With the development of pulsed EPR methods, one can obtain longer distance constraints for materials and biological assemblies, typically between 20 to 80 Angstroms. Also, it's a natural choice for examining dynamics at the nanosecond time scale.

Sorry for my choice of words. I was thinking of body imaging, and I think most would agree that EPRI is in a preliminary research stage that is very small compared to MRI research and clinical practice.

 

[K^2]

I don't follow you here. Both T1- and T2-weighted images are typically taken with spin echoes. The interplay of the echo time TE and the repetition time TR determine the contrast.

You can measure the T2 with a single experiment, without any echo at all. Simply apply 90° pulse, and measure resonance. But yes, if you need to improve signal-to-noise, you can get the same result with an echo and long TR. T1, on the other hand, you can't measure without spin echo at all, because T2 is much shorter.

It's not clear to me, though, why you'd get the same tissue contrast in EPR as in MRI.

I would expect similar dependence of T1 and T2 on tissue type. I'd have to think about it to make sure I can construct a convincing argument. A lot of the same factors play role, though. Diffusion, for example, which contributes to T2 contrast of fluids, should be having similar impact on EPR T2. I'm not as certain about T1.

P.S. I have no experience with medical MRI, and I'm sure it shows. Everything I've done was with NMR spectrometers and some micro-imaging. Though, we did have a sample of a mouse spinal column that fit in the bore of the spectrometer, so the concepts aren't completely alien.

 

[dangus]

I apologize for the 10 million questions. Hopefully this will be my last. Can someone explain to me exactly how the 180° pulse used to convert a T2* -> T2 signal works. I understand the concept I think.

The analogy of two cars driving at two different constant speeds is often used. If they both start from the same spot and drive for x seconds, then stop, turn around, and drive for x seconds in the opposite direction, they will both arrive back at the start at the same time. They are now "in phase" again.

This makes sense to me. The problem though, is that I don't see how the 180 pulse actually accomplishes this. The protons get out of phase because each experiences a slightly different magnetic field and thus has a slightly different precession frequency. The 180 pulse though doesn't actually reverse the direction of spins, so I don't see how it will bring everything back into phase.

 

[marcusl]

Actually the 180 does reverse the direction of spin (that's what a 180 rotation does, after all), rephasing the spins into an echo. The video cartoon in Wikipedia's Spin Echo article is one of many website diagrams that shows how it works.

 

[K^2]

I think his confusion stems from the fact that precession direction doesn't change, unlike in the car analogy. But the thing is, if you flip the track, along with the cars, and have cars keep driving in the same direction, it's the same thing as if the cars turned around. That's a bit closer to what the 180° pulse does.

 

[marcusl]

Hmm, even that's not quite the best picture. Remember that the slowest car (spin) must end up closest to the finish line after the flip...
I think the wiki cartoon shows it clearly.

 

[dangus]

I think his confusion stems from the fact that precession direction doesn't change, unlike in the car analogy. But the thing is, if you flip the track, along with the cars, and have cars keep driving in the same direction, it's the same thing as if the cars turned around. That's a bit closer to what the 180° pulse does.

Correct. That was my confusion. After drawing some pictures and staring at it for a bit I see it now. The analogy of flipping the track around was actually quite helpful.

Thank you

 

[K^2]

I talked to an actual NMR theorist about all this. He set me straight on some off the issues. While my understanding of T2 contrast wasn't far off, the thing I overlooked with EPR is the frequencies. Basically, molecular movement that's relevant to T2 in NMR is irrelevant in EPR. So EPR T2 contrast is going to be effectively useless.

T1 contrast in EPR will give you some information about the tissues, and it might or might not be useful, but it would be different from information you'd get from T1 using NMR. Again, different frequencies mean that lattice relaxation happens due to completely different vibrational modes.

So marcusl was absolutely correct. As far as medical imaging goes, EPR wouldn't be able to replace NMR.

 

[marcusl]

Thanks for the new information, K^2. It's nice to know the reasons behind the differences.