I’ve been reading the wiki on MRI and am getting confused. It says it works by reading the location of hydrogen atoms in order to show the shape and structure of human innards. Hydrogen is a good element to choose I guess since we’re primarily made of DHMO. But can the MRI be used to read the location of different elements or even detect specific chemical compounds? It seems to me that if you set the imager to detect lead, one could determine if a person has lead poisoning just be running them through the machine and seeing how much lead they have and where it is.
MRI is based on Nuclear Magnetic Resonance (NMR):
http://en.wikipedia.org/wiki/NMR
NMR is indeed used to determine the chemical structure of compounds by chemists - this wiki article gives a brief explanation of how it works:
Because NMR is based on the way the nucleus of an atom responds to an external magnetic field it will only work on nucleii with an odd number of protons (as otherwise the protons “line up” in such a way that they cancel each other out and the nucleus doesn’t respond to the magnetic field). So although you are correct that other elements can be used (and carbon-13 and phosphorus are used, just not as commonly as hydrogen), you couldn’t use NMR to look for lead as it has an even number of protons.
Even if you were looking for something with an odd number, I’m not sure whether the technique would be sensitive enough when dealing with an entire human body (rather than a chemical in a solution). For example normal levels of lead in an adult are <0.5 umol/L and even very toxic levels are very rarely above say 3 umol/L - contrast that with hydrogen being 111 mol/L in pure water and not that much less in the human body ~100,000,000 more concentrated.
It’s actually fairly simple to detect lead poisoning based on blood tests - atomic absorption spectrometry or mass spectrometry of a blood sample can detect lead levels, as well as other trace elements if required - copper, zinc and selenium (mostly used to detect deficiency) and cadmium, mercury and aluminium (in cases of suspected poisoning) are all relatively easy to test for.
Hydrogen is in a lot more of the compounds in our body than just water. Every protein, fat, and carbohydrate you have is coated with a layer of hydrogen. It’s everywhere!
You could theoretically test for [sup]207[/sup]Pb, but MRI/NMR is not very sensitive. The signals are very weak. A better test for lead poisoning is with a mass spectrometer. In fact, you can very quickly test for all metals in a matter of seconds with the right instrument. An MRI/NMR is not suitable for detecting low concentrations of elements.
You can do some impressive things with NMR, but doing the same things with an MRI is not likely. You cannot use an MRI to determine a chemical structure. There would be way too much noise from other compounds. You can get certain types of protons (hydrogen atoms) to “light up”. There may be some neat tricks you could do with an MRI, but I don’t know how many things translate from NMR to MRI.
Beg pardon? I thought that MRI was just the same thing as NMR, but with the scary N-word removed from the acronym. Care to elaborate?
Yes. The biggest difference between NMR and MRI is the complexity of the subject. Looking at a single protien in solution by NMR is extremely complicated, but there are various tricks to decoding the structure. For example, you can pick a specific carbon in the [sup]13[/sup]C NMR spectrum, then irradiate it in such a way as to get all of the protons bonded to it make a signal. Also, you could get all of the protons near that carbon spacialy (not through bonds but through space) give a signal. Given a protien with a few hundered amino acids you can eventually work out the structure.
With a human body, there is no way you are going to be able to direct a single carbon in a specific protien to do this. It is just too complicated. Also, the NMR experiments I’m describing work on the timescale of days and weeks* with extremely advanced equipment. How long do you think a patient can sit still?
ETA: * Literally one of these experiments takes days and weeks. We’re talking years to figure the structure of the protien.
MRI is NMR, with the ability to spatially resolve the signal.
While MRI could in principle be used with any nucleus it is pretty insensitive (having to resolve 3-D info). The only practical nucleus it can use is H in water.
Yea, I forgot that taking a simple [sup]13[/sup]C NMR takes at least ten minutes for a concentrated sample. Still, the human body is chock full of carbon, and unlike with an NMR spectrum, you really aren’t looking for spectral resolution in the MRI. I’d be suprised if it hadn’t been tried in some form.
How stronng to the magnets for MRI’s get in terms of MHz for [sup]1[/sup]H? You could tell me Telsas, but I would have to hunt down the conversion to get something meaningfull. The stronger the magnet, the more sensitive the machine is. I’m guessing they aren’t at the 900 MHz level yet, since MRI’s don’t have their own building.
Another way to look at it: Chemists use NMR like they use absorption spectra; you get something looking like this NMR spectra or this visible light spectra. Arrangements of the peaks tell you about the chemical nature of a compound you’re studying. This is done with very pure, small samples.
MRI is then more analogous to a digital image, where simpler measurements are taken across many points in space. In the body, different types of tissue will respond differently to the magnetic fields, and thus you can see structures.
Admittedly, this is a huge oversimplification.
I think the problem is that the MRI looks at differences in relaxation time not chemical shift. The relaxation time of Hs is certainly influenced by its environment (and added contrast agent). I suspect the relaxation time of the Cs is less strongly influenced by its external environment
I wasn’t aware that MRI looked at relaxation time rather than frequency (I know it’s the same shift, but I don’t like comparing different nuclei by chemical shift.) In that case, [sup]13[/sup]C MRI is next to impossible since I think the relaxation time is incredibly long.
All the MRI procedures that I’ve heard of show the person being put into the hole of a toroid-shaped magnet. Is it necessary to do this, or can the magnetic effects be used “outside” the magnet. In all the grade school demos of magnets, they show the lines of force extending outward from each pole of a bar magnet. Is it theoretically possible to have the person scanned while laying next to a magnet? And if it is, is it simply a matter a efficiency or design that they use a toroidal magnet?
It is criticle that the magnetic feild be as uniform as possible. That is the reason for the tube. Outside the tube, the magnetic feild will not be uniform. I’m guessing everytime they run an MRI they spend a few minutes shiming the magnetic feild to get it right. The computer could be automatically doing it, but you wont get as good a feild that way. It’s possible that the shim isn’t as criticle in an MRI as it is in an NMR. If I want a really good spectrum for a low sensitivity nucleous, I will spend a half an hour just fixing the magnetic feild.
MRIs are built for a very specialized goal. The construction and software does not lend them to being condusive to basic NRM analysis. A M1-Abrams and a Mini are based on the same basic principle, but you wouldn’t use one for the purpose of the other.
You haven’t seen me drive.
in seconds rather than milliseconds for H, so incredibly is relative
Isn’t it always relative? So when your scans take 1000 times as long and your dealing with 1% [sup]13[/sup]C so you have to multiply the number of scans by 100 to get an equivalent S/N ratio, that turns a 15 minute MRI into a week long event I think the patient will agree that it is incredible ;).
I make no claim towards the accuracy of any of the numbers I gave there. I’m just pointing the practical side of several orders of magnitude difference in relaxation. I can get a good [sup]1[/sup]H NMR in 8 scans, but I need 128 for scans just to get a crude [sup]13[/sup]C. If I wanted to let the carbon relax so I can get a quantitative ratio, the experiment would take several hours. Quaternary carbons would be right out.
This isn’t strictly true. It’s more than possible to do nmr on nuclei with an even number of protons, for example 13C, 29Si, 77Se, 129Xe, 183W etc. etc. and that’s only nuclei with spin I=1/2. The issue is the number of nucleons; compounds with even numbers of both protons and neutrons have nuclear spin I=0, they are thus insensitive to nmr, those with even numbers of protons and odd numbers of neutrons or vice versa have I=n/2, half integer nuclear spin and are nmr active. Finally, nuclei with odd numbers of both have I=n, integer nuclear spin. These are relatively rare, but do exist for example; 2D and 14N.