Entering the exam room you're asked if you're carrying anything metal one last time even as you walk through a metal detector. You lie down on the exam table and try to find a comfortable position as the QALO ring wearing technician checks the machine's status and inserts your IV. Headphones go on for ear protection and this time you hope they're working well. Now you're given a panic button and the the table slides you into the magnet's bore. If all goes well you'll be out in thirty to forty minutes and there will be a rich three dimension image of a part of your body that needed a close look. Assuming you aren't a child or claustrophobic it all takes place without anesthesia.
Magnetic Resonance Imaging is relatively new making use of several branches of physics and a computationally easy version of the Fourier Transform mentioned a few posts ago. The basic technique although it gets into quantum mechanics, isn't too difficult to grasp if you think of a classical model.
Remember playing with tops and toy gyroscopes? Start one spinning and carefully place it on the floor. It will rotate around tracing out a cone around the axis of its spin. There's an angular momentum associated with the spin which, acting with the gravitational pull on the top, causes it to precess tracing out the shape of a cone. Don't worry about the details if you haven't encountered the physics, but just keep the image of the spinning toy tracing the cone.
Protons - the tiny positively charged particles in the nucleus of every atom, act like tiny magnets. They have a magnetic moment. It makes them act like the spinning top. Turn on a magnetic field and the magnetic moment will rotate around in that cone-like motion. The property was noticed before quantum mechanics was fully developed and is called spin. Although nothing is physically spinning, the model is useful for this discussion.
Consider the protons in a piece of your body. In the absence of a magnetic field their magnetic moments are scattered about randomly cancelling each other's contributions. You don't have a net magnetism. Turn on a strong magnet and something happens - they start to precess and line up. Quantum mechanics tells us a proton only has two possible orientations - pointing with the magnetic field and pointing against it. The configuration with the north pole of the proton pointing towards the south pole of the field is a lower energy state than the one pointed the other way. Add energy to the system and more magnetic moments will move into the high energy state.
The Earth's magnetic field is a feature probably necessary for life to have evolved. Depending on your location it's strength is between 0.3 and 0.6 gauss. A tesla is a more commonly used unit and is ten thousand gauss. Refrigerator magnets are a few hundred times stronger. The MRI machine you're getting into is usually 1.5 to 3.0 tesla or T, magnets in the accelerator ring at CERN produce 8T and you can levitate a frog (not the kindest thing in the world) in a 16T field. But 3.0T, the standard high end MRI in the US, is a serious and potentially dangerous magnet.1
The magnetic field is arranged to run along the axis of the bore of the magnet - from your head to your feet. The most proton rich element is hydrogen and we have a lot of it in our body - about sixty percent of you, ten percent of your mass, is hydrogen. Hydrogen atoms are simple - just a proton for a nucleus and an electron. Now consider the precession of the atom's magnetic moments. The frequency of precession - how fast that cone shape is traced - is directly proportional to the external magnetic field.2 Double the field and the frequency doubles. At 3T we have a lot of hydrogen protons precessing at 127.74 MHz. Most of the lined up fields in your body are pointing opposite the MRI's field, called Bo, with some pointing with it. Impressive but not very useful yet.
We'll see why in a moment, but the trick is to get the spins to synch up we can create a net field that has a component transverse to Bo. In other words pointing away from the axis that runs down your body. Next the energy of each of the protons is increased so more spins are in the high energy state - namely north pole to the MRI's north field rather than the other way around. Adding the energy into those spins has to be done without boiling you alive. For that it makes sense to think about music.
You've probably played with tuning forks. Imagine two that ring at middle C and another at the A just down from it. Bang one of the A tuning forks and the other A will start ringing. The C fork? ... crickets. Resonance is an efficient way to transfer vibrational energy. The same is true for the protons. Just "ring" them magnetically at the appropriate frequency and you can efficiently transfer enough energy to move some of the magnetic moments. If you do it right they'll all be in phase too ... all of those little vector arrows will be traveling around together. A transverse component to the magnetic field of your protons is starting to appear - order is coming to the system.
(excuse my crude drawing ... this works a lot better on a blackboard or with hands)
At some point half are pointing with Bo and half against. The components of the magnetic moments parallel Bo cancel, leaving only a transverse component rotating around your core at the precession frequency. Now there is a magnetic field pointing out of your body rotating around you at nearly 128 MHz. You're a rotating magnetic beacon. Now we have the resonance part of MRI.
If you move a magnet through a coil of wire an electric current flows through the wire. Receiving coils are placed in the MRI's bore to pick up the signal from your rotating magnetic field. It doesn't mean very much at this point, but a few more tricks and we get the imaging part of MRI.
Turn off the exciting field. The spins in the high energy state lose energy on average and move to the lower energy state and the transverse magnetic field decays. In fact it decays a different rate for different tissues - muscle is different from fat for example.3 The problem is it isn't localized. To remedy that there are an array of gradient coils. Their field is much weaker than Bo - usually a few percent of a Tesla - but they're still strong enough to change the net field along in the magnet changing the resonant frequency a bit. By sweeping the exciting frequency and moving the strength of the gradient coils you make smaller regions of the body transverse while the rest isn't contributing. You "ring" small regions of the body and then carefully listen before moving on to ring other regions. The sort of task a computer can handle.
This richly changing field appears in the from analog signal from the receiving coils. The frequency and signal strength information is transformed into position and intensity information with a Fourier Transform. Additionally the relaxation rates of the signals have a lot of rich information about tissue type - eg .. tumor or regular tissue. The individual chunks that were scanned become an array of volume pixels or voxels and we have our image. It's usually read out in slices along each of three axis of your body.
A couple of issues about the ride. It is classified as minimally invasive as it don't do the damage cutting does. It heats the tissue up a bit, but not enough to worry about. If the resonance part didn't work you'd have to be heated past the point where you boiled, so yay for resonance! It is loud - often painfully so. The gradient coils may only be a few hundredths of a T, but they have to be switched on and off quickly and the motion produces a loud banging noise. Somewhat louder than a jackhammer at very close range. Past the pain limit. Protective headphones vary - on my last ride they worked poorly. Something to ask about if you have a ride. Claustrophobia is a more issue for some. A normal full body bore machine is 60 centimeters in diameter - about two feet. The largest full bores are 70 cm - a bit under 28 inches. Can you live in a very loud tube like that? If not you're anesthetized - not recommended if you can help it. There are quieter and more open machines, but they tend to be have less powerful magnets that produce fuzzier images that may or may not be diagnostic.
Contrast agents, or " dyes", are injected at different parts in the procedure to change the relaxation time in tumors producing images with better contrast. Gadolinium is commonly used. In small doses it's safe although there are some side effects to consider (one is a chilling effect when injected. My experience has been that it's common to make a full series of images without the dye and then inject it by IV and make a second run.
The experience isn't fun, but fantastically better than the alternative. fMRIs are a different beast with their own tricks - perhaps another post. And a bit on the name. Technically it's really a form of nuclear resonance, but one of the first major companies to get involved was GE. They were sensitive to public reaction to the word nuclear and somehow MRI stuck. It's also accurate, but marketing and branding are powerful.
__________
1 The big MRI magnets have to be superconducting ... you can't build a large and strong enough permanent magnet and the heat produced by a non-superconducting magnet would render it useless and terribly expensive. MRIs are a major user of liquid helium .. it's precious stuff and the world supply is an issue.
A MRI magnet is strong enough that stray metal objects can fly into it at extremely high speed - lethal speeds. Oxygen tanks have flown in and killed so extreme caution is taken.
2 The precession frequency is called the Larmor frequency.. it's simple something called the gyromagnetic ratio times the magnetic field. Atoms have individual gyromagnetic ratios. In a 1 T field hydrogen is 42.58 MHz so a 3T magnet has them precessing at just under 128 MHz .. a bit above the FM band on your radio.
3 There's another relaxation time, but I'd spend a few pages talking about it... for now just understand that there are two relaxation rates that are part of the signal when the energizing field is turned down.
__________
un-recipe corner
Many of you are into holiday cooking so I'll refrain from adding to the pile today, but rather offer a video of Merlin's Swan. watchmaker+jeweler at the extreme
and something in the omg category... amber with a dinosaur feather! The linked paper is outside the journal's paywall and is quite readable for sciency types.
Comments