So, how does that MRI machine work? There are plenty of books out there that will enlighten you on theory behind MR image. It takes tremendous amount of time to obtain mastery of all concepts related to MRI. People spend their entire careers learning theories and creating new ones. In addition to basic concepts, each MRI scanner is a little bit different in itself, and each new sequence is the result of many peoples’ work. Fortunately, most of the MRI sequences are based on basic acquisitions such as spin echo or gradient echo techniques. Below, you will find an attempt to explain physics of MRI in the simplest possible terms. If you have ever taken college-level physics and chemistry courses, you should be able to follow along.
One fundamental principle required to understand MR imaging is the concept of electromagnetic energy. Radio receiver in your car accepts radiofrequency (RF) waves, which represent one form of electromagnetic energy. For now, think of MR coils as “antennas” that transmit and receive radiowaves.
How do we get organic tissue to receive and transmit RF energy? Protons (H+) in living tissues are able to accept electromagnetic energy in the form of RF pulse. This RF pulse “excites” the protons to a higher energy level. As the protons “relax”, energy is emitted back to the antennas in MR machines.
This concept would work at a single proton level, but the amount of signal transmitted back would be too small to be detected. In real life, each proton is positioned close to numerous other protonss and the signals coming from each individual proton would very quickly cancel each other out. Let’s dig a little deeper: each proton in the body acts like a small magnet with its own magnetic field. While each proton can receive and emit back the RF energy, any sort of coherent signal generated by such experiment is quickly cancelled out by neighboring protons. One analogy is a room of people talking at the same time. While each individual person is transmitting meaningful information, the sound of entire room makes absolutely no sense.
So, the first trick is to get the protons to talk in unison with each other. When placed into external magnetic field, protons align parallel to it. There is almost an equal chance that a proton orients itself at 0 or at 180 degrees to the main magnetic field. The keyword here is “almost,” because fortunately for us, a few more protons align with the magnetic field rather than 180 degrees to it. Explanation for this phenomenon requires understanding of quantum physics, so let’s not go there for this brief introduction. These extra few protons (actually quite a lot of them) create a net magnetic field that we can now manipulate. The number of protons aligned with the magnetic field varies with strength of the magnet. Utilization of a 3T magnet results in much better signal-to-noise ratio compared to a 0.5T magnet.
Next task is to get the protons to emit enough energy for the MR coils (antennas) to pick it up. Unlike static pictures on this page, each proton actually rotates around its axis (precesses) like a wobble. Turns out that the signal generated by each rotating proton is much stronger if we make protons precess in unison with each other at 90 degrees to the main magnetic field. RF pulse from transmit MR coil does just that.
The signal generated by protons precessing in unison and at 90 degrees to main magnetic field is the strongest signal we can get. However, 90 degree rotating frame is a high energy state, a state in which protons will not stay forever. There are two types of relaxation. First, there is relaxation to align back with the main magnetic field (T1 relaxation), Second, there is de-phasing in the transverse plane (90 degree plane). Each individual proton precesses at slightly different speed. After a while, the signal from protons in transverse plane degenerates as protons start precessing out of phase with each other (T2 relaxation).
It takes different amount of time to loose strength of the signal from T1 and T2 relaxations. By manipulating manner and frequency in which we apply RF pulses (TR), and by changing time to start of signal acquisition after RF has been applied (TE), we are able to produce T1-weighted or T2-weighted images.
We can also manipulate the strength of the magnetic field at various locations by applying additional gradient magnetic fields. Variations in axis and timing of these additional gradient magnetic fields results in predictable variations in frequency and phase of signal coming from different locations. The RF signal that is being emitted from tissues is a complex wave that contains contributions from numerous frequencies, phases and amplitudes.
It is helpful to have an idea of what Fourier transform does at this point. An oversimplified explanation would be the following: Fourier transform takes complex signal that contains infinite number of possible sinusoids (or circles with various frequences, phases, and amplitudes) and breaks it down into individual components.
A myriad of possible variations in the RF pulses and in magnetic fields results in a complex matrix of signals coming back from the tissues to the RF receiver coils. Using sophisticated mathematical manipulations, these signals are converted into variety of images. Each particular set of images contrasts specific property of the tissues to our advantage.
We are not quite at the point where MRI can substitute for direct histological tissue evaluation. Slowly, but surely the resolution of the image is getting improved, however. Anatomical and functional information that is transformed into images is becoming crisper by the day, if not by the hour.
This overly abbreviated introduction most likely leaves you with more questions than answers. A simple internet search for other sites dealing with MRI education will produce a number of links to very good websites and videos put together by physicists and others.