When a patient is placed in a strong magnetic field such as is used for MR imaging, a tiny amount of magnetization from the protons in his or her body is created. The exquisite sensitivity of the MR imaging process allows us not only to perceive this magnetization, but also to manipulate it in ways that give us information about the tissues themselves. A "net magnetization" results from the added microscopic magnetization vectors from all the tissue protons within the sensitive volume of the receiving RF coil. (There is also magnetization in the body outside the sensitive volume of the RF coil, but because this is not received as part of the MR signal, it is generally ignored). This net magnetization is aligned by default (i.e. in its "relaxed" state), with the direction of the polarizing, or main, magnetic field, B₀.

All MR imaging techniques require an initial "excitation" of this net magnetization, i.e., the addition of energy to protons in the body in a way that tips the net magnetization away from the main field toward the transverse plane. This magnetization has a tendency to realign with the main field after a time, or "relax" back to its longitudinal starting state. The excitation of this magnetization away from the longitudinal axis is accomplished by using an RF pulse tuned to the resonant frequency of the protons. This resonant frequency is proportional to the polarizing field strength; at 1.5 Tesla, the resonant frequency is approximately 63 MHz. T2 (and T2*) effects relate to the behavior of this magnetization when it is in the transverse plane, and T1 describes the rate at which it relaxes back to its low-energy state, i.e. aligned with the main magnetic field.

In a simple pulse sequence, the net magnetization is tipped 90° into the transverse plane, where it precesses (i.e. rotates) at its resonant frequency around the direction of B⁰. This precessing magnetization will create an electric current, i, in a properly positioned RF coil that is tuned to the protons' resonant (or "precession") frequency. (See Fig.1.12) The resulting electric current is an MR signal that can be collected and processed to form an image.

Figure 1.12 The precessing transverse component of tissue magnetization, M(t), creates a current, i(t), in the RF coil that is detected as the MR signal.

Figure 1.12 shows the behavior of the net magnetization in time. The vector that corresponds to the net magnetization is precessing in the transverse plane at a rate determined by its resonant frequency. As the magnetization vector continually rotates through the plane of the RF coil, it interacts with the electrons in the coil, and causes them to slosh alternately backward and forward. This is the cause of the current which goes from positive to negative at the precession frequency of the magnetization, and is detected as the MR signal*. The degree to which the precessing magnetization is able to push the electrons backward and forward in the RF coil and create this alternating current is referred to as the strength of "coupling" between the tissue and the coil. Ideally, this coupling is as large as possible so that the maximum current is produced in the coil for the amount of magnetization that is created.

Figure 1.12 and the depiction of the net magnetization as a single vector require some explanation. The net magnetization that is depicted in this figure is actually a vector sum of all the magnetization within the tissue to which the RF coil is sensitive. Magnetization at different locations within the tissue can behave differently, depending on the particular local magnetic environment of the protons at these locations. The precession frequency of a proton is determined by its specific local magnetic environment, so is usually slightly different from the idealized resonant frequency corresponding to the main field strength. It is actually this variation in precession frequencies for local magnetic environments that creates contrast between different tissues, and is also the foundation of the imaging process itself. Often, physicists will speak of "ensembles of spins". This is a notation that refers to a group of protons that has the same resonant frequency. Throughout this discussion, the magnetization corresponding to an ensemble of spins will be referred to as a magnetization component. The magnetization components can be thought of as smaller magnetization vectors, each of which corresponds to a group of protons with a common precession frequency (these add to give the detected net (or "total") magnetization.

The "driving" effect of the RF pulse causes the magnetization components to precess at the same frequency with the same phase during the application of the pulse.^† The net magnetization from these protons is a maximum immediately after the RF pulse is applied. After the RF pulse is ended, each of the individual components begins to revert to its own individual resonant frequency. The coherence between these individual components decreases as time goes on (i.e. they fall farther out of step with each other), and the net magnetization decreases. The effect on the received MR signal is a decay of its amplitude over time. The time for the MR signal to decay to 37% of its initial value (i.e. the time constant for the decay) is referred to as T2*.

In an imaging pulse sequence, imaging gradients are used to tag the magnetization in different voxels with specific resonant frequencies. These frequency signatures are used to map the MR signal back to its correct anatomical location in the reconstructed image. The imaging process can be thought of as a coarse compartmentalization of the magnetization into discrete components corresponding to individual voxels. However, within any tissue voxel, there exists a secondary spread of resonant frequencies. This spread of resonant frequencies results from differences in the magnetic fields in each of the individual protons' environments. The magnetic field at the location of an individual proton is the result of adding up the individual magnetic field contributions from many different sources and is not uniform across the entire voxel. Changes in any of the individual field components from location to location within a voxel results in a spread of resonant frequencies within the same voxel.

The sources of magnetic field at the level of individual protons are many; however, a first approximation can be arrived at by considering only the most important contributions. First, any main magnetic field imperfections contribute to the loss of coherence of the transverse magnetization. Second, even in a perfectly created scanner, the main field intensity would never be completely uniform on the sub-voxel level in a clinical imaging situation. Regions of slightly different main field strength exist throughout each voxel due to differences in tissue magnetic susceptibility. Magnetic susceptibility is a parameter that describes the extent to which a substance will magnetize when placed in a uniform magnetic field. At boundaries between tissues of different susceptibilities, magnetic field gradients are created. These gradients cause protons within the same voxel to precess at different frequencies, and therefore contribute to a faster T2* decay of the MR signal. The presence of metal, or an interface with air also creates spatial variations of field strength due to the extreme difference of magnetic susceptibility with tissue. Third, protons on different molecules have different resonant frequencies due to their "chemical shifts",( i.e. different local magnetic fields created by different configurations of neighboring electrons within the molecules themselves). All of these three contributors to the T2* decay are static in time (at least for the length of time between the excitation pulse and the acquisition window), and are referred to specifically as T2' effects.

In addition to these static T2' effects, there exist other contributing factors to the observed T2* decay that arise from more ephemeral interactions. Within each of the macroscopic magnetic environments that make up a voxel, there exists another layer of magnetic interactions that causes the many millions of protons that may exist there to fall out of sync with each other. These interactions are known as T2 effects, and occur on the molecular and submolecular level. Each proton feels the magnetic fields that are created by neighboring protons and other magnetic nuclei such as Gadolinium (Gd³⁺). The individual resonant frequency of a single proton at any particular moment depends on that proton's immediate magnetic environment. This causes the individual protons to precess at different frequencies and to lose step (phase coherence) with their neighbors, even though their macroscopic magnetic environment may be the same. Because the protons are constantly in motion, the immediate magnetic environment of each proton changes throughout the excitation-to-acquisition time. Generally speaking, the more mobile the protons are in a tissue, the more similar their individual precession frequencies will be to an average frequency. Tissues with mobile protons will therefore have longer T2s (i.e. slower T2 decay rates) compared to tissues with bound or restricted protons (assuming that there is no external contrast agent like Gd3+ present). Synovial fluid, edema, and necrotic cores of tumors are examples of tissue components with long T2. Tissues with Type I collagen such as fibrocartilage (meniscus, labrum and tendons) tend to have shorter T2s.

---

* Actually, the MR signal is detected as a voltage, rather than a current, but these quantities are related simply by the electrical resistance of the coil.

^† An analogy for frequency and phase is the lock-steip march of soldiers: not only do they all march with the same number of steps per minute (frequency), but they also all raise their right feet at exactly the same time (phase).