Fat protons have a long T2 relaxation time, less however than that of water protons. Therefore, on T2-weighted spin-echo (SE) images, the signal intensity of fat is lower than that of rich in water proton structures, i.e., nondegenerated disc, muscle, and most disease entities affecting the bone marrow [6]. Fatty marrow signal intensity is similar to or slightly lower than that of subcutaneous fat. Red marrow has an intermediate signal intensity on T2-weighted SE images, which is lower than that of most abnormal bone marrow processes. Differentiation of red and yellow marrow is not as easy as on T1-weighted images.

On fast spin-echo (FSE) sequences, the signal intensity of fat remains relatively high unless fat suppression techniques are employed, rendering these sequences not useful for the study of bone marrow as both normal fat-containing marrow and water-rich lesions will manifest with increased signal intensity. Lesion conspicuity is, therefore, greatly increased on fat-suppressed T2-weighted sequences (Figs. 2.8 and 2.9). On T2-weighted images with fat saturation, foci composed almost entirely of fat cells, such as focal fatty marrow or Modic II degenerative endplate changes, will be of lower signal intensity compared to those which contain red marrow alone or both red and yellow marrow (Figs. 2.10 and 2.11).

Fat suppression may be achieved with application of either a chemically selective fat-saturation technique or an inversion recovery technique [1]. The chemically selective fat-saturation technique consists of application of a radiofrequency pulse which selectively suppresses the signal of fat protons. The smaller the chemical shift between fat and water protons (i.e., the lower the magnetic field applied), the more difficult it is to selectively suppress fat. Field inhomogeneity (presence of metal, interfaces of tissues with pronounced differences in proton density such as air and bone, off isocenter imaging, large field of view) will cause insufficient suppression of fat; increasing the bandwidth or setting the frequency direction parallel to a metallic prosthesis may render the susceptibility artifact less pronounced. On T2-weighted images with chemically selective fat saturation, yellow marrow signal intensity will be lower than that of muscle, red marrow will have approximately the same signal intensity with muscle, and most pathologic lesions tend to be hyperintense to muscle [1].

Short inversion time inversion recovery (STIR) techniques apply a 180° pulse to invert magnetization of all tissues being imaged. When the inverted longitudinal magnetization of fat has reached the null point after a delay time called inversion time (TI), a 90° RF pulse is applied generating signal from all, other than fat, tissues. Optimal TI for fat suppression is 130–170 ms for 1.5 T scanners. Because nonfatty tissues recover their longitudinal magnetization more slowly than fat, the signal-to-noise ratio (SNR) of this sequence is less than that of the chemically selective fat-saturation techniques. Contrast resolution is, however, very good because tissues with long T1 and T2 relaxation times appear very bright [7]. Although with this sequence fat suppression is more homogeneous, signal from tissues with similar relaxation times to fat (blood, paramagnetic contrast, mucoid or proteinaceous material, melanin) may be suppressed.

Many diagnostic dilemmas with MR imaging of the bone marrow arise in the presence of red marrow; red marrow signal intensity may be close to that of abnormal lesions on both T1- and T2-weighted sequences. Chemical-shift (in-phase and opposed-phase) imaging may help establish the presence or absence of red marrow by exploiting the fact that water protons precess slightly faster than fat protons [1, 8].

Inside the MR scanner, spins precess at frequencies determined by the strength of the net magnetic field (B_net). The main component of B_net is the scanner’s static magnetic field (B₀), but the electronic environment in tissues under study may also contribute to B_net by producing local magnetic fields. The difference in precessional frequency conferred by the environment is known as the chemical shift [9]. The chemical shift most relevant to clinical imaging is the one between fat protons and water protons [10]. The additional local field generated by the electron clouds of fat molecules causes spins within adipose tissue to precess slightly slower than spins within water. This small difference in resonant frequency between protons in fat and protons in water is 3.5 ppm (parts per million), and it corresponds to 224 Hz at a magnetic field strength of 1.5 T and to 447 Hz for 3 T units [9]. Because water precesses faster, there will be times after excitation when fat and water spins will be in-phase (i.e., their transverse magnetization vectors will point to the same direction) and times when they will be 180° out-of-phase (i.e., their transverse magnetization vectors will point to opposite directions). At 1.5 T, fat and water protons are in-phase and out-of-phase every 2.25 ms. In other words, they will be out-of-phase 2.25 ms after excitation, in-phase at 4.5 ms, out-of-phase at 6.75 ms, and so forth. On out-of-phase images, in voxels which contain both water and fat, there is a cancellation of signal because their respective transverse magnetizations are in opposed phase. Conversely, on in-phase images, the transverse magnetization vectors of water and fat protons add together to produce a strong signal.

One of the major clinical applications of chemical shift is in bone marrow imaging. Since normal red marrow is composed of almost equal amounts of water and fat (about 40 % each, see Sect. 1.​1), it will suffer significant signal loss on out-of-phase images (Fig. 2.12). This is helpful in distinguishing between red marrow and infiltrated marrow since the latter does not contain fat and will, therefore, not demonstrate signal loss (Figs. 2.13 and 2.14). Caution is warranted in older individuals, in whom, because of abundant, age-related fatty marrow, there may be no obvious signal loss on visual inspection of opposed-phase images (although signal loss is noted when regions of interest (ROIs) are placed and signal intensity measurements are obtained) (Fig. 2.15). Opposed-phase images must always be interpreted in conjunction with other sequences (T1, STIR), since a lesion that does not show signal loss may simply represent a focal area of fatty marrow (Fig. 2.16).