9.2 Three-Dimensional Imaging

Three-dimensional images combine a series of CT slices into a surface-rendered volume. The most widely used technique is shaded surface display (SSD). Here, threedimensional (3D) volume data are represented in a two-dimensional plane, displaying spatial relationships with visual depth cues. The computer algorithm determines which pixels within the volume data are displayed and how they are spatially related to other pixels in the volume set. In SSD, surfaces are modeled as a number of overlapping polygons, with surface shading added, and a virtual light source is computed for each. More sophisticated programs allow the surface models to be repositioned and manipulated. With surface rendering algorithms, interior structures are not visible (Fig. 9.4).

Volume rendering is a technique by which selected surfaces can be defined by a threshold density and overlying tissues can be made semitransparent. Transparency and colors are used to represent specific volumes. This technique allows 3D reconstructions that allow exceptional evaluation of skull anomalies, fractures, and other bony lesions.

Fig. 9.4 T hree-dimensional reconstructed CT image of a patient with fibrous dysplasia of the left face and orbit

Magnetic resonance imaging offers several advantages over CT for orbital disease [6]. Because of the low resonance signal generated from bone, soft tissue visualization in the region of the orbital apex, optic canal, and cavernous sinus is not degraded by dense surrounding bone as in CT scans [4, 11, 12]. However, because of the low signal generated by bone and foreign bodies, these structures are not well imaged on MRI. Manipulation of resonance signals from various tissues provides contrast variability and a level of tissue differentiation unobtainable with any CT technique. This is particularly useful for neural tissues such as the optic nerve and brain. Surface coil technology, improvements in signal-to-noise ratios, and techniques for suppressing the high-fat signal on T1-weighted (T1-WI) images have greatly improved visualization of many orbital lesions [14–17].

The major component of the MRI system is the magnet that provides the primary polarizing field. Located within the bore of the magnet are gradient coils that provide the spatial localization information during the imaging process. Within the gradient coils are the radio-frequency (RF) antennae (“coils”), which transmit the RF energy to the tissues and receive the returning resonance signals. The use of smaller surface coils placed immediately over the area of interest increases the signal strength and increases the sig- nal-to-noise ratio. These permit acquisition of the highresolution images of modern scanners. However, such coils are limited in the depth of penetration they can image, and they are associated with some artifact.

The generation of a magnetic resonance signal depends on the presence of magnetic isotopes of common elements in biological tissues. The atom most frequently imaged is the ubiquitous hydrogen nucleus, or proton [13]. All protons are normally in a state of axial spin. This spinning charged particle generates a magnetic field, with north and south poles. Under normal conditions, all the nuclei in a given volume of tissue are randomly oriented, but when placed within a strong external magnetic field the individual protons align with the external magnetic direction (Fig. 9.5a). Most of the axes of individual protons lie at various small angles to the external magnetic moment, and they are equally distributed 360° around it. Like spinning tops, these inclined axes wobble, or precess, around the mean magnetic direction (Fig. 9.5a). The rotating axes therefore describe a conical surface with angular momentum determined by the strength of the external magnetic field and by an intrinsic property of the particular type of atomic nucleus. The resultant angular velocity of precession is called the Larmor frequency.

When this system is exposed to an external RF pulse at the Larmor frequency, energy is absorbed by the atomic

nuclei, and the spinning nuclei move into higher energy levels. The angular orientation of their axes with respect to the external magnetic direction increases, and in so doing they tilt away from the magnetic axis and into a plane perpendicular to it (Fig. 9.5b). In addition, the individual atomic axes group to one side of the external magnetic direction. When the RF signal is turned off, the precessing nuclei return to equilibrium by giving up energy to the environment at the specific Larmor frequency. Return to equilibrium occurs by two simultaneous decay, or relaxation, processes, which are detected as resonance signals.

9.3.1 The T1 Constant

During the T1 relaxation, the nuclear axes realign into an orientation parallel to the external magnetic direction as the spinning protons gradually give up their absorbed energy to the environment [18] (Fig. 9.5c, d). The time

Mz

T1 relaxation time

Time

Fig. 9.6 T he T1 relaxation or decay is represented as a timedependent asymptotic curve as energy is given up to the environment

required for completion of this process is an exponential rate called the T1 time (Fig. 9.6). It is influenced by the interaction of the proton with other atoms bound to the

molecular lattice, by temperature, and by viscosity of the tissue. At any specific time following the RF pulse, the total amount of energy given up by the spinning protons depends on the rate at which the T1 relaxation occurs. Tissues with a short T1 constant, such as fat, give up more resonant energy per unit time and therefore appear brighter on the final MR (magnetic resonance) image than tissues with longer T1 constants, such as muscle. This is the basis for contrast intensity, and specific orbital tissues will demonstrate characteristic T1 signal intensities (see Table 9.3).

time, and is influenced by the tiny magnetic fields generated around adjacent spinning nuclei (Fig. 9.8). As with T1 constants, biochemical differences between tissues confer slightly different T2 relaxation times to their protons. At any specific time following the RF pulse, tissues with long T2 constants, such as vitreous, maintain a greater transverse vector component than tissues with short T2 constants, such as muscle. This greater transverse vector produces a higher signal and is therefore brighter on the final MR image.

9.3.2 The T2 Constant

Immediately following the RF pulse, the atomic nuclei are grouped on one side of the mean magnetic axis (Fig. 9.7a). As they rotate, they generate an RF signal as they cut across the external magnetic field and thus generate a small alternating current voltage. During the T2 relaxation, the atomic nuclei redistribute themselves evenly 360° around the external magnetic field direction (Fig. 9.7b). As they do so, the strength of this induced signal decreases because of the increasing canceling vectors. The time for complete decay of this signal (i.e., even distribution of magnetic moments) is the T2, or spin–spin relaxation

9.3.3Creating the MR Image

The signals generated by the T1 relaxation and the T2 decay are measured by RF detectors. They will detect in mass fashion all similar signals at the Larmor frequency, regardless of their specific location within the tissue. Spatial encoding of resonant signals from particular small volumes of tissue is necessary for the creation of a meaningful two-dimensional image. This is achieved by deformation of the external magnetic field using gradient coils, such that the protons in every small volume of examined tissue (voxel) has a unique magnetic field strength and therefore a unique Larmor frequency. The detected Larmor frequency therefore will identify the precise

T2relaxation time

Signal

Time

Fig. 9.8 T he T2 relaxation is represented as a time-dependent asymptotic curve as energy is given up and the signal decays to zero

location of the signal, and a topographic image can be created.

The final MR image is determined by the proton density and by the variations in the T1 and T2 decay constants of specific tissue components. The T1 and T2 resonance signals can be manipulated by application of various pulsed sequences, thus altering the way the signals are collected. The MR image can therefore be weighted in favor of the T1 or the T2 information (Fig. 9.9a, b). In a T1 image, the vitreous is imaged as a dark hypointense signal compared to fat, which shows a bright hyperintense signal. On a T2 scan, the vitreous is typically bright, and the fat is dark (Fig. 9.10a, b). Pathologic lesions in the orbit often show distinctive T1 and T2 imaging characteristics that can help distinguish them from other lesions (Table 9.4) [31].

Gadolinium is a rare earth element with paramagnetic properties. In the presence of an external magnetic

a

b

Fig. 9.9 (a) Coronal T1 MRI image of a patient with a lymphoma of the medial right orbit; the lesion is isointense to normal muscle. (b) Axial T2 MRI scan of a different patient showing a lateral orbital lymphoma that is homogeneous and slightly hyperintense to muscle