Because of the isotropic nature of acquisition, the volumetric data set can be sectioned nonorthogonally to provide nonaxial two-dimensional planar images referred to as multiplanar reformation (MPR). MPR modes include linear oblique, curved planar, and serial transaxial reformations. Several anatomic structures are not particularly well visualized and represented as displayed in orthogonal planes, and oblique reformatting can be useful in these instances (Fig. 12-6). Oblique images are most often used to transect the mandibular condyle. Curved planar images are generated by manually drawing a planning line or spline by selecting multiple nodes along the centerline corresponding to the jaw arch on an appropriate axial image; this creates a “simulated” or reconstructed dental panoramic image. Because clinicians are familiar with them and they are distortion-free, panoramic MPR reconstructions are useful for jaw evaluation. Such reconstructions must be thick enough to include the entire mandible to avoid missing disease. Serial transaxial images provide sequential images perpendicular to an arbitrary linear oblique or curved planar MPR. They are most commonly called cross-sectional images. The slice width (slice thickness) of the image and the distance between adjacent cross-sectional images (interslice interval) can often be defined. Cross-sectional images are optimal for examining teeth and alveolar bone.

The slice thickness of orthogonal or MPR images can be “thickened” by increasing the number of adjacent voxels included in the display (Fig. 12-7). This creates an image slab that represents a specific volume of the patient, referred to as a ray sum. The thickness of the slab is usually variable and determined by the thickness of the structure to be imaged. Full-thickness perpendicular ray sum images can be used to generate simulated projections, such as lateral cephalometric images (Fig. 12-8). In contrast to conventional radiographs, these ray sum images are without magnification and parallax distortion. However, this technique uses the entire volumetric data set, and interpretation is negatively affected by “anatomic noise”—the superimposition of multiple structures—also inherent in conventional projection radiography.