Radiographs

The usefulness of conventional radiographs is primarily in the extradural space. Conventional radiographs are relatively inexpensive and easy to perform. They provide a convenient means of assessing alignment and gross bony integrity and can also be used for purposes of localization and evaluation of movement. They are capable of demonstrating the general changes involving various types of arthritis and disk space narrowing. According to recent appropriateness criteria, radiographs are considered sufficient for the initial evaluation of recent significant trauma, osteoporosis, or back pain in individuals older than 70 years.¹ Substitution of digital radiography for conventional film-screen techniques has provided more flexible handling of images (e.g., duplication, transmission, postprocessing), as well as expedited cycle time. This latter advantage is extremely useful in the intraoperative environment because the results are available immediately after exposure without the need for traditional processing.² The disadvantages of conventional radiographic techniques are that they are able to visualize only bony structures and are relatively insensitive to soft tissue changes within the disk, ligaments, nerves, and paraspinal tissues. Bone marrow involvement must be significant and far advanced before it is demonstrable on conventional radiographs.

Isotope Bone Scans

Isotope bone scans are most useful for evaluation of the extradural space. They are a moderately sensitive test for detecting the presence of tumor, infection, or occult fractures of the vertebrae but are rarely specific for diagnosis. The yield is very low in patients with normal radiographs and laboratory studies and highest for those with known malignancy. These scans are contraindicated in pregnancy. They are an important tool for surveying the entire skeleton for metastatic disease. To further evaluate sites positive on isotope bone scans, single-photon emission computed tomography (SPECT) and positron emission tomography (PET) are often used in conjunction with computed tomography (CT) and are capable of improving specificity and anatomic localization, particularly in patients with malignancies.

Computed Tomography

CT without contrast enhancement is most useful for evaluation of the extradural space. Depending on the window and center setting, CT can discriminate among various tissues by differences in attenuation. CT is readily available and can provide high-resolution bony detail and enough soft tissue information to accurately assess degenerative disk disease and associated changes or sequelae such as herniation, stenosis, and facet arthritis. CT is equivalent to magnetic resonance imaging (MRI) in the diagnosis of a lumbar herniated disk and stenosis but less accurate in depicting marrow changes and soft tissue structures such as paravertebral lesions and intraspinal structures or pathology. The primary disadvantages of CT are the overlap in attenuation with normal and abnormal soft tissue pathology and, as with conventional radiography, ionizing radiation. Recent technical advances in CT have led to the production of multidetector devices that can provide larger areas of coverage in shorter examination times. In addition, thinner sections can be used to allow the acquisition of isotropic three-dimensional (3D) data sets, which can be reformatted in any plane without loss of resolution. Subsequent 3D multiplanar reformatted images are particularly useful for the evaluation of bony trauma and assessment of fusion postoperatively, pseudoarthrosis, and instrumentation. CT remains an important tool and is the first alternative when MRI is contraindicated or unavailable.

CT Myelography

The introduction of contrast material into the intrathecal space provides both extradural and intradural information. Myelography is able to outline the nerve roots, spinal cord, and its coverings. Intradural and extradural disease can be identified and characterized on the basis of the character and morphology of its effect on the contrast column (Fig. 18-1). Plain-film contrast-enhanced myelography is rarely performed on its own and is almost always used in conjunction with CT. Nonionic contrast material has completely replaced its oil-based predecessor and is usually introduced into the lumbar region at L3 or L4 or cervically at C1-2. Plain-film conventional myelography may be useful for general anatomic appraisal in patients with complex deformities or when implanted hardware or electronic devices preclude the use of MRI or CT. CT myelography provides excellent delineation of intradural structures such as the spinal cord and nerve roots and their relationship to the soft tissue and bony surroundings. It is useful in distinguishing bone from soft tissue changes (e.g., osteophytes and disk herniations), which may be inseparable on plain-film myelography (Fig. 18-2). Multidetector CT of the entire spinal axis can be performed in 20 to 30 seconds. With intrathecal contrast media, subsequent multiplanar reformatted images can provide complex 3D assessment of anatomic and pathologic abnormalities with myelographic-like images. More recently, various forms of dual-energy multidetector CT have been introduced that can subtract out bone in both single slices and 3D multiplanar reformatted images (Fig. 18-3).

FIGURE 18-1 A, Diagrammatic representation of extradural (ED), intradural extramedullary (ED-EM), and intramedullary (IM) lesions of the spine. Purple, spinal cord; orange, mass; tan, subarachnoid space. B, Oblique view from a cervical myelogram demonstrating an extradural defect (arrow) from a herniated disk at the C6-7 level with cutoff of the nerve root. The diagram demonstrates the typical concave inward appearance of an extradural defect. C, Anteroposterior (AP) radiographs from a thoracic myelogram demonstrating an intradural extramedullary lesion. Contrast agent was introduced into the subarachnoid material below the lesion on the left and at C1-2 to outline the superior extent of the lesion on the right. One can see displacement of the cord silhouette to the left. D, AP views from a thoracic myelogram demonstrating an intramedullary mass at the conus. The contrast column in the subarachnoid space is splayed around the intramedullary mass. CT myelography demonstrates the enlarged conus.

FIGURE 18-2 Coronal and sagittal multiplanar reformatted CT myelographic images obtained from the axial data set. Hypertrophic degenerative changes of the vertebral body margins and uncinate processes are producing extradural indentations on the subarachnoid space.

FIGURE 18-3 Dual-energy subtraction CT myelography. Primarily acquired axial and multiplanar reformatted coronal images through the lumbar region from a CT myelographic study were obtained with a CT scanner that uses two x-ray tubes. Different exposure factors using 80 and 140 kV produce data sets on which subtraction can be performed on the basis of attenuation differences related to the exposure factors.

The disadvantages of myelography/CT myelography relate primarily to the introduction of contrast media into the subarachnoid space, use of ionizing radiation, and limited soft tissue discrimination. Myelography is an invasive procedure associated with postprocedural headaches (10% to 15%) and rare but possible complications such as nerve damage, formation of an iatrogenic epidermoid, infection, and arachnoiditis. The incidence of headaches severe enough to require treatment (e.g., epidural blood patch) is approximately 1%.

Magnetic Resonance Imaging

MRI has become the initial imaging technique for most patients with significant signs and symptoms of spinal disorders. It provides excellent contrast among different types of soft tissue. MRI without exogenous contrast material is capable of evaluating the extradural, intradural extramedullary, and intramedullary spaces. MRI is noninvasive and without the risks associated with ionizing radiation. Anatomic imaging can be performed in different planes, as well as used for volume acquisitions. It is the most sensitive modality and most specific technique for identifying abnormalities of the spinal cord, nerve roots, cerebrospinal fluid (CSF) space, soft tissues, and bone. When used with gadolinium complexed with diethylenetriaminepentaacetic acid (DTPA), MRI has even greater sensitivity for intramedullary disease, inflammatory changes, and reparative processes such as may be seen with postoperative changes and trauma. MRI is particularly efficacious for the evaluation of so-called red flag diagnoses such as osteomyelitis, neoplasia, and trauma.

Traditional clinical imaging has emphasized orthogonal T1- and T2-weighted imaging for morphologic assessment of the diskovertebral complex. These sequences also provide an evaluation of the changes in signal intensity associated with degenerative disk disease. Fast spin echo (SE) T2-weighted images have replaced conventional T2-weighted images because of their shorter acquisition times, but they provide no increased diagnostic advantage. Short tau inversion recovery (STIR) or fat-suppressed T2-weighted images have been added because they are more sensitive to marrow and soft tissue changes (Fig. 18-4). Although these standard sequences remain the mainstay of diagnostic MRI of the spine, new techniques continue to be evaluated in the hope of providing stronger correlation between imaging findings and patient symptoms. The utility of many of these techniques for the routine evaluation of degenerative disk disease remains unknown, and the number of patients in whom they have been evaluated remains small. Nevertheless, these approaches may be important for redefining the direction of spinal imaging from strictly anatomic imaging to imaging that combines more physiologic and functional information.³ Techniques that have been evaluated include assessment of spinal motion (dynamic imaging, kinetic assessment, or axial loading), diffusion imaging (water or contrast agents), MR neurography, spectroscopy, functional MRI of the spinal cord, and ultrashort–echo time imaging. Of the variety of techniques available, only MR neurography and dynamic imaging have expanded beyond the experimental phase to demonstrate specific clinical utility (albeit in niche populations).

FIGURE 18-4 MRI sequences. Sagittal T1-weighted spin echo, T2-weighted spin echo, and short tau inversion recovery (STIR) images through the lumbar spine demonstrate the different contrast obtained with different techniques. On the T1-weighted image, the vertebral body marrow has higher signal intensity than the intervertebral disk, primarily because of the fat content of the marrow space. On the T2-weighted spin echo image, the intervertebral disk has higher signal intensity than the vertebral body because of prolonged relaxation time related to both bound and unbound water species. This change in signal is probably a reflection of the health of the proteoglycans more than the total water content. The STIR sequence demonstrates decreased signal intensity of the marrow space and is the sequence that is most sensitive to marrow involvement. Note the degenerative changes at L4-5 and L5-S1. The decreased signal intensity of a degenerative disk is better noted on the T2-weighted and STIR sequences. Note the loss of the internuclear cleft as the degree of degeneration increases, as at the L5-S1 level.

Dynamic Imaging

The utility of dynamic spine MRI remains unclear, in part because of the varied methods used. Methods used to date include axial loading in the supine position by means of a harness attached to a nonmagnetic compression footplate with nylon straps that can be tightened or the use of an upright open MRI system that allows flexion-extension imaging (Fig. 18-5).^4,5 Dynamic MRI has been used to evaluate for the presence of occult herniations, which may not be visible or be less visible when the patient is supine, to measure motion between spinal segments, and to measure canal or foraminal diameter when subjected to axial loading.^3,5–8 Hiwatashi and colleagues evaluated 200 patients with clinical symptoms of spinal stenosis and found 20 with detectable differences in caliber of the dural sac on routine and axial-loaded studies.⁷ In 5 of these selected patients, all three neurosurgeons involved in the clinical evaluation changed their treatment recommendation from conservative management to decompressive surgery. Although a subset of patients may benefit from this type of evaluation, the benefit appears to be small for the added machine time and patient discomfort.

FIGURE 18-5 Sagittal T2-weighted spin echo images through the cervical spine in flexion (A) and extension (B). Note the decreased anterior-posterior canal diameter on the extension image (7 versus 4 mm) at C3-4.

Cerebrospinal Fluid

CSF flow studies with cine MRI provide cardiac-gated gradient phase-contrast images that are displayed qualitatively as CSF flow images in a closed loop kinematic format. This same data set can be displayed quantitatively on a graph or in numerical format as analysis of values of flow velocity and volume flow rate. These images demonstrate the pulsatile motion of CSF, not the bulk flow. Pulsatile flow is a result of expansion of the brain during systole and relaxation during diastole and is therefore bidirectional, with CSF flowing caudally during systole and cranially during diastole. Typically, the signal intensity of normal CSF flow in these studies is demonstrated as hyperintense during systole, where there is caudal (downward) flow, and hypointense during diastole, where there is cranial (upward) flow. These images are usually displayed in the sagittal and axial planes. In general, the degree of pulsatile flow diminishes as it proceeds caudally. Clinically, these studies are most often used for the evaluation of CSF flow in patients with Chiari I and II malformations and for the assessment of syrinx cavities (Fig. 18-6).

FIGURE 18-6 A-D, Sagittal T1-weighted, T2-weighted, and cerebrospinal fluid (CSF) flow studies of the craniovertebral junction in a patient with a Chiari I malformation. Note that the majority of CSF flow is within the basal cisterns and ventral to the cervical cord. Only a small amount of CSF flow is identified dorsally (arrow).

Artifacts and Contraindications

Instrumentation, metal used as part of surgical procedures, and implanted electronic devices represent challenges for both CT and MRI. In the case of CT, metal can create beam-hardening artifacts that obscure adjacent soft tissue and osseous structures (Fig. 18-7). Multidetector studies with isotropic voxels and software used to decrease beam-hardening artifact can often decrease the image degradation caused by implanted metal.²⁸ Different types of metal produce different types of artifacts on MRI that may make the examination uninterpretable. The term magnetic susceptibility describes the manner and amount by which a material becomes magnetized in a magnetic field. Nonferrous magnetic metals may produce local electrical currents induced by the changing field, which causes distortion of the field and artifacts. These artifacts take two main forms: geometric distortion and signal loss secondary to dephasing (Fig. 18-8). Different techniques are more susceptible to these artifacts, and gradient echo images in particular are especially sensitive to differences in magnetic susceptibility and field homogeneity. Although metals can cause artifacts that render the examination uninterpretable, certain indwelling devices may be a contraindication to the entire examination. Certain types of aneurysm clips, cardiac valves, and vascular devices can result in harm to patients during an MRI examination. Some implantable devices may cease to function properly after an MRI study, and others may be a source of heating with deleterious biologic effects and burns. Readers are referred to publications and websites that track information related to contraindications.²⁸

FIGURE 18-7 Beam-hardening artifact on axial CT secondary to pedicle screws.

FIGURE 18-8 Metallic artifacts seen on sagittal (A) and axial (B) T1-weighted images through the thoracic spine in a patient with steel rods. There is a geometric artifact with complete distortion of the image as a result of the stainless steel. Sagittal T1-weighted (C), sagittal gradient echo (D), axial gradient echo (E), and axial CT myelographic (F) images. On the sagittal T1 image, a small extradural defect is noted at the C5-6 level (white arrow). This defect appears more prominent on the gradient echo image (black arrow). On the axial gradient echo image this defect appears midline. Axial CT myelography shows postsurgical changes but no evidence of an extradural defect. These changes are secondary to a susceptibility effect produced by small shards of metal that are often noted on MRI after surgery but are too small to be seen on CT or plain films. The susceptibility artifact can be mistaken for a recurrent disk herniation or osteophyte. Sagittal T1-weighted (G), sagittal T2-weighted (H), and gradient echo (I) images of the lumbar spine in a patient with a metallic intradiscal fixation device at L4-5 and L5-S1. Note that the susceptibility artifact is more prominent on the T2-weighted image (white arrows) than on the T1-weighted sequence (black arrows). It is most prominent on the gradient echo image because of its heightened sensitivity to field inhomogeneity and susceptibility changes.

The primary disadvantages of MRI are related to its cost and safety concerns with respect to electronic devices or implants. The safety issues include heating, dislodgment, and malfunction, as well as image distortion in the presence of metal. Claustrophobia remains a significant problem that sometimes requires machines with a larger bore and sedation or general anesthesia.

Intravenous contrast agents are commonly used with both CT and MRI examinations. Both iodinated (CT) and paramagnetic substances (MRI) are cleared by the kidney and should be used with caution or not at all in patients with impaired renal function. Although this has been common knowledge with the iodinated contrast media used for CT, recent evidence has also documented various complications related to the use of paramagnetic contrast agents in patients with impaired renal function. Despite being uncommon, nephrogenic systemic fibrosis, a late serious adverse reaction to gadolinium, has been well documented.^28,29

Spinal Angiography

Spinal angiography can be performed with MRI (magnetic resonance angiography [MRA]), CT (CT angiography [CTA]), and catheter-based digital subtraction angiography with selective arterial catheterization. The latter is still regarded as the most accurate technique for the detection and pretreatment characterization of vascular malformations of the spinal cord and meninges. Although probably associated with the highest risk among procedures for spinal imaging, the complication rate has been greatly reduced with the use of microcatheters and more experienced users. Definitive treatment of vascular lesions by endovascular means is now more common, and the examination can be better directed by information from diagnostic MRA and CTA studies, which can help focus the catheter examination. The addition of MRA to standard MRI does not significantly alter the sensitivity, specificity, and accuracy of detection of spinal malformations but often provides detection of the level of a fistula or vascular malformation. CTA of the spinal axis is relatively easy to perform and is also useful in identifying the level of involvement.

Ultrasound

The use of ultrasound in the spine is limited because of difficulty penetrating bone. Its main value is intraoperative for additional characterization and localization of disease such as spinal cord tumors or in pediatric applications. Intraoperatively, high-frequency probes that improve spatial resolution are commonly used because of the ability to directly approach the region of pathology.

Thermography, Diskography, CT Diskography

Most expert panels believe that thermography is too nonspecific to be of significant value for the evaluation of spinal disorders. Diskography is a long-standing provocative technique that by nature of direct stimulation, is thought by some to be able to identify painful and concordant disks.^1,30 Diskography and CT diskography are still used, especially when other imaging modalities have failed to localize the cause of pain. The morphologic information is not as critical as reproduction of a patient’s characteristic pain. Although some data support its use, prospective well-controlled trials in support of its prognostic value remain absent from the literature.

Terminology

The term degeneration includes any or all of the following: real or apparent desiccation, fibrosis, narrowing of the disk space, diffuse bulging of the annulus beyond the disk space, extensive fissuring and mucinous degeneration of the annulus, defects and sclerosis of the end plates, and osteophytes at the vertebral apophyses. Conventional radiographs are able to identify disk space narrowing, bony changes, and malalignment. CT can further identify disk herniation and provide better morphologic characterization of various types of stenosis. MRI provides the greatest spectrum of morphologic findings. With MRI, degenerative changes are manifested as disk space narrowing; loss of T2-weighted signal intensity from the intervertebral disk; the presence of fissures, fluid, vacuum changes, and calcification within the vertebral disk; ligamentous signal changes; marrow signal changes; osteophytosis; disk herniation; malalignment; and stenosis.

Degenerative Disk Changes

The major cartilaginous joint (amphiarthrosis) of the vertebral column is the intervertebral disk. Each disk consists of an inner portion, the nucleus pulposus, surrounded by a peripheral portion, the annulus fibrosus. The nucleus pulposus is eccentrically located and is closer to the posterior surface of the intervertebral disk. With degeneration and aging, type II collagen increases outwardly in the annulus, and there is greater loss of water from the nucleus pulposus than from the annulus. This results in a loss of the hydrostatic properties of the disk, with an overall reduction in hydration in both areas to about 70%. In addition to water and collagen, the other important biochemical constituents of the intervertebral disk are the proteoglycans. The individual chemical structures of the proteoglycans are not changed with degeneration, but their relative composition is. The ratio of keratin sulfate to chondroitin sulfate increases, and there is a diminished association with collagen, which may reduce the tensile strength of the disk. The decrease in water-binding capacity of the nucleus pulposus is thought to be related to the decreased molecular weight of its nuclear proteoglycan complexes (aggregates). The disk becomes progressively more fibrous and disorganized, with the end stage represented by amorphous fibrocartilage and no clear distinction between the nucleus and annulus.^31–33 On T2-weighted images, central disk signal intensity is usually markedly decreased and at distinct variance to that seen in unaffected disks of the same individual. Work with T2-weighed SE sequences³⁴ suggests that MRI is capable of depicting changes in the nucleus pulposus and annulus fibrosus associated with degeneration and aging based on the loss of signal intensity (Fig. 18-9). In work with cadaver spines of various ages, absolute T2 measurements correlated more closely with the glycosaminoglycan concentration than with the absolute water content. Thus, the signal intensity may not be related to the total amount of water but rather the state of the water. At present, the role that specific biochemical changes (proteoglycan ratios, aggregation of complexes) plays in the altered signal intensity is not well understood. Given that the T2 signal intensity in the disk appears to track the concentration and regions with a high percentage of glycosaminoglycan more than the absolute water content, it seems likely that the health and status of the proteoglycans are major determinants of signal intensity.³⁵ It has been proposed that annular disruption is the critical factor in degeneration and that when a radial tear develops in the annulus, there is shrinkage with disorganization of the fibrous cartilage of the nucleus pulposus and replacement of the disk by dense fibrous tissue with cystic spaces (Fig. 18-10).^36–40 Annular tears, also properly called annular fissures, are separations between annular fibers, avulsion of fibers from their insertions on the vertebral body, or breaks through fibers that extend radially, transversely, or concentrically and involve one or many layers of the annular lamellae. The term tear or fissure describes the spectrum of such lesions and does not imply that the lesion was caused by trauma (see Fig. 18-1). Although it is established that annular disruption is a sequela of degeneration and is often associated with it, its role as the causal agent of disk degeneration has not been proved. MRI is the most accurate anatomic method for assessing intervertebral disk disease. The signal intensity characteristics of the disk on T2-weighted images reflect changes caused by aging or degeneration. A classification scheme for lumbar intervertebral disk degeneration has been proposed that has reasonable intraobserver and interobserver agreement.⁴¹ To date, however, there has been no correlation between disk changes on MRI and a patient’s symptoms. With loss of water and proteoglycans, the nucleus pulposus is desiccated and friable with yellow-brown discoloration. Its onion skin appearance begins to unravel, and cracks, clefts, or crevices appear within the nucleus and extend into the annulus fibrosus. Fissuring, chondrocyte generation, and formation of granulation tissue may be noted within the end plate, annulus fibrosus, and nucleus pulposus of degenerative disks and are indicative of attempts at healing.⁴⁰ Radiolucent collections (vacuum disk phenomena) representing gas, principally nitrogen, occur at sites of negative pressure produced by the abnormal spaces.⁴² The vacuum phenomenon within a degenerated disk is represented on SE images as areas of signal void (Fig. 18-11).⁴³ Although the presence of gas within the disk is usually suggestive of degenerative disease, spinal infection may (rarely) be accompanied by intradiscal or intraosseous gas.⁴⁴ As intervertebral osteochondrosis progresses, there may be calcification of the disk. Calcification has usually been described on MRI as a region of decreased or absent signal intensity. The loss of signal is attributed to a low mobile proton density, as well as, in the case of gradient echo imaging, to its sensitivity to the heterogeneous magnetic susceptibility found in calcified tissue. There is, however, variability in the signal intensity of calcium with various sequences, and the type and concentration of calcification are important factors. Hyperintense disks on T1-weighted MRI may be secondary to calcification (Fig. 18-12).⁴⁵ For concentrations of calcium particulates of up to 30% by weight, the signal intensity on standard T1-weighted images increases but then subsequently decreases.^46,47 These data probably reflect particulate calcium reducing T1 relaxation times by a surface relaxation mechanism. Hyperintensities that are affected by fat suppression techniques have also been noted within intervertebral disks and are thought to be related to ossification with lipid marrow formation in severely degenerated or fused disks.

FIGURE 18-9 Degenerative disk. There is mild disk space narrowing (arrow) and loss of signal intensity on T2-weighted images of the L4-5 disk.

FIGURE 18-10 Annular tear seen on parasagittal (A) and axial (B) T2-weighted images through the L4-5 disk. Note the high signal intensity in the outer annulus/longitudinal ligament complex, which represents the area of annular disruption (arrows).

FIGURE 18-11 A and B, Vacuum phenomenon. Note the decreased signal intensity within the L4-5 disk, which is better seen on the gradient echo image (white arrow) than on the spin echo image (black arrow).

FIGURE 18-12 Intervertebral disk calcification (arrows). Note the high signal intensity on T1-weighted (A) and T2-weighted (B) images at the L5-S1 level.

Degenerative Marrow Changes

Changes in signal intensity of the vertebral body marrow adjacent to the end plates of degenerated disks are a long recognized and common observation on MRI of the lumbar spine.^48,49 However, despite a growing body of literature on this subject, their clinical importance and relationship to symptoms remain unclear.⁵⁰

These marrow changes appear to take three main forms. Type I changes consist of decreased signal intensity on T1-weighted images and increased signal intensity on T2-weighted images (Fig. 18-13). They have been identified in approximately 4% of patients scanned for lumbar disease,⁴⁹ in approximately 8% of patients after diskectomy,⁵¹ and in 40% to 50% of chymopapain-treated disks, which may be viewed as a model of acute disk degeneration.⁵² Histopathologic sections of disks with type I changes show disruption and fissuring of the end plate and vascularized fibrous tissue within the adjacent marrow that prolongs T1 and T2 times. Enhancement of type I vertebral body marrow changes is seen with the administration of gadolinium, and at times the changes extend to involve the disk itself and are presumably related to the vascularized fibrous tissue within the adjacent marrow. Type II changes are represented by increased signal intensity on T1-weighted images and isointense or slightly hyperintense signal on T2-weighted images (Fig. 18-14). They have been identified in approximately 16% of patients on MRI. Disks with type II changes also show evidence of end plate disruption, with yellow (lipid) marrow replacement in the adjacent vertebral body resulting in a shorter T1 time. Type III changes are represented by decreased signal intensity on both T1- and T2-weighted images and correlate with extensive bony sclerosis on plain radiographs. The lack of signal in type III changes no doubt reflects the relative absence of marrow in areas of advanced sclerosis (Fig. 18-15). Unlike types III, types I and II changes show no definite correlation with sclerosis on radiography.⁵³ This is not surprising when one considers the histology; the sclerosis seen on plain radiographs is a reflection of dense woven bone within the vertebral body, whereas the MRI changes are more a reflection of the intervening marrow elements.

FIGURE 18-13 Type I degenerative marrow changes. Sagittal T1-weighted (A) and T2-weighted (B) images demonstrate decreased signal intensity of the L5 vertebral body adjacent to the degenerated disk space on T1 and increased signal intensity on T2.

FIGURE 18-14 Type II degenerative marrow change (arrows). There is increased signal intensity along the opposing margins of L5 and S1 on both T1-weighted (A) and T2-weighted (B) images.

FIGURE 18-15 Type III degenerative marrow change (arrows). There is decreased signal intensity of the opposing vertebral body margins at L4-5 on both T1-weighted (A) and T2-weighted (B) images.

Similar marrow changes have also been noted in the pedicles (Fig. 18-16). Although originally described as being associated with spondylolysis, they have also been noted in patients with degenerative facet disease and pedicle fractures.^54,55 The exact mechanism by which these marrow changes occur is not known. The association of these marrow changes with degenerative disk disease, facet changes, and pars and pedicle fractures suggest that they are a response to biomechanical stress.

FIGURE 18-16 Pedicle hyperintensity/pars interarticularis fracture. Parasagittal T1-weighted (A), T2-weighted (B), and short tau inversion recovery (STIR) (C) images demonstrate high signal intensity within the pedicle of L4, which is best appreciated on the STIR sequence (arrow). Parasagittal (D) and sagittal (E) CT multiplanar reformatted images demonstrate a subacute fracture through the pars at L4-5.

Of these three types, type I changes appear to be more fluid and variable, a reflection of ongoing underlying pathologic processes such as continuing degeneration with associated changing biomechanical stresses. Of the three types, type I is most often associated with ongoing low back symptoms.^56–60 In a longitudinal study, the incidence of new degenerative marrow changes was 6% over a 3-year period, with most being type I.⁵⁸ In a study of nonoperated patients with low back pain, Mitra and associates found that 92% of type I changes converted either wholly or partially to type II (52%), became more extensive (40%), or remained unchanged (8%).⁵⁷ There was an improvement in symptoms in patients in whom type I changes converted to type II.

Some diskography studies in patients with degenerative marrow changes have suggested that type I marrow changes are invariably associated with painful disks.^61,62 Others have failed to reproduce this association,^63,64 and thus the relationship between degenerative marrow changes and diskogenic pain remains unproved.

Multiple authors have observed a variety of inflammatory mediators in association with degenerative marrow changes. Burke and colleagues observed an increase in proinflammatory mediators such as interleukin-6, interleukin-8, and prostaglandin E₂ in the disks of patients with type I marrow changes who were undergoing fusion for low back pain.⁶⁴ Ohtori and coworkers found that the cartilaginous end plates of patients with type I marrow changes had more protein gene product (PGP) 9.5 immunoreactive nerve fibers and cells immunoreactive for tumor necrosis factor (TNF) than did patients with normal end plates.⁶⁵ PGP 9.5 immunoreactivity was seen exclusively in patients with diskogenic low back pain. TNF immunoreactivity in end plates with type I marrow changes was higher than in those with type II marrow changes. The authors concluded that type I marrow changes represent a more active inflammation mediated by proinflammatory cytokines whereas type II and type III changes are more quiescent.⁵⁰ In a study of infliximab, a monoclonal antibody against TNF-α, Korhonen and associates found that it was most effective when there were degenerative type I marrow changes at the symptomatic level.⁶⁶ Nevertheless, the relationship of degenerative marrow changes to immunobiologic and cellular response mechanisms, although probably important, remains unclear.

In a study by Toyone and colleagues, 70% of patients with type I marrow changes had segmental hypermobility versus 16% with type II changes.⁵⁶ Probably the greatest indication that these marrow changes, particularly type I, are related to biomechanical instability is based on observations after fusion. Chataigner and collaborators suggested that patients with type I marrow changes have much better outcomes with surgery than do those with isolated degenerative disk disease and normal or type II marrow changes.⁶⁷ In addition, conversion of type I marrow changes to either normal or type II was associated with higher fusion rates and better outcomes. Other studies support the contention that persistence of type I changes after fusion suggests pseudarthrosis and is associated with persistent symptoms. Conversely, conversion of type I marrow changes to either normal or type II is associated with higher fusion rates and better outcomes (Fig. 18-17).^68–70 The conclusion is that fusion produces greater stability, reduces biomechanical stress, and accelerates the course of type I marrow changes toward improvement.

FIGURE 18-17 Conversion of type I marrow changes. Preoperative (A and C) and postoperative (B and D) images of the lumbar spine in a patient after lumbar fusion demonstrate typical type I marrow changes at the L4-5 level preoperatively (arrows). Postoperatively, the type I changes have converted to type II changes. The decreased signal on T1 is converted to an increased signal representing lipid marrow conversion. The increased signal intensity on T2 has resolved.

As further support that these fluid marrow changes reflect biomechanical stress, we have seen similar marrow conversion in the pedicles of vertebral bodies associated with symptomatic pars and pedicle fractures, as well as in those with severe degenerative facet joint disease. Pedicle marrow change to a normal or type II appearance is associated with improvement in symptoms.⁷¹

Degenerative Facet and Ligamentous Changes

The superior articulating process of one vertebra is separated from the inferior articulating process of the vertebra above by a synovium-lined articulation, the zygapophyseal joint. Like all diarthrodial synovium-lined joints, the lumbar facet joints are predisposed to arthropathy with alterations of the articular cartilage (Fig. 18-18). With disk degeneration and loss of disk space height, the increased stress on the facet joints with craniocaudal subluxation results in arthrosis and osteophytosis. The superior articular facet is usually more substantially involved. Facet arthrosis can result in narrowing of the central canal, lateral recesses, and foramina and is an important component of lumbar stenosis. However, it has been proposed that facet arthrosis may occur independently and be a source of symptoms on its own.^72,73 Synovial villi may become entrapped within the joint with resulting joint effusions. The mechanism of pain may be related to nerve root compression from degenerative changes of the facets or direct irritation of pain fibers from the innervated synovial linings and joint capsule.⁷³ Osteophytosis and herniation of synovium through the facet joint capsule may result in synovial cysts, although the cause of these facet joint cysts is unclear. There is a more straightforward relationship of synovial cysts with osteoarthritis and instability of the facet joints than with degeneration of the intervertebral disk alone. In a review of patients with degenerative facet disease, synovial cysts occurred at anterior or intraspinal locations in 2.3% of patients and at posterior or extraspinal locations in 7.3% (Fig. 18-19).⁷⁴ The important ligaments of the spine include the anterior longitudinal ligament, the posterior longitudinal ligament, the paired sets of ligamenta flava (connecting the laminae of adjacent vertebrae), the intertransverse ligaments (extending between transverse processes), and the unpaired supraspinous ligament (along the tips of the spinous processes). Because these ligaments normally provide stability, any alteration in the vertebral articulations can lead to ligamentous laxity with subsequent deterioration. Loss of elastic tissue, calcification and ossification, and bone proliferation at sites of ligamentous attachment to bone are recognized manifestations of such degeneration. Excessive lordosis or extensive disk space loss in the lumbar spine leads to close approximation and contact of the spinous processes and to degeneration of intervening ligaments.^75,76 Histologically, granulomatous reaction and perivascular cellular infiltration characterize the condition.

FIGURE 18-18 Degenerative facet changes (arrows) seen on T1-weighted (A), high-resolution CT (B), and T2-weighted spin echo (C) axial images through the L4-5 facets. Note the bony osteophyte along the anterior margin of the inferior facet at L4 on the right and the degenerative facet narrowing bilaterally. The bony changes are not as obvious on the T1- and T2-weighted images, but the findings of narrowing and asymmetrical soft tissue are clearly identifiable on MRI.

FIGURE 18-19 Synovial cyst demonstrated on axial T1-weighted (A), contrast-enhanced T1-weighted (B), and T2-weighted (C) images through the facets at L4-5. There are severe bilateral degenerative facet changes with distraction and fluid. On the T1-weighted image, there is an ill-defined soft tissue mass projecting medially off the left facet (arrow). On the contrast-enhanced T1-weighted image, this soft tissue mass (arrow) is now clearly outlined by peripheral enhancement. On the T2-weighted image this mass (arrow) is demonstrated to have high signal intensity suggestive of fluid.

Instability

Segmental instability can result from degenerative changes involving the intervertebral disk, vertebral bodies, and facet joints that impair the usual pattern of spinal movement and produce motion that is irregular, excessive, or restricted. It can be translational or angular. Spondylolisthesis results when one vertebral body becomes displaced relative to the next most inferior vertebral body. The most common types are classified as degenerative, isthmic, iatrogenic, and traumatic. Degenerative spondylolisthesis is seen usually with an intact pars interarticularis, is related primarily to degenerative changes of the apophyseal joints, and is most common at the L4-5 vertebral level (Fig. 18-20). The predilection for degenerative spondylolisthesis at that level is thought to be related to the more sagittal orientation of the facet joints, which makes them increasingly prone to anterior displacement. Degenerative disk disease may predispose to or exacerbate this condition secondary to narrowing of the disk space, which can produce subsequent malalignment of the articular processes and lead to rostrocaudal subluxation.

FIGURE 18-20 Degenerative spondylolisthesis. Sagittal T1- and T2-weighted spin echo images (A and B) of the lumbar spine. Grade I spondylolisthesis of L4 on L5 (arrows) is present, as well as severe central canal stenosis and thickening of the posterior ligaments. Axial T1- and T2-weighted images (C and D) through the L4-5 level. Note the severe central canal stenosis, thickened posterior ligaments, and severe bilateral degenerative facet changes.

Herniation

Herniation refers to localized displacement of nucleus, cartilage, fragmented apophyseal bone, or fragmented annular tissue beyond the intervertebral disk space. The disk space is defined rostrally and caudally by the vertebral body end plates and peripherally by the outer edges of the vertebral ring apophyses, exclusive of osteophytic formations. The term localized contrasts with the term generalized, the latter being arbitrarily defined as greater than 50% (180 degrees) of the periphery of the disk.⁷⁷ Displacement, therefore, can occur only in association with disruption of the normal annulus or, as in the case of intravertebral herniation (Schmorl’s node), a break in the vertebral body end plate. Because details of the integrity of the annulus are often unknown, the diagnosis of herniation is usually made by observation of localized displacement of disk material beyond the edges of the ring apophyses—that is, less than 50% (180 degrees) of the circumference of the disk (Fig. 18-21). Localized displacement in the axial (horizontal) plane can be focal, or less than 25% of the disk circumference, or broad based, between 25% and 50% of the disk circumference. The presence of disk tissue circumferentially (50% to 100%) beyond the edges of the ring apophyses may be called bulging and is not considered a form of herniation. A disk may have more than one herniation. The term herniated disk does not imply any knowledge of etiology, relationship to symptoms, prognosis, or need for treatment. When data are sufficient to make the distinction, a herniated disk may be more specifically characterized as protruded or extruded. These distinctions are based on the shape of the displaced material. Protrusion is present if the greatest distance, in any plane, between the edges of the disk material beyond the disk space is less than the distance between the edges of the base in the same plane (Fig. 18-22). Extrusion is present when, in at least one plane, any one distance between the edges of the disk material beyond the disk space is greater than the distance between the edges of the base in the same plane or when no continuity exists between the disk material beyond the disk space and that within the disk space (Fig. 18-23). Extrusion may be further specified as sequestration if the displaced disk material has completely lost any continuity with the parent disk. The term migration may be used to signify displacement of disk material away from the site of extrusion, regardless of whether it is sequestrated (Fig. 18-24). Herniated disks in the craniocaudal (vertical) direction through a break in the vertebral body end plate are referred to as intravertebral herniations. Nonacute Schmorl’s node intrabody herniations are common spinal abnormalities regarded as incidental observations. They have been reported in 38% to 75% of the population.^78,79 Although intrabody herniations may occur as a result of end plate weakness secondary to bone dysplasia, neoplasms, infections, or any process that weakens the end plate or the underlying bone, most intrabody herniations probably form after axial-loading trauma, with preferential extrusion of nuclear material through the vertebral end plate rather than an intact and normal annulus fibrosus. It has been suggested that asymptomatic intrabody herniations may be traceable to a specific occurrence of acute nonradiating low back pain in the patient’s history, which supports the concept that intrabody herniations (Schmorl’s nodes) occur through sites of end plate fracture. Type I vertebral body marrow changes have been described surrounding acute interbody herniations.⁸⁰

FIGURE 18-21 A, Symmetrical bulging disk. B, Broad-based herniation. C, Focal herniation. D, Protrusion and extrusion.

FIGURE 18-22 Disk protrusion seen on sagittal (A) and axial (B) T2-weighted images of the lumbar spine. On the sagittal image, the disk extends beyond the vertebral body margins, which on the axial image shows the base to be broader than the posterior extent (arrows). Sagittal multiplanar reformatted (MPR) (C) and axial (D) CT images through the lumbar spine. The sagittal MPR image demonstrates an ill-defined soft tissue mass projecting posteriorly at the L4-5 disk level. The broad-based disk protrusion is better depicted on the axial image (white arrow). Axial gradient echo MRI (E) and CT (F) at the C5-6 level. Note the broad-based right-sided disk protrusion (arrow). Sagittal (G) and axial (H) T2-weighted images of the cervical spine demonstrating disk protrusion at C5-6 and disk extrusion with an inferior fragment at C6-7. The axial image through the body of C7 demonstrates the rounded soft tissue mass centrally in the anterior epidural space.

FIGURE 18-23 Extruded disk. Sagittal (A) and axial (B) T2-weighted images of the lumbar spine demonstrate disk extrusion (arrow) on the sagittal image with a narrow base. The axial image demonstrates posterior displacement of the S1 nerve root on the left as a result of this disk extrusion.

FIGURE 18-24 Disk extrusion and a free fragment seen on sagittal T1-weighted (A) and T2-weighted (B) images of the lumbar spine. There is disk extrusion and an inferior fragment (white and black arrows, respectively).

Stenosis

Spinal stenosis was defined in 1975 as any type of narrowing of the spinal canal, nerve root canals, or intervertebral foramina.⁸¹ Two broad groups have been defined: acquired (usually related to degenerative changes) and congenital or developmental. Developmental stenosis can be exacerbated by superimposed acquired degenerative changes (Fig. 18-25). In the acquired type, there has been no association between the severity of pain and the degree of stenosis. The most common symptoms are sensory disturbances in the legs, low back pain, neurogenic claudication, weakness, and relief of pain by bending forward. The imaging changes are in general more extensive than expected from the clinical findings.⁸² Patients with symptoms referable to spinal stenosis tend to have narrower spines than asymptomatic patients do. Although there appears to be a correlation between cross-sectional area and midsagittal measurements in patients with symptomatic spinal stenosis, absolute values and correlation between measurements and symptoms are lacking. The degree of stenosis is not static; extension worsens the degree of central and foraminal stenosis by 11%, whereas flexion appears to improve it by an average of 11%. Segmental instability, which can cause static and dynamic stenosis, is considered a cause of low back pain but is poorly defined.⁸³ Some evidence suggests that disk degeneration, narrowing of the spinal canal, and degenerative changes in the facets and spinal ligaments contribute to stenosis and that instability increases with age. Unfortunately, there do not appear to be reliable prognostic imaging findings that correlate with surgical success or even predict whether patients will benefit from surgery.⁸⁴

FIGURE 18-25 Lumbar canal stenosis noted on midline T1-weighted (A) and T2-weighted (B) images of the lumbar spine. There is diffuse central canal narrowing, evidence of a broad-based disk protrusion, and severe canal stenosis at L4-5. Contiguous axial T1-weighted images (C and D) of the L4-5 level demonstrate severe central canal stenosis (arrows).

Neoplasms

The key to diagnostic imaging evaluation and the differential consideration of tumors lies with their physical location or space (e.g., extradural, intradural, intramedullary), in addition to the patient’s medical history and age. Conventional radiographs are relatively insensitive and do not turn positive until more than 50% of the bone has been destroyed. Myelography has been used to determine the presence of intraspinal masses and their effect on the underlying or adjacent subarachnoid space, thereby permitting the location and degree of mass effect and compression to be determined. With water-soluble contrast media it is generally unnecessary to introduce contrast material both above and below a lesion to determine its full extent because even with severe blocks, contrast media can usually be maneuvered past the lesion and subsequent CT myelography and 3D formats can provide excellent anatomic delineation.

MRI is the mainstay of imaging in patients suspected of having a neoplasm. In the extradural space, MRI is capable of identifying paraspinal soft tissue masses and can assess marrow signal intensity, an important consideration in both localized and diffuse neoplastic disease. As with the previous entities, here we approach neoplasms from the space in which they are most likely to exist.

In the extradural space, the most common spinal neoplasms are metastatic lesions. Conventional radiographs usually remain negative until 50% to 70% of the bone has been destroyed. Radionuclide studies (technetium 99–labeled SPECT) have high sensitivity with larger lesions, especially if there is cortical involvement. MRI provides coverage of the spinal axis and is more sensitive than other modalities to the presence of marrow replacement. The characteristic MRI finding in lytic metastases is decreased signal intensity of the marrow space on T1-weighted images (Fig. 18-29). This finding may be solitary, multiple, or diffuse (Fig. 18-30). Typically in adults, the vertebral body marrow signal on T1-weighted images is brighter than that of adjacent disks. Reversal of this signal intensity ratio after the age of 40 is a cause for concern. Lesions on T2-weighted imaging are more variable and can range from hypointense to hyperintense. STIR sequences usually demonstrate high signal intensity, and the appearance on diffusion-weighted images is variable. The presence of epidural disease is easily evaluated. Major differential considerations from an imaging perspective are hematopoietic malignancies such as plasmacytoma, benign osteoporotic compression fractures, and necrosis of a vertebral body. Normal heterogeneous marrow signal intensity may also be somewhat confusing. In hematopoietic malignancies, the marrow signal intensity may be patchy or diffusely decreased. Blastic osseous metastases often demonstrate hypointensity on T1-weighted, STIR, and T2-weighted images because the sclerotic bony changes overwhelm the marrow signal intensity. More benign conditions that need to be considered in the differential diagnosis include hemangiomas, osteodystrophy, Paget’s disease, osteomyelitis, and severe degenerative marrow changes. Distinction between benign and malignant compression fractures on imaging can often be difficult (Fig. 18-31). The typical benign compression fracture has a relatively homogeneous decreased intensity of the vertebral body marrow signal and may have patchy areas of more normal marrow. The posterior elements are involved in less than 25% of cases. From a morphologic perspective, a compressed vertebral body demonstrates fragmentation and may be associated with intrabody disk herniations. Additional fractures are seen in 50% of patients. In contrast, malignant compression fractures show diffuse replacement of the vertebral body marrow signal intensity, which is almost complete in the presence of volume loss. The posterior elements are involved more than 90% of the time. The compressed body has convex rather than fragmented margins. Additional metastatic lesions are common. Compression fractures caused by remote trauma tend to have a more normal or fatty marrow signal intensity and can easily be distinguished from acute trauma or malignant infiltration.

FIGURE 18-29 Metastases demonstrated on a sagittal T1-weighted spin echo image of the lumbar spine. Patchy as well as focal replacement of the marrow space is seen at all visualized levels. There is complete marrow replacement of the L2 vertebral body with loss of height and a convex posterior extension of the vertebral body into the spinal canal.

FIGURE 18-30 Diffuse marrow replacement from metastatic disease visualized on sagittal T1-weighted (A) and T2-weighted (B) images and a posterior view from a whole-body technetium 99 bone scan (C). On the T1-weighted image there is diffuse decreased signal intensity of the visualized osseous structures. Note the higher signal intensity of the intervertebral disk than the vertebral body on this T1-weighted spin echo image. On the T2-weighted image there is diffuse decreased signal intensity of the marrow space. On the radionuclide study, diffuse increased activity is seen throughout the spinal column.

FIGURE 18-31 Vertebral body compression fractures from acute trauma (A), remote trauma (B), and malignant disease (C). The compression fracture caused by remote trauma demonstrates hyperintense signal within the vertebral body from increased lipid marrow content and is easy to distinguish from subacute or acute changes. Acute trauma and malignancy may be difficult to differentiate.

MRI is the imaging examination of choice and the most sensitive imaging study for the detection of marrow involvement and paravertebral tumor extension. It has, in most cases, replaced myelography for the evaluation of spinal cord compression because it is able to evaluate areas between the myelographic block. It is 95% accurate in detecting metastatic compression of the cord and cauda equina.^96–98

Multiple myeloma is an important differential consideration in patients with abnormal marrow signal intensity. On conventional radiographs, it is often seen as diffuse osteopenia or multiple lytic lesions. CT may demonstrate lytic lesions, cortical disruption, or extraosseous soft tissue; however, these findings are nonspecific and may be related to any number of etiologies. On MRI, the marrow signal intensity may have a normal focal, diffuse, or heterogeneous pattern (Fig. 18-32) with foci of decreased intensity on T1-weighted images. There is concomitant osteopenia in 85% of patients and multiple lytic lesions in 80%. Vertebral fractures in patients with multiple myeloma may appear benign with a fragmented appearance. Bone scintigraphy is notoriously insensitive and detects less than 10% of lesions.^99,100

FIGURE 18-32 Multiple myeloma. Diffuse patchy heterogeneous marrow signal intensity of the visualized osseous structures is seen on both T1- (A) and T2-weighted (B) images. A more focal area of marrow replacement is noted in the spinous process of L1. There is loss of vertebral body height at L2 and L3.

Hemangiomas, on conventional radiographs, demonstrate coarse vertical bony trabeculae. On axial CT, these lesions are often manifested as a “corduroy” pattern of thickened trabeculae with intervening low-attenuation fat. On MRI, the intraosseous portions often appear mottled and have increased signal intensity on T1 images secondary to adipose tissue interspersed among thickened bony trabeculae. On T2-weighted images, both intraosseous and extraosseous tumor demonstrates increased signal intensity, presumably related to the more vascular and cellular components of the tumor (Fig. 18-33). Although this is the typical appearance of an intraosseous hemangioma, variable signal intensity on T1- and T2-weighted images is not uncommon. Incidental intraosseous hemangiomas are a common finding on MRI of the spine. The primary differential consideration is another normal incidental finding, a lipid marrow rest. Vertebral body hemangiomas can extend extraosseously and, particularly in the thoracic region, can produce severe cord compression (Fig. 18-34).

FIGURE 18-33 Benign intraosseous hemangioma demonstrated on sagittal T1-weighted (A) and T2-weighted (B) images of the mid lumbar spine. On T1, there is an ovoid heterogeneous but predominantly increased signal intensity replacement of the normal marrow signal. On the T2-weighted image there is heterogeneous increased signal intensity in the same location. Axial T1-weighted (C) and T2-weighted (D) images through the hemangioma within the L2 vertebral body. The T1-weighted changes are a reflection of the fat within the hemangioma, and the T2-weighted changes are a reflection of the more vascular elements. Note the somewhat speculated core, which is characteristic of a hemangioma.

FIGURE 18-34 Extraosseous extension of a vertebral body hemangioma visualized on sagittal multiplanar reformatted (A) and axial source (B) images through the midthoracic spine. The striated osseous appearance on the sagittal image and the spiculated appearance on the axial image are characteristic of an intraosseous hemangioma. Note the extension into the pedicles and posterior elements on the axial image. Sagittal T1-weighted (C), contrast-enhanced T1-weighted (D), T2-weighted (E), and short tau inversion recovery (F) images through the thoracic spine. Striated hyperintense changes within the vertebral body are noted on T1 and T2. There is contiguous vertebral body involvement, as well as extension into the posterior elements. On the contrast-enhanced T1-weighted and T2-weighted images, there is evidence of extraosseous extension and soft tissue enhancement in the anterior epidural space.

Other primary tumors of bone such as osteomas, osteoblastomas, aneurysmal bone cysts, giant cell tumors, osteochondromas, chondrosarcomas, and osteosarcomas are much less common than metastatic disease and occur as solitary lesions that involve both bone and soft tissue. Unlike plain-film radiography, myelography, and CT, which focus primarily on the bony architecture of the spinal canal and its adjacent soft tissues, MRI focuses on the vertebral marrow space and nearby soft tissues. The characteristics used for differential considerations are different between the two approaches. The designations osteolytic and osteoblastic have no meaning with MRI. However, certain primary lesions involving the vertebrae have been noted to present a unique appearance on MRI. For instance, aneurysmal bone cysts have a somewhat unique MRI appearance consisting of numerous well-defined cystic cavities surrounded by a rim of low signal intensity and multiple fluid levels. These changes, however, are not specific in that other bony lesions such as osteosarcoma, chondroblastoma, and giant cell tumor of bone may mimic aneurysmal bone cysts.

Chordomas are typically seen on conventional radiographs as radiolucent lesions or heterogeneous destructive masses, usually of the sacrum or vertebral body. On CT, chordomas also appear destructive and may have a large paraspinal soft tissue mass. Sclerosis is seen in almost half the cases. The sacral coccygeal region is the most common location (50%), followed by the spheno-occipital (35%) and vertebral bodies elsewhere (15%). On MRI, chordomas typically appear as heterogeneous hypointense to isointense lesions on T1 and hyperintense lesions on T2 with low-signal septations indicative of fibrous stroma (Fig. 18-35).

FIGURE 18-35 Chordoma demonstrated on sagittal T1-weighted (A), sagittal T2-weighted (B), and short tau inversion recovery (C) images of the sacral region. Coronal T1-weighted (D), contrast-enhanced T1-weighted (E), and fat-suppressed T2-weighted (F) images. There is a destructive mass involving the S2 body with an extraosseous soft tissue component involving the sacral canal and presacral soft tissues.

Intradural Extramedullary Tumors

Intradural extramedullary tumors include meningiomas, nerve sheath tumors, embryonal tumors, paragangliomas, leptomeningeal metastases, and meningeal cysts (arachnoid cysts). Fifty percent of intradural extramedullary lesions are meningiomas or nerve sheath tumors. They are typically isolated and well circumscribed with marked homogeneous enhancement.

Meningiomas are most commonly located in the thoracic region in a lateral or posterolateral location. On conventional radiographs the only finding may be focal intraspinal calcification. On CT, an isointense to hyperintense mass, relative to muscle, is the usual appearance on a nonenhanced study. After intravenous contrast administration there may be homogeneous enhancement. Adjacent bone may be hyperostotic. The MRI findings of an intraspinal meningioma are similar to those seen intracranially. On T1-weighted images the mass is isointense with spinal cord. On T2 the mass remains isointense or slightly hypointense in comparison to adjacent spinal cord. Rarely, it may be hyperintense. There is usually prominent homogeneous enhancement after contrast administration (Fig. 18-36). An enhancing dural tail, commonly seen intracranially, is less common intraspinally.¹⁰¹ The common differential consideration is a schwannoma or nerve sheath tumor, which often has higher signal intensity on T2-weighted images, may be cystic, and may be more commonly associated with hemorrhage. Other differential considerations are paraganglioma, epidermoid, arachnoid cyst, intradural metastasis, or lymphoma.

FIGURE 18-36 Meningioma seen on sagittal T1-weighted contrast-enhanced (A) and T2-weighted (B) images of the thoracic spine. An anterior extra-axial mass at the T10-11 level demonstrates homogeneous enhancement and is posteriorly displacing and markedly compressing the thoracic cord. There is a small superior dural tail noted (arrow). On the T2-weighted image, the mass is isointense with spinal cord. The hyperintense intramedullary signal within the thoracic cord presumably represents cord edema secondary to compression.

Schwannomas are characterized by a well-circumscribed intraspinal masses; they are most commonly located in an intradural extramedullary location. Fifteen percent may have an extradural component. An enlarged intravertebral foramen and thin pedicle are conventional radiographic signs of a long-standing mass. Schwannomas tend to be more hyperintense on T2-weighted images and have cystic changes or hemorrhage more often than meningiomas do. They are indistinguishable from solitary neurofibromas by imaging (Fig. 18-37).^102,103 Spinal involvement in neurofibromatosis is most common in type 2. Schwannomas in this condition may be solitary or multiple and can appear solid or cystic (Fig. 18-38).

FIGURE 18-37 Schwannoma visualized on axial CT (A), T1-weighted (B), T2-weighted (C), and T1-weighted contrast-enhanced (D) images through the L2 vertebral body. A large lobulated dumbbell mass (arrows) is expanding the neuroforamina and remodeling the adjacent vertebral body. The T1 signal is isointense with neural structures. There is a somewhat heterogeneous increased signal on T2 and homogeneous enhancement after contrast enhancement.

FIGURE 18-38 Sagittal cervicothoracic (A) and thoracolumbar (B) T1-weighted contrast-enhanced images demonstrate multiple intradural extramedullary enhancing masses that represent schwannomas in this patient with neurofibromatosis type 2.

Dermoid and epidermoid tumors are usually either intradural extramedullary (60%) or intramedullary (40%). The lower thoracic and lumbar regions are the most common locations. Conventional radiographs are generally normal but may demonstrate benign spinal canal widening with flattening of the pedicles and laminae. On CT these tumors are usually seen as well-demarcated masses with attenuation similar to that of CSF. The presence of calcification is more suggestive of a dermoid than an epidermoid tumor. Again, there may be focal osseous erosion or spinal canal widening. On MRI, dermoids are typically hypointense to hyperintense on T1 with variable signal intensities that reflect fat (hyperintense on T1) or calcium (decreased signal intensity on T1) (Fig. 18-39). Epidermoids are usually isointense on T1. Both tumors demonstrate increased signal intensity on T2-weighted images. Typically, these tumors do not enhance after contrast administration and may demonstrate restricted diffusion.

FIGURE 18-39 Dermoid. An intradural extramedullary mass (arrow) is displacing the conus anteriorly. This mass has variable signal intensity, with soft tissue, lipid, and heterogeneous regions noted. These lesions typically do not enhance.

Myxopapillary ependymomas represent 90% of filum tumors. They occur exclusively in the conus and filum terminale, are slow growing, and may fill the entire spinal canal. Conventional radiographs and CT may demonstrate vertebral body scalloping and canal enlargement (Fig. 18-40).

FIGURE 18-40 Myxopapillary ependymoma seen on sagittal T1-weighted (A), T2-weighted (B), and T1-weighted contrast-enhanced (C) images of the lumbar spine. The T1-weighted image shows a subtle mass filling the lumbar canal of L2 through L5. On the T2-weighted image this mass is noted to be heterogeneous with areas of both increased and decreased signal intensity. There is diffuse subtle enhancement after contrast administration, as well as evidence of enhancement more proximally in the region of the conus.

Leptomeningeal disease is silent on conventional radiographs and on myelography is manifested by filling defects of variable size within the subarachnoid space. Non–contrast-enhanced CT is usually normal unless there is coexisting extradural disease, as in the case of metastatic breast or lung carcinoma. On MRI, leptomeningeal disease is generally isointense with the spinal cord and nerve roots and, when extensive, may be manifested as a diffuse increase in the signal intensity of the CSF space on T1-weighted images with a so-called ground-glass appearance. Nerve roots may appear blurred or less distinct. On T2-weighted images, leptomeningeal disease is usually isointense with neuroelements or may be manifested as thickened nerve roots within the subarachnoid space. On contrast-enhanced T1-weighted images, there is obvious enhancement that may be diffuse, nodular, or linear. The most common cause is hematogenous dissemination from extracranial neoplasms such as adenocarcinoma (Fig. 18-41) of the lung or breast, melanoma, lymphoma, and metastases. In children and young adults, primitive neuroectodermal tumors are most commonly manifested as leptomeningeal disease. Differential considerations include inflammatory conditions such as hemangic meningitis, granulomatous meningitis (tuberculosis and sarcoidosis), congenital hypertrophic polyradiculoneuropathy, thickened nerve roots secondary to inflammatory conditions such as Guillain-Barré syndrome, polyneuropathy associated with acquired immunodeficiency syndrome, and chronic interstitial demyelinating polyneuropathy and arachnoiditis (see Fig. 18-41).¹⁰⁴

FIGURE 18-41 Leptomeningeal metastases demonstrated on T1-weighted sagittal (A) and contrast-enhanced T1-weighted sagittal (B) and axial (C) images. On the precontrast study, diffuse increased signal intensity of the cerebrospinal fluid space is seen in the distal lumbar region. After contrast administration there is enhancement of the surfaces of the distal thoracic cord and the traversing nerve roots in the lumbar region. An axial image through the region of the conus demonstrates enhancement of the surface of the distal cord and subarachnoid space.

Arachnoid cysts are typically intradural extramedullary loculated CSF collections that can be manifested as space-occupying lesions. Type I is an extradural meningeal cyst that contains no neural tissue (Fig. 18-42). Type II includes extradural meningeal cysts that contain neural tissue, and type III includes intradural meningeal cysts. Type II, or perineural cysts/Tarlov’s cysts, is more commonly encountered and contains nerve roots, often adherent to the cyst’s wall. Type III consists of spinal intradural meningeal cysts (arachnoid cysts), which are thought to arise from the diverticulum of the arachnoid. When long-standing, they may produce posterior vertebral body scalloping and thinning of the pedicles with a widened intrapedicular distance on conventional radiographs. On myelography, intradural cysts produce an intrathecal filling defect and extradural effacement of the subarachnoid space. Depending on their size, they may result in spinal cord compression. On CT, non–contrast-enhanced studies show isoattenuation with adjacent CSF, and there may be evidence of cord displacement. After intrathecal contrast administration an intradural arachnoid cyst may be difficult to visualize if it is opacified, but it otherwise appears as a filling defect with a mass effect. Delayed imaging with postmyelographic CT often results in opacification of the cyst with decreased attenuation in the generalized subarachnoid space. On MRI, CSF signal intensity is that of CSF that does not usually share the same degree of pulsations as the general subarachnoid space. T2-weighted images are equivalent to CSF but again may lack CSF flow signal changes. There is no enhancement after contrast administration, and the walls of the cyst may be impossible to resolve. Differential considerations include large degenerative cysts, dural ectasia, and spinal cord herniation.

FIGURE 18-42 Type I arachnoid cyst visualized on sagittal T1-weighted (A), T1-weighted contrast-enhanced (B), short tau inversion recovery (C), and T2-weighted (D) images through the midthoracic region. There is a large cerebrospinal fluid (CSF)-like intradural extramedullary mass that is displacing the cord anteriorly and does not enhance. Axial T1-weighted images before (E) and after (F) contrast administration demonstrate the large posterior located intradural extramedullary CSF-like collection.

Spinal cord herniation occurs through a defect in the dura, which is usually ventral. Clinically, these patients usually have unexplained chronic progressive leg pain and myelopathy. Imaging findings are those of anterior displacement of the thoracic cord with apparent expansion of the dorsal subarachnoid space. It is usually located in the midthoracic region. On myelography, there is either displacement of the cord or a focal deformity anteriorly. CT myelography is helpful in defining the deformity. MRI demonstrates cord displacement anteriorly, which is usually focal. The increased dorsal subarachnoid space can mimic the appearance of a type I arachnoid cyst (Fig. 18-43).^105,106

FIGURE 18-43 Thoracic cord herniation seen on coronal (A) and sagittal (B) multiplanar reformatted images from a CT myelogram. There is apparent displacement of the thoracic cord to the right and an increased cerebrospinal fluid dorsally. Axial CT myelography (C) and axial T2-weighted MRI (D) demonstrate herniation of the thoracic cord anteriorly and laterally on the right.

Intramedullary Disease and Tumors

Intramedullary abnormalities are silent on conventional radiographic and CT imaging unless nonspecific cord expansion has developed. MRI is clearly the study of choice and is able to characterize abnormalities on the basis of morphology, changes in signal intensity, the presence or absence of contrast enhancement, and anatomic location and extent. In general, the location, length, cord morphology, signal intensity characteristics, contrast character or absence of contrast enhancement, and blood by-products are important morphologic considerations for the differential diagnosis. Differential considerations of intramedullary disease are demyelinating disease, tumor, hydrosyringomyelia, infection, ischemia, vascular malformation, acute disseminated encephalomyelitis, sarcoid, and changes related to trauma.

In patients with acute transverse myelitis and the onset of acute motor, sensory, and autonomic dysfunction in the absence of preexisting neurological disease and spinal cord compression, the main differential considerations are multiple sclerosis, postinfectious myelitis, metabolic changes, paraneoplastic syndromes, and infection.

Ten percent to 20% of patients with multiple sclerosis have isolated spinal cord disease, most commonly affecting the cervical region in the dorsal lateral aspect of the cord (Fig. 18-44). These usually appear as well-circumscribed hyperintense lesions on T2 with homogeneous, nodular, or ring enhancement after contrast administration. Long-standing disease is often manifested as cord atrophy. Top differential considerations are other intramedullary diseases, neoplasms, or other causes of acute transverse myelitis.

FIGURE 18-44 Multiple sclerosis demonstrated on T1-weighted (A), T2-weighted (B), contrast-enhanced sagittal T1-weighted (C), and axial contrast-enhanced T1-weighted (D) images. The T1-weighted image demonstrates patchy decreased signal intensity. There is increased signal in this region on the T2-weighted study. Patchy enhancement of the upper cervical cord is seen after contrast enhancement. The axial image after contrast administration demonstrates posterior lateral enhancement of the cord.

The most common primary neoplasms of the spinal cord are ependymomas and astrocytomas. In general, it is usually impossible to differentiate between the two with imaging. Typically, primary cord tumors demonstrate expansion of the cord, low signal intensity on T1, and high signal intensity on T2 with variable contrast enhancement. Ependymomas are more likely to have associated blood by-products and large satellite cysts (Fig. 18-45). When long-standing, ependymomas may have the conventional radiographic findings of canal widening and posterior vertebral body scalloping. T1-weighted images demonstrate an isointense to slightly hypointense spinal cord mass, which may or may not have associated hemorrhage. T2-weighted images demonstrate hyperintense signal, often associated with intratumoral cysts similar to CSF in signal intensity. The majority of these lesions enhance in a homogeneous fashion, although nodular peripheral enhancement is possible.¹⁰⁷ The imaging findings of astrocytoma are similar to those of ependymomas, and typically most enhance (Fig. 18-46). Patients with ependymomas are typically older, hemorrhage is more common, and the prevalence in the lower thoracic region is higher. Cystic or necrotic components are more common in ependymomas.

FIGURE 18-45 Ependymoma seen on sagittal T1-weighted (A), sagittal T2-weighted (B), and sagittal T1-weighted contrast-enhanced (C) images of the thoracic spine. The multilocular cystic mass within the substance of the thoracic cord with variable signal intensity is suggestive of cyst and blood by-products. There is evidence of cystic changes or cord edema within the upper thoracic cord and inferior to the mass that is best depicted on the T2-weighted image. On the T1 contrast-enhanced study, homogeneous and peripheral enhancement of the central portion of the mass is apparent.

FIGURE 18-46 Cervical astrocytoma demonstrated on sagittal T1-weighted (A), T1-weighted contrast-enhanced (B), and T2-weighted (C) images through the upper cervical spine. There is diffuse enlargement of the cervical cord from C2 through C5. After contrast administration, loculated and cystic regions of enhancement are present. The T2-weighted image shows both focal hyperintensity and diffuse increased signal intensity throughout this enlarged intramedullary mass.

Less common intramedullary tumors include hemangioblastoma, oligodendroglioma, ganglioglioma, lymphoma, metastatic disease, lipoma, schwannoma, or dermoid/epidermoid tumor.¹⁰⁸

Thirty percent of patients with hemangioblastomas have von Hippel-Lindau syndrome, which may be associated with pancreatic cysts, renal cell carcinoma, renal cysts, and retinal angiomas. The thoracic region is more commonly involved than the cervical. Most are associated with a cyst. Typically, the hemangioblastoma is an expansile cord mass that is hypointense on T1 and hyperintense on T2. After contrast administration, there is usually homogeneous intense enhancement of the nodule or mass. Generally, extensive cord edema extends from the region of the mass, and in some cases hydrosyringomyelia is present. The angiographic findings typically demonstrate an intense prolonged vascular stain that may be associated with arteriovenous (AV) shunting and enlarged spinal arteries (Fig. 18-47). Hemangioblastomas may be single or multiple.¹⁰⁹

FIGURE 18-47 Hemangioblastoma visualized on sagittal T1-weighted (A), T1-weighted contrast-enhanced (B), and T2-weighted (C) images of the thoracolumbar junction. There is an intramedullary mass associated with adjacent cord edema or a syrinx. Homogeneous enhancement of this intramedullary mass is seen after contrast administration. The T2-weighted image demonstrates multiple areas of focal decreased signal intensity representing flow voids from the increased vascularity of the lesion. An axial T1-weighted image (D) demonstrates the associated cyst within the adjacent thoracic cord. An anteroposterior view from a spinal arteriogram (E) demonstrates an enlarged artery of Adamkiewicz feeding the hypervascular mass at the thoracolumbar junction.

Intramedullary cystic cavities may be idiopathic, congenital, associated with congenital abnormalities such as Chiari malformations, secondary to trauma, postoperative, secondary to arachnoiditis, or found as a satellite cyst associated with a tumor (Fig. 18-48). Pathologically, these cavities are lined by either neural fibrous or ependymal elements, in distinction to intratumoral cysts, which are characteristically lined by neoplastic cells. Differentiation requires contrast material; intratumoral cysts often exhibit peripheral enhancement, whereas hydrosyringomyelia or satellite cysts associated with tumors may not.¹¹⁰

FIGURE 18-48 Syringomyelia seen on sagittal T1-weighted (A), T2-weighted (B), and cerebrospinal fluid (CSF) flow studies (C and D) of the cranial vertebral junction. There is a syrinx cavity within the cervical cord and a Chiari malformation with downward herniation of the tonsils. The CSF flow studies demonstrate an absence of flow dorsal to the upper cervical cord. There is absence of flow in the region of the cisterna magna as well.

Congenital Abnormalities

Recent advances in neuroimaging of the pediatric spine have dramatically improved our ability to evaluate children noninvasively. Plain films, ultrasound, CT, and MRI are the mainstay. Spinal cord tumors and infection are diagnosed earlier and more easily in children, and those with complex congenital anomalies of the brain and spinal cord can be completely evaluated both before and after surgery. Plain films of the spine are most helpful in evaluating trauma, spinal instability, scoliosis, and spina bifida. Certain primary bone tumors of the spine are better detected initially on routine radiographs.

Ultrasound of the spinal contents in infants with a skin dimple or nevus when there is no or minimal neurological deficits is useful as the first imaging modality after plain films. It is also useful intraoperatively for precise localization of intramedullary tumors, cysts, or syringomyelia. It may be performed through surgical bony defects or congenitally absent posterior elements.

CT is excellent in assessing bone pathology and, with intrathecal contrast enhancement, in delineating cord anatomy. As with other disease states, MRI provides the most accurate soft tissue, bone, and fluid depiction of anatomy and pathology. In a child, however, greater preparation and teamwork are required because motionless examination is critical. General anesthesia or sedation is often needed in the younger age groups.

Spinal dysraphism refers to a spectrum of disorders in which there is defective midline closure of neural, bony, or other mesenchymal tissues. The “open” dysraphic states include myelocele, myelomeningocele, hydromyelia, Chiari II malformations, hemimyelocele, and myeloschisis. The closed dysraphic states include entities such as dermal sinus, lipomyelomeningocele, tight filum terminale, meningocele, myelocystocele, diastematomyelia, neurenteric cyst, slit notochord, and developmental tumors such as spinal lipomas.

Patients with spinal dysraphism usually have cutaneous stigmata. Plain-film findings may show formation or segmentation anomalies, congenital scoliosis or kyphosis, canal widening, or osseous defects. Multislice CT with multiplanar reformatting is helpful in definitive preoperative bone assessment. As noted earlier, in infants with a skin dimple or nevus and no or minimal neurological deficits, ultrasound is of value as the first imaging modality after plain films and is also used for intraoperative surgical guidance. MRI is the definitive modality for diagnosis, surgical planning, and follow-up.

In addition to ultrasound, MRI has also proved useful in fetal imaging of congenital abnormalities. For disorders of the neural tube that have been diagnosed with ultrasound, MRI can provide additional information in determining the level and contents of a suspected myelomeningocele and demonstrating the presence and severity of associated Chiari II malformations and ventriculomegaly in the presence of open spinal neural tube defects. MRI can be used to assess the morphology of the brainstem and cortex and provide useful prognostic information (Fig. 18-49).

FIGURE 18-49 Sagittal T2-weighted image through a fetus demonstrating spinal dysraphism with herniation of the thecal sac (arrow). Resolution is not adequate for determining whether neural elements are present.

Open dysraphic states are usually closed surgically within 24 to 48 hours of birth. Subsequently, patients are referred for imaging to evaluate any change in neurological status that may be related to Chiari II malformations or other associated abnormalities or after myelomeningocele closure or shunted hydrocephalus (Fig. 18-50).

FIGURE 18-50 Myelomeningocele demonstrated on sagittal (A) and axial (B) T1-weighted images through a dysraphic defect in the distal region through which both neural elements and cerebrospinal fluid have herniated. The placode is adherent dorsally, and the nerve roots are splayed more ventrally.

Distinction between hydromyelia and syringomyelia is difficult if not impossible to make by imaging studies, and hence the term hydrosyringomyelia is more commonly used. MRI demonstrates intramedullary CSF signal intensities within the spinal cord, which may be diffusely enlarged with sacculations or septations.

MRI signal intensity characteristics are useful in the assessment and differentiation of a series of disorders. In lipomyelomeningocele or lipomas, MRI demonstrates the relationship of a T1 hyperintense lipoma and neural elements and is useful for surgical planning (Fig. 18-51).

FIGURE 18-51 Lipomyelomeningocele seen on sagittal T1-weighted (A), fat-suppressed T1-weighted (B), and T2-weighted (C) images of the lumbosacral region. A tethered spinal cord terminates in a lipomeningocele in the lumbosacral region that is communicating with a presacral lipoma. A syrinx cavity is present in the tethered cord in the upper lumbar region.

Dermal sinuses, or epithelial tracks extending from the skin surface into deeper tissues, often end in a dermoid/epidermoid or lipoma. The patient may have infection secondary to an abscess. On MRI a dermal sinus generally appears as a relative hypointensity (Fig. 18-52). Uninfected dermoid/epidermoid tumors do usually not enhance and may exhibit restrictive diffusion. They may be hypointense to CSF on MRI sequences, but their characteristics are variable. If infected, paramagnetic contrast material may be helpful in that it usually enhances.

FIGURE 18-52 Dermal sinus visualized on sagittal T2-weighted (A) and T1-weighted (B) images of the lumbar region. A sinus track (black arrow) passing intraspinally is associated with a tethered cord at the L3 level. There is a syrinx cavity within the cord at the L1 level (white arrow).

Neurenteric cysts are typically seen as a posterior mediastinal mass associated with vertebral body abnormalities on plain films. They are usually found in the lower cervical or thoracic region. On MRI the signal intensity is variable, depending on the contents of the cyst. If serous, they may appear similar to CSF, but if they contain mucoid secretions, they may be hyperintense on T1-weighted MRI.

Diastematomyelia, the most common neurenteric or split notochord entity, may require several imaging modalities for full anatomic delineation (Fig. 18-53). Plain films may show spina bifida, intersegmental laminar fusion, anomalies of the vertebral bodies, and kyphoscoliosis. For septal definition, axial SE T2-weighted MRI, axial gradient echo T2-weighted MRI, CT, or CT myelography may be necessary. Other commonly associated abnormalities, including a thickened filum, developmental tumors (lipomas, dermoid/epidermoid), and hydromyelia, can be depicted.

FIGURE 18-53 Anteroposterior radiograph of the thoracolumbar region (A) and an axial CT myelogram (B) demonstrating spinal dysraphism with widening of the interpedicular distance in the lumbar region and anomalies of the vertebral bodies and posterior elements. Axial T1-weighted MRI at a slightly higher level (C) demonstrates a septum separating two spinal cords. There appears to be two separate dural coverings.

There may be a bone or cartilaginous spur within the cleft in the cord. For accurate definition, a combination of sagittal, coronal, and axial MRI, CT, or CT myelography may be necessary. The conus is located below the L2 level in more than 75% of patients, and there is often an associated thickened filum. Hydromyelia is present in approximately 50% of patients, and the spinal column is almost always abnormal.

Vascular Malformations of the Spine and Spinal Cord

Spinal vascular malformations are a heterogeneous group of non-neoplastic vascular abnormalities that account for 3% to 16% of spinal mass lesions. The current classification system is based on the angioarchitecture and hemodynamics of the lesion as defined by spinal angiography. The major groups of spinal vascular malformations include spinal-dural arteriovenous fistulas (SDAVFs) (Fig. 18-54), spinal cord arteriovenous malformations (SCAVMs) (Fig. 18-55), perimedullary spinal cord arteriovenous fistulas (SCAVFs) (Fig. 18-56), and cavernous malformations (Fig. 18-57).¹¹¹

FIGURE 18-54 Type I dural arterial venous fistula seen on T1-weighted contrast-enhanced (A), T2-weighted (B), and magnetic resonance angiographic (MRA) (C) images of the distal thoracic region. Multiple enhancing vessels are noted along the surface of the thoracic cord on the T1 contrast-enhanced image. Diffuse increased hyperintensity is apparent within the cord on the T2-weighted image, as well as multiple small focal flow voids along the surface. MRA demonstrates an enlarged draining venous plexus. Axial T1-weighted (D) and T2-weighted (E) images of the thoracic cord demonstrate the decreased signal intensity on T1 and increased signal intensity on T2 within the spinal cord centrally secondary to the cord edema. Sagittal (F) and coronal (G) multiplanar reformatted images from a CT angiogram of the thoracic cord. Note the prominent radicular vessel leading to the dural fistula, which then shunts into the venous plexus on the surface of the thoracic cord (arrow). A spinal arteriogram (H) demonstrates a dural fistula fed from an intercostal artery that shunts into the venous plexus along the surface of the cord.

FIGURE 18-55 Type II arteriovenous malformation (AVM) shown on sagittal T2-weighted images (A and B) and a spinal arteriogram (C) of the cervical cord. The sagittal T2-weighted images demonstrate an intramedullary mass with heterogeneous signal at the C2 level and flow voids along the surface of the cervical cord. A lateral view from the spinal arteriogram demonstrates opacification of the AVM nidus fed from the anterior spinal artery.

FIGURE 18-56 Type IV arterial venous malformation (perimedullary fistula) visualized on T1-weighted (A), T1-weighted contrast-enhanced (B), and T2-weighted (C) images at the level of the distal thoracic cord. On the T1-weighted image there is subtle decreased signal intensity within the thoracic cord suggestive of edema. On the contrast-enhanced T1-weighted image, multiple enhancing vessels are seen along the surface of the thoracic cord. On the T2-weighted image there is increased signal within the cord secondary to edema and a focal low signal that presumably reflects compressed perimedullary vessels. An anteroposterior spinal arteriogram (D) and CT angiogram (E) demonstrate filling of the vascular malformation on the surface of the cord by a radicular artery. In contradistinction to type I, shunting in this pathology is at the cord level and not at the dura.

FIGURE 18-57 Cavernous angioma shown on sagittal T1-weighted (A) and T2-weighted (B) images at the level of the cervical medullary junction. There is a heterogeneous intramedullary mass on the T1-weighted image with a rim with decreased signal intensity. On T2, there is a heterogeneous increased central core surrounded by a larger area of decreased signal intensity reflecting signal loss related to a susceptibility effect from remote blood by-products.

SDAVFs are the most common spinal vascular malformation and represent up to 80% of all spinal malformations. Males are affected in 80% to 90% of cases, and they are usually seen initially in the fourth or fifth decade of life. Anatomically, SDAVFs are arteriovenous shunts in the dura, most commonly adjacent to the intervertebral foramen or in the nerve root sleeve. The arterial supply usually arises from a dural branch of the radicular artery and drains directly into the pial veins of the cord via an intradural vein. This abnormal drainage pattern results in venous hypertension and spinal cord edema. Hemorrhage with SDAVFs is rare. There are four basic types of arterial vascular malformations of the spinal cord: I, AV fistulas between dural branches of the spinal ramus of the radicular artery and an intradural medullary vein; II, intramedullary glomus malformation; III, extensive juvenile malformation; and IV, intradural perimedullary AV fistulas.

MRI findings of SDAVFs include cord edema and engorgement of the pial veins. Abnormal signal intensity is often seen in the cord on T1-weighted, proton density, and T2-weighted images, although these findings are nonspecific and may occur with inflammatory and demyelinating conditions as well. A more specific finding is the presence of dilated pial vessels, best seen on T2-weighted images as flow voids within the high-signal CSF surrounding the cord.

Although MRI may be very suggestive of SDAVF, the “gold standard” for diagnosis remains spinal angiography. Selective angiography can confirm the diagnosis and provides additional information such as the level and side of the fistula, information that is essential for treatment. Occasionally, the location of the fistula may be remote from the site of greatest edema; therefore, multiple levels should be examined to ensure that the fistula is not missed. More recently, CTA has emerged as a useful, noninvasive adjunct for the diagnosis of SDAVF. In many cases the level of the fistula can be determined from CTA and be used to guide the angiographer to the appropriate levels to study selectively.

Intradural SCAVMs are thought to be congenital lesions. They occur most often in the thoracolumbar region of the cord but may arise from any level. The nidus of the AVM is located on or within the parenchyma of the spinal cord itself. The arterial supply arises from either the anterior or posterior spinal arteries. Dilated draining veins are frequently seen extending both rostrally and caudally from the nidus. Two subtypes are classically described. The more common “glomus”-type SCAVM consists of a relatively compact nidus confined to the spinal cord itself. In “juvenile”-type SCAVM, the nidus involves the spinal cord with extramedullary and extraspinal extension. Hemorrhage, either subarachnoid or intramedullary, occurs in approximately half the patients with SCAVMs.

MRI is the study of choice for noninvasive imaging of SCAVMs. Enlarged arterial vessels and an intramedullary nidus are seen as flow voids on MRI. Evidence of previous hemorrhage can be seen in many cases. T2-weighted sequences may show perinidal increased signal representing edema, gliosis, or areas of infarction. CTA may be helpful in delineating the location of the feeding vessels. However, conventional spinal angiography of SCAVMs continues to be the gold standard and is considered mandatory before any intervention. Additional information that may be obtained only from spinal angiography includes the location of feeders and draining vessels, the presence of perinidal aneurysms, and the location of normal blood vessels rostral and caudal to the lesion.

Perimedullary AV fistulas are direct arterial-to-venous shunts located on or within the spinal cord. Arterial feeders arise from the anterior or posterior spinal arteries. They are thought to be congenital, usually arise in the second to fourth decades, and are manifested as progressive radiculomyelopathy predominantly in the lower extremities. Spinal subarachnoid hemorrhage is not uncommon.

MRI is the imaging modality of choice for the diagnosis of SCAVFs. The presence of enlarged feeding and draining vessels demarcated by flow voids is the characteristic finding. Cord signal abnormalities and evidence of previous hemorrhage may be present. The lack of a true nidus may help differentiate SCAVFs from SCAVMs. Once a diagnosis is suspected based on MRI findings, spinal angiography is recommended for confirmation of the diagnosis and further characterization of the angioarchitecture for planning treatment.¹¹⁵

Cavernous malformations are slow-flow vascular malformations with no AV shunting. They are usually intraparenchymal and can occur throughout the spinal cord, cauda equina, and filum terminale. In most cases they are well-demarcated lesions with surrounding hemosiderin deposition. MRI is the imaging modality of choice for cavernomas and is often diagnostic because of their relatively unique imaging characteristics. A rim of low signal intensity usually surrounds the lesion and represents iron storage products. The lesion is seen as heterogeneous signal abnormalities within the cord on both T1- and T2-weighted images secondary to blood products of various ages within the lesion. Angiography is not helpful in the diagnosis of cavernous malformations (see Fig. 18-57).^116–118

Cord infarction or ischemia is a devastating entity best identified by MRI. Patients generally have a relatively rapid onset of paraparesis or quadriparesis. The most common causes of spinal cord infarction are related to embolic disease resulting from therapeutic intervention, aortic dissection, and hematologic diseases. MRI usually demonstrates nonspecific increased signal intensity on T2-weighted images involving variable lengths of the cervical or thoracic cord and may demonstrate restricted diffusion. T1-weighted images are often normal or demonstrate slight expansion of the cord (Fig. 18-58).

FIGURE 18-58 Spinal cord infarction noted on sagittal (A) and axial (B) T2-weighted images through the lower thoracic cord. The T1-weighted images were normal. The T2-weighted images demonstrate increased signal intensity within the distal thoracic cord.

Magnetic Resonance Imaging

Frequently, the use of gadolinium can aid in the distinction between “normal” postoperative changes and pathology. Scar tissue consistently enhances and is often associated with retraction of the thecal sac and the absence of a mass effect (Figs. 18-59 and 18-60). It may, however, show a mass effect in some situations, and it may be contiguous with the disk space. Disk, in contrast, does not usually enhance centrally within the first 20 minutes after the injection of paramagnetic contrast medium. Peripheral enhancement is common both preoperatively and postoperatively (Fig. 18-61). Disk enhances centrally if delayed images are obtained. This distinction can aid in the diagnosis of recurrent/residual disk herniation.

FIGURE 18-59 Axial T1-weighted (A), contrast-enhanced T1-weighted (B), and T2-weighted (C) images through the L4-5 disk in a patient after left-sided laminectomy and diskectomy. The immediate preoperative study demonstrates aberrant soft tissue (arrows) in the anterior and left lateral epidural space. This enhances in a relatively homogeneous fashion after administration of contrast material. There is mild deformity of the thecal sack. Eight weeks postoperatively, there has been reexpansion of the thecal sack and some retraction of the aberrant soft tissue, which still continues to demonstrate homogeneous enhancement. This is epidural fibrosis.

FIGURE 18-60 Scar tissue seen on axial T1-weighted images before (A) and after (B) contrast enhancement in a patient after a left-sided laminectomy. On the precontrast T1-weighted axial image there is aberrant soft tissue (arrow) in the left anterior epidural space. After contrast administration there is relative homogeneous enhancement of this aberrant soft tissue (arrow) without evidence of a mass effect. This is scar tissue.

FIGURE 18-61 Recurrent disk herniation demonstrated on axial and sagittal T1-weighted (A), contrast-enhanced T1-weighted (B), and T2-weighted (C) images through the lumbar spine in a patient after a right-sided laminectomy at the L5-S1 level. On the pre–contrast-enhanced T1-weighted axial and sagittal images there is aberrant soft tissue (arrows) in the right anterior and lateral epidural space. After the administration of contrast material, there is mild peripheral enhancement of aberrant soft tissue (arrows) surrounding a nonenhancing core.

Primary differential considerations, besides scar and disk, are epidural hematoma and abscess (Fig. 18-62). In the acute period an epidural hematoma may not have peripheral enhancement, and the signal intensity characteristics of the soft tissue mass may be that of blood by-products. Fortunately, the characteristic signal intensity of blood by-products that one is accustomed to from brain imaging does always translate to the spine. In particular, the high signal intensity noted on T1 is not as frequently encountered. Epidural abscesses usually have peripheral enhancement and in the absence of osseous changes may be difficult to distinguish from recurrent disk herniation or hematoma.

FIGURE 18-62 Postoperative epidural hematoma (arrows) visualized on sagittal T1-weighted (A), T2-weighted (B), and contrast-enhanced T1-weighted (C) images of the lumbar spine in a patient after laminectomy. A large soft tissue mass (arrows) is situated in the anterior epidural space behind the body of L3. It has soft tissue signal intensity on T1 but markedly decreased signal intensity on T2. After administration of contrast agent there is minimal peripheral enhancement. Note the posterior laminectomy defect. This was found to be an epidural hematoma at surgery.

The imaging findings of arachnoiditis are a reflection of the adhesion and clumping of nerve roots after inflammation. The first type is manifested by clumping of individual traversing nerve roots within the thecal sac (Fig. 18-63). The second type is a featureless sac in which individual nerve roots are not identified (Fig. 18-64). There is often absent or blunting of nerve root sleeve filling. The third type demonstrates soft tissue masses that cause filling defects within the subarachnoid space and may block the flow of myelographic contrast media. These findings of arachnoiditis are evident on myelography, CT myelography, and MRI. The different appearances of arachnoiditis are probably points on a spectrum rather than different entities or stages. There may be minimal nerve root and dural enhancement after contrast administration or no enhancement at all. Differential considerations include crowding of nerve roots with stenosis, leptomeningeal seeding of neoplasms, or meningitis^119,120 (Fig. 18-65).

FIGURE 18-63 Group I arachnoiditis. Axial T1-weighted (A) and T2-weighted (B) images of the lumbar spine show evidence of a laminectomy defect. Portions of the traversing nerve roots are clumped posteriorly (arrows).

FIGURE 18-64 Group II arachnoiditis demonstrated on an axial T1-weighted spin echo image through the lumbar spine. There is a laminectomy defect and a thin rim of soft tissue signal intensity on the thecal sac that appears adherent peripherally. No individual traversing nerves are identified centrally (arrows).

FIGURE 18-65 Group III arachnoiditis. Sagittal multiplanar reformatted CT myelogram (A), axial CT-myelogram (B), and a T1-weighted spin echo image (C) show complete blockage of the flow of contrast material at the L2-3 level. A laminectomy defect is seen together with evidence of calcification along the posterior margin of the thecal sac. The axial CT myelogram demonstrates a soft tissue mass within the thecal sac, as evidenced by soft tissue signal intensity on the axial T1-weighted image.

In patients with metallic implants, screw artifact can make visualization of the foramina and nerve roots very difficult, although the central canal can still be seen adequately in most cases. Metal artifact can be minimized by using a variety of techniques, including fast SE sequences, smaller voxel size, enlargement of the field of view, or the use of higher readout bandwidth. Moreover, geometric distortion occurs along the frequency in the coded direction. Frequency in coding directed parallel to the axis of an implant improves image quality except at the tip.

Computed Tomography

CT and CT myelography are essential tools for imaging postoperative patients. Screw artifact can be minimized to allow greater visualization of the foramina and lateral recesses. The presence of pseudarthrosis in spinal fusion patients is best assessed with fine-cut CT. Postprocessing of the images with sagittal and coronal reconstructions can aid in the assessment of fusion. Hardware placement and integrity can also be assessed in this manner. Myelography can enhance the ability to visualize compressive and intrinsic lesions of the neural elements.

Plain Films

Although largely supplanted by CT for the assessment of spinal fusion, plain radiographs can be very useful in postoperative patients. They are a cheap, fast, effective way to assess the position and competence of spinal hardware. Moreover, 3-ft anteroposterior and lateral films are essential in the evaluation of deformity and correction of deformity. Dynamic imaging with flexion and extension images is also helpful in the assessment of spinal stability and the presence of pseudarthrosis.

Computed Tomography

CT scanning has been shown to be significantly more sensitive and more time efficient than plain films for detection of cervical fractures in the setting of acute spinal trauma. The widespread availability of high-quality CT scanners, the ability to rapidly acquire images, and the ability to construct multiplanar and 3D images make CT the ideal screening test for cervical fracture. The sensitivity of CT scanning for acute cervical fractures ranges from 90% to 99% with specificities of 72% to 89%. The sensitivity of plain films in the acute setting ranges from 39% to 94%.^122–124 Moreover, a number of studies have demonstrated limitations in the ability of plain films to detect injuries to the upper cervical spine and occipital condyles.^125,126 Because of its greater sensitivity and wide availability, CT is quickly becoming the initial screening modality for bony injury in the cervical spine.

CT has been shown to be superior to plain radiography in assessing fractures of the thoracic and lumbar spine as well. Campbell and coauthors¹²⁷ reported that 20% of burst fractures diagnosed by CT were misdiagnosed as stable wedge compression fractures on plain films. CT is better at detecting fractures of the dorsal elements, malalignment, and intracanalicular fragments.

Plain Films

CT has largely supplanted plain-film imaging as the modality of choice for the evaluation of osseous injury to the spine. However, plain films can be very useful in the evaluation of trauma patients, particularly when CT is unavailable.

Plain films are also critical in the evaluation of instability in the absence of bony injury. Stability of the cervical spine is best assessed with dynamic imaging that includes flexion and extension views. An increase in the atlantodental interval or greater than 3.5-mm horizontal displacement of the vertebral body between flexion and extension can be indicative of spinal instability. This examination should be performed only on alert, cooperative patients without neurological injury or radiographic evidence of unstable spinal injuries. Frequently, cervical mobility is limited by pain and muscle spasm at the time of the initial injury. Flexion and extension views may be more helpful when performed in a delayed fashion 7 to 10 days after the injury.¹²⁸

Magnetic Resonance Imaging

MRI provides the best evaluation of soft tissue pathology and is the only means of directly evaluating the spinal cord. The information obtained is often complementary to evaluation of the bony structures by CT. Information about disk herniations, hematoma formation, and ligamentous and muscular injury can be instrumental in determining the appropriate treatment for the patient. MRI is indicated in a trauma patient with a neurological deficit or when there is suspicion of a soft tissue or vascular injury. STIR sequences can detect bone edema and aid in differentiating the acuity of fractures. Heavily T2-weighted sequences can be used to evaluate for nerve root avulsions and pseudomeningocele development. MRA and fat-saturated T2 sequences can be used to screen for vascular injuries. Ligamentous and soft tissue injuries are best visualized on fat-saturated T2 images. The normal anterior and posterior longitudinal ligaments are seen as continuous hypointense lines along the ventral and dorsal aspects of the vertebral bodies. In the presence of soft tissue injury, areas of increased T2 signal or discontinuity of the ligament may be seen. MRI is the modality of choice for the evaluation of a variety of posttraumatic conditions, including myelomalacia, cord tethering, syrinx formation, and the presence of dural AV fistulas.¹²⁸

Although MRI can be very useful in the setting of acute trauma, it has not gained widespread use because of a variety of factors. The requirement for special ventilators and monitors can often make it difficult or even impossible to image critically ill trauma patients. MRI in patients with bullet fragments within the spine remains controversial. Most bullets are nonferrous; however, the composition of the embedded projectile is rarely known in the acute setting. There is a theoretical risk that a ferrous fragment may become mobile in the magnetic field and result in greater damage to surrounding structures, although this has never been reported. Finally, limited access to MRI scanners and qualified technicians decreases the usefulness of MRI in the acute setting at many centers.