Limbus vertebra was first described by Schmorl in 1927 (Schmorl and Junghanns 1971) and results from the herniation of the nucleus pulposus into the body of the adjacent vertebra near its anterior margin, causing separation of a triangular bony fragment (Fig. 9.3). This has been well demonstrated on diskographic procedures, where intradiscal contrast is seen to extend between the fragments of a limbus vertebra. It is postulated to arise from mechanical stress factors during flexion, similar to a Schmorl node (Fig. 9.4), which occurs in the central portion of the endplate rather than marginally (Hellstadius 1949). It is most commonly seen in the anterosuperior corner of a single vertebral body in the mid-lumbar spine. Posterior herniations are rare but can be the cause of lumbosacral radiculopathy and cauda equina compression, especially in adolescents and young adults (Goldman et al. 1990). The radiographic appearance of a limbus vertebra may mimic a fracture, infection, or tumor, particularly in children, where it can appear as an irregular and apparently destructive process along the anterior margin of the vertebra. In adults, the identification of a limbus vertebra or Schmorl node is less difficult as they commonly display sclerotic and well-defined margins, in contrast to the appearance of diskitis or acute fracture. The ability to recognize this condition is important to avoid unnecessary further investigations or biopsy (Ghelman and Freiberger 1976).

The first two cervical vertebrae have unique anatomical morphologies compared to the remaining vertebrae. Developmental anomalies can result in variant anatomy, which are often mistaken for fractures in the setting of spinal trauma. It is thus important for the radiologist to be familiar with these anomalies. The most commonly seen anomaly in the atlas is a posterior cleft or rachischisis, being found in 4 % of adult autopsy specimens. This can be seen on radiographs and usually confirmed on CT as a well-corticated linear defect in the midline of the posterior arch. Rarely, there is coexistence of an anterior cleft, resulting in a “split atlas” (Smoker 1994) (Fig. 9.5).

Developmental anomalies of the axis involve the odontoid process in the vast majority of cases and may similarly be mistaken for fractures. A persistent ossiculum terminale or Bergmann ossicle results from the failure of fusion between the terminal ossicle and the rest of the odontoid process, which normally takes place by the age of 12 years. It appears as a small triangular or pyramidal ossicle just above the odontoid process and can thus mimic a type I odontoid peg fracture. Differentiation is normally made on CT, where well-corticated margins of the ossicle are demonstrated. An os odontoideum is an independent osseous structure seen in place of the normal odontoid peg above the body of the axis (Fig. 9.6). This can simulate a type 2 odontoid peg fracture. The differentiating features lie in the outline of the upper border of the body of the axis, as well as the anterior arch of the atlas. Os odontoideum is usually associated with a well-corticated convex upper border of the body of the axis, as well as a hypertrophied anterior arch of C1, which appears rounded on lateral or sagittal views rather than its usual semilunar shape. A type 2 odontoid peg fracture, on the other hand, shows a flat, non-corticated upper border of the body of the axis and a normal anterior arch of C1 (Smoker 1994) (Fig. 9.7).

Filum terminale lipoma or fatty filum terminale (Fig. 9.8) is a normal variant seen as a thin linear strip of fatty tissue in the dorsal half of the thecal sac, extending for a varying length below the conus medullaris to the sacrum, in an asymptomatic patient. It demonstrates absence of contrast enhancement as well as low signal on fat-suppressed images. It is differentiated from a tethered cord by the normal position of the conus medullaris and the absence of thickening of the filum terminale. In a symptomatic patient, intraspinal lipoma needs to be considered. This is characterized by a larger lipomatous mass, with an associated thickening of the filum terminale of >2 mm (Ross et al. 2004). Postmortem studies have reported a 4–6 % incidence of occult fibrolipomas of the filum terminale in what were thought to be otherwise normal spinal cords (Brown et al. 1994). Epidural lipomatosis is a separate entity commonly associated with obesity and chronic steroid use. Diffuse proliferation of fatty tissue within the dural sac can cause neural compression, producing neurologic symptoms as well as back pain (Tehranzadeh et al. 2000).

Ventriculus terminalis or terminal ventricle is the cystic dilatation of the distal central spinal cord canal, sometimes referred to as the fifth ventricle. It is located within the conus medullaris with cerebrospinal fluid (CSF) signal intensity on MRI, typically 2–4 mm in diameter and rarely exceeding 2 cm in length. It is often an incidental finding and is rarely the cause of symptoms. It is important to distinguish this from cystic neoplasm or hydrosyringomyelia on imaging. The presence of calcification, septation, nodule, enhancement, eccentric location, or signal changes in the adjacent spinal cord would be considered suspicious for a cystic neoplasm. Hydrosyringomyelia is only restricted to the distal cord in 2.5 % of cases and usually extends for a longer length compared to the terminal ventricle. A terminal ventricle may be indistinguishable from a small localized syrinx; however, the subsequent management for either remains close monitoring of neurologic symptoms and interval follow-up imaging. One must be wary of truncation or phase ghosting artifacts which can also mimic the appearance of a dilated terminal ventricle or syrinx (Coleman et al. 1995; Ross et al. 2004; Ciappetta et al. 2008).

Compression fractures of the vertebral bodies are seen with increasing incidence in the elderly population, with osteopenia or osteoporosis being the most common cause. The spine is also a common site for metastatic disease in the same age group, accounting for up to 39 % of all bone metastases and which may result in pathological fractures (Yuh et al. 1989). Chronic, nonneoplastic compression fractures are easily recognized on MRI by their lack of abnormal marrow signal. However, an acute or subacute fracture may have imaging features similar to a malignant compression fracture, making differentiation between the two difficult. This is of great clinical significance as it may impact on the staging, treatment plan, and prognosis of a patient with a known primary malignancy (Jung et al. 2003).

There are several useful imaging signs described in the literature that can assist the radiologist in the differentiation of osteoporotic from malignant fractures. In osteoporotic fractures, there is usually preservation of some normal fatty marrow signal within the vertebral body (Fig. 9.9). Most vertebral metastases do not result in compression fractures until the entire body is infiltrated by tumor, causing structural weakening from bone destruction (Yuh et al. 1989). The involvement of the pedicles or posterior elements is also a useful sign. Benign compression fractures rarely extend into the pedicle or posterior elements. Fluid or gas within the vertebral body can be seen as vacuum clefts on CT or conventional radiographs and is usually a sign of a benign fracture as it implies the presence of a cavity that is not filled with neoplastic tissue (Baur et al. 2002). The ability to discern a discrete fracture line of low T1 and T2 signal intensity within the vertebral body has also been reported to be a feature of osteoporotic fracture. The presence of an associated soft tissue component, including epidural mass and particularly a focal paraspinal mass, is suggestive of a neoplastic process (Fig. 9.10). Involvement of the cervical or upper thoracic (T1–T5) spine is more often seen in metastatic fractures. A convex posterior border of the fractured vertebra is somewhat helpful; however, it can also be seen in up to 20 % of osteoporotic fractures (Jung et al. 2003; Griffith and Guglielmi 2010).

The presence of a known primary malignancy is not a useful parameter, as up to one-third of vertebral fractures in this group of patients are due to osteoporosis and not metastases (Fornasier and Czitrom 1978). Administration of intravenous gadolinium contrast agent is also of limited use as both acute osteoporotic and neoplastic fractures can show enhancement. Several studies have demonstrated the value of using diffusion-weighted imaging (DWI) in differentiating benign from pathological fractures of the vertebra (Raya et al. 2006; Karchevsky et al. 2008). Neoplastic fractures tend to show increased DWI signal due to higher cellularity and reduced extracellular fluid component. However, this appearance may be confounded by the presence of “T2 shine-through” effect. The use of the apparent diffusion coefficient (ADC) as a quantitative measure of diffusivity of water molecules has been proposed to address this. Several studies have found a cutoff value of more than 1.0 × 10⁻³ mm²/s as a predictor of osteoporotic fracture (Chan et al. 2002; Balliu et al. 2009; Tang et al. 2007). The described features are summarized in Table 9.1. The application of a combination of these signs and imaging techniques aids the radiologist in more accurate discrimination between benign and pathological fractures of the spine. However, a biopsy may still be required for definitive diagnosis in equivocal cases.

Image interpretation of a postsurgical spine is complex and challenging. Many factors have to be taken into consideration, including the type of surgery performed, time since the procedure, implants used, preoperative imaging findings, as well as the nature of the current symptoms (Van Goethem et al. 2002). The presence of metallic implants further complicates matters when imaging with CT or MRI, producing streak or susceptibility artifacts which obscure vital structures and may even render the study uninterpretable or nondiagnostic. Several techniques have been employed to reduce the beam-hardening artifacts from metallic prostheses on CT. These include the use of high peak voltage and tube current, as well as narrow collimation and thin sections during image acquisition. The use of thick sections with lower kernel values during reformatting can also reduce streaking (Thakkar et al. 2012). The early days of MRI of the instrumented spine were plagued by severe artifacts, limiting its use. This has improved in recent times with the use of modern pulse sequences, as well as with the introduction of titanium implants which resulted in reduced artifacts (Viano et al. 2008). When imaging with MRI, the following parameters may help to attenuate the effect of metallic prostheses:

Spinal infections are an important aspect of spine imaging and account for 2–4 % of all skeletal infections (Maiuri et al. 1997). They may arise from distant septic foci via hematogenous spread or by direct extension from infections in adjacent tissues. Urinary tract infections are a well-known source of infection. Direct inoculation can also occur as a result of trauma or following spinal surgery or procedures. Staphylococcus aureus is the most common organism isolated, accounting for at least a third of all cases (Mylona et al. 2009; Yoon et al. 2010). The lumbar spine is most often affected (50 %), followed by the thoracic spine (35 %), with the remaining 15 % occurring in the cervical spine (Van Tassel 1994). MRI is the modality of choice for the assessment of suspected infectious spondylitis due to its high sensitivity and specificity (Tins and Cassar-Pullicino 2004). However, spondylitis may show atypical features, which make accurate diagnosis difficult. Furthermore, degenerative and noninfectious inflammatory conditions may have similar imaging features and can mimic spondylitis. Tuberculous spondylitis occasionally demonstrates distinctive features which enable differentiation from pyogenic spondylitis (Hong et al. 2009).