Journal Logo


Magnetic Resonance Imaging of the Pediatric Spine

Khanna, Jay A. MD; Wasserman, Bruce A. MD; Sponseller, Paul D. MD

Author Information
Journal of the American Academy of Orthopaedic Surgeons: July 2003 - Volume 11 - Issue 4 - p 248-259
  • Free


Magnetic resonance is an excellent modality for imaging pathologic processes in the pediatric spine. It allows high-resolution views of not only osseous structures (including the vertebral body, spinal canal, and posterior elements) but also soft-tissue structures (including the spinal cord, intervertebral disk, and nerve roots). Magnetic resonance imaging (MRI) can show these structures in various planes using different pulse sequences that allow optimal characterization of the tissues in and around the pediatric spine. Indications for MRI in children (<18 years) are gradually expanding as technology improves. Properly interpreting MRI scans in these age groups depends on understanding the MRI appearance of the normal pediatric spine anatomy at various stages of development. For entities such as spinal dysraphism, left thoracic curves, and juvenile scoliosis, specific recommendations can help clinicians use MRI effectively.

MRI Techniques

The major factors that influence the MRI appearance of various tissues are the density of protons in the tissue, the chemical environment of the protons, and the magnetic field strength of the scanner. Unlike computed tomography (CT), which produces images based on the density of various tissues, MRI produces images based on free water content and on other magnetic properties of water, yielding superior soft-tissue contrast.

Various sequences are produced by manipulating the strength of the radiofrequency (RF) pulses, the interval between the pulses, the repetition time (TR), and the echo time (TE), that is, the time between applying the RF pulse and measuring the signal emitted by the patient. By manipulating these variables, the images can be weighted to emphasize the T1, T2, gradient-recalled echo, or proton density characteristics of a tissue. T1-weighted images allow evaluation of anatomic detail, including that of osseous structures, disk, and soft tissues. T2-weighted images are used primarily to evaluate the spinal cord and to enhance lesion conspicuity. A gradientrecalled echo sequence typically is used when thin axial images are needed, such as for evaluating foraminal narrowing in the cervical spine, because its three-dimensional acquisition allows for very thin sections.

Standard pulse sequences for spinal imaging include spin echo T1-weighted images and fast spin echo (FSE) T2-weighted images. The FSE technique allows acquisition of scans without prolonged imaging times. Because cerebrospinal fluid (CSF) is bright on T2-weighted images and the spinal cord retains its intermediate signal, the images maximize the contrast between CSF and neural tissue, allowing optimal delineation of the spinal cord and nerve roots. T2-weighted images are very sensitive to pathologic changes in tissue, including any processes in which cells and the extracellular matrix have an increase in water content. This pathologic change is usually shown as an increase in signal intensity on T2-weighted images.

The signal from fat may be suppressed by a variety of techniques, including chemical saturation of its signal or application of an inversion pulse, and imaging at a short time of inversion (TI) when there is no fat signal present (short TI recovery [STIR]). Chemical suppression typically is used in sequences that result in high fat signal, such as FSE T2-weighted images or postcontrast T1-weighted images. Fat suppression is of little value for noncontrast T1-weighted images because the signal from most pathologic lesions, whether inflammatory, neoplastic, or infectious, is often low and better visualized because of contrast against the adjacent bright fat signal. Fat suppression on postcontrast T1-weighted images of the vertebral body is useful in adults who have fatty transformation of marrow. Fat-suppressed images may be particularly useful for evaluating ligamentous injuries or lesions involving the paraspinal tissues. The usefulness of STIR imaging is more limited because the imaging parameters are restricted and cannot be optimized to maximize contrast between adjacent tissues of interest.

Gradient-recalled echo images appear to be T2-weighted because CSF is relatively bright; however, parenchymal lesions typically are more conspicuous on FSE T2-weighted images. The gradient-recalled echo sequence is sensitive to local inhomogeneities of the magnetic field, and signal loss is exaggerated in the presence of such inhomogeneities. Field inhomogeneities may be caused by metallic implants (eg, pedicle screws or paraspinal rods), differences in the magnetic susceptibilities of adjacent tissues (eg, air-tissue interfaces), and paramagnetic substances (eg, gadolinium). Bloodbreakdown products cause local field distortions resulting in signal loss, making this technique very sensitive for the detection of blood.

Open MRI systems are being used more frequently, especially for children. These systems have notably lower field strengths than do closed systems and therefore usually produce studies of inferior overall quality, especially of the spine. However, open MRI systems allow easier access to the sedated or otherwise compromised patient. Young patients and patients with claustrophobia have access to parents and the environment, making the procedure less intimidating. However, whenever possible, spinal MRI should be done using closed, 1.5-T systems.

Pediatric Sedation Protocols

Sedation is often required for successful MRI in young children. Many studies have evaluated specific sedation protocols.1,2 The American Academy of Pediatrics (AAP) has published guidelines for the elective sedation of pediatric patients,3,4 but compliance with these guidelines is not mandatory. The AAP has stated that careful medical screening and patient selection by knowledgeable medical personnel are needed to exclude patients at high risk of lifethreatening hypoxia.4 Also, monitoring using AAP guidelines is necessary for the early detection and management of life-threatening hypoxia.3 The AAP recommends that before an examination in which sedation is to be used, children from newborn to age 3 years take nothing by mouth for 4 hours and those aged 3 to 6 years take nothing by mouth for 6 hours.4

Pediatric sedation practices vary, but a few agents are common to most protocols. Oral chloral hydrate is often recommended for children younger than 18 months. However, its use is controversial because of its variable absorption, paradoxical effects, and nonstandardized dosing. Older children usually receive intravenous pentobarbital with or without fentanyl. Although studies have reported successful administration of sedatives by trained nurses,1,2 an anesthesiologist's expertise can be beneficial for patients with substantial comorbidities, including cardiopulmonary disease, skeletal dysplasias, neuromuscular disease, and abnormal airway anatomy. Because of the potential risks of anesthesia and sedation in children, there is a trend toward referring those who require sedation to hospitals with pediatric anesthesiologists.

An important consideration after sedation for pediatric MRI is the need for strict adherence to established discharge criteria, including return to baseline vital signs, level of consciousness close to baseline, and ability to maintain a patent airway.5 Because of the inherent risks of sedation, alternative techniques have been devised, including sleep deprivation and rapid, segmental scanning. The latter permits acquisition of highquality images without the use of sedation.

Normal MRI Anatomy

Appreciating normal MRI anatomy (Fig. 1) is essential for understanding and predicting the MRI appearance of pathologic processes.6

Figure 1 A,
Figure 1 A,:
Sagittal T1-weighted MRI scan of a normal lumbar spine in a 2-year-old boy shows the rectangular shape of the vertebral bodies. The conus medullaris is seen at the L1-L2 level (arrow). B, T2-weighted image shows the long, thin appearance of the intervertebral disk. C, Sagittal T1-weighted scan of a normal lumbar spine in a 10-year-old girl. D, T2-weighted scan. Lordosis is normal. The posterior elements are well formed, with a resultant decrease in the canal diameter. E, Sagittal T1-weighted scan of a normal lumbar spine in a 16-yearold girl shows dark CSF (thin arrow), the conus medullaris at the L1-L2 level (open arrow), and the basivertebral channel (arrowhead). Note the normal rectangular appearance of the vertebral bodies and the lumbar lordosis compared with the 10-year-old girl. F, Sagittal T2-weighted scan shows bright CSF (thin arrow) and a bright nucleus pulposus (arrowhead).

Adolescents and Adults

The lumbar spine is more frequently imaged than the cervical and thoracic area in both children and adults. In adolescents and adults, the lumbar spinal canal appears round proximally and triangular distally. The lumbar facet joints, best visualized in the axial plane, are covered with 2 to 4 mm of hyaline cartilage. This cartilage can be well visualized with FSE and gradient-recalled echo pulse sequences. The epidural space and ligaments also should be evaluated carefully. Epidural fat is seen as high signal intensity on T1-weighted images; the ligamentum flavum shows minimally higher T1-weighted signal compared with the other ligaments. The conus medullaris is usually located at the L1-L2 level. The traversing nerve roots pass distally from the conus medullaris and extend anteriorly and laterally, exiting laterally underneath the pedicle and extending into the neural foramen. The intervertebral disk, consisting of the cartilaginous end plates, anulus fibrosus, and nucleus pulposus, normally shows increased T2-weighted signal in its central portion. CSF, well imaged as low T1-weighted and high T2-weighted signal, often can be used to determine the type of pulse sequence that is being used. CSF pulsations often create artifacts that degrade the image in the lumbar spine; these artifacts must not be mistaken for a pathologic process.

The cervical spine shows a mild lordosis on sagittal images. On axial images, the spinal canal is triangular, with the base located anteriorly. A dark band at the base of the dens is a normal variant that is a remnant of the subdental synchondrosis and should not be mistaken for a fracture. In adults, the facet joints are small and triangular, whereas in children they are large and flat. The spinal cord is elliptical in cross section in the cervical spine. There is a difference in signal between the normal gray and white matter of the spinal cord. This signal heterogeneity should not be mistaken for intramedullary pathology. The intervertebral disks are similar in appearance to, but smaller than, those seen at the thoracic and lumbar levels. An important anatomic feature of the cervical spine is the prominent epidural venous plexus, which is not present in the thoracic or lumbar spine.

The thoracic vertebral bodies are relatively constant in size, and the spinal canal is almost round. Abundant epidural fat is present posteriorly, but there is less anteriorly than in the lumbosacral region. The cord is more round than in the cervical or lumbar regions, and the cord segment lies two to three levels above the corresponding vertebral body. The intervertebral disks are thinner than the disks in the lumbar spine. The appearance of the CSF is more variable in the thoracic spine than in the lumbar region because of more prominent CSF pulsations, but on T1-weighted images, it is commonly seen as a region of low signal dorsal to the spinal cord. This artifact is often most severe at the apex of curves, including the thoracic kyphosis. Certain techniques can minimize this artifact, including gating to the pulse or cardiac cycle.


Differences Between the Pediatric and the Adult Spine

The MRI appearance of the growing spine is complex. Substantial changes occur in the vertebral ossification centers and the intervertebral disks, changing the overall appearance of the spine markedly, especially between infancy and age 2 years.7 In general, the vertebral ossification centers are incompletely ossified early in childhood, and the disks are thicker and have a higher water content than those in adults. The spinal canal and neural foramina are larger, and there is less curvature. In addition, the overall signal intensity of the vertebral bodies is lower than that of the adult spine on T1-weighted images because of the abundance of red (hematopoietic) marrow relative to yellow (fat) marrow in the pediatric, adolescent, and young adult spine.

Full-Term Infant

In the newborn, the overall size of the vertebral body is small relative to the spinal canal, and the spinal cord ends at approximately the L2 level. The lumbar spine does not exhibit the usual lordosis and is straight. The vertebral bodies show a markedly low signal intensity on T1-weighted images, with a thin, central, hyperintense band that likely represents the basivertebral plexus. The spongy bone of the ossification center is ellipsoid rather than rectangular and often mistaken for disk. The intervertebral disk is relatively narrow and often contains a thin, bright central band on T2-weighted images that represents the notochordal remnants.6,7

Age 3 Months

At age 3 months, the osseous component of the vertebral body has increased and the amount of hyaline cartilage has decreased, giving the vertebral bodies a rectangular appearance. The ossification centers begin to gain in signal intensity, starting at the end plates and progressing centrally. The neural foramina have not substantially changed at this age, remaining relatively large and ovoid.6,7

Age 2 Years

At age 2 years, the spine has begun to show its normal sagittal alignment, most likely because of weight bearing (Fig. 1, A and B). The ossified portion of the vertebral body increases substantially and begins to assume its adult appearance, with nearcomplete ossification of the pedicles and the articular processes. The disk space and nucleus pulposus become longer and thinner. The cartilaginous end plate has decreased in size and is often difficult to identify. The neural foramen also begins to take its adult appearance as its inferior portion narrows.7

Age 10 Years

At age 10 years, sagittal alignment resembles that of an adult (Fig. 1, C and D). Ossification of the vertebral bodies and posterior elements is nearly complete, with a resultant decrease in the spinal canal diameter. The vertebral bodies also develop concave superior and inferior contours. The nucleus pulposus becomes smaller at this age and spans approximately half the disk space in the sagittal plane. The neural foramina continue to narrow inferiorly.6

The Conus Medullaris

In early fetal life, the spinal cord extends to the inferior aspect of the bony spinal column.6 Because the vertebral bodies grow more rapidly longitudinally than the spinal cord does, by birth the conus medullaris is repositioned in the upper lumbar spine. It is important to note the location of the conus medullaris on every pediatric spine MRI study (Fig. 1, A and E). A conus medullaris level below the L2-3 interspace in children older than 5 years is abnormal and indicates possible tethering.8,9 Saifuddin et al10 reviewed the MRI findings in 504 normal adult spines and found that the average position of the conus medullaris was the lower third of L1 (range, middle third of T12 to upper third of L3).

Pathologic Processes in the Pediatric Spine


Infectious processes involving the pediatric spine include osteomyelitis, diskitis, and epidural and paraspinal abscess.11-13 In general, the MRI signal characteristics of infection include a region of low T1 and high T2 signal intensity in bone and soft tissue.

In identifying vertebral osteomyelitis, MRI is more sensitive than conventional radiographs or CT and more specific than nuclear scintigraphy.14,15 Marrow edema can be detected on precontrast, fat-suppressed, FSE T2-weighted images. Postgadolinium enhancement of the disk and adjacent vertebral bodies on postcontrast, fat-suppressed, T1-weighted images helps confirm the diagnosis. The specificity of MRI for infection is higher in children than adults because one of the primary confounders, degenerative arthritis, is not part of the differential diagnosis. Differentiating osteomyelitis from neoplastic disease is a common dilemma; generally, infectious processes are more likely to cross and destroy intervertebral disks than are neoplastic conditions.

Diskitis is seen as a disruption of the normally well-defined diskvertebral borders on T1-weighted images and as an increase in signal of the disk on T2-weighted images.12 On T2-weighted images, diskitis may obliterate the normally seen horizontal cleft within the intervertebral disk. The abnormal signal seen in infectious diskitis is associated classically with surrounding soft-tissue inflammation and reactive end-plate changes. Primary diskitis is more likely to develop in children than adults because of the greater blood supply to the disk. Secondary diskitis after diskography or surgery is more likely to develop in adults.

Epidural abscesses are rare, but when they do develop, it is usually after surgery or vertebral osteomyelitis. Epidural abscesses are diagnosed based on the MRI findings of a collection in the epidural space and the appropriate clinical setting.11 Gadoliniumenhanced T1-weighted images often show a peripheral rim of enhancement that represents the abscess wall.

Paraspinal abscesses occur adjacent to the spinal column, most commonly in the paraspinal musculature. They may be secondary to a primary infection in the spine or may arise spontaneously in the paraspinal musculature. These abscesses may be seen as retropharyngeal abscesses in the cervical spine, paraspinous or retromediastinal abscesses in the thoracic spine, or psoas abscesses in the lumbar spine. The MRI characteristics of paraspinal abscesses include a welldefined wall and peripheral enhancement on postgadolinium, T1-weighted images.


MRI can be used to evaluate the pediatric spinal trauma victim who has an abnormal neurologic examination or is unresponsive. The patient is first evaluated with conventional radiographs, which may be normal, even in a child with a neural deficit. Although CT allows for better evaluation of osseous detail and displaced fractures, MRI provides improved evaluation of nondisplaced fractures because of its ability to detect marrow-signal abnormalities.

Spinal cord injury without radiographic abnormality (SCIWORA) is a well-defined entity seen in the pediatric age group.16,17 The characteristic hypermobility and ligamentous laxity of the pediatric bony cervical and thoracic spine predispose children to this type of injury.16 The elasticity of the bony pediatric spine and the relatively large size of the head allow the musculoskeletal structures to deform beyond physiologic limits, which results in cord trauma followed by spontaneous reduction of the spine.16

As with other types of spinal cord injuries, the most important predictor of outcome is the severity of neurologic injury. A patient with a complete neurologic deficit after SCIWORA has a poor prognosis for recovery of neurologic function. The role of MRI in SCIWORA syndrome is to define the location and the degree of neural injury, rule out occult fractures and subluxation that may require surgical intervention, and evaluate for the presence of ligamentous injury. T2- weighted images typically show increased signal in the cord, vertebral body, or ligaments. The increased T2 signal in the cord is compatible with edema and can range from a partial, reversible contusion to complete transection of the cord.

Two other traumatic entities can occur in children, usually as the result of participation in sports. The first is acute disk herniation. This is often a fracture with a hingelike displacement of fibrocartilage and slipping of the entire disk with vertebral end-plate fracture rather than extrusion of a disk fragment from the nucleus, as is seen in adults.18 Such avulsion fractures are often occult on conventional radiographs and are better detected with CT and MRI.18 Axial MRI scans demonstrate the fracture fragment as an area of low signal intensity protruding into the spinal canal, and sagittal images demonstrate a low signal intensity region in the shape of a Y or 7 on all pulse sequences.18

The second entity is a spondylolysis as a cause of back pain in young athletes. MRI, however, is not the optimal method for evaluating spondylolysis. CT offers increased spatial resolution and the ability to accurately define the osseous defect, whereas radionuclide imaging can demonstrate increased radiotracer activity in the region of the defect.


MRI is the modality of choice for evaluating neoplasms in and around the pediatric spine.19 An effective and commonly used approach is to classify the lesion as extradural, intradural-extramedullary (Fig. 2), or intradural-intramedullary (Fig. 3). With this anatomic classification system, the primary role of the MRI examination is to define the location of the suspected neoplasm, which is best achieved with axial and sagittal T1-and T2-weighted images. Once the lesion has been classified, the T2-weighted images can be used to characterize the lesion further. Specifically, the degree of surrounding edema and tissue infiltration and the presence or absence of a cystic component can be determined. Next, postgadolinium enhancement images should be compared with unenhanced T1-weighted images. The final step in obtaining a diagnosis is to correlate the imaging findings with the patient's age and other criteria to narrow the differential diagnosis.

Figure 2
Figure 2:
A schwannoma in an 8-year-old boy. A, Sagittal T1-weighted MRI scan shows an intradural-extramedullary mass impressing on the anterior cervical cord at the C5 level (arrow). B, Axial T2-weighted image shows the lesion herniating through the right C5-C6 neural foramen (arrows).
Figure 3
Figure 3:
An astrocytoma in a 6-year-old boy. A, Sagittal T1-weighted MRI scan shows an intradural-intramedullary lesion within the spinal cord at the T3-T5 levels (arrow). B, Sagittal T2-weighted image shows the partially cystic nature of the lesion. C, Axial T2-weighted image confirms that the lesion (arrow) is within the center of the spinal cord.

Spinal Dysraphism

Spinal dysraphism is a general term used to describe a wide range of anomalies resulting from incomplete fusion of the midline mesenchyma, bone, and neural elements. The osseous abnormalities consist of defects within the neural arch with partial or complete absence of the spinous processes, laminae, or other components of the posterior elements. MRI has been shown to be the best modality for evaluating spinal dysraphism.20,21

A classification system has been proposed for evaluating a patient with a suspected spinal dysraphism (Table 1).21 The differential diagnosis can be narrowed to one of three types: spinal dysraphism with a back mass either covered or not covered with skin, or with no back mass. The final diagnosis then can be made based on the lesion's MRI characteristics.

Table 1
Table 1:
Classification of Spinal Dysraphism

Myelomeningocele is the most common form of spinal dysraphism (Fig. 4). It usually presents in the lumbosacral region (although it can be seen at higher levels) as a back mass not covered with skin. The mass may or may not be covered by leptomeninges containing a variable amount of neural tissue. The sac herniates through a defect in the posterior elements of the spine. The spinal cord usually contains a dorsal cleft, is splayed open, and is often tethered within the sac.21 Progressive scoliosis is seen in 66% of patients with myelomeningocele, Arnold-Chiari type II malformation in 90% to 99%, diastematomyelia in 30% to 40%, and syringohydromyelia in 40% to 80%.22 Scarring can occur at the surgical site after sac closure, so it is important to monitor these patients for signs and symptoms of tethered cord syndrome.

Figure 4
Figure 4:
A myelomeningocele in a 6-year-old girl. A, Sagittal T1-weighted MRI scan shows a low-back mass contiguous with the contents of the spinal canal (arrows). B, T2-weighted image shows that the mass is filled with high-signal-intensity fluid, compatible with CSF (arrows). C, Axial T1-weighted image confirms that the mass communicates with the spinal canal through a defect in the posterior elements (arrows).

Of the entities presenting with a skin-covered back mass in the presence of spinal dysraphism, lipomeningocele is the most common.6,21 The lipomeningocele consists of lipomatous tissue that is continuous with the subcutaneous tissue of the back and also insinuates through the dysraphic defect and dura and into the spinal canal. The spinal cord often contains a dorsal defect at the level of the lipomatous tissue and may be tethered at this level. The essential MRI feature of this lesion is that the lipomatous tissue follows the signal characteristics of subcutaneous fat on all pulse sequences, including fatsuppressed pulse sequences.

Occult spinal dysraphism presents without a back mass. Diastematomyelia is characterized by a sagittal splitting into two segments of the spinal cord, conus medullaris, or filum terminale, often in the thoracic or lumbar spine. The dural tube and arachnoid are undivided in approximately half these patients; clinical findings are rare, and surgery is not indicated. In the remaining patients, the dural tube and arachnoid are completely or partially split at the level of the spinal cord cleft, which results in tethering of the cord and subsequent clinical symptoms. Coronal T1- and T2-weighted images best define the sagittal split in the cord; the findings should be confirmed on axial images.

Another entity often seen in patients with spinal dysraphism is syringohydromyelia, or a syrinx (Fig. 5). A syrinx is a longitudinal cavity within the spinal cord that may or may not communicate with the central canal. Attempts to explain the etiology include developmental, traumatic, inflammatory, ischemic, and pressurerelated causes. Sagittal MRI scans show a linear, low T1 and high T2 signal intensity within the parenchyma of the spinal cord.

Figure 5
Figure 5:
A large syrinx involving the entire spine in a 2-year-old boy. A, Sagittal T1-weighted MRI scan shows the syrinx to be largest at the level of the lower thoracic spine (arrows). Axial T1-weighted (B) and T2-weighted (C) images confirm that the syrinx is located within the center of the spinal cord.

Gibbs artifact, or truncation artifact, can mimic a syrinx on sagittal images (Fig. 6). Gibbs artifact is seen on sagittal T1- and T2-weighted images as a linear region of altered signal intensity in the center of the spinal cord. Thus, it is important to evaluate serial axial T1- and T2- weighted images to confirm findings. Gibbs artifact results from not using a sufficiently high spatial frequency for sampling data. It can be corrected by using a higher-resolution matrix.

Figure 6
Figure 6:
A 5-year-old girl had a history of neck and arm pain. A, Sagittal T2-weighted MRI scan shows a long linear region of high signal intensity within the center of the cervical spinal cord (arrow). This finding can easily be mistaken for a syrinx. B, Sagittal T1-weighted image also suggests low signal intensity in the same region but fails to show a syrinx, demonstrating normal cord anatomy. C, Axial T2-weighted image also demonstrates normal anatomy. These findings are compatible with a Gibbs artifact.

Chiari Malformations

Chiari malformations are seen frequently in patients with spinal dysraphism. Chiari type I malformations consist of cerebellar tonsillar ectopia, in which the cerebellar tonsils extend below the level of the foramen magnum. The common measurement for the degree of herniation of the tonsils below the foramen magnum is 5 mm. Mikulis et al23 reported a variation by age in the upper limit of normal: 6 mm in the first decade of life, 5 mm in the second and third decades, and 3 mm by the ninth decade. In Chiari I malformations, the brainstem is spared and the fourth ventricle remains in its normal location. Chiari I malformations are associated with syringohydromyelia, craniovertebral junction anomalies, and basilar invagination. Chiari II malformations are more advanced and consist of downward displacement of the brainstem and inferior cerebellum into the cervical spinal canal, with a decrease in size of the posterior fossa.

Tethered Cord Syndrome

Tethered cord syndrome is seen in a substantial number of patients with spinal dysraphism, especially those who have undergone surgical closure of the defect.24,25 During fetal life, the spinal cord extends to the sacrococcygeal level. Because of the rapid growth of the vertebral column after birth, the cord ascends to the L1-L2 level in the newborn. During the formation of a spinal dysraphic defect such as myelomeningocele, the open neural elements often attach to the peripheral ectoderm, resulting in spinal cord tethering. After surgical closure of the sac, there is a tendency for the spinal cord to become adherent at the repair site. As the child grows, this adherence may tether the cord and prevent cephalad cord migration, with eventual symptoms. Thus, in patients with spinal dysraphic and related conditions, including myelomeningoceles, myeloceles, lipomeningoceles, and diastematomyelia, tethered cord should be ruled out as the potential cause of any deterioration in neurologic function.

MRI has been proposed as the initial, and possibly only, imaging study for a patient with a suspected tethered spinal cord.9 Sagittal images should be evaluated to determine the level of the conus medullaris (Fig. 7). A conus level below the L2-L3 interspace in children older than 5 years is abnormal and an indication of possible tethering.8,9 In addition, the tethered cord is often displaced posteriorly in the spinal canal. Other findings include lipoma or scar tissue within the epidural space and increased thickness of the filum terminale.9 Although MRI can determine whether a spinal cord is anatomically tethered, these findings should be correlated with the patient's symptoms and serial physical examinations before surgical release is considered.

Figure 7
Figure 7:
A 14-year-old boy had a history of lipomeningocele. After surgical resection, bowel and bladder dysfunction and new lower-extremity paresthesias developed. A, Sagittal T2-weighted image shows the conus medullaris extending to approximately the L4 level and the filum terminale extending to the S1 level (arrow), compatible with tethered cord syndrome. B, Axial T2-weighted image at the L4 level shows the cord located posteriorly within the thecal sac (arrow). C, Axial T2-weighted image at the L5 level shows the placode (thin arrow) with a right-side nerve root (thick arrow) coursing anteriorly and laterally.

Controversies in MRI of the Pediatric Spine

MRI of the pediatric spine remains controversial in several conditions, including scoliosis and tethered cord syndrome, as well as with spinal instrumentation. Safety is also a concern.


The use of MRI imaging in scoliosis is primarily to detect intraspinal abnormalities, which are more frequently associated with uncommon curve patterns such as left thoracic curves, an abnormal neurologic examination, or young age at presentation.26-30 Recently, Do et al26 concluded that MRI is not indicated before spine arthrodesis in a patient with an adolescent idiopathic scoliosis curve pattern and a normal physical and neurologic examination.

One area of particular controversy is back pain in the presence of scoliosis. In a retrospective study of 2,442 patients, Ramirez et al31 found that a left thoracic curve or abnormal result on neurologic examination best predicted an underlying pathologic condition. They found a significant association between back pain and age older than 15 years (P < 0.001), skeletal maturity (P < 0.001), postmenarcheal status (P < 0.001), and history of injury (P < 0.018). The authors concluded that it is unnecessary to perform extensive diagnostic studies on every patient with scoliosis and back pain. MRI should be reserved for patients with infantile or juvenile scoliosis, left thoracic curves, or abnormal neurologic findings. Because coronal views are especially useful in evaluating patients with scoliosis, they should be a part of the routine imaging protocol.

Tethered Cord Syndrome

The rate of MRI in tethered cord syndrome remains controversial. When MRI demonstrates a tethered cord, a choice between surgical and nonsurgical treatment must be made. Although anatomic tethering of the cord is detected easily on MRI, indications for surgery depend on the clinical history and results of serial physical examinations.

Imaging in the Presence of Implants

MRI of the spine in the presence of instrumentation is generally safe but is limited by the image artifacts the implants produce. The pulse sequence used for imaging titanium produces less degradation from artifact because it is less ferromagnetic than stainless steel (Fig. 8).32,33 Thus, titanium may be the better choice of implant in a patient who may require follow-up with MRI. However, with appropriate imaging techniques, clinically useful information can be obtained safely in the presence of both types of implants.34 Specialized pulse sequences such as the metal artifact reduction sequence (MARS) can help reduce the degree of tissue-obscuring artifact produced by spinal hardware and improve image quality compared with conventional T1-weighted spinecho pulse sequences.35

Figure 8
Figure 8:
A 6-year-old boy had a history of high-grade astrocytoma. A, Anteroposterior radiograph 6 weeks after resection, multilevel laminectomy, and posterior spinal arthrodesis from T4 to L3 with titanium pedicle screws, hooks, and rods. B, Midline sagittal postgadolinium T1-weighted MRI scan allows visualization of the canal contents with minimal artifact from the pedicle screws (arrows). C, Parasagittal postgadolinium T1-weighted image shows a rod (thick arrow) and pedicle screw (thin arrow). Neither obscures the MRI scan. D, Axial postgadolinium T1-weighted image also shows the pedicle screws (arrows) and a patent spinal canal.

MRI Safety

MRI may be contraindicated in patients with ferromagnetic implants, materials, or devices because of the risk of implant dislodgement, heating, and induction of current.36 Shellock et al36 reviewed and compiled the results of more than 80 studies and described the ferromagnetic qualities of 338 objects, including 30 orthopaedic implants, materials, and devices. They found that most orthopaedic implants are made from nonferromagnetic materials and therefore are safe for MRI procedures. Another concern is that of safety within the MRI suite. Areas surrounding and within the suite should be carefully monitored for the presence of ferromagnetic equipment that may act as a projectile and injure the patient or hospital personnel. A recent report described a series of projectile cylinder accidents when ferromagnetic nitrous oxide or oxygen tanks were in the MRI suite.37 Other equipment (eg, intravenous pumps, hospital beds, handheld instruments) also should be compatible with MRI.


MRI is an excellent modality for advanced imaging of the pediatric spine. A basic understanding of the normal MRI appearance of the spine at various ages, the signal characteristics of various pathologic changes, and the differential diagnosis of spinal pathology can help the clinician correlate the history and physical examination with MRI findings to establish the most likely diagnosis.


1. Beebe DS, Tran P, Bragg M, Stillman A, Truwitt C, Belani KG: Trained nurses can provide safe and effective sedation for MRI in pediatric patients. Can J Anaesth 2000;47:205-210.
2. Sury MR, Hatch DJ, Deeley T, Dicks-Mireaux C, Chong WK: Development of a nurse-led sedation service for paediatric magnetic resonance imaging. Lancet 1999;353:1667-1671.
3. Vade A, Sukhani R, Dolenga M, Habisohn-Schuck C: Chloral hydrate sedation of children undergoing CT and MR imaging: Safety as judged by American Academy of Pediatrics guidelines. AJR Am J Roentgenol 1995; 165:905-909.
4. American Academy of Pediatrics Committee on Drugs: Guidelines for monitoring and management of pediatric patients during and after sedation for diagnostic and therapeutic procedures. Pediatrics 1992;89(6 pt 1):1110-1115.
5. Malviya S, Voepel-Lewis T, Prochaska G, Tait AR: Prolonged recovery and delayed side effects of sedation for diagnostic imaging studies in children. Pediatrics 2000;105:e42. Available at Accessed May 2, 2003.
6. Goske MJ, Modic MT, Yu S: Pediatric spine: Normal anatomy and spinal dysraphism, in Modic MT, Masaryk TJ, Ross JS (eds): Magnetic Resonance Imaging of the Spine, ed 2. St. Louis, MO: Mosby-Year Book, 1994, pp 352-387.
7. Sze G, Baierl P, Bravo S: Evolution of the infant spinal column: Evaluation with MR imaging. Radiology 1991;181: 819-827.
8. Barson AJ: The vertebral level of termination of the spinal cord during normal and abnormal development. J Anat 1970;106:489-497.
9. Moufarrij NA, Palmer JM, Hahn JF, Weinstein MA: Correlation between magnetic resonance imaging and surgical findings in the tethered spinal cord. Neurosurgery 1989;25:341-346.
10. Saifuddin A, Burnett SJ, White J: The variation of position of the conus medullaris in an adult population: A magnetic resonance imaging study. Spine 1998;23:1452-1456.
11. Auletta JJ, John CC: Spinal epidural abscesses in children: A 15-year experience and review of the literature. Clin Infect Dis 2001;32:9-16.
12. du Lac P, Panuel M, Devred P, Bollini G, Padovani J: MRI of disc space infection in infants and children: Report of 12 cases. Pediatr Radiol 1990;20:175-178.
13. Modic MT, Feiglin DH, Piraino DW, et al: Vertebral osteomyelitis: Assessment using MR. Radiology 1985;157:157-166.
14. Fernandez M, Carrol CL, Baker CJ: Discitis and vertebral osteomyelitis in children: An 18-year review. Pediatrics 2000; 105:1299-1304.
15. Miller GM, Forbes GS, Onofrio BM: Magnetic resonance imaging of the spine. Mayo Clin Proc 1989;64:986-1004.
16. Kriss VM, Kriss TC: SCIWORA (spinal cord injury without radiographic abnormality) in infants and children. Clin Pediatr (Phila) 1996;35:119-124.
17. Pang D, Pollack IF: Spinal cord injury without radiographic abnormality in children: The SCIWORA syndrome. J Trauma 1989;29:654-664.
18. Banerian KG, Wang AM, Samberg LC, Kerr HH, Wesolowski DP: Association of vertebral end plate fracture with pediatric lumbar intervertebral disk herniation: Value of CT and MR imaging. Radiology 1990;177:763-765.
19. Walker HS, Dietrich RB, Flannigan BD, Lufkin RB, Peacock WJ, Kangarloo H: Magnetic resonance imaging of the pediatric spine. Radiographics 1987;7:1129-1152.
20. Altman NR, Altman DH: MR imaging of spinal dysraphism. AJNR Am J Neuroradiol 1987;8:533-538.
21. Byrd SE, Darling CF, McLone DG, Tomita T: MR imaging of the pediatric spine. Magn Reson Imaging Clin N Am 1996;4:797-833.
22. Modic MT, Yu S: Normal anatomy, in Modic MT, Masaryk TJ, Ross JS (eds): Magnetic Resonance Imaging of the Spine, ed 2. St. Louis, MO: Mosby-Year Book, 1994, pp 37-79.
23. Mikulis DJ, Diaz O, Egglin TK, Sanchez R: Variance of the position of the cerebellar tonsils with age: Preliminary report. Radiology 1992;183:725-728.
24. Hall WA, Albright AL, Brunberg JA: Diagnosis of tethered cords by magnetic resonance imaging. Surg Neurol 1988; 30:60-64.
25. Heinz ER, Rosenbaum AE, Scarff TB, Reigel DH, Drayer BP: Tethered spinal cord following meningomyelocele repair. Radiology 1979;131:153-160.
26. Do T, Fras C, Burke S, Widmann RF, Rawlins B, Boachie-Adjei O: Clinical value of routine preoperative magnetic resonance imaging in adolescent idiopathic scoliosis: A prospective study of three hundred and twenty-seven patients. J Bone Joint Surg Am 2001;83:577-579.
27. Evans SC, Edgar MA, Hall-Craggs MA, Powell MP, Taylor BA, Noordeen HH: MRI of ‘idiopathic’ juvenile scoliosis: A prospective study. J Bone Joint Surg Br 1996;78:314-317.
28. Gupta P, Lenke LG, Bridwell KH: Incidence of neural axis abnormalities in infantile and juvenile patients with spinal deformity: Is a magnetic resonance image screening necessary? Spine 1998; 23:206-210.
29. Mejia EA, Hennrikus WL, Schwend RM, Emans JB: A prospective evaluation of idiopathic left thoracic scoliosis with magnetic resonance imaging. J Pediatr Orthop 1996;16:354-358.
30. Schwend RM, Hennrikus W, Hall JE, Emans JB: Childhood scoliosis: Clinical indications for magnetic resonance imaging. J Bone Joint Surg Am 1995;77: 46-53.
31. Ramirez N, Johnston CE, Browne RH: The prevalence of back pain in children who have idiopathic scoliosis. J Bone Joint Surg Am 1997;79:364-368.
32. Rudisch A, Kremser C, Peer S, Kathrein A, Judmaier W, Daniaux H: Metallic artifacts in magnetic resonance imaging of patients with spinal fusion: A comparison of implant materials and imaging sequences. Spine 1998;23:692-699.
33. Rupp R, Ebraheim NA, Savolaine ER, Jackson WT: Magnetic resonance imaging evaluation of the spine with metal implants: General safety and superior imaging with titanium. Spine 1993;18:379-385.
34. Lyons CJ, Betz RR, Mesgarzadeh M, Revesz G, Bonakdarpour A, Clancy M: The effect of magnetic resonance imaging on metal spine implants. Spine 1989; 14:670-672.
35. Chang SD, Lee MJ, Munk PL, Janzen DL, MacKay A, Xiang QS: MRI of spinal hardware: Comparison of conventional T1-weighted sequence with a new metal artifact reduction sequence. Skeletal Radiol 2001;30:213-218.
36. Shellock FG, Morisoli S, Kanal E: MR procedures and biomedical implants, materials, and devices: 1993 update. Radiology 1993;189:587-599.
37. Chaljub G, Kramer LA, Johnson RF III, Johnson RF Jr, Singh H, Crow WN: Projectile cylinder accidents resulting from the presence of ferromagnetic nitrous oxide or oxygen tanks in the MR suite. AJR Am J Roentgenol 2001;177:27-30.
© 2003 by American Academy of Orthopaedic Surgeons