Obtunded Patient

Controversy remains as to the appropriate imaging evaluation of obtunded trauma patients. Unrecognized injuries can result in devastating consequences, including paralysis and death. The risk of neurologic sequelae is 10-fold greater in patients with occult cervical spine injuries compared with those whose injuries are identified on initial screening. Conversely, unwarranted prolonged spinal immobilization has its own risks and complications, including pressure ulcers, venous thrombosis, and respiratory deterioration, as well as limits to mobility and central venous access sites.

The vast majority of evidence indicates that the number of unstable cervical spine injuries is exceedingly low in the setting of a negative MDCT examination. Hogan and colleagues performed a retrospective study on 366 obtunded trauma patients who underwent MRI after negative CT scan findings on a 16-detector MDCT scanner. The reported negative predictive value for ligament injury was 98.9% (362 of 3666) with none of the injuries involving more than one column or deemed mechanically unstable for a 100% negative predictive value for unstable cervical spine injuries. Como and colleagues prospectively studied 115 obtunded trauma patients with MRI after negative screening MDCT examination findings. Acute injuries were found in six patients; however, the findings did not result in an alteration in management or require surgical intervention. Muchow and colleagues performed a meta-analysis determining the predictive value of MRI to evaluate cervical spine trauma. They found that there was a high false-positive rate, but the negative predictive value was 100%, indicating that negative MRI findings excluded possibility of cervical spine injury. Despite several other studies demonstrating low false-negative rates and high sensitivity for MDCT, the subject of supplemental MRI remains a matter of debate because Menaker and colleagues concluded that negative MDCT findings are insufficient in this patient population. In this study group, 18 patients were found to have abnormalities on MRI after normal CT scans, 14 requiring extended collar immobilization and two necessitating surgery. Further research regarding the need for additional imaging in the setting of negative CT examination findings, as well as the timing of supplemental imaging is required to optimize the treatment of these patients, avoid extended immobilization, and provide a cost-effective algorithm.

The Pediatric Patient

Concerns regarding the adverse health effects associated with radiation exposure have prompted further evaluation of imaging protocols and indications, particularly in the pediatric population. Unfortunately, the national trend has been an exponential increase in use of CT examinations over the past 2 decades. Direct epidemiologic evidence has shown an increased risk of cancer after the organ dose delivered from common CT studies performed on two or three body parts (30–90 mSV). In comparing single-detector CT examinations with radiographs of the cervical spine, Rybicki and colleagues found a 14-fold increase in radiation dose to the thyroid gland—26 mGy for CT compared with 1.8 mGy for radiography. Muchow and colleagues estimated a lifetime increased relative risk of thyroid cancer of 25% in adolescent girls receiving a single screening cervical spine CT.

The current recommendation for children younger than the age of 14 years is to perform radiographs rather than CT as the initial screening examination. The reasoning is that cervical spine injuries are relatively rare in this age group and commonly involve the upper cervical spine, a region that can be adequately evaluated with radiographs in children. The same holds true for the thoracic and lumbar spine unless the patient has already undergone a CT examination of the chest, abdomen, and pelvis. In this scenario, CT reconstructions from the source data can be used. Cross-sectional imaging should be reserved as a supplement and problem-solving tool. Interestingly, a 2006 study evaluating 1692 pediatric trauma patients observed a substantial increase in use of CT (9%–21%) in this patient population based on alterations in hospital protocol during the two phases of the study without an increase in sensitivity in detecting spinal injuries.

Although several studies have shown the NEXUS criteria to be reliable in children, a multicenter study of 12,537 pediatric trauma patients identified four predictors associated with a higher incidence of cervical spine injury: A GCS score of less than 14, Glasgow Coma Score EYE (GCS _EYE ) score of 1, involvement in a motor vehicle crash, and age older than 2 years. In these circumstances, CT may be the optimal initial imaging study. After age 14 years, the spine has developed fully; therefore, these patients should be treated in the same manner as adults. CT examinations, particularly in pediatric patients, should be optimized to reduce the radiation dose while maintaining image quality. Dose reduction strategies include using automatic exposure control based on patient size and body habitus and tube current modulation.

Geriatric Patient

Elderly patients present a diagnostic dilemma because fractures can be difficult to detect both clinically and with imaging because of underlying degenerative changes superimposed on osteopenia. The injury pattern also differs from the pattern in younger individuals with a higher proportion of injuries involving the upper cervical spine often caused by low-impact trauma such as a fall from standing height.

Although studies have shown that clinical prediction rules developed for the adult population (NEXUS and CCR) can be applied to patients older than 65 years of age, the most important predictors of injury included neurologic deficit, head injury, and high-impact trauma. CT remains the primary modality for assessing elderly patients suspected of having spinal trauma, but MRI may add vital information under certain circumstances. In equivocal cases in which a vertebra appears wedged without a discrete fracture line evident on CT and the patient is symptomatic at that level, MRI can help differentiate an acute from a chronic fracture by the presence of bone marrow edema.

Role of Magnetic Resonance Imaging

Magnetic resonance imaging is the modality of choice in the evaluation of the soft tissues, including the spinal cord, ligaments, intervertebral discs, musculature, and vasculature. MRI, however, is not well suited in the evaluation of osseous injuries, particularly those involving the posterior elements, because of the relative lack of cancellous bone in these structures. Improved sensitivities have been reported for injuries involving the vertebral body (37%–100%) compared with those involving the posterior elements (12%–45%). Although the sensitivity of MRI remains inadequate to replace CT in the evaluation of an osseous injury, the specificity is reported to be around 95%. MRI is the preferred means to determine the acuity of osteoporotic fractures when indicated.

Debate continues, however, regarding MRI’s role because of the lack of a standardized imaging protocol across medical centers, cost, difficulty in positioning patients with multisystem trauma within the bore, and evidence indicating that detection of occult injuries on MRI rarely results in alteration of medical management. After a normal CT examination, the incidence of abnormal MRI findings is approximately 15%, with only 0.3% of patients requiring the need for surgical intervention.

Despite the small number of unstable cervical spine injuries detected with MRI in patients with normal CT examination findings, MRI is advocated in particular clinical circumstances, including in patients with neurologic impairment, progressive neurologic deficits, a change in neurologic status, or severe unexplained pain ( Table 30-1 ).

In this patient population, unrecognized injuries identified on MRI examinations resulted in either urgent surgical intervention or prolonged collar immobilization. Menaker and colleagues performed a retrospective study on 203 patients, 18 of whom had injuries only detected with MRI, 16 of whom ultimately required a change in management. MRI is ideally suited to diagnose and detect the presence of traumatic disc herniations, extramedullary hematomas, ligamentous injuries, and acute spinal cord injuries, as well as to characterize any associated central canal or foraminal stenosis.

Knowledge regarding the integrity of the ligamentous structures will aid in characterizing the extent of injury and potential for mechanical instability. MRI protocols use multiple sequences exposing differences in the hydrogen concentration of the soft tissues to allow for exquisite anatomic detail and identification of pathologic processes. Although the parameters vary widely depending on the field strength, coil design, and software capabilities of the MRI system, protocols should be tailored to provide sufficient clinical data in a timely fashion. Standardizing protocols and limiting imaging sequences reduce the time needed to perform exams in medically unstable patients.

Magnetic Resonance Imaging Protocols

Sagittal T2-weighted and proton density images delineate the ligaments from the surrounding soft tissues and bone, as well as areas of periosteal stripping or discontinuity. The ligaments appear as low signal intensity structures on all sequences because of a relative lack of mobile hydrogen atoms compared with the surrounding structures. Injuries are diagnosed by either identifying focal disruption of the ligament often with fluid seen tracking within the ligamentous gap or identifying attenuation of a ligament ( Fig. 30-1 ). The ligamentous stump can have an associated avulsed fracture fragment as well. The accuracy of MRI in diagnosing a ligamentous injury varies among several studies with reported sensitivities ranging from 46% to 71% for the anterior longitudinal ligament, 43% to 93% for the posterior longitudinal ligament, 67% for the ligamentum flavum, 36% to 100% for the interspinous ligament, and 89% for the supraspinous ligament.

Sagittal T2-weighted magnetic resonance images demonstrating ligamentous injuries. A, Extension injury with discontinuity of the anterior longitudinal ligament (ALL) and fluid tracking through the torn portion of the ALL into the C5-C6 intervertebral disc space (arrow) . Formation of associated prevertebral soft tissue swelling and anterior hematoma (thick arrows) is noted. B, Flexion injury with disruption of the ALL (arrow) and ligamentum flavum (thick arrow), stripping of the posterior longitudinal ligament (dashed arrow), and hyperintense signal within the interspinous space (star) consistent with tearing of the interspinous ligaments.

The majority of MR protocols also include a sagittal T1-weighted sequence to delineate normal anatomy as well as assess the bone marrow, which should appear brighter than the adjacent intervertebral disc. Fracture lines will typically appear dark on the T1 sequences.

Other commonly used sequences include gradient echo sequences, short tau inversion recovery (STIR), and isotropic 3-D volumetric sequences such as vastly interpolated projection reconstruction (VIPR) imaging. Sagittal gradient echo sequences optimize detection of areas of hemorrhage caused by bloom artifact. This artifact is the result of susceptibility and dephasing on sequences that do not use a 180-degree radiofrequency refocusing pulse; the end result is that small areas of hemorrhage “bloom” and become more conspicuous to the reader. STIR sequences provide robust fat saturation and increased contrast between areas of edema and normal soft tissue as can be seen in cases of cord contusions or bone marrow edema surrounding fractures ( Fig. 30-2 ). Isotropic 3-D volumetric sequences such as VIPR allow reconstructions in any imaging plane after a single acquisition. The literature to date, however, has not shown these additional sequences to provide information that would alter patient management.

Mid sagittal T2-weighted fat saturation ( A ) and short tau inversion recovery (STIR) ( B ) images of the cervical spine after a flexion injury demonstrating focal hyperintense signal (arrows) within the central cord in keeping with edema. STIR sequences provide more robust fat saturation and increased soft tissue contrast accentuating regions of increased signal intensity.

Sagittal T2-weighted sequences also provide the highest correlation with patient prognosis in the setting of an acute spinal cord injury because the level of a cord laceration or transection or extent of edema and hemorrhage within the spinal cord can be determined. In the acute or subacute period (1–7 days), edema is seen on the T2-weighted sequences as areas of hyperintense signal within the cord, and hemorrhage appears hypointense ( Figs. 30-2 and 30-3 ). Four signal patterns have been described: normal, single-level edema, multilevel edema, and mixed hemorrhage and edema. Patient outcomes and the potential of restoring function directly relate to the MRI cord signal with patients recovering fully with normal signal characteristics on MRI regardless of initial neurologic status, but the most severe outcomes are associated with the hemorrhagic pattern. Interestingly, the greater extent and length of edema have been shown to lead to poorer average improvement; however, the same has not been shown with hemorrhage.

Severe traumatic cord injury of the thoracic spine. Axial ( A ) and sagittal ( B ) T2-weighted images reveal areas of mixed signal intensity within a partially transected thoracic cord. Areas of low signal intensity (arrows) represent areas of hemorrhage.

Axial sequences, either T2 or T1 weighted, have not been found to have prognostic value but characterize clinically relevant lesions such as acute disc herniations and extramedullary hemorrhage, as well as assess with better accuracy the amount of cord compression, central canal stenosis, and foraminal stenosis ( Figs. 30-4 and 30-5 ). The incidence of disc herniation on initial MRI approaches 36% with a higher proportion of concomitant injuries to the posterior ligament complex (64%) compared with the anterior longitudinal ligament (37%). Doran and colleagues reported a high incidence of traumatic disc herniation in the setting of facet dislocations, both unilateral and bilateral. The degree of canal stenosis, underlying cord signal changes, and evidence of impingement of the adjacent nerve roots must be documented, particularly in patients requiring surgical intervention, to optimize surgical treatment and neurologic recovery. Optional coronal sequences can be obtained to aid in the evaluation of the craniocervical junction and characterize the alignment of dens and occipital condyle fractures.

Acute epidural hematoma. Magnetic resonance images of the lumbar spine in a patient involved in a motor vehicle accident show an epidural hematoma anteriorly (arrows), which indents the thecal sac. The age of the blood products is determined by the signal characteristics, which are predominantly hyperintense in respect to the conus.

Subacute epidural hematoma. A large epidural hematoma (arrow) with internal intermediate signal intensity, isointense to the adjacent nerve roots, results in central canal stenosis and narrowing of the left lateral recess best depicted by the axial T2-weighted sequences ( A ). The hematoma not only indents the thecal sac but results in clumping of the traversing nerve roots (arrowhead). B, Sagittal T2-weighted sequence. C, Sagittal T1-weighted sequence.

Extramedullary hemorrhage is an uncommon consequence of spinal trauma occurring in approximately 1% to 2% of cases (see Figs. 30-4 and 30-5 ). Epidural hematomas most commonly involve the cervical and thoracic spine and are thought to be venous in origin (see Figs. 30-4 and 30-5 ). Typically located dorsally because of the relationship of the dura ventrally with the posterior longitudinal ligament and local pooling within valveless, thin-walled dorsal epidural veins, it has been postulated that a sudden increase in intravenous pressure leads to rupture of these veins. Younger patients and patients with fused spinal segments such as seen in ankylosing spondylitis and diffuse idiopathic skeletal hyperostosis (DISH) are at highest risk for the development of an epidural hematoma after a traumatic event. Conversely, subdural hematomas more commonly occur ventral to the cord. The diagnosis can be difficult as the hematoma evolves over time ( Figs. 30-6 and 30-7 ). In the hyperacute stage (1–6 hours), intact red blood cells contain oxyhemoglobin, which appears hyperintense on the T2-weighted sequences. At 1 to 3 days, oxyhemoglobin deoxygenates to deoxyhemoglobin with a paramagnetic effect, resulting in areas of hemorrhage appearing dark on the T2-weighted sequences compared with the spinal cord. Within 3 to 7 days, deoxyhemoglobin converts into intracellular methemoglobin, which remains dark on the T2-weighted sequences, but now appears bright rather than isointense on the T1-weighed images. Posterior fossa subdural hematomas are associated with craniocervical dissociation, which requires critical evaluation.

Subtle acute cervical subdural hematoma. Given similar signal characteristics to the subjacent cerebrospinal fluid and lack of compression of the thecal sac, acute subdural hematomas can be difficult to diagnose. Typically, they present as hyperintense lesions (block arrows on the sagittal ( A ) and axial ( B ) T2 fat-saturated sequences) within the dural sac silhouetting and typically displacing the adjacent spinal cord (arrow) . As evident on the axial T2-weighted sequence ( B ), the cervical cord is eccentrically located secondary to mass effect from the subjacent subdural hematoma. Lack of direct continuity with the adjacent osseous structures is another imaging clue to this diagnosis.

Sagittal T1- ( A ), sagittal T2- ( B ), and axial T1- ( C ) weighted sequences demonstrating a subdural hematoma (arrows), which is clearly distinct from the epidural fat (asterisk) . In addition, intrathecal hemorrhage is seen involving the distal cord (arrowheads) . The blood products are hyperintense in signal compared with the adjacent cord.

Although the research is lacking, based on the experience of several trauma centers, the optimal time interval between injury and MRI examinations is within 72 hours. This recommendation is based on the natural progression of MR signal changes beyond this time interval. The length of cord edema increases by one vertebral level for each 1.2-day delay in imaging within the first 5 days. Single-level edema will progressively resolve within 3 weeks, and the initial hypointense signal of hemorrhage will transform to hyperintense signal between 1 to 2 weeks as deoxyhemoglobin is converted to extracellular methemoglobin. Additionally, soft tissue edema progressively resolves, resulting in reduced visibility of subtle ligamentous injuries.

Role of Flexion and Extension Views and Dynamic Fluoroscopy

The literature does not provide sufficient evidence to support the use of either flexion and extension views or dynamic fluoroscopy in the detection of cervical spine ligamentous injuries. Often the examination is inadequate with Bolinger and colleagues reporting visualization of the cervicothoracic junction in only 4% of fluoroscopic studies and Anglen and colleagues reporting 28% (236 of 837) of flexion and extension radiographs to be technically inadequate. Duane and colleagues compared flexion and extension views with MRI in 271 patients. Radiographs were nondiagnostic in approximately 30% of cases and did not facilitate treatment. Additionally, flexion and extension views resulted in increased cost and prolonged immobilization. A follow-up study showed that flexion and extension views failed to identify MRI-confirmed injuries. Freedman and colleagues concluded that the technique for flexion and extension views is unreliable and reported four false negatives in seven patients with injuries. Furthermore, neurologic deficits have been described after dynamic fluoroscopy with the development of quadriplegia in one patient. Given the low sensitivity for ligamentous injury detection, low rate of interpretable studies, high false-positive rates, and potential to induce neurologic deficits, the role of flexion and extension views or dynamic fluoroscopy is limited and should not be performed in obtunded patients.

Far less data are available concerning the indication for imaging the thoracolumbar spine with flexion and extension views; however, the same principles apply. In addition to low sensitivities, isolated unstable ligamentous injuries to the thoracolumbar spine are extremely rare in the absence of fractures. As is the case for the cervical spine, neurologic deficits indicate the need for imaging the symptomatic level with MRI.

Cervical Spine

A systematic approach should be applied when evaluating either radiographs or cross-sectional imaging (MRI and CT) to detect subtle injuries. Radiographic assessment of the cervical spine includes standard three views (anterior-posterior [AP], lateral, and open mouth), although the vast majority of injuries are detected on the lateral view. Given the inherent difficulty in visualizing the cervicothoracic junction secondary to obscuration by the overlying soft tissues and osseous structures, a swimmer’s view is often necessary to demonstrate the anatomy to the T1 level. CT examinations should include reconstructions of the axial images in both the coronal and sagittal planes. Regardless of the modality, a checklist should be used when evaluating imaging studies to ensure high diagnostic accuracy. Each of the views will be discussed separately below, focusing on radiologic signs and measurements used in their interpretation.