The craniocervical junction: embryology, anatomy, biomechanics and imaging in blunt trauma

The lateral zone of the first cervical sclerotome develops into the posterior arch of the C1 vertebra while the lateral zone of the second cervical sclerotome develops into the arch of the axis (C2 vertebra; Fig. 2). The anterior arch of the C1 vertebra develops from a small mesenchymal off-shoot ventral to both the notochord and the axial segment of first cervical sclerotome called the hypochordal bow. [The hypochordal bow of the proatlas gives rise to a small osseous midline tubercle attached to the ventral surface of the basiocciput below the anterior margin of the foramen magnum (basion) and frequently visible on CT (and MRI) of the mature normal craniocervical junction].

The apical ligament is derived from the axial proatlas. The alar ligaments and the transverse ligamentous component of the cruciform (cruciate) ligament develop from the axial component of the first cervical sclerotome.

The membranous C2 vertebra in the early part of the first trimester consists of three median segments: the apical dental segment, the basal dental segment and the body of the axis. These three segments chondrify at 6 weeks of gestation but remain segregated by the upper and lower dental synchondroses [8‐11]. Ossification occurs in three stages: the first ossification centre appears in the body of the the C2 vertebra at 4 months of gestation; the second wave of ossification appears at 6 months of gestation as two ossification centres, one on each side of the basal dental segment—these two centres integrate and fuse at birth when they should have begun to show some attempt at osseous fusion but the non-ossified cartilage may still be evident radiologically into the 6th year of life (Fig. 3a–f); the third wave of ossification appears in the apical dens at 3 to 5 years of age—ossification of the apical dens and fusion across the upper dental synchondrosis is not complete until adolescence (Fig. 3a–f) [8‐11].

The occipital bone encompasses the foramen magnum and extends from the clivus anteriorly to the lambdoid suture posteriorly. The occipital condyles angle medially and inferiorly from the posterior to anterior: this angulation limits the mobility of the atlanto-occipital joints (i.e. C0-C1 joints), particularly in axial rotation compared with the atlanto-axial joint (i.e. C1-C2 joint) [12‐14]. The predominant movements at the atlanto-occipital joint are flexion and extension. Lateral flexion at the atlanto-occipital joint is significantly limited by the contralateral alar ligament.

The atlanto-axial joints (i.e. C1-C2 joints) allow mobility in flexion, extension, axial rotation and, to a lesser degree, lateral flexion as a result of the biconvex and inherently unstable construct of the joint; it is the ligaments (transverse ligament and alar ligaments) related to this particular articulation which stabilise the joint complex. In the event of traumatic disruption of these ligaments, the atlanto-axial joints are poorly equipped to tolerate axial rotation. This is in stark contrast to the atlanto-occipital joints which are less affected by ligamentous injury [15, 16].

These paired ligaments attach the axis to the base of the skull (Figs. 1 and 4) and originate from the posterior surface of the upper third of the dens and typically travel caudocranially (in 50 % of cadeveric dissections) or horizontally (in 50 % of subjects); the exact insertion point of the alar ligaments has been subject to some contention with researchers variably describing insertion on the medial aspect of the occipital condyles or the anterolateral aspect of the foramen magnum [14, 17‐21]. Each ligament is narrowest at its origin and comparatively wider at its insertion giving it a “V-shaped” configuration [20]. The alar ligaments limit axial rotation and lateral flexion on the contralateral side and, apart from the transverse ligament, are the strongest stabilisers of the atlas preventing anterior displacement in the event of rupture of the transverse ligament. Together with the transverse ligament (transverse atlantal ligament; see below), the alar ligaments are primary stabilisers of the craniocervical junction.

The cruciform ligament is composed of transverse and vertical parts which form a cross behind the odontoid peg (Figs. 1 and 4) [14, 16, 19]. The vertical component which is relatively weak and offers no discernible craniocervical stability consists of a cranially orientated longitudinal band which inserts on to the upper surface of the clivus between the apical ligament and tectorial membrane and a caudally directed band which inserts on to the posterior surface of the body of the axis.

The transverse ligament (sometimes termed the transverse atlantal ligament) of the cruciform ligament complex is arguably the most important ligament in the body. It is the largest, thickest and, crucially, the strongest of the craniocervical junction ligaments (and, in fact, the strongest ligament in the entire spine) and, therefore, a primary stabiliser of the craniocervical junction. It arches behind the odontoid peg attaching to a tubercle arising from the medial aspect of each lateral mass of the atlas (Figs. 1 and 4). The transverse ligament is central to stability of the craniocervical junction, fixing the odontoid peg firmly to the posterior aspect of the anterior arch of the atlas. A synovial capsule is situated between the odontoid process and the transverse ligament. The tectorial membrane, epidural fat and dura are located posterior to the transverse ligament. The transverse ligament serves as the major stabiliser of the atlanto-axial articulation: it permits rotation at the atlanto-axial joints while, at the same time, the alar ligaments will prevent excessive rotation. Tears of the transverse ligament typically occur laterally at the attachment to the tubercle on the atlas.

This thin structure represents an upward extension of the posterior longitudinal ligament (Fig. 1). It forms the posterior border to the supraodontoid space or apical “cave” [14, 22] and runs posterior to the cruciform ligament. It extends cranially to the clivus (as far cranially as the spheno-occipital synchondrosis) and caudally to the posterior surface of the body of the axis. It attaches as far laterally as the hypoglossal canals and, at the level of C0-C1, merges with the atlanto-occipital capsular ligaments (Arnold’s ligaments). The cranial portion of the membrane is adherent to and anatomically indistinguishable from dura [14, 23].

The capsular ligaments of the atlanto-occipital and atlanto-axial joints (which are paired synovial joints) are typically described as thin and loose.

This ligament extends from the tip of the odontoid process to the basion and is situated between the anterior atlanto-occipital membrane and the cruciform ligament (Fig. 1); it is surrounded by fat, connective tissue and a venous plexus which accounts for the slightly variable signal characteristics of this supraodontoid space or “apical cave” on MR imaging. It may be absent in up to 20 % of subjects based on cadaveric dissections undertaken by Tubbs et al. [14, 19, 22, 24]. Despite it being renowned, this ligament probably offers little, if any, significant contribution to craniocervical junction stability.

This thin structure attaches the anterior aspect of the atlas to the anterior rim of the foramen magnum (Fig. 1) and is located immediately posterior to the prevertebral muscles [14, 19, 25]. It forms the anterior wall of the supraodontoid space (which is very discernible on MRI assessment owing to its contents of fat and veins) which also houses the alar and apical (and Barkow) ligaments. It serves to limit atlanto-occipital extension at the craniocervical junction.

Although believed to play very little part in stability of the atlantooccipital articulation, the posterior atlanto-occipital membrane is important because it is highly visible on MRI assessment and also exhibits some specific anatomical features which may be misinterpreted on imaging as traumatic disruption. This broad ligament attaches the posterior arch of the atlas to the posterior margin of the foramen magnum and is continuous with the posterior atlantoaxial membrane and, subsequently, the ligamentum flavum [14, 19, 25] (Fig. 1). Laterally, it may extend over the capsules of the atlanto-occipital joints. Posteriorly, it is related to the rectus capitis posterior minor muscle and, anteriorly, to the dura mater. Interdigitations with both the dura mater and the related rectus capitis posterior minor muscle parenchyma may be present in this ligament; additionally, a connective tissue bridge (exhibiting increased vascularity) joining the rectus capitis posterior minor muscle to the spinal dura is frequently present, particularly in the midline [14, 26‐28]. This myoligamentous complex (comprising the posterior atlanto-occipital membrane, interspinous ligament, ligamentum nuchae, rectus posterior major and minor muscles and obliquus capitis supripr and inferior muscles) adds further stability to the craniocervical junction [29]. An important consideration in trauma of this component of the craniocervical junction is the vertebral artery which pierces the posterior atlanto-occipital membrane and then the dura mater before entering the posterior fossa.

This is a cephalic extension of the supraspinous ligament and extends from the spinous process of the C7 vertebra attaching to the inion of the occipital bone. A sturdy structure, it limits hyperflexion of the cervical spine [14].

There are a number of other anatomically defined small “accessory” ligaments related to the craniocervical junction in addition to the more substantial ligaments outlined above (Table 1). While demonstrable on cadaveric dissection, both the identification of these smaller ligaments (or need to identify these ligaments) on clinical MRI assessment as well as the contribution of these structures to overall craniocervical junction stability is limited and, as such, injury to these ligaments in isolation is probably of low clinical significance. However, individual ligamentous injury at the craniocervical junction infrequently occurs in isolation and, therefore, as an example, MRI evidence of injury to the apical ligament, which is an inconsistent ligament on normal cadaveric dissection and contributes little to craniocervical stability, is still a valuable MRI finding when present as it indicates the likelihood of other coexisting more clinically relevant osseo-ligamentous injuries which should then be sought on the MRI sequences available.

When the hypochordal bow of the fourth occipital sclerotome (proatlas) fails to integrate, an ossified remnant may be evident at the caudal end of the basi-occipit called the condylus tertius or the third occipital condyle (Fig. 5a and b). It is usually single but may be multiple and may form an arthrosis or pseudoarthrosis with the odontoid process or the anterior arch of the atlas. A well-corticated margin and a site typical of the embryological location of the hypochordal bow as well as the occasional association with an os odontoideum (see below) will aid radiological distinction from fracture [30].

Absence of the posterior arch of the atlas is rare and usually isolated but can be associated with bilateral atlanto-axial subluxation or “offset”, mimicking Jefferson’s fracture [30]. Developmental clefts of the arch of the atlas are more common. Such rachischisis is more common posteriorly [30, 31]; the vast majority are midline (97 %; Fig. 6). Less commonly, a (postero)lateral rachischisis may be observed through the region of the sulcus of the vertebral artery (3 %); this may be unilateral (Fig. 7) or bilateral (Fig. 8).

Anterior arch rachischisis is rare (occurring in less than 0.1 % of autopsy dissections) and is typically associated with a posterior arch rachischisis in which case it may be termed a “split atlas” (Fig. 9) [30‐32].

The persistent ossiculum terminale results from failure of fusion of the secondary ossification centre (“terminal ossicle”) to the remainder of the odontoid process which has usually occurred by 12 years of age. It may be confused with a (type I) odontoid fracture. Identification of a smooth corticated margin is, again, central to discriminating the two aetiologies [5‐11, 30, 32].

First described in 1886 by Giacomini and derived from the Latin os meaning bone and odontoideum meaning tooth-form, this represents a separate ossicle with a smooth cortical border lying superior to a small hypoplastic dens and body of the axis in the location of the odontoid process (Fig. 10) and may simulate a (type II) odontoid fracture [5‐11, 30, 32]. It still remains contentious if the os odontoideum represents a post-traumatic acquired anomaly or a truly congenital anomaly. The anterior arch of the atlas may be rounded and hypertrophic in contrast to the normal anterior arch morphology. While the (type II) odontoid fracture is typically associated with a flattened uncorticated sharp margin to the adjacent body of the axis and normal morphology to the anterior arch of the atlas, the os odontoideum exhibits a well-corticated convex upper margin and rounded hypertrophic anterior atlas arch (Fig. 10).

Rarely, nodular calcification/ossification can be seen in the alar ligament which can mimic type III fracture of the occipital condyle or type I fracture of the odontoid process (see below). While rare, this can occasionally present as a imaging diagnostic dilemma in the unconscious or intubated and ventilated polytrauma patient with high-risk injuries for craniocervical trauma. The calcification usually presents as nodular relatively well-circumscribed calcification/ossification in the region of the alar ligament on CT imaging (Fig. 11) [33].

These fractures account for only 2 % of cranial fractures but the associated mortality is high, estimated at between 24 % to 80 % because of the proximity to the brainstem and the high incidence of neurological injury (particularly cranial nerve VI) and vascular injury [34]. The characteristic patterns described are transverse, oblique and longitudinal. The transverse and oblique patterns typically result from lateral blunt force impact or crush injuries, and associated cranial nerve injury and internal carotid injury has been described. The longitudinal fractures result typically from an axial loading mechanism through the vertex and may be associated with vertebrobasilar vascular injury and brainstem infarction.

Basiocciput fractures may be seen in conjunction with craniocervical injury including the rare entity of the retroclival epidural haematoma (Fig. 12a–d). The retroclival epidural haematoma seems to be more common in the paediatric population; the reduced stability of the paediatric craniocervical junction because of smaller occipital condyles and a more horizontally orientated atlanto-occipital articulation presumably predisposes the infant craniocervical junction to this injury. The relatively reduced adherence and strength of the paediatric tectorial membrane may allow traumatic detachment and disruption of local vascular structures such as the basilar venous plexus, the dorsal meningeal branch of the meningohypophyseal trunk and a meningeal branch of the ascending pharyngeal artery (which anastomoses with the meningohypohyseal trunk, inferolateral trunk and arterial arcades related to the odontoid process) leading to blood accumulation in the retroclival area. Regardless, in the adult and paediatric populations, the retroclival epidural haematoma indicates traumatic injury of the sturdy tectorial membrane and suggests a traction/distraction injury mechanism [18, 34].

Until the emergence of CT technology, occipital condyle fractures were considered rare but, in fact, were probably significantly under-reported. The first report of an occipital condyle fracture may date back to 1817 when Sir Charles Bell described the case of a patient who was well until the time of discharge when he reached down to pick up his belongings and died suddenly—an occipital condyle fracture was identified at post-mortem and was presumed to have compressed the medulla [35, 36]. The reported incidence of occipital condyle fractures ranges from 4 % to 19 % [35, 37‐40].

On the medial aspect of the occipital condyle, there is a tubercle for attachment of the alar ligament. The hypoglossal canal is an important anatomical relation located above the middle third of the occipital condyle and transmits the hypoglossal nerve (cranial nerve XII), a meningeal branch of the ascending pharyngeal artery and an emissary vein. An important lateral relation to the occipital condyle is the jugular foramen which transmits cranial nerves IX–XI, the internal jugular vein, the inferior petrosal sinus and the posterior meningeal artery [41‐44].

Clinical presentation of occipital condyle fractures is variable but concomitant head injury is a frequent finding. The anatomical location of the occipital condyles means that the brainstem, the lower cranial nerves (cranial nerves IX–XII) and venous and arterial vessels are at particular risk in the event of fracture. Lower cranial nerve palsies (including Collet–Sicard syndrome where all of the cranial nerves IX–XII are affected) may be acute in two-thirds of cases [35, 41, 45‐51]. Delayed cranial nerve palsies may result from migration of fracture fragments or proliferation of fibrous tissue. Vascular complications related to occipital condyle fractures include internal carotid artery and vertebral artery traumatic dissection, arteriovenous fistulae of the posterior inferior cerebellar artery and Wallenberg syndrome (lateral medullary syndrome) [35, 49].

CT assessment is mandatory to establish or confirm the diagnosis. MRI allows evaluation of the associated ligaments and, in particular, integrity of the alar ligament, transverse ligament and the tectorial membrane; it also allows evaluation of the relationship of the fractured segment on surrounding structures, particularly the brainstem and neurovascular structures.

The most widely accepted classification system of occipital condyle fractures was described by Anderson and Montesano and incorporates the probable mechanism of injury and the potential risk of resultant instability [52]. This system describes three types of occipital condyle fractures. Type I is an impaction type fracture resulting in comminution of the occipital condyle with or without minimal fragment displacement and the mechanism is believed to be axial loading similar to Jefferson’s fracture; this fracture type is considered stable because the tectorial membrane and contralateral alar ligament are intact (Fig. 13). However, bilateral lesions may clearly be unstable. Type II fracture (Fig. 14) is part of a more extensive basioccipital fracture involving one or both occipital condyles and is associated with intact tectorial membrane and alar ligaments, preserving stability. The type III fracture (Fig. 15) is an avulsion fracture resulting in medial fragment displacement into the foramen magnum; in this fracture type, the contralateral alar ligament and tectorial membrane may have been stressed resulting in partial tear or complete disruption and it is thus considered potentially unstable.

This injury is associated with both high mortality and significant neurological morbidity: the force required to cause atlanto-occipital dislocation is such that the injury often proves fatal and, therefore, antemortem imaging of this traumatic injury is uncommon [53‐58]. Under normal anatomical conditions, the convex occipital condyles sit within the concavity of the lateral masses of the atlas. The joint is surrounded by a loose capsule. In the paediatric population, the articulating surface of the atlas at this joint is less concave, probably contributing to the greater incidence of atlanto-occipital dislocation in this group [57, 58]. The most important ligaments for stability of the atlanto-occipital articulation are the cruciform and alar ligaments and the tectorial membrane. These ligaments are underdeveloped in the paediatric population, further contributing to the incidence of this injury in this patient group.

The Traynelis classification is largely descriptive and divides atlanto-occipital dislocation into three types which are determined by the direction of dislocation of the occipital condyles: type I injuries represent anterior displacement of the occipital condyles in relation to the atlas (Fig. 16); type II injuries are distraction injuries with vertical displacement of the occipital condyles in relation to the atlas (Fig. 17a and b); type III injuries describe posterior displacement of the occipital condyles relative to the atlas [58‐60]. Both the basion-dens interval (which is abnormal if greater than 10 mm in the adult and greater than 12 mm in the paediatric patient) and the condyle-atlas interval (which is abnormal if greater than 2 mm in the adult and greater than 5 mm in the paediatric patient) can be used to identify abnormality at the atlanto-occipital articulation; additionally, anterior displacement of the posterior margin of the odontoid peg and body of the axis relative to the basion greater than 12 mm or posterior displacement of the posterior margin of the same relative to the basion greater than 4 mm represents an abnormal relationship [58‐60].

Estimated to account for 25 % of craniocervical injuries, the most common causes are motor vehicle accidents and falls [29, 61‐65]. Atlas fractures can occur in isolation but are frequently associated with fractures of the axis and the subaxial cervical spine and may also be associated with rupture of the transverse ligament and closed head injury [29, 61‐64]. Complications related to vertebral artery dissection injury and lower cranial nerve palsies (IX–XII) have been reported [66]. Cervico-medullary parenchymal injury occurs more frequently when fractures of the atlas coexist with axis or subaxial cervical spine injury and are typically associated with transverse ligament disruption [61, 62, 64, 67] (Fig. 18a–c).

Initially described by Sir Geoffrey Jefferson, atlas fractures can be classified as type I which are fractures involving the posterior arch alone, type II fractures which involve the anterior arch alone, type III (the classical Jefferson fracture) which are bilateral posterior arch fractures associated with a unilateral or bilateral anterior arch fracture, type IV which involve the lateral mass and type V (Fig. 17a and b) which are transverse fractures of the anterior arch [29, 68, 69]. A crucial stabiliser of the atlanto-axial articulation is the transverse ligament and the integrity of this ligament determines the stability or instability of atlas fractures.

Post-mortem studies have revealed more detail of the biomechanics of such fractures [70, 71]: atlas fractures occur primarily from an axial loading mechanism; when the axial loading occurs through the occiput, distraction of the lateral masses of the atlas occurs resulting in increased radial stress on the ring of this vertebra which subsequently “fails” or fractures at the weakest points which are the junction points between the anterior and posterior arches with the lateral masses (i.e. four in total); transverse ligament injury is common with atlas fractures but, crucially, transverse ligament injuries can occur without bone injury (Fig. 21a–c). The remaining key craniocervical ligaments are usually spared unless there are associated fractures of the occipital condyles. On plain radiographic and CT imaging, an atlanto-dental interval of greater than 3 mm in an adult and greater than 5 mm in a child, particularly in the absence of any evidence of fracture, warrants MRI assessment of the craniocervical junction as this is highly suggestive of transverse ligament disruption.

A cautionary note is the radiographic normal variant of “pseudospread” of the atlas which generally occurs in children under 7 years of age and in which the ossified lateral masses of the atlas are projected beyond the ossified articular surface of the axis (C2 vertebra) [72]. This may give the false impression of a Jefferson’s fracture and has, hence, been termed “pseudo-Jefferson’s” by some authors. It is due predominantly to the differential faster growth rate of the atlas compared to the axis in infancy with the axis growth rate eventually catching up with that of the atlas. CT evaluation will clarify this false positive radiographic impression confirming an intact atlas.

The incidence of neurological deficit and acute mortality associated with fractures of the axis approach 8.5 % and 2.4 %, respectively [73‐76]. In addition, the incidence of neurological deficit is significantly higher in combination atlas-axis fractures than when either of these fractures occurs in isolation [73, 77] (Fig. 18a–c).

Ligamentous injury at the craniocervical junction may occur in both the paediatric and adult population following trauma despite the absence of fracture of the osseous structures of the craniocervical junction (Figs. 21a–c and 22a–c). Such ligamentous injury may be acutely symptomatic and may have longer term sequelae if not identified acutely and managed appropriately, albeit frequently conservatively. There are potential medico-legal implications where failure has occurred to identify and manage such injuries expeditiously.

Traumatic alar ligament injuries typically occur near the condylar insertion. Alar ligament failure predisposes to excessive axial rotation with resultant compression of or dissection injury to the vertebral artery and damage to the spinal accessory nerves [14]. Damage to the alar ligaments is most typically secondary to high-energy blunt trauma as might occur in motor vehicle accidents. Contained alar ligament injury has also been implicated in the symptomatology related to whiplash injury [21].

MRI evidence of injury of less clinically relevant ligaments such as the apical ligament or effusion in or around the supra-odontoid space (not to be confused with normal-variant inconsistency of the apical ligament or venous plexus high signal in the supra-odontoid space, respectively) should prompt a search for strain injury, partial tear or full-thickness rupture affecting more clinically relevant stabilising ligaments of the craniocervical junction (Table 1) in spite of absence of any fracture abnormality on CT assessment.

CT is the initial imaging modality of choice in the setting of acute traumatic injury of the craniocervical junction and cervical spine. An appropriately thin-section axial source data-set of 0.75 mm is recommended from which appropriate-resolution multi-planar reformats can be acquired (ideally, 2 mm or less). Coronal and sagittal reformats are mandatory, particularly given the availability or multi-detector CT technology and scrutiny of the axial source data-set as well as liberal use of angled/oblique reformats is also recommended. Imaging evaluation of soft tissue ligamentous injury at the craniocervical junction requires appropriate MR imaging. Inclusion of an isovolumetric T2-weighted sequence allows orthogonal post-processing assessment for enhanced radiological evaluation of the integrity of the key ligamentous structures as well as the location of any telling traumatic peri-ligamentous, intra-articular and periarticular effusions (Table 2).

Atlanto-axial rotatory subluxation (AARS) is relatively rare in the adult population, occurring more commonly in the paediatric population. In adults, the most common cause is trauma. If the facets become locked, then the term atlanto-axial rotatory fixation (AARF) is often applied as the deformity is irreducible. Some two-thirds of the normal rotation that occurs in the cervical spine is derived from the atlanto-axial articulation. However, there is a biomechanical trade-off for this degree of mobility which is stability. It is also important to bear in mind that when the head is rotated normally, the neighbouring segments of the vertebral arteries are structurally affected: the ipsilateral vertebral artery will be kinked and the contralateral vertebral artery will be stretched. Hence, vertebral artery injury or insufficiency is a potential complication associated with AARS. Clinically, patients may demonstrate torticollis and adopt the cock-robin position of the head (because of the apparent descriptive resemblance of a robin listening for a worm in the ground) and occipital pain may occur as a result of compression of the greater occipital nerve or the C2 nerve root. Patients may also experience vertigo, nausea, tinnitus and visual disturbances, possibly related to haemodynamic compromise of the vertebral artery. Spasmodic torticollis (sternocleidomastoid muscle spasm) should be distinguished clinically from AARS—in the former, the shortened sternocleidomastoid on the side contralateral to the direction of head rotation is creating the force producing the deformity, whereas in the latter, the lengthened sternocleidomastoid muscle on the side ipsilateral to the direction of head rotation demonstrates spasm in an attempt to correct the deformity [83]. CT (with or without incorporation of a dynamic protocol in the setting of trauma if fracture is or is not demonstrated on the static CT study) and MRI (to assess soft tissue injury including ligamentous injury) are recommended. Probably the most widely used classification of AARS/AARF is that created by Fielding and Hawkins which classifies the deformity in to four types: type I is rotatory subluxation/fixation without anterior displacement of the atlas (i.e. the atlanto-dental interval is less than 3 mm); type II is rotatory subluxation/fixation with anterior displacement of the atlas of 3–5 mm; type III is rotatory subluxation/fixation with anterior displacement of the atlas greater than 5 mm; type IV is rotatory subluxation/fixation with posterior displacement [83].

The centralisation of the management of severe polytrauma into specialised trauma centres as well as utilisation of aggressive screening criteria has seen a significant rise in the incidence of documented cerebrovascular injury in blunt trauma patients, particularly when applied to patients with an injury severity score greater than or equal to 16 [84‐91]. In spite of the relative infrequency of blunt carotid and vertebral artery injury, it is the potentially devastating complications of such injuries, when present, which makes pre-empting such vascular trauma an essential consideration of the radiologist. Ischaemic events do not infrequently present with some delay after a latent asymptomatic period [85, 87, 92, 93]. Traumatic vascular dissection usually begins with a traumatic intimal tear or intramural haematoma: intimal disruption results in platelet aggregation and subsequent thrombus formation which may narrow or occlude the affected vessel or embolise distally; the intimal flap may be stripped distally creating a false lumen and compromise haemodynamic stability; intramural haematoma within the media may propagate distally (cranially), narrow or occlude the vessel or expand the adventitia producing a traumatic false aneurysm (pseudoaneurysm). Traumatic arteriovenous fistulae may also occur as a consequence of such injuries. In the acute setting, CT angiography is the imaging modality of choice performed from the aortic arch through to and including the circle of Willis as vessel injury may be remote from other sites of non-vascular trauma. Miller et al. [91] spurned the Memphis criteria and Biffl et al. [87, 94] developed the Denver criteria which provide some predictive imaging risk factors for blunt traumatic carotid and vertebral artery dissection injury, of which fractures of the craniocervical junction (C0-C2) are included (Fig. 23). Early identification of blunt traumatic cerebrovascular injury allows appropriate early implementation of antithrombotic or anticoagulant therapy during the asymptomatic period post-trauma and has been shown to decrease the incidence of post-traumatic stroke and improve final neurological outcome [85, 87, 92, 93].