27. Trauma CT

Cranial trauma on CT is read as the surface and intracranial record of two distinct loading regimes: direct contact loading, which fractures bone and produces focal coup injury, and inertial loading (linear acceleration–deceleration and, most destructively, angular/rotational acceleration), which tears bridging veins, drives the brain against bony irregularities to produce contrecoup contusion, and shears the gray–white interface to produce diffuse axonal injury. Noncontrast CT in true Hounsfield units is the undisputed first-line modality because acute extravascular blood is intrinsically hyperdense ($+50$ to $+80\,\mathrm{HU}$) against hypodense brain and CSF, and because bone, air, and hemorrhage — the three things that change acute management — are all directly visible. The reader's task is not merely detection but mechanistic inference: a fracture is interrogated for the vessel or dural structure its path threatens, and every hemorrhage is assigned a compartment that predicts its source vessel and natural history.

The governing question of who needs the scan is answered by validated rules rather than reflex imaging. The Canadian CT Head Rule (Stiell, 2001) achieves near-100% sensitivity for neurosurgically important lesions in minor head injury (GCS 13–15) using high-risk criteria (GCS $<15$ at two hours, suspected open/depressed or basilar skull fracture, $\geq 2$ episodes of vomiting, age $\geq 65$) and medium-risk criteria (retrograde amnesia $\geq 30$ min, dangerous mechanism). The interpretive lexicon then partitions intracranial hemorrhage by anatomy. The epidural hematoma is a biconvex (lentiform) collection between bone and dura that respects sutures but can cross dural reflections, classically arising from a temporoparietal fracture lacerating the middle meningeal artery; its arterial source explains the lucid interval and its capacity for rapid, lethal expansion, and the intraclot swirl sign of unretracted fresh blood marks active extravasation. The subdural hematoma is a crescentic, suture-crossing collection from torn bridging veins, the prototypical lesion of the atrophic elderly and anticoagulated brain, whose subacute phase transits isodensity to cortex and is then detectable only by sulcal effacement and gray–white 'buckling' on dedicated subdural windows. Traumatic subarachnoid hemorrhage layers peripherally in convexity sulci and along the tentorium and is chiefly a biomarker of injury severity, to be distinguished from the central, basal-cistern, circle-of-Willis distribution of aneurysmal SAH. Cortical contusions cluster at the inferior frontal and anterior temporal surfaces and characteristically 'blossom' over hours, mandating short-interval re-imaging.

Management and prognosis are quantified. Clot thickness, midline shift (measured at the septum pellucidum, with $>5\,\mathrm{mm}$ significant), basal-cistern status, and hematoma volume by ABC/2 map onto the Brain Trauma Foundation surgical thresholds (an epidural $>30\,\mathrm{cm}^3$ or a subdural $>10\,\mathrm{mm}$ thick or with $>5\,\mathrm{mm}$ shift warrants evacuation) and onto the Marshall and Rotterdam CT classifications that predict outcome. Secondary injury — the herniation cascade and the ischemic consequences of raised intracranial pressure under the Monro–Kellie constraint — is the principal preventable cause of death and is tracked by cisternal effacement and evolving shift. The dominant failure modes are beam-hardening at the petrous apices and vertex (which both fabricates and conceals fractures and thin extra-axial blood), the in-plane undisplaced fracture missed without reformats, satisfaction of search after one obvious lesion while a contrecoup contusion or second collection is overlooked, and false reassurance from a near-normal CT in a deeply comatose patient whose injury is diffuse axonal and demands MRI.

Thoracic injury is the principal cause of death in roughly a quarter of trauma fatalities, and contrast-enhanced MDCT has displaced the chest radiograph and catheter aortography as the central diagnostic instrument because it simultaneously resolves the pleural space, lung parenchyma, mediastinum, great vessels, airways, diaphragm, and skeleton in a single breath-hold acquisition. The injury that dominates the interpretation is blunt thoracic aortic injury (BTAI), a deceleration lesion produced when differential motion between the relatively mobile aortic arch and the tethered descending aorta concentrates shear at the aortic isthmus just distal to the left subclavian origin at the ligamentum arteriosum — the site of $>90\%$ of injuries in survivors reaching imaging. The mechanistic chain runs from intimal tear, through disruption of media (with the adventitia containing the bleed as a contained pseudoaneurysm), to free rupture. The sentinel CT sign is mediastinal hematoma, particularly periaortic blood that effaces the normal mediastinal fat plane; the direct signs are an intimal flap, intraluminal filling defect, abnormal aortic contour, pseudoaneurysm, and pseudocoarctation. These map onto the Society for Vascular Surgery grading scheme (Lee, 2011): grade I intimal tear, grade II intramural hematoma, grade III pseudoaneurysm, and grade IV free rupture — a hierarchy that directly governs management, with grade I often managed nonoperatively and grades III–IV triaged toward thoracic endovascular aortic repair (TEVAR). The decisive mimic is the ductus diverticulum, a smooth, gently sloping contour bump at the isthmus with obtuse margins and no associated hematoma, distinguished from a pseudoaneurysm's acute angles and intimal irregularity.

The same scan must systematically exclude the rest of the spectrum. Pneumothorax, including the occult pneumothorax invisible on supine radiograph, is found anteromedially and at the lung bases; tension physiology is suggested by mediastinal shift, diaphragmatic depression, and an enlarged hemithorax and is a clinical-radiologic emergency. Hemothorax appears as dependent high-attenuation pleural fluid ($35$–$70\,\mathrm{HU}$), and a focus of higher attenuation within it may indicate active bleeding. Pulmonary contusion is non-segmental, geographic ground-glass and consolidation that appears within hours and resolves over days, whereas pulmonary laceration produces a rounded cavity that may fill with blood (hematocele) or air (pneumatocele) and evolves more slowly. Tracheobronchial injury is suggested by a large persistent pneumothorax with continuous air leak, the 'fallen lung' sign, and focal airway-wall discontinuity. Diaphragmatic rupture — far more often left-sided — is recognized by the 'collar' or 'dependent viscera' sign and herniation of abdominal contents, and is notoriously under-detected acutely. Injuries within the cardiac box raise the concern for pericardial hemorrhage and tamponade (hemopericardium, distended pericardial sac) and blunt cardiac injury.

Bayesian prioritization is driven by mechanism and the mediastinal sign: a high-energy deceleration (frontal motor-vehicle collision, fall) with periaortic hematoma makes BTAI the leading diagnosis until the aortic contour is proven normal, and the absence of any mediastinal hematoma is strong negative evidence. The dominant failure modes are mistaking a ductus diverticulum or motion/pulsation artifact for an aortic injury (ECG-gating and recognizing the obtuse, hematoma-free contour resolve it), missing occult pneumothorax and subtle diaphragmatic injury (countered by deliberate lung-window and multiplanar review), and satisfaction of search after the dramatic aortic finding while a tracheobronchial or esophageal injury is overlooked.

Contrast-enhanced MDCT is the reference standard for the hemodynamically stable patient with blunt abdominal trauma, and its central conceptual advance over the older laceration-grade paradigm is the recognition that the vascular status of the injury — active extravasation versus contained vascular lesion versus bland hematoma — drives management more directly than the depth of the parenchymal tear. Active arterial extravasation appears as an irregular, ill-defined focus of contrast (typically $\geq 90$–$100\,\mathrm{HU}$, approaching the attenuation of an adjacent artery) that increases and disperses on delayed imaging; a contained injury — pseudoaneurysm or arteriovenous fistula — is a rounded focus that follows the blood pool and washes out in parallel with it; and a bland hematoma is non-enhancing. This triad, best resolved with multiphase (arterial, portal-venous, and delayed) or split-bolus technique, is the pivot between angioembolization, operative control, and nonoperative observation.

The spleen is the most commonly injured solid organ. The AAST 2018 organ-injury scale (Kozar, 2018) was specifically revised to incorporate vascular injury into the grade: a contained vascular injury (pseudoaneurysm/AVF) elevates an injury to grade IV, and active intraperitoneal extravasation defines grade V, formally codifying the CT-driven shift toward splenic artery embolization in high-grade injuries that would historically have gone to splenectomy. The liver, the second most injured, is graded on the same updated scale; the danger is hemorrhage from the hepatic veins and retrohepatic IVC and delayed complications (biloma, pseudoaneurysm, hemobilia), and active extravasation again triages toward angioembolization. The kidney is graded with explicit attention to the collecting system: a delayed (excretory-phase) acquisition is mandatory to detect urinary extravasation (urine leak), which converts a parenchymal laceration into a collecting-system injury (AAST grade IV) and changes management. The pancreas, retroperitoneal and frequently injured by direct epigastric compression against the spine, is the great occult injury — early CT may show only subtle peripancreatic fluid or fat stranding, and the critical determinant is integrity of the main pancreatic duct, whose disruption (a deep laceration $>50\%$ of gland thickness) escalates grade and mandates intervention; MRCP/ERCP is often needed.

The most error-prone diagnosis is blunt bowel and mesenteric injury (Bates, 2017), where the findings are frequently subtle and surgical delay is lethal. The specific signs are discontinuity of the bowel wall, extraluminal gas (best on lung/wide windows), extraluminal enteric contrast, active mesenteric arterial extravasation, and abrupt termination or beading of mesenteric vessels; the less specific but corroborating signs are focal bowel-wall thickening and abnormal mural enhancement, mesenteric fat stranding, mesenteric hematoma, and free intraperitoneal fluid in the absence of a solid-organ injury — the last a classic clue that mandates a deliberate search for a bowel injury. Bayesian reasoning weights mechanism heavily: a lap-belt 'seatbelt sign,' handlebar impact, or a transverse abdominal-wall contusion raises the prior for both bowel/mesenteric and pancreatic injury. The dominant failure modes are suboptimal contrast timing fabricating or hiding extravasation, mistaking unopacified bowel or physiologic free fluid for injury, partial-volume and streak artifact obscuring the pancreas and duodenum, and satisfaction of search after a conspicuous splenic or hepatic laceration while the subtle, more dangerous hollow-viscus injury is missed; the countermeasures are multiphase/delayed imaging, lung-window review for free gas, and a mandatory second look at the bowel and mesentery whenever free fluid is unexplained.

Pelvic-ring trauma is fundamentally a problem of hemorrhage: the bony pelvis is a closed osseofascial container surrounding a dense arterial and a high-capacitance venous plexus, and a disrupted ring both lacerates these vessels and loses its tamponade, so that exsanguination — venous in the majority, but arterial in the most lethal — is the proximate cause of death. CT interpretation therefore couples the mechanical pattern of ring failure to the vascular consequence, and triages on hemodynamic status. The mechanics are organized by the Young–Burgess classification, which reads the fracture morphology as the signature of the force vector. Anteroposterior compression (APC) opens the symphysis and sacroiliac joints like a book (the 'open-book' injury); higher APC grades progressively disrupt the anterior and then posterior sacroiliac ligaments, enlarge the pelvic volume, and are strongly associated with arterial injury and massive transfusion. Lateral compression (LC), the most common pattern, drives horizontal pubic-ramus fractures and sacral impaction/buckle fractures and tends toward venous and cancellous bone bleeding. Vertical shear (VS) produces cranio-caudal displacement of a hemipelvis through vertical fractures and complete ligamentous disruption and is the most unstable. The interpreter localizes the anterior lesion (symphyseal diastasis, ramus fractures), the posterior lesion (the biomechanically decisive sacroiliac and sacral injury), and quantifies displacement and pelvic-volume expansion.

The vascular read is the management driver. On contrast-enhanced CT, active arterial extravasation — most often from branches of the internal iliac artery (superior gluteal, internal pudendal, obturator) — appears as a focus of high-attenuation contrast ($\geq 90\,\mathrm{HU}$) that enlarges on delayed phase, in contrast to a contained pelvic hematoma. The presence and size of a pelvic hematoma and the demonstration of a 'blush' predict the need for angioembolization. The WSES pelvic-trauma classification and guidelines (Coccolini, 2017) integrate the anatomic lesion with hemodynamic status into a management algorithm: the unstable patient with a pelvic-ring injury bypasses prolonged imaging for immediate hemorrhage control (mechanical stabilization with a binder, preperitoneal pelvic packing, angioembolization, and increasingly resuscitative endovascular balloon occlusion of the aorta, REBOA, in zone III), whereas the stable or transient responder is characterized by CT with a blush triaging to angioembolization. The associated injuries that CT must actively seek are bladder rupture (extraperitoneal, conforming to the perivesical space, versus intraperitoneal, outlining bowel loops — distinguished by CT cystography), urethral injury in males, and rectal and vaginal lacerations marking an open pelvic fracture.

Bayesian prioritization uses pattern and mechanism: an APC open-book or a vertical-shear injury in a hypotensive patient makes arterial hemorrhage the leading concern and lowers the threshold for angiography, while an isolated LC pattern in a stable patient more often reflects self-limited venous/bony bleeding. The dominant failure modes are beam-hardening and streak artifact across the dense pelvic bones and from positioning that obscure sacral and acetabular fractures (countered by multiplanar and 3D reformats and bone windows), missing the posterior (sacroiliac/sacral) injury that defines instability while anchoring on the obvious anterior ramus fractures, misclassifying bladder rupture without dedicated cystographic distension, and the cognitive error of attributing all pelvic blood to venous oozing and thereby underestimating an embolizable arterial source.

Spinal trauma imaging has shifted decisively to CT as the primary screening modality for the bony column in the adult blunt-trauma patient, displacing radiography because thin-section MDCT with multiplanar reformats resolves fractures, alignment, and canal compromise with far greater sensitivity, while MRI is reserved for neurologic deficit, suspected ligamentous or cord injury, and the obtunded patient in whom ligamentous integrity cannot be cleared. The mechanistic framework reads fracture morphology as the signature of the loading vector — flexion, extension, axial compression/burst, distraction, rotation, and shear — and the central interpretive question is stability, which depends jointly on the bony injury, the integrity of the posterior ligamentous complex (PLC), and the neurologic status. Two validated, morphology-plus-ligament-plus-neurology systems govern reporting and management. The Thoracolumbar Injury Classification and Severity Score (TLICS) (Vaccaro, 2005) scores injury morphology (compression, burst, translation/rotation, distraction), PLC integrity (intact, indeterminate, disrupted), and neurologic status, with a summed score $\leq 3$ favoring nonoperative management, $\geq 5$ favoring surgery, and $4$ equivocal. The Subaxial Cervical Spine Injury Classification (SLIC) (Vaccaro, 2007) applies the same triad — morphology (compression, burst, distraction, translation/rotation), disco-ligamentous complex integrity, and neurology — to C3–C7 with an analogous operative threshold. For the craniocervical and upper cervical regions and for comprehensive characterization, the AOSpine classifications provide the contemporary unifying morphologic grammar (type A compression, type B tension-band/distraction, type C translation).

The CT search pattern is systematic: assess alignment on sagittal reformats (the anterior and posterior vertebral body lines, the spinolaminar line, and interspinous widening that betrays PLC disruption), inspect each vertebral body and posterior element for fracture, evaluate canal compromise by retropulsed fragments, and scrutinize the craniocervical junction (atlanto-occipital and atlantoaxial relationships) and the cervicothoracic junction, both classic blind spots. Specific high-yield patterns include the axial-load burst fracture with retropulsion, the flexion-distraction Chance fracture (a horizontal osseous and/or ligamentous failure through the posterior elements and body, the prototypical lap-belt 'seatbelt' injury strongly associated with hollow-viscus and mesenteric injury), the unstable flexion teardrop fracture, and translational injuries.

A distinct and frequently missed consequence of cervical-spine and skull-base trauma is blunt cerebrovascular injury (BCVI) — intimal injury, dissection, pseudoaneurysm, or occlusion of the carotid and vertebral arteries — which carries a substantial stroke risk and is screened for with CT angiography according to the Denver criteria (mechanism and injury-pattern triggers such as cervical-spine fractures involving the transverse foramina, fractures of C1–C3, Le Fort II/III facial fractures, basilar skull fractures involving the carotid canal, and a seat-belt or hanging mechanism) and graded by the Biffl scale (grade I luminal irregularity/dissection with $<25\%$ narrowing through grade IV occlusion and grade V transection), the grade guiding antithrombotic versus endovascular management. The dominant failure modes are the in-plane fracture missed without sagittal/coronal reformats, beam-hardening at the cervicothoracic junction and shoulders obscuring lower cervical levels, underappreciating PLC and ligamentous injury on CT alone (the indication for MRI), and — most consequentially — failing to recommend CTA for BCVI when a Denver-criterion injury pattern is present, a satisfaction-of-search error in which the conspicuous fracture is reported while its vascular implication is ignored.

In the severely injured patient, CT is no longer a series of regional examinations but a single integrated diagnostic act — the whole-body or 'pan-scan' CT — that has become the central instrument of the resuscitation phase. The protocol is a contrast-enhanced acquisition from the vertex through the pelvis (commonly a noncontrast head and cervical spine followed by an intravenous-contrast chest–abdomen–pelvis, with arterial and portal-venous timing achieved by split-bolus or multiphase technique and delayed images where collecting-system or active-bleeding questions arise), reconstructed into thin sections with multiplanar and three-dimensional bone and vascular reformats. Its appeal is mechanistic and probabilistic: high-energy transfer distributes injury unpredictably across body regions, clinical examination is unreliable in the intubated or obtunded patient, and a single comprehensive dataset both detects occult injury and provides the spatial map for damage-control prioritization.

The evidence base is mature and must be cited with nuance. The REACT-2 randomized controlled trial (Sierink, 2016) compared immediate total-body CT with a selective, conventional imaging work-up in patients with suspected severe trauma and found no overall mortality difference, while total-body CT shortened time to diagnosis. The correct inference is therefore not that pan-scanning saves every life, but that it accelerates definitive diagnosis and is justified in the patient with a high-energy mechanism and a meaningful pretest probability of multi-region injury — and conversely that it should not be reflexively applied to the low-risk, clinically evaluable patient, in whom the radiation-dose cost (a whole-body study delivers an effective dose on the order of $10$–$20\,\mathrm{mSv}$, with attendant stochastic risk that is most consequential in the young) is unjustified. Dose stewardship — appropriate patient selection, iterative and deep-learning reconstruction, automatic exposure control, and avoidance of unnecessary repeat phases — is an explicit responsibility of the protocol, the modern expression of the ALARA principle in trauma.

The interpretive challenge of the pan-scan is as much cognitive as perceptual. The single greatest hazard is satisfaction of search: the dramatic, immediately life-threatening injury (the ruptured aorta, the massive hemoperitoneum) captures attention and the reader stops, leaving the second and third injuries — the cervical-spine fracture, the bowel injury, the contralateral pneumothorax — undiscovered. The disciplined countermeasure is a fixed, exhaustive search pattern that surveys every region and every reconstruction (soft-tissue, lung, and bone windows; arterial and venous phases; multiplanar and 3D) regardless of how compelling the index injury is, complemented by structured reporting and the deliberate practice of re-interrogating the regions statistically associated with the injury already found (the seat-belt sign that links a Chance fracture to a mesenteric injury; the transverse-foramen fracture that mandates CTA for blunt cerebrovascular injury). The findings are then synthesized into an injury-severity picture — the constellation of organ-injury grades, vascular extravasation, hemorrhage volume, and spinal stability — that drives the choice between damage-control surgery, angiographic hemorrhage control, and staged definitive management. The dominant failure modes, beyond satisfaction of search, are anchoring and framing by the most visually arresting injury, premature closure before the survey is complete, motion and contrast-timing degradation across a long acquisition in an unstable patient, and the overarching error of treating a single CT as a static endpoint when serial imaging is required to capture the evolving secondary injuries — blossoming contusions, delayed solid-organ bleeding, and progressive ischemia — that the resuscitation exists to interrupt.