21. Neurotrauma CT

The skull fracture is the surface readout of transmitted impact energy and, more importantly to the interpreter, the anatomic predictor of which intracranial complication is likely to follow; it is rarely dangerous in itself but reorganizes the pretest probability of the lesions that are. When the kinetic energy delivered to the calvarium exceeds the elastic tolerance of bone, the vault fails along lines determined by the force vector and the local biomechanics: a focused high-energy load over a small area produces inbending and a depressed fracture, whereas a diffuse load produces a propagating linear fracture that radiates from the point of contact along paths of least resistance, frequently crossing the squamous temporal bone, the thinnest segment of the vault. The reader's task is to detect the fracture, classify it, and—critically—infer its vascular and dural consequences. On CT this demands the bone window (roughly $WW\,2500$–$4000 / WL\,300$–$600$) and thin-section ($\leq 1\,\mathrm{mm}$) acquisition with multiplanar and, where available, three-dimensional reformations, because an undisplaced fracture in the axial plane may parallel the section and vanish, while a vascular groove or a normal suture may masquerade as a fracture.

The central discriminations are fracture-versus-suture-versus-vascular-groove and the recognition of features that escalate management. A fracture is typically sharply lucent, non-corticated, straight or angular, runs in an unexpected direction, and crosses sutures and vascular grooves; a suture is corticated, serrated, bilaterally symmetric, and located at a known site (coronal, lambdoid, squamosal); a vascular groove (notably the middle meningeal) is corticated, branching, and follows a stereotyped course. The features that change disposition are (i) a fracture line that crosses the middle meningeal artery groove or a dural venous sinus, which raises the probability of epidural hematoma or sinus injury; (ii) depression greater than the thickness of the adjacent inner table, an operative threshold because of dural laceration and cortical injury; (iii) involvement of the skull base, where a fracture traversing the carotid canal, the foramina, or the petrous temporal bone implicates the internal carotid artery, the facial and vestibulocochlear nerves, and the tegmen; and (iv) signs of an open or basilar injury—intracranial air (pneumocephalus), an air-fluid level in a sinus, opacified mastoid air cells, or CSF leak. Basilar fractures are a Bayesian linchpin: their clinical surrogates (periorbital and mastoid ecchymosis, hemotympanum, CSF rhinorrhea/otorrhea) appear in decision rules precisely because the fracture marks energy delivered to vascular and neural structures.

The governing question—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 fracture, $\geq2$ vomiting episodes, age $\geq65$) and medium-risk criteria (retrograde amnesia $\geq30$ min, dangerous mechanism); suspected fracture is itself an entry criterion, underscoring the fracture-as-sentinel logic. The dominant failure modes are technical and cognitive: beam-hardening streak at the petrous apices and the vertex both fabricates and conceals fractures (mitigated by bone-window multiplanar review and recognizing the artifact's morphology), the in-plane undisplaced fracture is missed without coronal/sagittal reformats, and satisfaction of search—stopping at a single obvious vault fracture—causes the second base-of-skull fracture or the underlying epidural to be overlooked. The disciplined report names the fracture, states whether it crosses a vascular structure, quantifies depression, and explicitly searches the structures the fracture path threatens.

The epidural (extradural) hematoma is the archetypal surgical emergency of neurotrauma, and its imaging, mechanism, and natural history are tightly coupled to a single anatomic fact: the dura is firmly adherent to the inner table except where it is anchored at the cranial sutures, so blood accumulating between bone and dura must dissect this potential space against high resistance, producing a biconvex (lentiform) collection that does not cross sutures but can cross dural reflections (the falx and tentorium). The classic mechanism is a temporoparietal fracture lacerating the middle meningeal artery, a high-pressure arterial source that drives the strongly hyperdense, expanding lentiform clot; venous epidurals from torn dural sinuses or diploic veins occur particularly at the vertex (superior sagittal sinus) and posterior fossa (transverse sinus), are lower-pressure, and may be sutural-crossing and slower. The arterial substrate explains the textbook lucid interval: the patient is initially neurologically intact because the brain has compensatory reserve (Monro-Kellie buffering), then deteriorates precipitously as the arterial hematoma exhausts that reserve and drives the steep limb of the intracranial elastance curve—an unforgiving tempo that makes prompt recognition and evacuation life-saving and frequently restorative of excellent outcome when performed before brainstem injury.

The CT signature is highly specific. Acute clot measures $+50$ to $+80\,\mathrm{HU}$ in a biconvex shape limited by sutures, typically over the temporoparietal convexity with an overlying fracture present in the large majority of adult cases. Two density features carry management weight. A swirl sign—an irregular region of low attenuation within the hyperdense clot—denotes hyperacute, unretracted, actively extravasating blood and predicts continued expansion and worse outcome; its presence should shorten the interval to surgery and to re-imaging. Heterogeneous and mixed-density collections likewise signal an evolving rather than stable lesion. Volume is estimated by ABC/2 (the ellipsoid approximation $V\approx A\times B\times C/2$), and clot thickness and midline shift are measured directly. The Brain Trauma Foundation surgical-management guidelines (Bullock, 2006) operationalize these numbers: an epidural exceeding $30\,\mathrm{cm}^3$ should be evacuated regardless of GCS, whereas a smaller clot ($<30\,\mathrm{cm}^3$, $<15\,\mathrm{mm}$ thickness, $<5\,\mathrm{mm}$ shift) in a patient with GCS $>8$ and no focal deficit may be managed nonoperatively with serial CT and close observation; coma with anisocoria mandates immediate evacuation. The reader therefore reports not only the diagnosis but the precise metrics that select the operation.

Differential reasoning is usually straightforward but has instructive edges. The principal alternative is subdural hematoma, separated by morphology (crescentic, suture-crossing, reflection-respecting) and source vessel; the discrimination occasionally blurs at the convexity, where multiplanar review and attention to suture and falx boundaries resolves it. Bayesian priors are powerful here: a young patient, a temporal fracture, an arterial bleed, and a lucid interval push hard toward epidural, whereas an elderly anticoagulated patient with a crescentic convexity collection favors subdural. The characteristic mimics are a venous epidural mistaken for subdural (and vice versa) and, at the vertex, a sinus-related epidural that axial images underrepresent—coronal reformats are essential. The failure modes are anchoring on the obvious convexity lesion while missing a contrecoup contusion or a second extra-axial collection (satisfaction of search), underappreciating the swirl sign and thereby the expansion risk, and beam-hardening at the skull base obscuring a posterior fossa venous epidural whose small volume carries outsized danger because of brainstem proximity and the risk of rapid obstructive hydrocephalus.

The subdural hematoma occupies the potential space between the dura and the arachnoid and is, mechanistically, a disease of the bridging veins—the cortical veins that traverse the subdural space to drain into the dural venous sinuses. Rapid head deceleration or rotational acceleration displaces the brain relative to the fixed dura and stretches these tethered veins beyond their tensile limit; because the bridging vein is a low-pressure, thin-walled conduit, the resulting hemorrhage spreads diffusely over the convexity as a crescentic (concavo-convex) film that, unconstrained by sutures, extends along the entire hemisphere and is limited instead by the dural reflections, layering along the falx and tentorium. This anatomy is the inverse of the epidural's and is the basis of the morphologic discrimination. The bridging-vein substrate also explains the epidemiology and the danger demographics: cerebral atrophy in the elderly and in alcoholics stretches and exposes the bridging veins (and enlarges the subdural space into which blood can accumulate before symptoms appear), and anticoagulation/antiplatelet therapy both increases incidence and shifts the clot toward a lower-density, heterogeneous, expansion-prone collection—so that subdural hematoma in these populations is common, often follows trivial trauma, and is the classic delayed and recurrent intracranial bleed.

The CT appearance is governed by the same blood-product chemistry developed for parenchymal hemorrhage but applied to a thin extra-axial film, yielding a clinically vital staging. An acute subdural is hyperdense ($+50$–$+80\,\mathrm{HU}$), crescentic, and suture-crossing; a subacute subdural transits isodensity to cortex (typically over one to three weeks), the single most treacherous moment in extra-axial imaging; and a chronic subdural is hypodense, sometimes with a hematocrit level and an enhancing neomembrane, with acute-on-chronic rebleeds producing layered densities. The isodense subacute collection is dangerous precisely because, lacking intrinsic contrast against brain, it is detected only by indirect signs—effacement of cortical sulci, medial displacement of the gray-white interface ('buckling'), and mass effect/midline shift disproportionate to a visible lesion—and this hazard is greatest when the collection is bilateral and symmetric, because the shift signs that would betray a unilateral isodense subdural are cancelled. Optimized subdural windowing (a wide intermediate window such as $WW\sim200 / WL\sim70$–$80$) is the targeted countermeasure: by spreading the gray scale across the dura-cortex transition it renders a thin or near-isodense film conspicuous against the inner table. Volume and shift are quantified as before, and surgical thresholds follow the BTF guidelines (Bullock, 2006): an acute subdural with thickness $>10\,\mathrm{mm}$ or midline shift $>5\,\mathrm{mm}$ warrants evacuation regardless of GCS, and in comatose patients additional criteria (GCS drop $\geq2$, asymmetric or fixed pupils, ICP $>20\,\mathrm{mmHg}$) lower the threshold further.

The Bayesian differential pivots on demographics and morphology: an elderly anticoagulated patient with a crescentic, suture-crossing, possibly bilateral convexity collection is subdural until proven otherwise, whereas a young patient with a temporal fracture and a lentiform clot is epidural. The instructive mimics are the isodense subdural (mistaken for normal brain or simply missed), a subdural hygroma (CSF-density traumatic subdural fluid) versus a chronic subdural hematoma, and prominent atrophic CSF spaces simulating a chronic collection—distinguished by the absence of mass effect and the presence of crossing veins within true subarachnoid space. The dominant failure modes are the bilateral isodense subdural producing 'pseudo-normal' symmetric effacement, satisfaction of search after a contralateral contusion is found, and reliance on a single brain window that flattens the dura-cortex contrast; the remedy is disciplined subdural windowing, active interrogation of sulcal and gray-white morphology, and multiplanar review along the falx and tentorium where interhemispheric and tentorial subdurals hide.

Traumatic subarachnoid hemorrhage (tSAH) is the most common form of intracranial hemorrhage after head injury and functions chiefly as a biomarker of injury severity and of the forces delivered to the brain surface, rather than as a primary surgical target. Mechanistically it arises from the rupture of small pial and cortical vessels and from the leakage of superficial contusions into the subarachnoid space; the blood layers within the CSF cisterns and sulci where the injuring force was greatest, so its distribution is itself diagnostic. The defining contrast is with aneurysmal SAH, and the distinction is both anatomic and Bayesian. Traumatic SAH characteristically lies peripherally—over the cerebral convexities, within the sulci adjacent to contusions, and along the tentorium and interhemispheric falx—and is frequently accompanied by other trauma stigmata (fracture, contusion, extra-axial collection). Aneurysmal SAH is central, filling the basal cisterns (suprasellar, perimesencephalic, sylvian) in a pattern that mirrors the circle of Willis, often with a dominant clot pointing to the ruptured vessel. When a patient presents after a fall or collision, the prior probability strongly favors traumatic SAH; but the crucial clinical caveat is that a ruptured aneurysm can cause a fall, so a centrally distributed, basal-cistern-predominant 'traumatic' SAH—especially if the trauma seems trivial or the history is unclear—should lower the threshold for CT angiography to exclude an underlying aneurysm.

The CT signature is hyperattenuation ($+50$–$+70\,\mathrm{HU}$, falling as blood is diluted and cleared) replacing the normally hypodense CSF, so that the affected sulci and cisterns appear filled and white. The search pattern is systematic and high-yield because tSAH is easily missed: interrogate the sylvian fissures, the interpeduncular and perimesencephalic cisterns, the sulci over both convexities (including the vertex, where partial-volume averaging with bone is worst), the interhemispheric fissure along the falx, and the dependent occipital horns of the lateral ventricles for layering intraventricular extension. Subarachnoid windowing and thin sections improve the detection of a thin layer of sulcal blood. The presence and burden of tSAH and intraventricular hemorrhage are not cosmetic findings: they are explicit components of the Rotterdam CT score (Maas, 2005), where they add prognostic weight, and they predict communicating hydrocephalus from impaired CSF resorption at the arachnoid granulations and, in larger volumes, posttraumatic vasospasm.

The Bayesian differential extends beyond aneurysm to the mimics of SAH, which are a notorious source of error. Pseudo-subarachnoid hemorrhage is apparent hyperdensity of the basal cisterns produced not by blood but by diffuse cerebral edema: as the brain swells and the cisternal CSF is effaced, the engorged superficial vessels and the relatively denser brain make the cisterns look falsely hyperdense, classically in hypoxic-ischemic injury and severe diffuse edema; the attenuation is usually lower than true blood ($<+30$–$40\,\mathrm{HU}$), the 'hemorrhage' is symmetric and conforms to effaced cisterns rather than layering in sulci, and the clinical context (cardiac arrest, global anoxia) is the tell. Other mimics include leaked intrathecal or recently administered intravenous contrast (resolved by dual-energy material decomposition or a noncontrast comparison), purulent meningitis filling sulci, and beam-hardening artifact at the skull base simulating interpeduncular blood. The dominant failure modes are therefore overcalling pseudo-SAH as hemorrhage (and conversely dismissing real thin convexity tSAH as artifact), missing isolated vertex or interhemispheric tSAH on axial images without coronal reformats, and—most consequentially—failing to question whether a 'traumatic' basal-cistern SAH is in fact aneurysmal. The expert reads the distribution first, assigns the mechanism probabilistically, quantifies the burden for the Rotterdam score, and names the downstream risks of hydrocephalus and vasospasm.

Traumatic intraparenchymal hemorrhage spans a spectrum from the small cortical contusion to the large coalescent intracerebral hematoma, and its hallmark biomechanical feature—distinguishing it from hypertensive or amyloid bleeding—is its predilection for the surfaces where the brain strikes bony irregularities. Acceleration-deceleration drives the brain against the anterior and inferior frontal lobes (against the orbital roofs and crista galli) and the anterior and inferior temporal lobes (against the sphenoid ridge and petrous bone), so contusions cluster at these sites both at the point of impact (coup) and, characteristically, diametrically opposite it (contrecoup), the latter generated as the brain rebounds and as negative-pressure cavitation tears surface vessels. The injured cortex shows a mixture of petechial hemorrhage, edema, and necrosis; on CT this is the 'salt-and-pepper' pattern of small hyperdense foci ($+50$–$+70\,\mathrm{HU}$) admixed with hypodense edema in a peripheral, gyral, frontotemporal distribution—an appearance and location that, taken together, are nearly specific for trauma and immediately separate contusion from a centrally located hypertensive ganglionic bleed or a lobar amyloid hemorrhage.

The defining clinical behavior of traumatic contusion is its tendency to 'blossom': the lesion enlarges and becomes more hemorrhagic over the first hours to days as secondary injury propagates, perilesional edema accumulates, and the avalanche of small-vessel rupture extends the bleed—exactly the hematoma-expansion physiology of spontaneous ICH, but here predictable from the traumatic mechanism. This temporal evolution has two direct consequences for practice. First, an early CT systematically underestimates the eventual injury, so a near-normal or minimal initial scan in a patient with significant mechanism or coagulopathy mandates short-interval re-imaging (commonly at 6–24 hours, sooner if neurologic decline), and the report should explicitly warn of expansion risk rather than offer false reassurance. Second, expansion predictors imported from the ICH literature apply: heterogeneous density, the swirl sign of unretracted fresh blood, fluid-fluid (hematocrit) levels indicating coagulopathy, and—on CT angiography—a spot sign of active extravasation all raise the probability of growth and lower the threshold for aggressive blood-pressure and coagulation management and for surgery. Volume is tracked by ABC/2, and a coalescent traumatic hematoma exceeding $\sim50\,\mathrm{cm}^3$, or a smaller deep/posterior-fossa lesion producing mass effect, enters the surgical-evacuation conversation under the BTF parenchymal-lesion guidelines (Bullock, 2006), integrated with GCS, pupillary findings, and cisternal status.

The Bayesian differential is anchored by location and company: peripheral, gyral, frontotemporal, multifocal, and accompanied by fracture, tSAH, or extra-axial blood points to traumatic contusion; central/ganglionic favors hypertensive ICH; lobar in an elderly normotensive patient raises amyloid angiopathy; and a hemorrhage out of proportion to a trivial mechanism, or in an atypical (non-contrecoup) site, should prompt the question of an underlying lesion that bled and caused the fall—a hemorrhagic neoplasm or vascular malformation—best pursued with contrast-enhanced imaging or follow-up. The characteristic mimics are hemorrhagic transformation of an ischemic infarct (vascular-territory rather than gyral-contusional distribution), a hemorrhagic tumor (nodular enhancement, disproportionate edema, atypical site), and a cavernous malformation (popcorn morphology, complete hemosiderin rim on MRI). The dominant failure modes are anchoring on the initial scan and missing blossoming (the remedy is mandated short-interval re-imaging), satisfaction of search after one obvious contusion while a contralateral contrecoup lesion or DAI is overlooked, and partial-volume/beam-hardening artifact at the frontal and temporal skull base masking the very surface contusions that are most common—best countered with thin sections, multiplanar reformats, and a deliberate inspection of the inferior frontal and anterior temporal cortex.

Diffuse axonal injury (DAI) is the most consequential discordance in all of neurotrauma imaging: it is the leading cause of persistent vegetative state and severe disability after closed head injury, yet the acute CT is frequently normal or grossly underwhelming relative to the depth of coma, because the primary lesion is microstructural axonal disconnection that CT cannot resolve. The mechanism is rotational (angular) acceleration-deceleration rather than direct impact: when the head undergoes rapid rotation, tissues of differing density and inertia—gray versus white matter, and brain versus the tethering dura and falx—move at different rates, generating shear strain at their interfaces. Axons, which are viscoelastic and tolerate slow stretch but fail under rapid loading, undergo cytoskeletal disruption, impaired axonal transport, focal swelling (the classic axonal retraction ball), and ultimately disconnection (secondary axotomy) over hours to days. The lesions therefore concentrate exactly where shear is maximal and in a reproducible, severity-graded gradient that the interpreter must commit to memory.

The Adams grading scheme (Adams, 1989), derived from the postmortem distribution and still the organizing framework, maps anatomic depth onto severity: grade 1 involves the lobar gray-white junction (parasagittal frontal and temporal white matter, internal/external capsules); grade 2 adds the corpus callosum, especially the splenium and body, often abutting the falx; and grade 3 adds the dorsolateral rostral brainstem (the midbrain and upper pons, near the superior cerebellar peduncles), the deepest and most ominous tier, strongly associated with prolonged coma and poor outcome. On CT, DAI is suggested only when small petechial hemorrhages ($\leq5$–$15\,\mathrm{mm}$ hyperdense foci) appear at these characteristic sites, or when diffuse swelling effaces the sulci and cisterns; but these hemorrhagic foci are merely the visible 'tip,' marking a vastly larger burden of nonhemorrhagic shear injury that is invisible to CT. The signature insight, and the one that protects the patient, is the clinico-radiologic dissociation: a patient with GCS 3–6 and a near-normal CT has DAI until proven otherwise, and this discordance is itself the indication to escalate. MRI is the reference standard, with susceptibility-weighted imaging (SWI/GRE) exquisitely sensitive to the microhemorrhages and diffusion-weighted imaging detecting the nonhemorrhagic cytotoxic foci; diffusion tensor imaging quantifies the white-matter disruption that drives the prognosis. CT's role is therefore to exclude a surgical lesion and to raise the suspicion that mandates MRI, not to grade DAI.

Bayesian reasoning here is driven almost entirely by mechanism and clinical-imaging mismatch rather than by a conspicuous finding. A high-speed motor-vehicle collision or a fall from height, a low GCS, and the absence of a mass lesion that explains the coma constitute a strong prior for DAI; the demonstration of even a single petechial focus at the gray-white junction, splenium, or dorsolateral midbrain moves the posterior decisively and prompts grading. The principal mimics on CT and MRI are cerebral microbleeds of chronic hypertension or amyloid angiopathy (which favor the basal ganglia/thalamus and lobar cortex-subcortical junction respectively, rather than the callosal and brainstem shear sites and lack the trauma context), cerebral fat embolism (a scattered microhemorrhage pattern with a characteristic latency and systemic features), and diffuse vascular injury. The dominant failure modes are the gravest in neurotrauma: false reassurance from a normal CT, in which the reader equates a clean scan with a benign injury and fails to recommend MRI in a comatose patient; satisfaction of search, in which an obvious extra-axial hematoma absorbs attention while the punctate callosal and brainstem foci that change the prognosis are never sought; and undercalling diffuse swelling, where bilateral cisternal effacement is the only CT trace of catastrophic axonal injury. The countermeasure is a fixed habit of correlating the CT with the GCS, deliberately inspecting the splenium and dorsolateral brainstem on every trauma scan, and recommending MRI whenever the deficit outruns the CT.

Primary injury—the fracture, the hematoma, the contusion, the sheared axon—is fixed at the moment of impact; secondary brain injury is the cascade of subsequent, potentially preventable insults that unfolds over hours to days and that, more than the primary lesion, determines whether the patient lives and how well. Its imaging is the imaging of the consequences of a mass within a closed box, governed by the Monro-Kellie doctrine: because the adult cranium is rigid, the sum of brain, blood, and CSF volume is constant, and any added volume—hematoma, contusional edema, or the cytotoxic and vasogenic edema that accumulates around injured tissue—must be offset by displacing CSF and venous blood. This buffering keeps intracranial pressure (ICP) nearly flat over an early compensated phase, but the reserve is finite and the pressure-volume relationship is exponential, $P=P_0\,e^{kV}$, so once CSF and venous compensation are exhausted each additional milliliter drives a steep, decompensated rise in ICP. The two lethal currencies of secondary injury follow directly: mechanical displacement (herniation) and ischemic failure, the latter because cerebral perfusion pressure $\mathrm{CPP}=\mathrm{MAP}-\mathrm{ICP}$ collapses as ICP climbs, and when impaired autoregulation can no longer hold cerebral blood flow constant the brain becomes ischemic on top of its trauma.

The CT findings form a graded hierarchy that the reader interprets as a readout of position on the elastance curve and of impending herniation. Local mass effect (sulcal effacement, ventricular compression) gives way to midline shift, measured at the septum pellucidum, with $>5\,\mathrm{mm}$ generally significant; progressive shift produces the herniation syndromes, each with a stereotyped CT appearance and vascular complication. Subfalcine herniation slides the cingulate gyrus under the falx and may compress the anterior cerebral artery (medial frontal infarction). Uncal/transtentorial herniation pushes the medial temporal lobe over the tentorial edge, effacing the suprasellar and perimesencephalic cisterns, compressing the third cranial nerve (a blown pupil) and the posterior cerebral artery (occipital infarction), and producing the false-localizing Kernohan-notch hemiparesis from contralateral peduncle compression and brainstem Duret hemorrhages from stretched perforators. Central (downward) herniation caudally displaces the diencephalon, and tonsillar herniation through the foramen magnum compresses the medulla and is rapidly fatal. Obliteration of the basal cisterns is among the most reliable and ominous CT signs of critical mass effect, and obstructive or communicating hydrocephalus adds a further treatable volume load. These features are precisely the variables of the Marshall classification (diffuse injury I–IV by cistern status and shift, plus evacuated/non-evacuated mass lesions) and the Rotterdam score (Maas, 2005), which quantify mass effect and hemorrhage into validated outcome predictors and structure the report.

Management is imaging-driven and the findings map onto specific interventions codified in the BTF severe-TBI guidelines (Carney, 2017). Cisternal effacement, shift, and refractory ICP elevation drive escalation through osmotherapy, CSF drainage, and ultimately decompressive craniectomy, after which CT monitors for external herniation, contralateral or remote hematoma 'unmasked' by decompression, and evolving infarction. The ischemic limb is increasingly assessed with CT perfusion and CT angiography: perfusion maps disclose regions of reduced cerebral blood flow (CBF) and prolonged mean transit time at risk before frank infarction is visible on noncontrast CT, characterize posttraumatic vasospasm (with CTA), and help distinguish viable oligemic tissue from established infarction—an extension of the core/penumbra logic of stroke imaging into the traumatic setting. The failure modes are consequential and recurrent: underweighting infratentorial and intraventricular mass effect, where small volumes decompensate rapidly and skull-base beam-hardening obscures the cisterns that matter most; misjudging shift in the atrophic brain, where generous CSF reserve permits a large lesion with little shift (and the young brain, conversely, decompensates at small volumes); satisfaction of search after the primary lesion, missing the evolving infarct or the new contralateral hematoma on follow-up; and reading a single early scan as definitive when secondary injury is by definition a process over time. The disciplined approach quantifies and localizes mass effect, assigns a Marshall/Rotterdam category, names the herniation pattern and its vascular territory at risk, and—through serial CT and, where indicated, perfusion—tracks the secondary cascade that the entire enterprise of neurocritical care exists to interrupt.