Imaging in Coma and Brain Death

Abstract

In the comatose patient, urgent diagnosis can be a critical priority if appropriate interventions are going to be performed promptly. In many cases, imaging investigations will form a core component of this assessment. In others, where clinical criteria allow confident diagnosis, imaging may still be of significant benefit in providing confirmatory information and may also provide clinically useful prognostic data. In the critically ill comatose patient, confirmation of a diagnosis of brain death may be required. Although this diagnosis is based on clinical criteria, imaging has long been used to provide adjunct supportive information. In recent years, there has been an increased interest in the use of imaging to support a diagnosis of brain death as functional imaging modalities have improved. In this chapter, we will initially review the role of imaging in supporting diagnosis and prognostication in patients suffering from coma. We will discuss the optimal imaging strategies, specific disorders, and specific imaging findings which might help with differential diagnosis and prognostication. We will then discuss the role of imaging in supporting the diagnosis of brain death.

1. Introduction

Coma is marked by a lack of awareness and response to external stimuli, patients cannot be roused. The neurological mechanism underlying arousal is complex but appears to be based on the ascending reticular activating system (ARAS) in the pons and midbrain, which produces ascending efferent projections to the thalamus and from there to the cerebral cortex, which processes outputs from the ARAS, generating awareness. Injury to any of these areas can result in impaired consciousness. Lesions affecting cortical function bilaterally or isolated lesions affecting the brainstem can all give rise to coma [1]. Coma can be broadly classified as one of three types based on pathogenesis. These are structural abnormalities, general inhibition of neuronal function, and psychogenic aetiology [2].

Imaging investigations play an important role in the diagnosis of many causes of coma. Computed tomography (CT) is commonly performed as a first imaging investigation and valuable additional information can be provided by magnetic resonance imaging (MRI). CT is the most common initial imaging investigation; however, not every comatose patient requires a brain CT. In many cases, a metabolic cause, such as diabetic ketoacidosis, may be clinically diagnosed, obviating the need for imaging. In a series of large retrospective studies, 42–58% of non-traumatic coma patients had CT on admission. In a study of patients with coma of metabolic origin, 23% of patients had CT, which was abnormal in less than 6% [3]. In patients with a subsequently confirmed structural lesion, 90% had CT, which was abnormal in 84% of cases [4, 5]. However, imaging investigation should be undertaken urgently in patients where clinical assessment suggests a possible structural injury, those with preceding head trauma, and where the diagnosis is unclear. MRI is generally more sensitive than CT and can provide additional information in many cases. However, MRI is time-consuming and is more difficult in a comatose patient, requiring life support and detailed monitoring. CT, therefore, remains the first investigation with MRI as a potentially valuable adjunct in most cases. However, CT can miss treatable causes of coma, such as basilar artery occlusion, early infarction, particularly thalamic and brainstem stroke, venous sinus thrombosis, posterior reversible encephalopathy syndrome (PRES), and pontine myelinosis [1]. CT angiography (CTA) and CT perfusion (CTP) imaging can be used to supplement information from anatomical CT scans. They will extend the ability to diagnose early stroke and other vascular abnormalities. Despite this, MRI can still provide valuable additional information in some cases and should be considered if the diagnosis is uncertain. In some specific clinical situations, particularly ischemic stroke, early MRI provides sufficient additional information that it is a standard first-line investigation in some centres.

2. Disease-specific imaging findings

2.1 Traumatic brain injury (TBI)

TBI is common, with 250,000 hospitalizations and 50,000 deaths annually in the US [6]. It can produce a wide range of long-lasting problems, including cognitive deficits, mental health issues and physical problems, such as headaches and sleep disturbance. In the long term, it has been associated with an increased risk of Alzheimer’s disease and Parkinsonism. Severe TBI can result in coma either in the acute phase or later in the progression of brain injury [7].

CT is an essential step for investigating severe TBI and is aimed initially at identifying focal injuries that might benefit from neurosurgical intervention. These include extradural or subdural haemorrhage, intra-cerebral haemorrhage and changes indicating a significant abnormality in ICP, such as loss of CSF space and brain herniation. Approximately 10% of patients with severe TBI require neurosurgery following initial CT [8, 9].

However, CT is insensitive at identifying cerebral contusion and diffuse axonal injury (DAI), which are present in most severe or fatal cases. DAI is a major underlying mechanism of long-term and severe disability in TBI survivors. As a result of this insensitivity to DAI, CT has minimal value in predicting outcomes following severe TBI [10].

Post-mortem examination following death from TBI shows the most common secondary injury is cerebral ischaemia. Reduction in blood flow due to changes in intracranial pressure or traumatic occlusion of small arterial vessels can cause extensive infarction. Contrast-enhanced CT with dynamic acquisition can produce maps of regional perfusion, a technique known as CT perfusion imaging (CTP). Many authors have shown significant changes in perfusion immediately after and in the long-term follow-up of TBI survivors [11]. CTP is also predictive of poor functional outcome based on global hypoperfusion in the early phases of TBI and the presence or absence of perfusion abnormalities in the frontal lobe [12].

In contrast to CT, MRI has a high sensitivity for demonstrating DAI [13]. DAI, resulting from shearing injuries of the brain, is commonly accompanied by the presence of microscopic hemorrhages. These can be seen on MRI and particularly on sequences sensitive to local variations in the magnetic fields produced by changed haemoglobin products. These T2* weighted sequences are more sensitive to microhemorrhages than conventional MR. The development of susceptibility-weighted imaging (SWI) sequences has provided routine acquisition protocols that are far more sensitive for the detection of micro hemorrhage than even standard T2* weighted approaches [14]. SWI provides high-resolution 3D information that uses the same information as T2* weighted images but supplements this with information on phase shifting secondary to the presence of paramagnetic molecules. SWI sequences have exquisite sensitivity to the presence of deoxyhemoglobin, hemosiderin, and intracellular methemoglobin [14]. The number and size of SWI lesions and the presence of lesions in the brainstem have been shown to predict poor outcomes at 6 and 12 months (Figure 1) [13].

Figure 1.

Images from a 43-year-old man, comatose after a major traumatic brain injury following a road traffic accident. A: T2 weighted image showing bilateral encephalomalacic changes in the frontal lobes (red arrow). B:FLAIR image showing associated gliosis (scarring, red arrows). C: SWI image showing haematoma in the walls of the encephalomalacic areas. D-F: SWI imaging showing extensive haemorrhage in the midline of the midbrain (D, red arrow), pons (e.g., red arrow), cerebellar vermis (F, red arrow) and in the cerebellar hemispheres (E and F, yellow arrows).

2.2 Mass lesions

A space-occupying lesion will compress adjacent brain structures. Because the skull is a fixed volume this will cause displacement of CSF from the cranial cavity and a reduction in blood volume, particularly in the cerebral veins. As the tumor grows, the volume of the interstitial fluid space will also be reduced. These changes are associated with increases in intracranial pressure (ICP), which will reduce cerebral perfusion pressure, eventually reducing cerebral blood flow [1]. The mass effect can be exacerbated by tumour-associated cerebral oedema or by occlusion of ventricular CSF flow leading to a “trapped ventricle” [15]. Identifying secondary hydrocephalus is of prime importance since it can lead to acute deterioration of consciousness level, which drainage procedures can treat. In addition to localized mass effects, adjacent normal brain tissue will be displaced, which may herniate below the rigid dural membranes of the falx cerebri and tentorium cerebelli [16]. It has long been taught that herniation of the uncus of the temporal lobe, compressing the midbrain, was a principal etiology of coma. However, it has now been shown that lateral displacement of the cerebral hemispheres is more closely related to the level of consciousness [17]. Brain swelling, herniation and raised intracranial pressure can cause arterial compression or critical reduction in regional cerebral blood flow, leading to secondary infarction. Compensatory mechanisms can be overwhelmed far more quickly with rapid progression of mass effect, such as is seen in an acute intracranial haemorrhage, whereas a slowly growing tumour of similar size may not produce the same severity of symptomatology.

Where coma results from a mass lesion, CT will usually be diagnostic in identifying the mass itself and the severity of cerebral displacement and herniation. MRI will be indicated in most cases; it is likely to provide further information as to the nature and anatomical location of the mass, and diffusion-weighted imaging will provide any areas of early secondary ischemic change due to arterial compression.

2.3 Ischaemic Stroke

When a diagnosis of ischaemic stroke is suspected, imaging should be available urgently and rapidly. CT remains the most used primary investigation [18, 19]. None contrast CT rules out intracranial haemorrhage, mass lesion, or other differential diagnosis, providing adequate information to base a decision on the use of thrombolysis. Thrombosis within one of the intracranial arteries may be seen as an area of high attenuation, indicating arterial occlusion [20]. Early CT may be relatively insensitive with subtle signs such as loss of grey-white matter differentiation and slight effacement of sulci reflecting cerebral swelling. Some early signs on CT are quite subjective and can be assessed more objectively using the Alberta Stroke Program Early CT Score (ASPECTS) [21]). Automatic AI-based systems to support clinicians in early stroke identification are becoming common (Figure 2) [22]. CTA of the intracranial arteries and the major arteries of the neck should also be performed urgently to identify arterial occlusion and guide the use of thrombectomy [23]. CTP can be performed to identify areas of potentially salvageable and non-salvageable ischaemia (Figure 3) [24]. Some centres use rapid MRI protocols specifically for stroke, which include diffusion-weighted imaging (DWI). DWI is exquisitely sensitive to stroke at a very early stage, and adding SWI sequences will identify small areas of micro haemorrhage, which could indicate embolic disease.

Figure 2.

A: unenhanced CT showing abnormal high attenuation in the left middle cerebral artery (dense MCA sign, red arrow). B: subtle area of low attenuation in the left insula (red arrow). C: printout from artificial intelligence-based analysis software showing areas of abnormality from the non-enhanced CT and providing automated ASPECTS score.

Figure 3.

CTP study in the case illustrated infigure 2, showing the time to maximum contrast enhancement (Tmax), mean transit time (MTT), cerebral blood flow (CBF) and cerebral blood volume (CBV). The red arrow indicates an area of reduced CBV and CBF representing unsalvageable brain tissue. The white arrow shows an area of reduced CBF but maintained CBV indicating potentially salvageable tissue.

2.4 Dural Venous Sinus Thrombosis (DVST)

DVST accounts for 1–2% of strokes in adults. It can be secondary to an infection adjacent to major dural venous sinuses, compression by tumour, pregnancy, dehydration, or hypercoagulable states [25]. It may result in brain swelling with raised intracranial pressure, cerebral infarction, coma, and death. CT may demonstrate focal brain swelling and show the thrombosed dural venous sinus as an area of hyperdensity and as a filling defect on enhanced images. On MR images, swelling with oedema can be seen, particularly in grey matter. The thrombus may be seen as a hyperintense signal on T1 weighted images, lying within the vein. Standard MRI sequences have a relatively low sensitivity (79%) and specificity (89%), so the threshold to perform vascular imaging must be low [26]. Diagnosis relies on vascular imaging with magnetic resonance angiography (MRA) or CTA demonstrating obstruction of major veins. Demonstration of haemorrhage in the area drained by the obstructed veins or thrombosis within the deep cerebral venous system is associated with poor outcomes (Figure 4) [27].

Figure 4.

26-year-old man presenting with epilepsy and coma. A: T2 weighted image showing extensive swelling and high T2 signal in the right hemisphere (red arrow). B and C: FLAIR images showing the extent of signal abnormality. D: magnetic resonance arena grant showing no filling of the distal superior sagittal sinus or transverse sinuses. E: T2 weighted image showing thrombus within the superior sagittal sinus (red arrow).

2.5 Intracranial and Subarachnoid Hemorrhage

CT has a high sensitivity for acute blood within the CSF or brain tissue. Due to its ease of access and speed, CT is the first line investigation for suspected subarachnoid or intra-cerebral haemorrhage [28]. MRI can detect subarachnoid blood with the same sensitivity as CT if FLAIR sequences are performed but is rarely used as a first line investigation [29].

CT is also highly sensitive to intracerebral haematoma and is a preferred primary investigation where this is suspected [28]. MRI is more sensitive to the presence of small haemorrhagic lesions and may, rarely, identify lesions that are missed by CT, in addition, it is more sensitive to the detection of chronic haemorrhage and for the demonstration of underlying structural lesions.

Imaging is aimed at identifying patients who could benefit from acute intervention such as drainage of a large forebrain or cerebellar haematoma or who require CSF drainage due to hydrocephalus.

Identifying any structural source of haemorrhage is important and can be difficult, particularly where intra-cerebral haemorrhages have compressed adjacent abnormal vessels. CTA and MR are both highly sensitive for the demonstration of cerebral aneurysm and arteriolar-venous malformation (AVM) and CTA is commonly performed at the time of diagnosis to inform management [30].

Imaging can provide valuable prognostic information. A large baseline volume of haematoma, expansion, and intraventricular blood are associated with poor outcomes. Bleeding into the cortex has an improved outcome but is associated with a 3 to 4 times increase in risk of re-bleeding [31, 32]. In addition, 1/3 of intracerebral haematomas show growth within three hours of onset [33]. This is common in large haematomas and can be detected by routine early repeat imaging and a low threshold for repeat CT in the presence of clinical deterioration.

2.6 Hypoxic ischaemic encephalopathy (HIE)

HIE is usually a consequence of acute circulatory and respiratory failure. Common causes include cardiac arrest, carbon monoxide poisoning, asphyxiation, and drowning. Hypoxia affects the cortex and subcortical grey matter first since they are most oxygen-dependent. Hypoxia inhibits Na + and K+ membrane pumps failing cellular integrity with the release of glutamate and overstimulation of N-methyl-D-aspartate receptors, loss of intracellular calcium, and damage to the mitochondrial respiratory chain [34].

CT is relatively insensitive and is often normal in the early stages but may show mild loss of grey-white matter junction or evidence of diffuse oedema with brain swelling. A few of these patients will show the “reversal sign” where the CT attenuation of grey and white matter reverses during the first 24 hours after insult [35]. Linear areas of high attenuation may be seen in the cortex, reflecting cortical laminar necrosis [36].

MRI can show changes within an hour after insult [37]. DWI imaging may demonstrate diffusion restriction due to early cytotoxic oedema during the first 24 hours in the cortex and basal ganglia (Figure 5). Imaging over the next 24 to 48 hours, shows extension of the area of restricted diffusion, particularly in the cortex, and especially the peri-Rolandic and occipital cortices, and in the basal ganglia and cerebellum [38, 39]. The changes are almost invariably symmetrical and bilateral. These changes usually resolve after approximately one week. In the later stage, hyperintensities may be seen in the cortex on T1 weighted imaging after about two weeks, indicating cortical laminar necrosis. In the late phase, beyond two weeks increased T2 signal will appear in the damaged areas with the same distribution where reduced diffusion was initially identified [40]. These appearances are not completely diagnostic and may be mimicked by changes resulting from other conditions [37].

Figure 5

FLAIR i.mages in a 48-year-old woman following resuscitation from cardiac arrest. Note high signal in the caudate nucleus, putamen, globus pallidus and substantia nigra (red arrows).

MRI performed on days 2–4 can provide significant prognostic information [41]. Extensive changes in DWI or FLAIR images are associated with poor outcomes [42]. Signs of global cerebral oedema on early CT examination are also a poor prognosticator, having a false positive predictive rate for death of 0% in one study [43]. Whole brain DWI improves the prediction accuracy for outcome over neurological examination alone by 38% [41]. Following respiratory arrest, one study has shown that DWI lesion size greater than 20 mL accurately predicted poor outcomes (11). Another study showed that DWI 48 hours post cardiorespiratory arrest with the lesion volume growth of 20 mls had a false positive rate of 0% for poor outcome defined as a cerebral performance category 3–5 [44]. Another study [45] showed a false positive rate of 23% for predicting a cerebral performance category 3–5 based on the presence of cortical or basal ganglia lesions on MRI. A single small study looking at magnetic resonance spectroscopy also demonstrated high specificity for predicting poor outcomes [46].

2.7 Posterior Reversible Encephalopathy (PRES)

PRES is a rare condition resulting from the breakdown of cerebral autoregulation. It is most seen in eclampsia and pre-eclampsia but may be associated with hypertension, chemotherapy, organ transplantation, and some autoimmune diseases [47]. CT may be normal or show cortical hypodensity areas in the occipital and posterior parietal lobes. MRI may show oedema in watershed regions and areas the posterior cerebral artery supplies. More rarely, changes can be seen in the frontal and temporal watershed areas. Up to 25% of cases may show intraparenchymal haemorrhage [48], and CTA, MRA, and CTP may show areas of cerebral vasospasm and reduced perfusion. Imaging changes can regress over weeks. The extent of abnormality on MRI correlates with outcome, and non-survivors show significantly higher scores for lesions demonstrated on T2 and DWI images [49].

2.8 Sepsis Related Encephalopathy (SRE)

In severe systemic sepsis, the brain can be exposed to an inflammatory response. Sepsis may also be associated with ischaemic brain lesions, cerebral haemorrhage, and micro-abscesses. In severe cases, multifocal necrotizing leukoencephalopathy may be seen [50].

Neuroimaging is often normal, but when abnormalities occur, they take the form of multiple T2 high-signal lesions in the white matter, which may vary over time. Ischaemic lesions, particularly in the Centrum semiovale [51], and signs of peri-ventricular inflammation with sub-ependymal high T2 signal may be seen. If cerebral autoregulatory mechanisms are disturbed, then vasogenic oedema can occur [52].

2.9 Herpes Simplex Encephalopathy (HSE)

HSE is a direct cerebral infection caused by herpes simplex virus type I [53]. CT is usually normal in the first week, although subtle swelling of the temporal lobe may be identified. MRI is more sensitive and shows high signal on T2 weighted images with restriction of diffusion in the medial temporal lobes, base of the frontal lobes, insular cortex and insular in over 90% of patients (Figure 6) [54, 55]. Areas of involvement can show contrast enhancement. Brain swelling is commonly a significant feature with midline shift. Magnetic resonance can, very rarely, be normal. Lesions that can be identified on CT are predictive of poor outcome and prolonged disease course [56, 57] and the extent of lesions on MRI also predicts poor prognosis [55]. Measurement of the apparent diffusion coefficient (ADC) from diffusion-weighted images illustrates restricted diffusion in the acute phase of the disease, and this is associated with poor outcome at discharge [58].

Figure 6.

FLAIR images in a patient presenting with rapidly progressive headache, and confusion leading to coma. Images show swelling of the right temporal low inferior right frontal lobe together with the involvement of the insular cortex bilaterally and the contralateral medial temporal low. Confirmed diagnosis of herpes simplex encephalitis.

2.10 Autoimmune encephalopathies

Acute disseminated encephalomyelitis (ADEM) is a diffuse encephalomyelitis that can occur following or during systemic infection, following vaccination or spontaneously [59]. The mechanism is not understood, although there is evidence of an autoimmune basis. CT and MR may be normal, but in the acute phase, T2-weighted images may show lesions in the deep white matter, which have the imaging characteristics of demyelination and may enhance. Up to 30% will have lesions in the brainstem and spinal cord [60]. DWI shows restricted diffusion during the acute phase, gradually increasing, reflecting initial cytotoxic swelling and eventual cell death. The distribution and appearance of the lesions is indistinguishable from those seen in Multiple Sclerosis in most cases. However, they may be more extensive and involve cerebellar white matter and basal ganglia [61]. Lesions considered relatively typical of Multiple Sclerosis occur where small penetrating veins leave the brain at a point where white matter abuts CSF. These include the “lumpy-bumpy” lesions at the margins of the lateral ventricle [62] and callososeptal junction lesions [63]. These have not been reported in ADEM and may help to some extent in differentiation [64]. During recovery, lesions largely regress, but some may persist.

2.11 Metabolic encephalopathies

2.11.1 Hypoglycaemic and Hyperglycaemic Encephalopathy

Both hyper and hypo-glycaemic insult can cause a decline in cell membrane ATPase pump activity and be associated with the release of aspartate, cytotoxic oedema and neuronal death.

In hyperglycaemic encephalopathy, seizure activity is common, and some imaging findings are associated with this. Typical features associated with focal seizures are areas of subcortical low signal on FLAIR images, often associated with high signal in the adjacent cortex. These transient changes show complete or at least partial resolution [65]. CT may show hyperdense changes in the putamen with a corresponding increase in T1 weighted signal on MR but with no T2 signal changes (Figure 7) [66]. These changes have been associated with the relatively unusual complication of chorea or hemiballismus and probably reflect diffuse haemorrhage or calcification within the basal ganglia [67].

Figure 7.

T1 weighted MR scan showing high signal in the Putamen bilaterally in hypoglycaemic encephalopathy. Case contributed by Chee Kok Yoon.

In hypoglycaemic encephalopathy, low attenuation lesions can be seen on CT in the basal ganglia, cortex substantia nigra, and hippocampus [68]. MR will show a similar distribution of high signal abnormalities on T2W and FLAIR images with corresponding low- signal areas on T1 weighted images. MR also commonly demonstrates lesions in the posterior limb of the internal capsule [69]. These lesions persist on follow-up scans at one year or more. Abnormalities can be seen on DWI images with evidence of restricted diffusion in the grey and white matter (Figure 8). Extensive DWI changes seen in the cortex, hippocampus, deep white matter, and basal ganglia, which fail to regress on follow-up imaging, have been associated with poor outcomes (Figure 8) [69, 70].

Figure 8.

Diffusion-weighted images from a patient with hypoglycaemia. A bilateral high DWI signal is seen in the cortex. Contributed by Mohammad A. ElBeialy,Radiopaedia.org.

2.11.2 Uremic Encephalopathy

CT is usually normal although subtle attenuation changes may be seen in the basal ganglia and internal capsule bilaterally [71]. MRI shows hyperintensities on T2 weighted sequences and high signal on DWI in the same areas and sometimes in the cortex; these are usually associated with T1 hypo intensity [72]. Basal ganglia lesions are moderately distinctive showing expansion of the basal ganglia and particularly high signal in the external capsule and medullary laminae of the putamen and globus pallidus. This gives rise to a distinctive appearance often called the “lentiform fork sign” (Figure 9). Many of these findings regress following dialysis, but in those with recurrent uraemia, changes are progressive, and cerebral atrophy occurs.

Figure 9.

Uraemic encephalopathy showing the lentiform fork sign. Case contributed by Coenraad Hattingh,Radiopaedia.org.

2.11.3 Hepatic Encephalopathy

CT may be normal or show changes in the anterior pituitary, subthalamic nucleus and tectal plates [64]. MRI classically shows areas of high signal in the basal ganglia on T1 weighted images; hyperintensity of the globus pallidus is particularly characteristic and can be seen in most cirrhotic patients. It is believed that this is due to the position of manganese in the central nervous system with a preferential distribution to the globus pallidus (Figure 10) [73]. In addition, scanning in acute cases may show increased T2 signal in the periventricular white matter, internal capsule, corticospinal tracks, thalamus, and cerebral cortex. Reversible cerebral oedema can be seen in fulminant hepatic failure and predominantly affects the globus pallidus with significant changes in DWI and T2 signal [74, 75].

Figure 10.

Hepatic encephalopathy showing high signal in the basal ganglia on T1 weighted image (left) and extensive cortical high signal on DWI images (right). Case contributed by Andrew Dixon,Radiopaedia.org.

2.11.4 Hypernatraemic and hyponatraemic encephalopathy

Acute hyponatraemia produces an osmotic imbalance, causing water to flow across the blood–brain barrier resulting in cerebral oedema. Glial cells and neurons loose solutes, shifting fluid from intracellular to extracellular spaces [76]. Correction of hypo/hypernatraemia can result in osmotic demyelination, known as central pontine myelinosis (CPM). CT is usually normal but may rarely demonstrate low-intensity lesions in the pons (Figure 11). MRI will show a high signal in the central pons on T2 weighted images and DWI. Lesions may less commonly be seen in the midbrain, thalamus, hippocampus, caudate, putamen, and cortex. Interestingly, the extent and volume of signal abnormality on MRI is not predictive of outcome, often does not correspond to clinical deficit, and may persist after significant symptomatic recovery [77, 78].

Figure 11.

Central pontine myelinosis shown on T2 weighted (left) and diffusion weighted (right) images. Case contributed by Yuliia Solodovnikova,Radiopaedia.org.

2.11.5 Wernicke’s encephalopathy (WE)

Wernicke’s encephalopathy results from thiamine deficiency, usually secondary to alcohol abuse, malabsorption, increased metabolism, or hemodialysis. It results in osmotic imbalance with perineural oedema and neuronal swelling [79]. Lesions show a specific distribution explained by higher dependence on thiamine for metabolism. CT may show areas of low attenuation in the brainstem, diencephalon and third ventricle periventricular areas. On MRI, high signal on T2 weighted and DWI images is seen symmetrically in the medial thalami and periventricular region of the third ventricle (85%), the periaqueductal area (65%), the mammillary bodies (58%), the tectal plate (38%) in the dorsal medulla (8%) (Figure 12). Contrast enhancement of the mammillary bodies is much commoner in WE due to alcohol abuse. In the chronic phase, signal changes may regress, but there is commonly focal cerebral atrophy, particularly affecting the mammillary bodies [79, 80].

Figure 12.

Wernicke’s encephalopathy showing high signal in the peri-aqueductal grey matter (long arrow) and mammilary bodies (short arrow). Case contributed by Yves Leonard Voss,Radiopaedia.org.

3. Imaging in brain death (BD)

In the United States, brain death (BD) is defined by law as “the irreversible cessation of all functions of the entire brain” [81, 82]. However, these definitions differ between countries, and, for example, Canada and the United Kingdom require only irreversible cessation function in the brainstem, focusing on arousal and ability to sustain respiration. There has been ongoing debate concerning the diagnostic accuracy of current tests; although accuracy appears to be high, false positive diagnoses of BD do occur [83, 84]. Whatever the initial insult, brainstem death seems to follow a common pathway with increases in intracranial pressure (ICP), and reducing cerebral circulation causing anoxic brain injury. Once ICP reaches arterial levels, cerebral blood flow will cease. The brainstem is relatively resilient to anoxia and is the last to lose function [82]. The diagnosis of BD is based on careful clinical examination combined with an apnoea test to assess medullary function as carbon dioxide levels rise in the bloodstream.

Imaging has been used as an ancillary test for BD using several methods, all aimed at proving that there is no effective cerebral blood flow and/or capillary perfusion [85]. Digital subtraction angiography has been used for many years but is labour-intensive [86]. CTA and MRA have been employed [87, 88] but have been criticized for high false positive rates. Blood flow can also be checked using transcranial Doppler at the bedside, but this is very operator-dependent and cannot be used where there are skull defects or ventriculostomy [89]. There has been considerable discussion about the optimal imaging parameters that should be examined to support a diagnosis of BD. Even if there is large vessel flow, this does not indicate the presence of capillary perfusion, so a method that can also assess capillary perfusion is desirable [90]. The most common method has been Single Photon Emission Computed Tomography (SPECT) using a lipophilic agent such as technetium-90 m hexamethylpropyleneamine oxime (Tc99 HMPAO). This enables analysis of cerebral blood flow and uptake by metabolically active cells directly reflecting capillary perfusion. Radionuclide scanning with Tc99 HMPAO has high sensitivity and positive predictive value approaching or reaching 100% and specificity over 90% (Figure 13) [85] .

Figure 13.

Brain death following cardiac arrest. Left panel shows Tc99 HMPAO profusion images with no evidence of cerebral uptake. Right panel shows CT scan demonstrating severe cerebral swelling with complete loss of CSF spaces. Case contributed by Ryan Thibodeau,Radiopaedia.org.

There has recently been considerable interest in using CT angiography either alone or combined with CT perfusion imaging to demonstrate both blood flow and perfusion (Figure 14). In one study, which notably included patients who were examined for brainstem death but were not clinically diagnosed as dead, CTP showed a sensitivity of 97% with a specificity of 100%. CTP is easily available, does not require specific isotopes, and can be easily performed in any hospital. CTP has the advantage of giving specific anatomical data as to the distribution of loss of perfusion so that cases who have lost brainstem function due to isolated perfusional abnormality can be correctly identified [85].

Figure 14.

Results of CTA (a) and CTP (b) in a patient diagnosed with brain death. CTA shows filling of cortical branches of the right and left MCA (arrows) and was classified as negative, i. e. inconsistent with the diagnosis of BD. CTP reveals perfusion values below the thresholds for non-viable tissue and, contrary to CTA was interpreted as positive, i.e. consistent with the diagnosis of BD, from [91].

4. Conclusions

Imaging investigations form a vital component of the diagnosis of patients in coma. They may provide evidence for differential diagnosis and, equally importantly, valuable prognostic information. For some diseases, such as intracranial haematoma, frequent imaging follow-up can be helpful in avoiding early complications. In brain death, imaging provides very valuable adjunct data, although it is not as yet accepted into the principal criteria for the diagnosis of brain death. CTP, although relatively early in its introduction, appears to be particularly promising in this application.