Article Text

Download PDFPDF

Imaging in acute ischaemic stroke: pearls and pitfalls
  1. James Caldwell1,
  2. Manraj K S Heran2,
  3. Ben McGuinness1,
  4. P Alan Barber3,4
  1. 1 Department of Neuroradiology, Auckland City Hospital, Auckland, New Zealand
  2. 2 Department of Neuroradiology, Vancouver General Hospital, Vancouver, Canada
  3. 3 Department of Neurology, Auckland City Hospital, Auckland, New Zealand
  4. 4 Centre for Brain Research, University of Auckland, Auckland, New Zealand
  1. Correspondence to Dr James Caldwell, Department of Neuroradiology, Auckland District Health Board, 2 Park Rd, Grafton, Auckland 1142, New Zealand; jamesrc{at}


Prompt and accurate diagnosis is the foundation of acute ischaemic stroke care. Multiple positive endovascular thrombectomy trials in ischaemic stroke patients with large vessel occlusions have further emphasised this but also added complexity to treatment decisions. CT angiography is now routine for patients who present with an acute stroke syndrome around the world. Members of the neurology and stroke teams (rather than radiologists) are often the first doctors to lay eyes on the CT images and are best equipped to integrate the clinical picture with the imaging findings. A sound understanding of acute stroke imaging is therefore essential for clinicians who work with acute stroke patients. This review describes some pearls we have gleaned from our own experience in acute stroke imaging as well as some potential follies to be avoided.

  • Neuroradiology
  • stroke

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The landscape of acute ischaemic stroke treatment has changed dramatically since the publication of multiple positive endovascular thrombectomy trials in patients with large vessel occlusion.1

Neurologists and stroke physicians need to be able to identify ischaemic changes that may be subtle on early CT brain scans as well as to exclude haemorrhage and stroke mimics. The need to detect large vessel occlusions in patients presenting with stroke has resulted in a substantial increase in the use of CT angiography in the acute setting; many would now consider this to be part of standard care.2 While recognising a large vessel occlusion may be obvious, there are many nuances to interpreting CT angiography that are not. There are technical considerations, subtle findings and the fact that cerebrovascular anatomy is highly variable. In some centres, more advanced CT techniques such as multiphase CT angiography and CT perfusion imaging can help further to select patients for stroke treatments.3 4

MRI, especially diffusion-weighted imaging (DWI), is a powerful tool in the diagnosis of acute stroke but is less commonly used as first-line imaging in the acute setting, mainly due to issues with access.5 It is important for clinicians to have a clear understanding of the limitations of CT imaging and the situations where MRI may provide key additional information. This review illustrates some practical points in interpreting and using imaging for patients presenting with acute ischaemic stroke.

Non-contrast CT

CT brain scans are the workhorse of acute stroke imaging. CT is readily available in most centres, quick to perform and shows haemorrhage clearly. However, ischaemic changes on CT can be subtle in the short therapeutic windows for intravenous alteplase and endovascular therapy. These may include subtle loss of grey–white matter differentiation, particularly in the insular region or basal ganglia in those with occlusion of the M1 segment of the middle cerebral artery. There may also be a hyperdense middle cerebral artery segment. The degree of ischaemic change using the Alberta stroke programme early CT score (ASPECTS)6 can also help to select patients for endovascular therapy. ASPECTS is a 10-point topographic CT scan score where the middle cerebral artery vascular territory is divided into 10 segments and where one point is deducted for every region involved, so that lower scores indicate greater ischaemic change. Patients were excluded from the SWIFTPRIME study of endovascular thrombectomy if the ASPECTS score was less than 6 and from the REVASCAT study if the score was less than 7.7 8

CT angiography

CT angiography is now a standard investigation in centres where endovascular clot retrieval is available. In the setting of acute stroke, an intravenous contrast bolus is given, and imaging is performed from the aortic arch to the vertex. CT angiography shows the arteries of the neck as well as the circle of Willis and its branches; occlusion of these large arteries is usually readily identified. CT angiography uses a single rapid helical acquisition and the resulting images effectively represent a snapshot in time.

Carotid pseudo-occlusion

With CT angiography, if there is slow blood flow within a particular artery, the contrast bolus may not yet have reached and opacified the artery when the scan is acquired; this can easily be mistaken for an occluded vessel (‘pseudo-occlusion’). Occasionally, a severe stenosis of the proximal internal carotid artery (ICA) can produce sufficiently slow flow to give a pseudo-occlusion appearance on CT angiography. However, pseudo-occlusion is most common in the cervical ICA, either due to an embolus in the carotid terminus or a so-called tandem lesion with severe stenosis at the ICA origin along with an middle cerebral artery embolus.9

Correctly recognising a pseudo-occlusion is critical to providing appropriate treatment. In the context of an acute stroke syndrome, having an ICA terminus embolus or a tandem lesion puts a patient at high risk of developing a large stroke, and both of these are potentially treatable. In the Hermes collaboration meta-analysis of the major mechanical thrombectomy trials, patients with a tandem lesion benefited from immediate endovascular treatment.10

Figure 1 shows the case of an 80-year-old man with an acute left middle cerebral artery syndrome. CT angiography shows vague contrast density that fades in the proximal cervical ICA (figure 1A), representing a small amount of diluted contrast. The axial CT angiogram image shows no contrast opacification at the carotid siphon (figure 1B). The subsequent catheter angiography images immediately before endovascular therapy show that the cervical ICA and carotid siphon are patent but have sluggish flow. Figure 1F shows an embolus lodged at the carotid terminus, which is causing the slow flow.

Carotid pseudo-dissection

Slow arterial flow can also mimic dissection on CT angiography (‘pseudo-dissection’).11 In a slow-flow state, a small volume of contrast making its way into the cervical ICA tends to lie in the dependent portion of the vessel, due to its high density relative to blood. This can mimic the flame-shaped appearance of a dissection, as in the catheter angiogram image in figure 1C. With a pseudo-dissection, other features of a true dissection are absent, including expansion of the vessel and vessel wall thickening due to intramural haematoma.

Figure 1

CT angiogram and catheter angiogram images showing an example of a carotid pseudo-occlusion due to an embolus at the carotid terminus.

Carotid dissection

For comparison, figure 2 shows images of a 50-year-old man with true arterial dissection who presented acutely with right-sided weakness. CT angiography shows marked narrowing of the lumen of the left cervical ICA below the skull base (figure 2A) but with clear expansion of this vessel on the non-contrast CT with a subtle hyperdense crescent due to intramural haematoma (figure 2B). These findings are highly specific for arterial dissection. It is worthwhile looking specifically at the ICA just below the skull base on the non-contrast CT head scan for this finding when suspecting a carotid dissection.

Figure 2

CT angiogram (A) and non-contrast CT (B) images showing changes due to dissection of the left internal carotid artery below the skull base.

Practical points

When confronted with a patient with an acute anterior circulation stroke syndrome and a seemingly occluded cervical ICA on CT angiography, there are several useful questions to consider:

  1. Is there previous carotid imaging that is readily available for comparison?

  2. Is there calcified atheroma at the ICA origin suggesting possible underlying stenosis?

  3. Is there an abrupt cut-off of opacification at the proximal ICA on the CT angiogram (suggesting true occlusion) or does contrast density fade up the cervical ICA (suggesting pseudo-occlusion)?

  4. Is there hyperdensity at the ICA terminus or M1 segment on the non-contrast scan due to an embolus?

  5. Are the carotid terminus, M1 and A1 segments patent on the CT angiogram? If they are, then the patient’s stroke symptoms could be due to hypoperfusion of the anterior circulation secondary to an acute proximal ICA occlusion, rather than due to an embolus.

Where there is doubt as to whether a cervical ICA is truly occluded, a catheter angiogram remains the gold standard to answer this important question.

Use of a delayed post-contrast scan

If the clinician suspects a pseudo-occlusion while the patient is still on the CT table, it can be confirmed by performing another post-contrast CT scan from below the carotid bifurcations up to the vertex. If there is visible contrast filling the cervical ICA on this delayed scan, then the artery is patent. The delayed phase may also better outline the location and length of an embolus responsible for the slow-flow state. Note that contrast filling in the carotid siphon on a delayed CT can also be due to collateral filling rather than slow anterograde flow through the cervical ICA. This typically occurs in an occluded or near-occluded ICA due to atheroma at the carotid bifurcation. External carotid to ICA collaterals, particularly through the ophthalmic artery, become significant in slow-flow states.12

Performing a delayed phase CT scan should not be allowed to cause significant delays in acute stroke treatment decisions. For example, if a pseudo-occlusion is suspected on CT angiography in a potential endovascular therapy candidate and the patient is already off the CT table, then a diagnostic catheter angiogram in the interventional suite is the appropriate course of action. A catheter angiogram will show a slow-flow state more clearly than a delayed phase CT scan, as it is time resolved. Depending on the findings, endovascular therapy can then proceed immediately. In our practice, performing a delayed post-contrast CT scan to confirm a carotid pseudo-occlusion is a rare event, as the patient is often out of the CT scanner by the time a pseudo-occlusion is suspected.

Some centres use multiphase CT angiography (including one or more delayed phases) routinely for patients presenting with stroke, emphasising that the delayed phase is useful to assess collateral arterial supply to the brain at risk,3 which is discussed in more detail later.

Intracranial atheroma

Intracranial atherosclerotic disease is an important and sometimes under-recognised cause of stroke. It is more common in Asian populations and is associated with a high risk of recurrent ischaemic events.13 CT angiography demonstrates well the presence of intracranial atherosclerotic disease.14 However, in the context of a patient presenting with an acute stroke syndrome with a large vessel occlusion, it can be difficult to tell if there is an underlying stenosis due to atheroma. Clues to the underlying atheromatous disease are the presence of mural calcification at the site of the occlusion as well as calcified atheroma and segments of luminal narrowing elsewhere in the intracranial circulation. Occasionally, mural low density due to lipid content in soft plaque is appreciable at the site of an intracranial stenosis or occlusion due to atheroma.

Figure 3 shows images from a 60-year-old Asian man who presented with an acute basilar syndrome. The non-contrast CT shows hyperdensity of the mid to upper basilar artery due to thrombus (figure 3A and B) with CT angiography confirming occlusion in this location. Close examination of the non-contrast CT scan shows a segment of low density due to soft plaque in the arterial wall just proximal to the thrombus (figure 3B, arrow). Plaque with stenosis was confirmed in the interventional suite, where the patient was treated with basilar angioplasty and stenting to maintain vessel patency after mechanical thrombectomy (figure 3C is the catheter angiogram image pretreatment, and figure 3D is post-treatment). A midbasilar thrombosis is a red flag for underlying stenosis, as a purely embolic basilar occlusion more commonly occurs at the basilar bifurcation. In our experience, endovascular treatment of patients who have underlying stenosis due to intracranial atherosclerotic disease is more challenging. It may not become clear until after thrombectomy that a patient with a large vessel occlusion has an underlying arterial stenosis. At this point, if thrombectomy achieves vessel patency, then questions arise over the need for angioplasty or stenting; however, the jury is still out on whether these treatments give enduring benefit.15

Figure 3

CT and catheter angiogram images show a case of basilar thrombosis with underlying stenosis due to atheroma.


Decisions regarding treatment for acute ischaemic stroke require consideration of a large number of factors that are made under time pressure. The goal of therapy is to restore blood flow to brain that is still viable and not to brain that has already infarcted since this carries the risk of haemorrhagic transformation and cerebral oedema.16 Treatment decisions can be refined by correlating a patient’s clinical deficit with the location of changes on imaging.

Figure 4 shows CT images of a 63-year-old man who presented 4 hours after onset of a receptive dysphasia with no other neurological deficit. His non-contrast CT shows evolving infarction of the left superior temporal gyrus, adjacent temporal lobe and posterior insula (figure 4A shows sagittal non-contrast CT images and figure 4B shows the axial). CT angiography shows occlusion of the inferior M2 division (figure 4C). Although the infarct volume is relatively small and there is an M2 occlusion, Wernicke’s area already shows established infarction. Aside from the dysphasia, there is no other deficit to improve with treatment. The corollary to this example is the patient who has unfavourable treatment factors but has an eloquent area at risk, such as the primary motor cortex. In this context, a greater degree of treatment-related risk is reasonable given the likely poor outcome without treatment.

Figure 4

Sagittal and axial non-contrast CT images showing evolving infarction in the left perisylvian region, including the superior temporal gyrus, with occlusion of the inferior division of the middle cerebral artery on the sagittal CT angiogram image. 

MRI in acute stroke

Diffusion-weighted MRI remains the gold standard for diagnosing acute cerebral infarction. However, the time cost of obtaining an MRI is too great for most patients presenting with ischaemic stroke in most centres. In selected cases, MRI can provide key information over and above that of CT. In our hospitals, MRI is reserved as a problem solver, most commonly in one of the following situations:

Lacunar syndromes

CT will often show a recent lacunar infarct, depending on the location and age of the infarct, but is relatively insensitive compared with DWI, particularly in the brainstem. DWI is useful if there is diagnostic uncertainty regarding a lacunar syndrome. While lacunar infarcts are traditionally thought to be due to isolated thrombosis of a penetrating artery, up to a third of patients with a lacunar syndrome will have multiple ischaemic foci on DWI due to emboli.17 We reserve MRI for cases with atypical lacunar presentations or where emboli are suspected for other reasons. Figure 5 shows images from a 75-year-old man who presented with dysarthria and right-sided weakness. Non-contrast CT (figure 5A) shows partly calcified atheroma of the basilar artery without any apparent brainstem infarction, but DWI shows an acute left pontine infarct (figure 5B).

Figure 5

Axial CT and MR images showing the superiority of diffusion-weighted imaging in the diagnosis of brainstem infarction in this patient with an acute infarct in the left pons.

Acute infarction on a background of chronic small vessel disease

Background chronic small vessel ischaemic changes reduces the sensitivity of CT in detecting acute infarcts in the white matter, basal ganglia and brainstem. DWI is very useful in this situation. Figure 6 shows images from a 69-year-old man who had right-sided weakness and reduced cognition following coronary artery bypass surgery. The non-contrast CT scan shows changes easily attributable to chronic small vessel disease, although age indeterminate (figure 6A and B). DWI demonstrates that most of the low-density foci on the CT scan are due to scattered areas of acute infarction (figure 6C and D).

Figure 6

Axial CT images (A and B) show low density foci of uncertain age in the cerebral white matter. Diffusion-weighted images in the same patient (C and D) show these to be due to multiple recent embolic infarcts.

Basilar thrombosis

Patients with acute basilar thrombosis who are being considered for intravenous thrombolysis or endovascular therapy can be challenging due to uncertainty about the extent of established brainstem infarction. The time window for recanalisation of an occluded basilar artery remains unclear, and CT assessment of infarction in the brainstem is often limited by beam hardening artefact. In cases of acute basilar thrombosis where the time of symptoms onset is unclear or there is concern about established brainstem infarction on CT, an urgent DWI can be useful to assess for brainstem infarction and thereby select those patients who may benefit from treatment.

Stroke or not stroke?

In patients who present with neurological symptoms of recent onset that are not clearly localisable on examination and who have no explanation on CT imaging, a DWI sequence is highly useful to identify or exclude ischaemic stroke. One neurologist in our department has been known to say, ‘Put them in the truth machine’, in such circumstances.

Imaging to differentiate infarct from tissue at risk

Deciding which patients will benefit from treatment remains one of the greatest challenges in acute stroke care. There have been numerous methods described to try to distinguish the volume of infarcted brain from the tissue at risk, including CT perfusion, multiphase CT angiography and MRI with diffusion-perfusion imaging. Each of these techniques has advantages and disadvantages, and it is beyond the scope of this review to debate these in detail.

Both the EXTEND IA and SWIFT PRIME trials used CT perfusion imaging as part of their patient selection process. In CT perfusion imaging, areas of irreversibly infarcted brain show prolonged transit times along with reduced cerebral blood flow and cerebral blood volume values. However, there is little consensus on precise thresholds for these to differentiate infarct core from penumbra. Both EXTEND IA and SWIFT PRIME used an operator-independent postprocessing software to standardise the CT perfusion imaging interpretation.18

The ESCAPE trial used multiphase CT angiography when possible for patient selection. This technique involves performing a conventional CT angiogram from the aortic arch to the vertex and then additional skull base-to-vertex acquisitions in the midvenous and then late-venous phases. The data are used to assess visually for the extent of blood flow via pial collateral vessels to the ischaemic territory. In ESCAPE, patients needed to have collaterals that were graded as ‘moderate-to-good’, which was defined as the filling of 50% or more of the middle cerebral artery pial arterial circulation.

While both CT perfusion and multiphase CT angiography have been used successfully in large multicentre stroke trials for endovascular therapy, there is not yet clear evidence for the superiority of one over the other. In our experience (which is with CT perfusion rather than multiphase CT angiography), CT perfusion can be a helpful adjunct in stroke cases where decision making is difficult, particularly for cases where the extent of infarction on the non-contrast CT scan appears borderline. However, for cases where the non-contrast scan looks favourable (such as an ASPECTS score of 9 or 10), we rarely if ever find that the CT perfusion scan changes management. Both techniques increase the total radiation dose of the CT examination and increase the time the patient spends on the CT table.

Using these advanced imaging techniques helps to select patients who will get the greatest benefit from treatment, but it is worth considering that this may mean a small number of patients who would have benefited from treatment will be denied it. Remember that some of the recent positive trials did not use advanced imaging selection techniques.19

Arterial anatomical variants

The anatomical variants of the arteries of the head and neck are legion, but knowledge of some of the common arterial variants is valuable when interpreting acute stroke imaging. Occasionally, a congenitally hypoplastic or aplastic artery can be misinterpreted as an occlusion or dissection. Two key points can help to avoid such misinterpretations: (1) a hypoplastic vessel will be smooth in contour rather than irregular and (2) the size of the intracranial arteries is determined by the volume of the brain parenchyma that they supply. There are uncommon exceptions to these rules, for example, a hypoplastic vessel can also be diseased and an artery may rarely be large due to a high-flow state, as may occur with arteriovenous malformations.

Figure 7 shows images from a 34-year-old woman who presented with a cerebellar infarct where vertebral artery dissection was a consideration. The coronal reconstruction from an MR angiogram (figure 7A) shows a small right vertebral artery. The artery is small throughout its length without any calibre irregularity. Intracranially, this vessel essentially terminates as the posterior inferior cerebellar artery. This is a common variant that should not be mistaken for dissection. Note that the contralateral vertebral artery is large, as it is responsible for supply to most of the intracranial vertebrobasilar circulation. Figure 7B shows that the osseous vertebral foramen for the right vertebral artery is small; this is clear evidence that this vessel has always been small.

Figure 7

A three-dimensional reconstruction of an MR angiogram shows a hypoplastic right vertebral artery (A). Note the small bony foramen for this vessel on the axial CT angiogram image (B).

Figure 8 an axial MR angiogram image in a patient with a small basilar artery. This is due to the presence of bilateral fetal posterior cerebral arteries rather than to basilar disease. The basilar artery is small as the supply to the occipital lobes is from the carotid circulation.

Duplication or fenestration may occur with most of the intracranial arteries, but middle cerebral artery variants are relevant in acute ischaemic stroke. Early bifurcation of the middle cerebral artery is common and can result in confusing terminology. Most consider the transition from M1 to M2 segments as the first point of middle cerebral artery bifurcation. However, in this situation, some clinicians refer to prebifurcation and postbifurcation M1 segments, with the M2 segments defined as starting at the limen insulae. Relatively rare variants are the duplicated middle cerebral artery and the accessory middle cerebral artery, each with a prevalence of less than 3%.20

Figure 8

An axial MR angiogram image shows a small basilar artery that is due to the presence of bilateral fetal posterior cerebral arteries.

Stroke mimics

There are certain stroke mimics that can be diagnosed with specificity on imaging.

Figure 9 shows images from a 38-year-old woman who presented with transient left-sided weakness. Non-contrast CT and CT angiography showed no acute infarct or arterial occlusion. However, there was absence of contrast filling in the superior sagittal sinus: the ‘empty delta’ sign. While absence of contrast in the sagittal sinus on a CT angiogram may simply be due to the timing of the study—if the images are acquired in the early arterial phase prior to venous filling—in this case, there was contrast opacification of the transverse and sigmoid sinuses (Figure 9C, arrows on axial CT angiogram), indicating that the empty delta sign was not due to a timing issue. MRI confirmed a dural sinus thrombosis (Figure 9D, arrow on sagittal T1 image), and the patient was anticoagulated. This case was unusual in that the thrombosed segment of the sagittal sinus was not hyperdense on the non-contrast CT, as would usually occur in dural sinus thrombosis. The parsimonious among us will appreciate from this example that a CT angiogram that has venous opacification effectively represents a free CT venogram. Note that modern CT scanners are capable of acquiring a CT angiogram in the early arterial phase and may show little or occasionally no venous opacification (the authors do not wish to trigger a spate of false positive diagnoses of dural sinus thrombosis).

Seizure is one of the more common stroke mimics and can cause abnormalities on cerebral perfusion imaging. Seizure activity may cause regional increases or decreases in cerebral perfusion. In the context of an acute neurological event, an area of abnormality on a CT or MRI perfusion study that does not conform to a vascular territory and spares the white matter should trigger the possibility of seizure activity.21

Figure 9

A coronal non-contrast CT image is unremarkable (A). Axial CT angiogram images show no contrast enhancement of the superior sagittal sinus but enhancement at both sigmoid sinuses (B and C). A sagittal T1 MRI shows high signal thrombus in the superior sagittal sinus (D).


Advances in imaging techniques have significantly improved the diagnosis of acute stroke. While neuroradiologists may issue the final report, neurologists and neurology trainees are often the first to interpret these scans and must be equipped to use this information to make rapid decisions regarding increasingly complex treatments.

Key points

  • Beware the pseudo-occlusion or pseudodissection of the cervical internal carotid artery on CT angiography.

  • Some patients who present with an anterior circulation stroke syndrome with a seemingly occluded internal carotid artery on CT angiography will be candidates for intravenous thrombolysis and endovascular stroke treatment.

  • A catheter angiogram remains the gold standard to differentiate occlusion from pseudo-occlusion of the cervical internal carotid artery.

  • MRI can be particularly useful in lacunar syndromes, in patients in whom the clinical deficit is not explained by the CT findings and in patients with basilar thrombosis to assess the extent of brainstem infarction.

  • To differentiate a congenitally hypoplastic artery from a diseased vessel, remember that (A) a hypoplastic artery will have a smooth contour in contrast to a narrowed diseased vessel and (B) the size of the artery is determined by the volume of brain parenchyma that it supplies.


View Abstract


  • Contributors All the authors on this paper have contributed to writing and reviewing it.

  • Competing interests None declared.

  • Patient consent The article only uses deidentified radiology images and patients cannot be identified.

  • Provenance and peer review Not commissioned. Externally peer reviewed. This paper was reviewed by Bejoy Thomas, Kerala, India, and William Whiteley, Edinburgh, UK.

Other content recommended for you