Neuroimaging for Acute Ischemic Stroke

Emerg Med 39(01):9-15, 2007

The authors discuss keys to the appropriate selection and interpretation of magnetic resonance imaging, computed tomography, and ultrasound techniques for patients whose presentation is consistent with acute stroke.

By Fuhai Li, MD,  Sean J. Snodgress, MD,  Linda Gray-Leithe, MD, and Larry B. Goldstein, MD, FAAN, FAHA

Dr. Li is chief resident in the division of neurology, department of medicine, at Duke University Medical Center in Durham, North Carolina. Dr. Snodgress is chief resident and Dr. Gray-Leithe is an associate professor in the department of radiology there. Dr. Goldstein is professor of medicine (neurology), director of the Duke Center for Cerebrovascular Disease and the Duke Stroke Center, and senior fellow in the Center for Clinical Health Policy Research at Duke University. He is also a staff neurologist at the Durham VA Medical Center and a member of the emergency medicine editorial board.

Neuroimaging is a critical part of the emergency assessment of patients with acute stroke symptoms. It is required to reliably distinguish intraparenchymal hemorrhage from ischemic stroke and to help identify stroke mimics such as tumor. In addition, neuroimaging can help to determine a stroke’s vascular territory and can provide insights into its potential causes.

Both acute interventions and the appropriate use of secondary prevention strategies are dependent on imaging results. Extracranial carotid artery stenosis is one of the major treatable causes of early recurrent stroke and can be identified with several different types of imaging studies in the emergency department. Some types of acute imaging studies may also help to identify ischemic but potentially viable brain tissue, which may prove useful in targeting therapeutic intervention.

The purpose of this article is to provide emergency physicians with essential information regarding the variety of neuroimaging studies that can be used in the emergency department to help guide therapy for patients with acute stroke. We will also comment on particular features of these tests that can improve the quality of the results and communication with the consulting radiologist.

COmputed tomography

Computed tomography (CT) is available in most hospitals that provide emergency care. The critical role of brain CT in the evaluation of patients with suspected acute stroke is reflected in the requirement for its immediate availability as part of a hospital’s certification as a primary stroke center.

There are several different CT-based techniques with potential utility in an emergency setting. Routine CT-based diagnosis is based on differences in tissue density, expressed as CT values or Hounsfield units (HU). Low-density structures, including air, fat, and water, appear dark on routine brain CT, whereas high-density structures, such as bone, calcifications, and acute hemorrhage, appear bright. Reduced brain tissue density can be seen in patients with recent stroke because of initial cytotoxic and later vasogenic edema. Because these changes take time to develop, most CT scans obtained within the first hours after the onset of an ischemic stroke are frequently normal. A normal brain CT scan is therefore consistent with a clinical diagnosis of ischemic stroke.

Noncontrast CT. Noncontrast CT is the most commonly used neuroimaging technique for patients with suspected acute stroke because it is noninvasive, can be obtained rapidly, and is relatively inexpensive. It is also very sensitive for detecting intracranial hemorrhage, which is a key clinical finding. Beam-hardening artifact from surrounding bone can compromise visualization of structures in the posterior fossa. When a vertebrobasilar-distribution stroke is suspected, this limitation of CT can be minimized by alerting the technologist and requesting thin (1-mm thick) images through the posterior fossa or by altering the plane of the scan.

Noncontrast CT can readily detect parenchymal intracranial hemorrhage, including frank hematoma or an acute hemorrhagic transformation of an ischemic infarction (see image below).

Acute hemorrhagic transformation. Noncontrast computed tomography (CT) demonstrates hemorrhagic transformation (bright areas) after a right middle cerebral artery (MCA) distribution ischemic stroke.

Although noncontrast CT results are frequently normal in patients evaluated soon after symptom onset, subtle changes can often be recognized after two to four hours. These changes can include loss of the insular ribbon (see images below), obscuration of lenticular nuclei, loss of differentiation between cortical gray and white matter, sulcal effacement, and cortical hypodensity. Use of a narrow CT window and level setting (for instance, 30 to 50 HU) can increase the sensitivity for detection of early is-chemic changes from approximately 50% to 70%, but this is often evident only in comparison to the contralateral side. It should be noted that the reliability of the detection of these early ischemic signs can be limited (ranging from 45% to 85%) and is closely related to clinical experience. Depending on the circumstances, the emergency physician can review the CT scan online with a radiologist.

Two different CT windows. Although the loss of right insular ribbon (arrow) and obscured basal ganglia are visible in both images, the image on the right shows better visualization of findings with a different CT window.

A hyperdensity can sometimes be seen on noncontrast brain CT in patients with a clot in the middle cerebral artery (MCA), the so-called “dense MCA sign” (see image below, left). Comparison with the contralateral side can again be helpful. Slow flow and calcification in the artery can mimic the MCA sign, and it can occasionally be mistaken for a subarachnoid hemorrhage in the Sylvian fissure. When this is a concern, an absolute HU number greater than 43 in the M1 segment of the MCA and a HU ratio greater than 1.2 between the ipsilateral and contralateral M1 suggest true intraluminal thrombus. Dense dots in the distal MCA segments can indicate acute thrombus in these portions of the artery.

Hyperdensity. In the same patient shown above, the dense MCA sign is seen in the left image (arrow). A typical left MCA-distribution ischemic stroke is seen clearly two days later (right).

In addition to helping establish a diagnosis, early CT changes can have immediate therapeutic implications. Identification of these changes should prompt a careful review of the reported time of onset of the patient’s symptoms. Edema, mass effect, or involvement of more than one third of the MCA territory can identify patients at relatively increased risk of bleeding complications associated with tissue plasminogen activator, although, as a group, these patients still have better outcomes with treatment.

Contrast CT. Contrast CT is not usually performed in patients with suspected acute ischemic stroke because it adds little to immediate clinical decision-making. Localized arterial enhancement can be detected in the acute phase due to the slow outflow of contrast from the ischemic region. Parenchymal enhancement can be seen as early as several hours after the stroke, reflecting the effect of clot lysis and early reperfusion. Multifocal, linear, bandlike, and ring parenchymal enhancement can be seen in the subacute phase (three to seven days after the stroke) due to breakdown of the blood-brain barrier. A ring enhancement pattern after ischemic stroke may sometimes be confused with a mass lesion such as a neoplasm or abscess.

Computed tomographic angiography. Computed tomographic angiography (CTA) can be completed in less than one minute; additional time for data processing takes about 15 minutes. Images can be acquired in separate sequences from the aortic arch to the skull base for extracranial carotid and vertebral arteries and from the skull base to the cranial vertex for intracranial vasculature. The data from axial source images can later be reconstructed using three-dimensional volume rendering, maximum intensity projection, or multiplanar reformatting techniques.

Source images with CTA can be used to localize an arterial occlusion to help guide endovascular management in patients for whom this is a consideration. The area of hypoattenuation on CTA-source images correlates with the area of restricted diffusion on magnetic resonance imaging (MRI) and reflects the region of ischemic injury.

Computed tomography angiography also provides high spatial resolution images of the cervical and intracranial vessels, identifying arterial stenosis and occlusion. The sensitivity and specificity of CTA for severe carotid artery stenosis, which are usually more than 70%, can approach 100%. Agreement between the results of three-dimensional CTA and conventional angiography is approximately 95%. Computed tomography angiography can yield better visualization of arterial calcification and better delineation of atherosclerotic plaque morphology and ulceration than conventional angiography. It may be superior to magnetic resonance angiography (MRA) in identifying cervical carotid dissection, but it is inferior to MRA in delineating dissection at the level of the skull base due to beam-hardening artifact and bony interference. It is also useful in identifying intracranial aneurysms.

The disadvantages of CTA include contrast dye toxicity and its relative contraindication in patients with renal insufficiency as well as additional exposure to ionizing radiation. Periprocedural administration of N-acetylcysteine may prevent contrast-induced acute renal failure in patients with elevated serum creatinine levels. Depending on the clinical situation, either CTA or MRA can be obtained in the emergency department.

Dynamic contrast-enhanced perfusion CT. Perfusion CT is a functional brain imaging study that can reflect regional cerebral perfusion. Dynamic scanning is performed to measure temporal changes in the density of brain tissue resulting from rapid changes in the concentration of the contrast agent. Compared with perfusion-weighted MRI, perfusion CT has higher spatial resolution and can provide an absolute measure of cerebral blood flow. The difference between the area of brain tissue with reduced perfusion and the area with hypoattenuation on CTA-source images may indicate the ischemic penumbra, an area of ischemic but potentially viable brain tissue.

The disadvantage of perfusion CT is that it is limited to a restricted volume of brain and may be unable to adequately visualize posterior fossa structures. The sensitivity and specificity of perfusion CT for detecting areas of ischemia, however, can be as high as 95% and 100%, respectively.

In clinical practice, CTA and perfusion CT are obtained following noncontrast CT. Within minutes, the combination of these three techniques can help exclude stroke mimics and hemorrhage, identify the area of ischemia and tissue at risk, and help localize the site of vascular occlusion.

Magnetic resonance imaging

A variety of MRI techniques can be useful in the immediate evaluation of patients with suspected acute ischemic stroke. Magnetic resonance imaging provides excellent anatomical information, including images of posterior fossa structures. The detection of hemorrhage, however, is not as straightforward as with CT. The interpretation of some MRI sequences requires specialized expertise; they will be mentioned because they are part of the routine stroke MRI studies performed at many institutions..

Conventional MRI. Conventional MRI sequences include T1-weighted imaging (T1WI), proton density images, and T2-weighted imaging (T2WI). On T1WI, as with CT, cerebrospinal fluid (CSF) is hypointense (dark) and tissue with high lipid or protein content is hyperintense (bright). In contrast, with T2WI, CSF is hyperintense and tissue with high lipid or protein content is hypointense. Proton density images are intermediate, with CSF appearing gray.

Generally, conventional MRI is more sensitive in detecting early ischemic changes than CT. Hypointense changes in ischemic areas may not be evident on T1-weighted imaging for 12 hours (see table below), although, as with CT, early signs of ischemic injury can include sulcal effacement and obscuration of gray-white matter differentiation. The loss of vascular flow voids, reflecting either slow blood flow or thrombus, can be detected within minutes. Contrast-enhanced T1WI can show arterial, meningeal, or parenchymal enhancement after ischemic stroke. Focal vascular enhancement indicates slow blood flow in the ischemic region and can be seen as early as two hours after stroke onset. Meningeal enhancement likely reflects reactive hyperemia and can be seen after about 24 to 72 hours. Parenchymal enhancement represents blood-brain barrier breakdown or collateral circulation and is usually seen after two to six days, but occasionally as early as two hours after the onset of ischemic stroke.

Ischemic tissue on T2WI appears hyperintense due to water accumulation and typically does not appear until 6 to 12 hours after the onset of ischemia, although subtle patchy hyperintensity can be visible within two to three hours. Absence of a vascular flow void can be easier to identify on T2WI than T1WI.

Diffusion-perfusion weighted imaging. Diffusion-weighted MRI (DWI) is one of the standard MRI sequences obtained in patients with suspected stroke and is very useful in detecting areas of acute ischemic injury (see image below). It uses tissue water diffusion as a natural contrast agent. Areas of both restricted diffusion and increased T2 signal (representing older abnormalities) appear bright on DWI. An apparent diffusion coefficient (ADC) map is produced to eliminate the so-called “T2 shine-through” effect. This map will have no T2 contrast. The ADC in the ischemic region decreases rapidly after the onset of vascular occlusion, appearing as a dark area. The typical temporal evolution of signal changes on T1WI, T2WI, DWI, and ADC maps after ischemic stroke is given in the table (above).

Magnetic resonance imaging. Ischemic changes are evident on MRI sequences T1W1, T2W1, and DWI. The DWI sequence is more sensitive than either T1W1 or T2W1.

Diffusion-weighted MRI has a sensitivity of 88% to 100% and a specificity of 86% to 100% for detection of acute cerebral ischemia. A false-negative DWI can be due to a small volume of ischemic tissue and areas of minimal cerebral ischemia. False positives can be caused by conditions such as abscess, tumor, seizure, and acute multiple sclerosis plaques.

Like perfusion CT, perfusion-weighted MRI (PWI) can be used to detect areas of brain with relatively reduced blood flow. One role of PWI is in the identification of an ischemic penumbra. The area of tissue with a PWI lesion but no DWI abnormality (a so-called DWI/PWI mismatch) may indicate tissue that is still salvageable if reperfused. The clinical usefulness of identifying an area of DWI/PWI mismatch has not been fully established.

Other MRI sequences. Fluid attenuation inversion recovery (FLAIR) imaging can be thought of as T2WI without a CSF signal. Because CSF appears dark rather than bright, areas of ischemia in periventricular and cortical regions can be seen more easily on FLAIR than on T2WI.

One of the most important reasons to perform imaging studies on patients with suspected stroke is to rule out hemorrhage. Although an MRI can be useful for determining the age of a hemorrhage (see table, below), a major limitation of standard MRI sequences is their low sensitivity for the detection of acute bleeding. Gradient recalled echo (GRE) imaging is as sensitive as CT for detecting acute intraparenchymal hemorrhage, which appears as an area of hypointensity. The emergency physician needs to recognize that unless GRE images are obtained, MRI cannot reliably exclude acute hemorrhage.

Magnetic resonance angiography. Magnetic resonance angiography is another noninvasive technique that can be used to assess arterial stenosis and occlusion. It is frequently performed with conventional MRI, FLAIR, PWI, and DWI in patients with suspected acute stroke. Extracranial and intracranial angiography can be performed separately with noncontrast MRA and contrast-enhanced MRA (CEMRA) techniques.

Noncontrast MRA usually overestimates the degree of arterial stenosis, and false-positives can occur due to in-plane flow-related signal loss or turbulence at vessel bifurcations. If dissection is suspected, fat-saturated axial T1WI is performed to detect hyperintense crescent hematoma in the false lumen. The emergency physician should alert the radiologist if this diagnosis is a clinical consideration because fat-saturated axial T1WI imaging is not routinely performed.

Advantages of CEMRA include an increased signal within vessels even in the presence of turbulent in-plane flow or slow flow, reduced motion artifact, and coverage of a larger region. Because of these advantages, CEMRA is frequently used for imaging the extracranial precerebral vasculature.

Magnetic resonance angiography has been widely used in clinical practice to assess arterial stenosis or occlusion in patients with acute stroke. The sensitivity of CEMRA for the detection of carotid stenosis exceeding 70% is 93% to 94%; its specificity ranges from 85% to 100%. Both noncontrast and contrast-enhanced MRA tend to overestimate the degree of arterial stenosis due to signal loss caused by stenotic turbulence (in the case of noncontrast MRA) and artificial loss of a faint signal during maximum-intensity projection for CEMRA. Neither technique can reliably differentiate a very high-grade stenosis from a complete occlusion. Other non-invasive techniques such as CTA or conventional catheter angiography are needed for this purpose.

ultrasonographic techniques

Several different types of ultrasonographic techniques are available for the vascular evaluation of stroke patients. Doppler ultrasound is used to assess vascular hemodynamics and to diagnose vascular stenosis, based on blood flow velocity changes reflected by frequency shifts in both the extracranial and proximal intracranial arteries. Peak systolic velocity is most frequently used to measure the severity of stenosis, but the end-diastolic velocity and flow pattern also provide important information. Brightness-mode (B-mode) imaging allows anatomical and structural assessment based on the acoustic properties of the tissue. Duplex ultrasound refers to a combination of Doppler and B-mode imaging.

Extracranial carotid and vertebral ultrasonography. Duplex ultrasound with or without color flow imaging is most commonly employed to identify vascular stenosis, with B-mode imaging used to assess anatomical features of an atherosclerotic plaque such as ulceration and calcifications. Carotid duplex ultrasound has reported sensitivities of 81% to 98% and specificities of 82% to 89% for the detection of a hemodynamically significant stenosis of the extracranial internal carotid artery. While it has the advantages of being noninvasive, safe, and inexpensive, it also has several limitations. Its accuracy in identifying carotid stenosis is operator-dependent. As with MRA, it cannot reliably distinguish a complete occlusion from a very high-grade stenosis, and changes in flow characteristics related to a contralateral stenosis can lead to artifactual increases in flow velocities. However, it can be very useful in the emergency department in helping to exclude a significant extracranial carotid artery stenosis in patients with carotid-distribution transient ischemic stroke or stroke.

Transcranial Doppler ultrasonography. Transcranial Doppler (TCD) ultrasound and transcranial color-coded duplex ultrasound can be used to detect stenosis or occlusion of the intracranial arteries involving or near the circle of Willis. Its sensitivity and specificity for identifying severe stenosis or occlusion is 85% and 95%, respectively, in the internal carotid artery distribution and 75% and 85%, respectively, in vertebrobasilar distribution. This procedure has also been used to monitor the efficacy of thrombolytic therapy in acute stroke patients. It cannot, however, directly evaluate distal intracranial arteries. In addition, many patients lack the acoustic windows required for this procedure because of relative thickening of the calvarium, precluding insonation.

CONVENTIONAL CATHETER ANGIOGRAPHY

The quality of conventional angiography has significantly improved since the introduction of digital subtraction techniques. Digital subtraction angiography can delineate even small cortical branches of intracranial arteries. Although conventional angiography is still the gold standard for diagnosing arterial stenosis, it is used only in limited circumstances in the acute stroke setting. When there are equivocal or conflicting results with noninvasive techniques, conventional angiography may be needed to reach the appropriate diagnosis. As previously noted, this procedure remains the only viable technique for imaging distal intracranial arteries.

back to top

Suggested Reading Alexandrov AV: Ultrasound identification and lysis of clots. Stroke 35(11 Suppl 1):2722, 2004. Konig M: Brain perfusion CT in acute stroke: current status. Eur J Radiol 45(Suppl 1):S11, 2003. Marks MP: CT in ischemic stroke. Neuroimaging Clin N Am 8(3):515, 1998. Schaefer PW, et al.: Diffusion magnetic resonance imaging of acute ischemic stroke. Semin Roentgenol 37(3):219, 2002. Schellinger PD, et al.: Imaging-based decision making in thrombolytic therapy for ischemic stroke: present status. Stroke 34(2):575, 2003. Schramm P, et al.: Comparison of perfusion computed tomography and computed tomography angiography source images with perfusion-weighted imaging and diffusion-weighted imaging in patients with acute stroke of less than 6 hours’ duration. Stroke 35(7):1652, 2004. Wintermark M, et al.: Simultaneous measurement of regional cerebral blood flow by perfusion CT and stable xenon CT: a validation study. Am J Neuroradiol 22(5):905, 2001. Yu D, et al.: Magnetic resonance angiography in acute stroke. Semin Roentgenol 37(3):212, 2002.

Printable version of this article

E-mail this article