eMedicine Specialties > Neurology > Neuro-imaging

Magnetic Resonance Imaging in Acute Stroke: Differential Diagnoses & Workup

Author: Souvik Sen, MD, MS, FAHA,, Associate Professor of Neurology, Founding Director of UNC Hospital Stroke Center, Director of Neurovascular Residency, Department of Neurology, University of North Carolina at Chapel Hill
Contributor Information and Disclosures

Updated: Jul 15, 2009

Differential Diagnoses

Acute Stroke Management

Workup

Imaging Studies

Radiography

A regular radiograph may be indicated if any possibility exists of a metallic implant or foreign body, pacemaker, aneurysm clip, or recently implanted prosthetic heart valve.

MRI techniques - Diffusion-weighted imaging

DWI is sensitive to the microscopic random motion of the water molecule protons, a value known as the apparent diffusion coefficient (ADC), which is measured and captured by this type of imaging. The water molecules move in the direction of the magnetic field gradient; they accumulate a phase shift in their transverse magnetization relative to that of a stationary one, and this phase shift is directly related to the signal attenuation of the image.

Numerous studies have shown that ADCs in ischemic areas are lower by 50% or more than those of normal brain areas, and they appear as bright areas (ie, hyperintensities) on the DWI (see Media file 1). Studies have demonstrated that changes in the ADC occur as early as 10 minutes following onset of ischemia.

Cytotoxic edema appears following sodium/potassium pump failure, which results from energy metabolism failure due to ischemic insult; this occurs within minutes of the onset of ischemia and produces an increase in brain tissue water of up to 3-5%. Reduction in intracellular and extracellular water molecule movement is the presumed explanation for the drop in ADC values.

The diffusion of water molecules is guarded by biologic barriers in the brain tissue (eg, cell membranes and cellular organelles). The behavior of water molecules is not symmetric and may show uneven distribution of the ADC when measured in one direction; this uneven distribution may give a false impression of a lesion. ADC values are measured in several directions (3, 6, or more), and ADC maps are created to produce a direction-insensitive measurement of the diffusion. When ADC is measured in 6 or more directions, the diffusion motion of all the water molecules (ie, ADC tensor matrix) can be calculated to create what is called full diffusion tensor mapping, which can also be used to visualize white matter tracts.

Reduction in the ADC also occurs in other conditions such as global ischemia, hypoglycemia, and status epilepticus; it should always be evaluated in relation to the clinical condition of the patient.

Human studies demonstrated that damage in the areas showing decreased ADC levels is very rarely reversible (in contrast to that in animal models), although a few studies have indicated that intra-arterial thrombolysis may occasionally result in disappearance of the diffusion defect. The technique most commonly used to acquire the DWI is an ultrafast one, echo-planar imaging (EPI); this technique decreases scanning time significantly and eliminates movement artifacts.

The acute drop in ADC is gradually normalized to baseline at 5-10 days after ischemia (pseudonormalization); it even exceeds normal levels as time passes, helping in some cases to differentiate between acute, subacute, and chronic lesions.

DWI is very sensitive and relatively specific in detecting acute ischemic stroke. DWI findings have shown high levels of diagnostic accuracy; however, recent studies demonstrated that small brainstem lacunar infarctions may escape detection. Normal DWI in patients with strokelike symptoms should trigger further investigation for a nonischemic cause of the symptoms. DWI has been shown to reveal diffusion abnormalities in almost 50% of patients with clinically defined transient ischemic attacks (TIAs); it tends to be of higher yield at increasing time intervals from the onset of stroke symptoms.

From the clinical experience at the University of North Carolina at Chapel Hill Stroke Center, the following differential for areas of hyperintensity on DWI was generated:

• Subacute ischemic stroke - Usually takes 7-14 days for hyperintensity to subside
• Hemorrhagic stroke - Usually bright on T1-WI
• Multiple sclerosis plaque - Also bright on FLAIR and T2-WI
• Traumatic brain injury - History of trauma
• Brain abscess - Ring enhancement on contrast MRI
• Choroid plexus - Usually intraventricular in location, may be bilateral
• Epidermoid - Usually extra-axial in location
• Air-bone interface - Commonly bilateral, in the temporal bone
• T2 shine through – Associated with abnormal T2-WI and normal ADC map

MRI techniques - Perfusion-weighted imaging

With this technique, information about the perfusion status of the brain is available. The most commonly used technique is bolus-contrast tracking (other techniques include blood oxygen level and arterial spin tagging). The imaging is based on the monitoring of a nondiffusible contrast material (gadolinium) passing through brain tissue.

The signal intensity declines as contrast material passes through the infarcted area and returns to normal as it exits this area. A curve is derived from this tracing data (ie, signal washout curve), which represents and estimates the cerebral blood volume (CBV).

An arterial input function can be derived by measuring an artery in lower brain slices or by measuring gadolinium concentration that is proportional to the changes in T2 when gadolinium is used at low doses (<3 mg/kg). Based on this arterial input function, quantitative maps of cerebral blood flow (CBF), CBV, mean transit time (MTT), time to peak (TTP), and various other hemodynamic parameters can be obtained. Considerable debate surrounds the choice of which PWI parameter should be used. Most centers in the United States use time domain parameters such as MTT or TTP.

DWI and PWI together have been shown to be superior to conventional MRI both in early phases and also up to 48 hours after the onset of stroke. Using both DWI and PWI is very important because together they provide information about location and extent of infarction within minutes of onset; when performed in series, they can provide information about the pattern of evolution of the ischemic lesion. This information may be of great importance in choosing the appropriate treatment modality as well as in predicting outcome and prognosis. Several recent randomized clinical trials are selecting patients with diffusion-perfusion mismatch to test thrombolytic treatment alternatives beyond the standard 3-hour time window used for IV TPA. These studies have been reviewed by Davis et al.¹

The lesion usually enlarges on serial DWIs over a period of several days. It has been suggested that this enlargement can be halted if reperfusion (ie, resolution of original PWI lesion) occurs early enough. Lesions that are not large on initial PWI do not show this enlargement.

The diffusion-perfusion mismatch (see Media file 3), ie, the difference in size between lesions captured by DWI and PWI, usually represents the ischemic penumbra (see Media file 3), which is the region of incomplete ischemia that lies next to the core of the infarction. The ischemic penumbra is regarded as an area that is viable but under ischemic threat; it can be saved if appropriate intervention is promptly instituted. The viability of this region could extend up to 48 hours after the onset of stroke. Determining the volume of the ischemic penumbra may be very useful in identifying patients who would benefit from thrombolytic therapy and perhaps even conventional treatments such as carotid endarterectomy or blood pressure elevation. It could also aid in evaluating the risk/benefit ratio of using such treatments in stroke patients.

Limitations of diffusion-perfusion mismatch are mainly methodological and include (1) lack of anatomical match between diffusion and perfusion-weighted abnormality, (2) variable sensitivity of perfusion-weighted image based on T_max delay, (3) visual versus quantitative estimation of mismatch.

One limitation of these techniques is in detection of acute intracerebral hemorrhages; early studies demonstrated that susceptibility imaging could be sensitive in the detection of acute intracerebral hemorrhage. Gradient-recalled echo (GRE) imaging sequences demonstrated the most favorable sensitivity in detecting susceptibility dephasing associated with chronic intracerebral hemorrhages.

MRI still has some limitations in its application, namely, in patients with metal implants and acutely ill patients requiring close monitoring.

These new techniques, DWI and PWI, together represent the most exciting areas in MRI for their potential ability to detect early changes (ie, within minutes of the stroke). They are currently used in the evaluation of thrombolytic and neuroprotective therapy in acute stroke clinical trials.

MRI techniques – Blood oxygen level-dependent (BOLD) MRI

Oxygen extraction fraction (OEF) measured by positron emission tomography (PET) imaging is considered the criterion standard for imaging the ischemic penumbra in acute ischemic stroke. Until now, MR diffusion-perfusion imaging has been the only MR technique that measures this reversibly damaged brain area.

BOLD is a new technique that can be used to detect deoxyhemoglobin in the cerebral capillaries and veins as an MRI indicator of brain OEF. Recent evidence suggests that BOLD MRI might provide a better estimation of the ischemic penumbra in acute ischemic stroke compared with MR diffusion-perfusion mismatch.

Further validation of the technique is required to confirm its clinical value in imaging of acute ischemic stroke.

MRI techniques - Echo-planar imaging

EPI is a recent technique that can be used to visualize physiologic parameters in addition to measuring diffusion coefficients of the ischemic brain. Changes in brain oxygenation can be monitored by using gradient echo and EPI, in which deoxygenated blood acts as a susceptibility contrast agent.

EPI can be used in conjunction with bolus injection of intravenous paramagnetic agents to assess cerebral perfusion and functional changes in CBV.

In this technique, hypoperfused areas appear as hyperintense signals after injection of the contrast material. This technique is considered a way of reducing the traditionally long scanning time of MRI.

MRI techniques - Magnetic resonance spectroscopy

Magnetic resonance spectroscopy (MRS) is one of the recent advances in MR technology; it evaluates metabolic activity and concentration of certain metabolites in specified areas of the brain. Proton and phosphorus spectroscopic studies have been performed.

In proton spectroscopy, depression of N -acetyl aspartate, which is considered to be a marker of neurons, is the most consistent finding in acute stroke. This depression may occur within hours after the onset of stroke and continues through the subacute and chronic phases of the stroke, presumably because of loss of neurons.

Increase in levels of lactate is another important finding and has been attributed to anaerobic metabolism in ischemic tissue. Initial studies of other metabolites, such as choline and creatine, demonstrated decreases in their levels in acute stroke.

Phosphorus spectroscopy provides information about energy metabolism and pH, depletion of ATP, decrease of tissue pH, and increase of the ratio of inorganic phosphate to phosphocreatine, which has been reported in both human and animal studies.

Long acquisition times, weak signal, and low spatial resolution of this technique have limited enthusiasm for its use in the clinical management of cerebral ischemia; however, some studies have suggested that MRS results can have prognostic value in stroke.

MRI techniques - Magnetic resonance angiography

Magnetic resonance angiography (MRA) is very sensitive to flow and is based on the difference in signal between moving blood and stationary brain tissue; angiographiclike images of the cervicocranial vasculature are produced.

MRA images are a useful tool in identifying dissections, in that both the true and false lumen of the involved artery can be observed on the source images. Following is a brief description of the 2 basic techniques.

The three-dimensional (3D) time-of-flight (TOF) technique is based on flow-related enhancement; it is the preferred MRA technique. However, it has some limitations, especially flow signal dropout secondary to turbulent flow in the tortuous and stenotic vascular segments, which makes interpretation of stenosis in these areas difficult. These are common predilection sites for atherosclerosis. Also, in slow-flow regions, the spin saturation of the scan causes overestimation of stenosis. In contrast-enhanced studies, it provides more information than standard angiography, especially in detecting critical stenosis of extracranial vessels, but it is less reliable in intracranial critical stenosis. Always keep in mind that MRA is a flow-dependent technology; absence of flow signal does not mean literally a complete occlusion but rather that flow is below a critical value.

Two-dimensional (2D) TOF MRA also depends on the relative contrast between flowing blood and stationary tissue; it provides better images than 3D TOF in slow-flow regions. 2D TOF images correlate well with carotid angiography images in depicting cervical bifurcation disease. Its disadvantages, however, are the significant artifacts (eg, stepladder) that often occur, which may obscure vessel details, and the longer scanning time.

The modified TOF MRA technique, which uses multiple overlapping thin slab acquisitions (MOTSA), combines the advantages of 2D and 3D TOF techniques. It is very helpful in demonstrating severe stenosis, although the degree of stenosis might be slightly overestimated.

Two-dimensional phase-contrast (PC) MRA is a technique that is helpful specifically in differentiating slow and absent flow from normal flow; it captures only truly patent vessels. Other imaging sequences (eg, spin-echo sequence or gradient-echo sequence) should be used with PC-MRA to avoid missing lesions such as paravascular hematomas, which are not captured by PC-MRA. PC-MRA also has the disadvantage of signal loss due to turbulent flow in tortuous vessels.

Types of infarction

• Thromboembolic infarction: This is the most common form of infarction. Typically, it is observed on MRI as a wedge-shaped infarct in the particular vascular distribution. Recent data support the hypothesis that a single infarct in a vascular territory is more likely to be thrombotic than multiple infarcts, which are more likely to be embolic.
• Watershed infarction: This type of infarction occurs at the distal margins of specific arterial territories. It can occur both superficially and deep in the brain parenchyma. Common etiologies for this lesion include hypotension, cardiac and respiratory arrest, and proximal arterial stenosis or occlusion. MRI findings follow the pattern of incomplete thromboembolic ischemic infarction in T1 and T2 morphologic and signal changes, with early parenchymal enhancement suggesting early reperfusion. Recent studies show that this type of infarction could be more readily detected by using DWI.
• Lacunar infarction: These are small deep cerebral infarctions believed to be caused by intrinsic small-vessel disease secondary to lipohyalinosis and fibrinoid necrosis; they are most frequently observed in patients with hypertension or diabetes mellitus. Common sites for these lesions include basal ganglia, internal capsule, thalamus, brain stem, and cerebellum. MRI findings in these lesions follow the same pattern observed in thromboembolic infarction.
• Venous thrombosis and infarction: Occlusion of cerebral veins and venous sinuses is usually caused by systemic conditions, such as pregnancy, collagen vascular diseases, inflammatory bowel diseases, and hypercoagulable states, as well as local conditions such as infection, neoplasia, and trauma. Occlusion of the venous structure causes outflow obstruction and vascular congestion that results in parenchymal infarctions and hemorrhages.
  • Patients usually present in the late acute phase or in the subacute phase, which makes the diagnosis difficult because diagnosis at these stages depends on imaging studies.
  • MRI findings in these lesions include loss of venous flow void signal, absence of normal venous enhancement, and visualization of isointense to hyperintense signals within the venous channels on both T1 and T2 images. These variable patterns of enhancement are due to mixed blood products, which are present in the lesion.
  • These patterns are usually bilateral, do not respect arterial vascular territories, and have associated hemorrhage.
  • Three-dimensional phase contrast magnetic resonance venography (MRV) is the preferred technique in the evaluation of venous thrombosis.

MRI findings in acute stroke - Hyperacute phase (0-24 h)

DWI is able to detect ischemic changes within minutes of onset (see Media file 1). Reduced proton motion is detected as a decreased ADC.

Early in the process of cerebral ischemia, PWI, using first-pass contrast bolus injection or spin tagging the protons in the water in blood, reveals reductions of CBF and CBV and an increased MTT of blood through the brain (see Media file 2).

Matched diffusion- and perfusion-weighted abnormalities correlate with the region of infarction and are indicative of permanent neuronal death. Mismatched diffusion and perfusion abnormalities with the perfusion abnormality larger than the diffusion abnormality may be indicative of a region of reversible ischemic penumbra (see Media file 3).

A popular paradigm is to subject patients presenting beyond the 3-hour window to multimodal MRI to detect DWI/PWI mismatch, and those with mismatch may be candidates for stroke treatment. Recently, 2 clinical trials were completed that tested the hypothesis that patients with ischemic penumbra as detected by DWI/PWI mismatch may benefit from thrombolysis beyond the 3-hour treatment window. The DIAS-2 trial randomized patients presenting 3-9 hours with greater than 20% DWI/PWI mismatch to IV desmoteplase or placebo.² The EPITHET trial randomized patients presenting 3-6 hours with greater than 20% DWI/PWI mismatch to IV TPA or placebo.³ Both studies seemed to show lack of benefit, suggesting that patients selected based on DWI/PWI mismatch do not benefit from IV thrombolysis beyond the 3-hour window. The ECASS-3 trial did not use such MRI paradigm and has shown benefit of IV TPA in acute ischemic stroke 3-4.5 hours from onset.⁴

The trials EPITHET and DIAS-2 raise several technical issues with DWI/PWI mismatch, particularly with regards to (1) most appropriate DWI/PWI ratio, (2) most appropriate PWI threshold, (3) coregistration of DWI and PWI, and (4) online automated assessment of mismatch.

Newer MRI techniques are being developed to assess regions of reversible ischemia. Susceptibility-weighted imaging such as BOLD functional MRI (BOLD-fMRI) is considered to provide a surrogate for brain OEF measurement and is one such emerging technique.⁵

A few hours after stroke onset, a loss of arterial void signal is sometimes observed (30-40% of patients); it is best observed on T2-WI.

• At 2-4 hours, T1-WI shows subtle effacement of the sulci due to cytotoxic edema.
• At 8 hours, T2-WI shows hyperintense signal due to both cytotoxic and vasogenic edema.
• At 16-24 hours, T1-WI shows hypointense signal due to both cytotoxic and vasogenic edema.

Contrast-enhanced images show arterial enhancement followed by parenchymal enhancement. The arterial enhancement can be very early (in more than 50% of patients) and is due to slow blood flow; it typically disappears after 1 week.

Parenchymal enhancement differs in complete and incomplete infarctions. In complete infarction, it starts 5-7 days after the stroke and persists for several months. In incomplete infarctions, it can be observed within 2-4 hours and usually is more intense than in complete infarction.

Although conventional MRI sequences most often do not show evidence of stroke in the acute phase, conventional MRI may show signs of intravascular thrombus such as absence of flow void on T2-WI, vascular hyperintensity on FLAIR, and hypointense vascular sign on GRE sequence.

MRI findings in acute ischemic changes

Open table in new window

MRI findings in acute stroke - Acute phase (1-7 d)

In this phase, edema increases (edema maximizes at 48-72 h), and MRI signals become more prominent and well demarcated. The ischemic area continues to appear as an area of hypointensity on T1-WI and as a hyperintense area on T2-WI. Also, the mass effect can be appreciated in this phase.

In contrast-enhanced images, the arterial enhancement usually persists throughout the acute phase, while the parenchymal enhancement is usually appreciated at the end of this phase in complete infarction. In incomplete infarction, the parenchymal enhancement is usually earlier.

During this period, reperfusion occurs and both petechial and frank hemorrhage can be observed, typically 24-48 hours after the onset of the stroke. Usually, petechial hemorrhages cause the "fogging" phenomenon, due to hemoglobin degradation products, that masks the infarction on both T1-WI and T2-WI.

MRI findings in acute stroke - Subacute phase (7-21 d)

In this phase, the edema resolves and the mass effect becomes less appreciated; however, the infarcted areas still appear as a hypointensity on T1-WI and as a hyperintensity on T2-WI.

In contrast-enhanced images, the arterial enhancement is usually resolved by this time, and the parenchymal enhancement typically persists throughout this phase. The cortical parenchymal enhancement is usually in a gyriform pattern, while the subcortical enhancement is usually a homogenous central pattern.

MRI findings in acute stroke - Chronic phase (>21 d)

In this phase, the edema completely resolves, and the infarcted area still appears as a hypointensity on T1-WI and as a hyperintensity on T2-WI. Because of tissue loss in the infarcted area by this time, ex-vacuo ventricular enlargement and widening of the cortical gyri and fissures take place.

In contrast-enhanced images, parenchymal enhancement typically persists throughout this phase also; it usually disappears by 3-4 months.

MRI findings in TIA

A third to a half of the patients presenting with a TIA have lesions on DWI. A significant proportion of these patients may not reveal a corresponding lesion on T2-WI. PWI may be more sensitive but has not been adequately tested in patients with TIA. DWI-positive TIA lesions do necessarily show as infarction on follow-up MRI.

Although TIAs have been traditionally defined as transient (<24h) neurologic deficit of vascular origin, the advent of MRI has lead to reconsideration of the definition. Whether DWI-positive TIAs are to be regarded as stroke or TIA is unclear.

MRI in hemorrhagic stroke

GRE and EPI sequences have the ability to detect microbleeds that are clinically silent and not visualized by CT scanning or routine MR sequences. These microbleeds are visualized in a fifth to a quarter of patients with ischemic stroke and 5% of elderly asymptomatic individuals. The microbleeds depict hemosiderin deposit and have been histopathologically correlated with prior extravasations of blood. These microbleeds may represent bleeding-prone angiopathy and a higher rate of hemorrhagic transformation from anticoagulation, antithrombotic, and thrombolytic therapy.

GRE, EPI, and DWI (B0) are sensitive to detecting intraparenchymal hemorrhage (primary intracerebral hemorrhage and hemorrhagic transformation) in the hyperacute stages (first few hours), whereas the conventional T1-WI and T2-WI are sensitive in detecting subacute and chronic bleeding. FLAIR sequences may have a role in detecting extra-axial collections of blood (subdural hemorrhages). Having stated the above, the current guidelines do not advocate the use of MR in place of CT scanning to screen patients for thrombolysis.

• Search for CME/CE on This Topic »

• Magnetic Resonance Imaging (MRI) Introduction
• Magnetic Resonance Imaging (MRI) Preparation
• Imaging Center
• Stroke Overview
• Stroke Causes
• Stroke Treatment
• Stroke Center
• Stroke Symptoms