Stroke

Fall 2000
Volume 11, Number 2

 Next Article  

MR diffusion and perfusion imaging techniques, by virtue of their ability to identify and characterize acutely ischemic tissue, have the potential to significantly improve clinical decision making in and care of stroke patients. At the Colorado Neurological Institute, MR diffusion and perfusion imaging techniques have been integrated into the routine stroke imaging protocols. Although CT remains the initial imaging study in all patients that present with an acute neurologic deficit, it is used primarily to exclude the presence of hemorrhage and other stroke mimickers. Diffusion-weighted MR imaging is now performed in all patients in whom ischemic stroke is a diagnostic consideration.

Background. It is estimated that between 700 to 750 000 new or recurrent strokes occur each year in the US, making it the most common neurological problem encountered in this country.1 It accounts for more than 50% of hospital admissions for neurological disease and is the most common cause of long-term disability in the US.2

Because stroke and its associated complications are so commonplace, imaging tests, such as computed tomography (CT), magnetic resonance (MR) imaging, cerebral angiography, and a variety of nuclear medicine studies are frequently performed. While each of these imaging tests provides valuable diagnostic information, each also has well known limitations. One of the most significant limitations of these conventional imaging tests is the inability to reliably detect acute ischemic infarcts.

Recent advances in imaging, particularly in the field of MR imaging, have the potential to revolutionize both the imaging and care of acute stroke patients through the identification and characterization of acutely ischemic tissue. The recent acceptance of intravenous rt-PA as a viable therapy for infarcts within 3 hours of symptom onset 3, 4 and promising results with intraarterial rt-PA within the first 6 hours of an ischemic stroke5, have made reliable detection of hyperacute stroke within these time periods absolutely essential when use of this or other potentially dangerous thrombolytic agents are being considered. The goals of this article are to introduce readers to these new MR techniques and provide examples of how they can be used effectively for identification of very early ischemic strokes.

Conventional imaging in ischemic stroke. Historically, imaging tests have been used to study stroke patients not so much to detect ischemic or infarcted brain tissue, but rather to detect or exclude disease processes that can mimic the clinical presentation of ischemic stroke. This has occurred primarily for 2 reasons; 1) Imaging studies traditionally have been unable to reliably detect acute ischemic infarcts, and 2) The diseases that most commonly mimic the presentation of acute ischemic stroke, including primary intracranial hemorrhage, subarachnoid and subdural hemorrhages, and brain tumors, are readily detected with routine CT. For that reason, CT has become the single most common test used to image stroke patients.

In contrast to the high sensitivity for detection of “stroke mimickers”, CT scanning is mediocre at detecting early ischemic stroke. Initial reports suggested CT detected only 58% of ischemic strokes within the first 24 hours after symptom onset.6 More recent data obtained as part of the PROACT II thrombolysis trial has suggested that CT may actually detect 75% of hyperacute infarcts (ie, infarcts less than 6 hours old).7 In that study, the investigators found the most common abnormal CT findings to be edema in the insular ribbon (46%), cerebral cortex (43%), or lentiform nucleus (39%), or a hyperdense middle cerebral artery (34%). Despite this apparent improvement in the ability of CT to detect hyperacute infarcts, missing 25% of infarcts in the first 6 hours still raises questions about the reliability of CT.

Detection of very early infarcts is widely considered to be better with MR imaging than with CT because of its superior contrast resolution. This was confirmed in one early study that found an 82% detection rate of ischemic infarcts in the first 24 hours with MR versus 58% with CT.6 Improvements in MR scanners and MR pulse sequences since that time have improved the ability of MR to detect ischemic infarcts; in a series of 70 patients with 72 infarcts, Ricci et al, were able to detect 96% of the lesions with conventional MR scanning techniques in the first 24 hours.8 However, detection rates in the first 6 hours with conventional MR techniques remain poor. Furthermore, 30% of infarcts detected with conventional MR imaging can increase in size on follow-up studies6, which suggests that the initial study either underestimates the size of the infarct or that it doesn’t detect ischemic tissue that is at risk for subsequent infarction.

The poor sensitivity of conventional CT and MR for identifying hyperacute infarcts stems from their inability to detect the main pathologic changes occurring within that 6-hour time period: decreased cerebral blood flow and development of cytotoxic edema. The most common vascular change detected with CT scanning is the hyperdense vessel sign, which is the result of in situ thrombus or an embolus. Most commonly seen in the middle cerebral artery, it is present in only one-third of cases.7 (Fig 1) Even though MR imaging is more sensitive than CT to alterations in cerebral blood flow, it detected altered flow in only 50% of infarcts in the first 8 hours in one study. 9 (Fig 2)

Fig 1 Dense middle cerebral artery sign. The CT scan in figure 1 was obtained on a 51 year old male who presented to the emergency department with left sided weakness. CT scan revealed a hyperdense middle cerebral artery. At the time of this scan, the remainder of the CT was normal. The hyperdense middle cerebral artery, one of the earliest CT of ischemic infarction, is due to in situ thrombus or embolism. Although it is highly specific, it is too uncommon to be a reliable sign of infarction.

Fig 2 Absent carotid artery flow on MR. MR imaging is more sensitive to alterations in cerebral blood flow than CT scanning. Figure 2 demonstrates black signal in the left internal carotid artery (small white arrow); this black signal results from time of flight effects and indicates normal flow. The lack of flow signal in the right internal carotid artery (large white arrowhead) suggests the vessel is occluded or has very slow flow.

At the cellular level, the inadequate blood flow deprives the brain parenchyma of the oxygen and glucose needed to perform basic functions. Within minutes, the sodium-potassium pump fails allowing water and sodium to move from the extracellular to the intracellular fluid compartment. This change in water location, known as cytotoxic edema, causes total water content of the brain to increase by less than 3%.10 With CT and MR imaging, this small fluid shift can occasionally be seen as gyral swelling; in most instances however, the changes are too subtle to be reliably detected. Vasogenic edema, which occurs when disruption of the blood brain barrier permits water and macromolecules to enter the brain from the intravascular compartment, results in much larger fluid shifts than those seen in cytotoxic edema. Because of the large fluid shifts involved, vasogenic edema is more readily detected with conventional CT and MR. Unfortunately, this type of edema typically takes between 4 to 6 hours to develop once blood flow decreases to critical ischemia levels.10 If blood flow is completely disrupted, vasogenic edema can take even longer to develop.

Acute stroke imaging. Two new imaging techniques, MR diffusion imaging and dynamic susceptibility-weighted MR imaging (also known as MR perfusion imaging), are gaining acceptance as the premier stroke imaging techniques because they permit accurate, reliable diagnosis and characterization of ischemic strokes within the critical first 6-hour time period needed to initiate thrombolytic therapy. These techniques, which use very different approaches to gather physiologic rather than anatomic information about ischemic or infarcted tissue, are generating considerable excitement in the medical community because of their potential to improve clinical decision-making in and the care of stroke patients.

Fig 3 Normal diffusion-weighted image. In the normal brain, motion of water molecules during the acquisition of the diffusion images a causes decrease in signal intensity. This produces a relatively uniform gray appearance to the brain on trace diffusion-weighted images. Because water diffusion differs between gray and white matter, a slight difference in signal intensity can be seen.

Fig 4 Diffusion-weighted image in acute ischemic infarction. In ischemic infarction, inadequate cerebral blood flow causes energy requiring cellular processes, like the sodium/potassium pump, to fail. Sodium and water then rush into the cell; the restricted water motion that results from confining the water molecules to the intracellular space results in high signal on diffusion-weighted images. Such high signal intensity areas correlate well with areas of infarction on follow-up studies.

Diffusion Imaging. Diffusion imaging is a unique method of rapid MR scanning that produces images in which the signal intensity is dependent on the motion of water molecules in the brain.11 In normal brain tissue, water motion in the extracellular fluid is relatively unrestricted and randomly oriented. The MR diffusion sequence is designed in such a way that this random water motion produces signal loss on diffusion-weighted images; the result is a uniform gray appearance to normal brain tissue. (Fig 3) In contrast, disease processes that restrict the motion of water molecules produce high signal-intensity instead of the uniform gray intensity of normal tissue. In the case of acute cerebral infarction, the shift of water molecules from the extracellular fluid to the intracellular space (i.e., the development of cytotoxic edema) severely restricts the motion of water molecules confined within the cells, producing high signal intensity on diffusion-weighted images. (Fig 4) When viewed against the uniform gray signal of normal tissue, these areas of high signal intensity infarction become much more conspicuous than infarcts seen with more conventional MR techniques.8

What makes diffusion imaging so valuable in the stroke imaging armamentarium is its ability to reliably detect hyperacute ischemic infarcts when other imaging tests are still normal (Fig 5), its ability to help determine infarct age, and, its ability to distinguish small lacunar infarcts from chronic microvascular ischemic changes. Cytotoxic edema has been found to occur as early as one minute after vessel occlusion.10

Fig 5a Conventional MR and diffusion imaging in hyperacute infarction a) T2-weighted image of hyperacute infarction. Numerous studies have shown that T2-weighted images are insensitive to the detection of ischemic infarction in the first several hours. This is most likely because the fluid shifts that occur in the early stages of infarction are too small to detect with conventional MR. On the image shown, vague signal change is present in the superior portion of the left basal ganglia; because this is a common location for the small vessel ischemic changes that are seen in the normal aging brain, unequivocal diagnosis of acute infarction is not possible.

Fig 5b Diffusion-weighted image of hyperacute infarction. This image, obtained at the same time as that in Figure 5a, clearly demonstrates a high signal intensity infarction in the left basal ganglia. The improved sensitivity of diffusion-weighted images is due to their unique ability to detect the restricted motion of water that results from the development of cytotoxic edema.

In experimental models of cerebral infarction, abnormal signal within infarcted tissue has been detected on MR diffusion studies within 15 minutes of vascular occlusion.12 More importantly, diffusion imaging has proven to be greater than 95% sensitive and specific for detecting infarcts within the first 6 hours of symptom onset in humans, making it the most reliable method currently available for infarct identification.13

As the infarct evolves, cell membranes break down, releasing the intracellular water back into the extracellular space. Water motion becomes relatively unrestricted once again causing the high signal in the infarct on diffusion images to fade back to gray. Typically, this return to normal signal-intensity on diffusion-weighted images occurs 10 to 14 days following the onset of the stroke.14 This change in signal intensity can be useful when trying to determine the age of ischemic infarcts.

Perfusion imaging. Perfusion studies of the brain have been performed using nuclear medicine and positron emission tomography techniques for many years. With the development of ultra-fast scanning techniques, these studies are now possible with MR scanners where they have the added benefit of superior spatial resolution compared to the nuclear medicine techniques. The “bolus-tracking” MR technique, first described by Villringer, is the most commonly used type of MR perfusion study.15 In this technique, a bolus of gadolinium contrast is injected into a peripheral vein while the brain is imaged. As the bolus passes through the brain, it distorts the local magnetic field around blood vessels causing a decrease in signal- intensity on the MR images. The degree of change in signal intensity is proportional to level of cerebral blood flow. Because data is acquired dynamically as the contrast passes through the brain, the change in signal intensity over time can be mapped. (Fig 6) This data can then be mathematically manipulated to produce a variety of perfusion measurements including: relative cerebral blood volume (rCBV), relative cerebral blood flow (rCBF), relative mean transit time (MTT), time to peak (TTP), and more. (Fig 7)

Fig 6 MR perfusion time intensity curve. Data from bolus tracking MR perfusion studies are obtained dynamically as the contrast bolus passes through the brain. In regions with normal cerebral blood flow, the contrast causes a decrease in signal intensity; in regions with impaired flow, this signal intensity change is blunted. On the time intensity curve depicted, the bottom curve was obtained from an area of normal blood flow in the left hemisphere. The upper curve was obtained from the region of an acute right middle cerebral artery infarct. The lack of any significant change in signal intensity suggests near complete absence of flow in the region from which the curve was obtained.

Fig 7 Normal mean transit time perfusion map. Time intensity curves, such as that shown in Figure 6, can be mathematically manipulated to produce a variety of different perfusion parameters. The relative mean transit time map shown here and the relative cerebral blood volume map are very sensitive for detection of cerebral hypoperfusion. Cerebral blood volume maps are less sensitive for infarct detection, but appear to correlate best with final infarct size.

Cerebral perfusion changes in ischemic stroke have now been well characterized with MR perfusion imaging. In animals, changes in cerebral perfusion have been noted within 15 minutes of vascular occlusion, even when conventional MR images and occasionally diffusion images were normal.16 Of the various calculated perfusion parameters, rCBF values and rMTT changes have been found to be most sensitive for hyperacute infarct detection in humans. (Sorenson, Ueda) The poor sensitivity of rCBV maps probably reflects compensatory vasodilatation that accompanies decreased cerebral blood flow and ischemia. The change in rCBV on the other hand appears to be a more accurate predictor of final infarct volume than either rCBF or rMTT. 17, 18

The ischemic penumbra. Cerebral infarction is a dynamic process that starts with an alteration in cerebral blood flow. Normal cerebral blood flow in humans is approximately 50-60ml/100gm of brain tissue/minute. When flow decreases to 20-40ml/100gm/min, various degrees of neuronal dysfunction are noted electrophysiologically. Below approximately 10-15ml/100gm/min, damage to the neural tissue appears to become irreversible.

In humans, there is extensive collateral blood flow in the brain. This typically results in a gradation of perfusion changes within an ischemic lesion that range from mild to severe. In the central core of the infarct, the severity of hypoperfusion results in irreversible cellular damage occurred. Around this core, there is believed to be a region of decreased flow in which either the critical flow threshold for cell death has not yet been reached or the duration of ischemia has been insufficient to cause irreversible damage. This region, analogous to the stunned myocardium seen in myocardial infarction, has been called the “ischemic penumbra.” An imaging technique that accurately identifies this tissue at risk could have a tremendous impact on patient management. Because bolus-tracking MR perfusion studies are qualitative rather than quantitative, they cannot determine absolute CBF and therefore cannot distinguish regions of completed infarction from those that have yet to suffer irreversible damage. Similarly, diffusion imaging is known to underestimate final infarct size, suggesting it cannot identify the ischemic penumbra either.

However, several recent studies have shown: 1) Up to 80% of infarcts have diffusion- perfusion mismatches in the first 24 hours; 2) If the initial rCBF and rMTT lesion volumes are larger than the initial diffusion lesion volume, there is an increase in infarct size corresponding to the perfusion-diffusion mismatch; and, 3) A 6 second or greater delay in TTP values in the region surrounding the diffusion abnormality accurately predicts infarct progression.17, 19 , 20 These studies have tremendous implications for the treatment of patients with acute stroke. If there is no difference in the volume of the perfusion and diffusion lesions, there may be nothing to gain from intravenous or intraarterial thrombolysis because the ischemic tissue has likely progressed to complete infarction. In contrast, a perfusion-diffusion mismatch suggests there is a penumbra of salvageable brain tissue that may benefit from thrombolysis. (Fig 8)

Fig 8a The ischemic penumbra. Diffusion-weighted image obtained from a patient with right-sided weakness reveals a small infarct in the left corona radiata. Although diffusion imaging is very sensitive for infarct detection, it typically underestimates final infarct size.	Fig 8b Mean transit time map obtained at the same level reveals a much larger area of hypoperfusion (the area of increased signal intensity). The difference in volume between the diffusion and perfusion images is felt to represent the ischemic penumbra, the territory at risk for subsequent infarction.

Conclusion. Stroke imaging at the CNI. MR diffusion and perfusion imaging techniques, by virtue of their ability to identify and characterize acutely ischemic tissue, have the potential to significantly improve clinical decision making in and care of stoke patients.

At the Colorado Neurological Institute, MR diffusion and perfusion imaging techniques have been integrated into the routine stroke imaging protocols. Although CT remains the initial imaging study in all patients that present with an acute neurologic deficit, it is used primarily to exclude the presence of hemorrhage and other stroke mimickers. Diffusion-weighted MR imaging is now performed in all patients in whom ischemic stroke is a diagnostic consideration. It has become an indispensable tool to diagnose and exclude cerebral infarction, distinguish small acute infarcts from chronic microvascular ischemic changes, and to help determine infarct age. Both diffusion and perfusion imaging are performed in any patient in whom intraarterial thrombolysis is being considered to identify the ischemic penumbra. The goals of this approach are to: 1) Identify individuals with brain tissue at risk for infarction that will benefit from thrombolysis; and, 2) Prevent thrombolysis in those individuals with completed infarcts, or in those individuals who have no evidence of infarction. Using these advanced tools has enabled the CNI Stroke Service to remain at the forefront of care in stroke patients.

Address comments and questions to:

Peter E. Ricci, MD
Radiology Imaging Associates
3900 S. Wadsworth Boulevard, # 250
Lakewood, CO

 Next Article