Applied Radiology

Dr. Hanes received his medical degree from Emory University Medical School, Atlanta, GA in 1998. He is currently a third-year Radiology Resident at Emory University. He plans to begin a fellowship in Neuroradiology at Emory in 2003.

First-pass perfusion computed tomography (CT) and CT angiography (CTA) can evaluate brain ischemia and large vessel patency rapidly using the same imaging equipment often used in the initial evaluation of stroke. While magnetic resonance (MR) diffusion and perfusion techniques hold great promise in the setting of cerebral ischemia/infarction, this promise has not been realized, due in part to a lack of widespread 24-hour availability of MR equipment. This paper reviews the use of contrast-enhanced CT studies to provide needed information within a relevant time frame to tailor appropriate treatment for victims of cerebral vascular accidents. Reviewed studies will demonstrate that these techniques triage patients quickly for current treatment regimens and soon may provide the basis for thrombolytic treatment regimens outside of the 6-hour limit currently advocated.

Cerebral vascular disease is the second leading cause of death and the leading cause of permanent disability, affecting more than 600,000 people each year in the United States alone. ¹ In recent years, significant advances have been made in the treatment of stoke. In 1996, intravenous (IV) tissue plasminogen activator (t-Pa) became the first treatment approved for acute stroke by the U.S. Food and Drug Administration. This approval was based on the findings of the National Institutes of Neurological Disorders and Stroke (NINDS) t-Pa Study Group. This large, multicenter, placebo-controlled study showed long-term clinical benefit if IV t-Pa was given within 3 hours of onset of ictus to patients who were found not to have hemorrhagic stroke by a noncontrast computed tomographic (NCCT) head scan. ² There was a substantially increased risk of hemorrhagic transformation in the treated group, however, and in practice the criteria have proven very restrictive.

The International Stroke Trial Pilot Study of 984 patients found that only 4% of patients presented to the emergency department clearly within the 3-hour limit. Extending the cut-off to 6 hours included no more than 16% of cases. ^3,4 Another study indicated that up to 60% of cases present with an uncertain time of onset. ⁵

The Prolyse in Acute Cerebral Thromboembolism (PROACT II) study demonstrated a new role for imaging, while again illustrating the limitations of current treatment. ⁶ It showed benefit for proven occlusion of the M1 or M2 branches of the middle cerebral artery (MCA) with intra-arterial urokinase within 6 hours of onset, but of 12,323 patients screened for suspected stroke only 474 candidates underwent immediate angiography. Of these, only 180 patients met all criteria for treatment.

Attempts to extend the proven effective treatment window for IV t-Pa have yielded ambiguous data, and again imaging is central to the issue. The European Cooperative Acute Stroke Studies (ECASS) I and II failed to demonstrate benefit of IV thrombolysis within 6 hours. ^7,8 However, a sub-group analysis by the ECASS CT review panel found that 52 patients were given t-Pa inappropriately due to "misinterpretation" of CT scans. ⁷ The review panel concluded these patients showed CT findings of ischemic changes involving >33% of a single MCA territory, which was a criterion for exclusion from the treatment group. When these patients were excluded from the analysis, benefit within a 6-hour window was shown.

CT perfusion and CTA techniques will be discussed that have demonstrated both sensitivity and specificity for ischemic stroke. These provide information in a clinically relevant time frame to expand the pool of patients eligible for treatment protocols within, and perhaps beyond, the 6-hour window.

Pathophysiology

If cerebral perfusion is gradually reduced in a region, the first detectable threshold is the point at which the metabolic needs of electrical activity are no longer met and that portion of the brain becomes "silent" by electroencephalographic (EEG) monitoring and symptomatic for the patient. Measurements using ¹⁵ O-labeled water positron emission tomography (PET), considered the gold standard for absolute perfusion, and other modalities such as xenon-enhanced CT (Xe CT) and single-photon emission computed tomography (SPECT), have given consistent estimates of 15 to 25 mL/100 g tissue/min for this threshold. ⁹ If perfusion is not restored within approximately 6 hours, or if perfusion falls to 10 to 15 mL/100 g/min, the neuron can no longer maintain normal ATP-dependent Na+/K+ membrane ion pump activity and dendritic and perivascular astrocytic end-foot swelling occurs. ¹⁰ This so-called "cytotoxic edema" can be detected readily by MRI and is partially or completely reversible if perfusion is normalized within 3 to 6 hours. Beyond this period, or if perfusion falls below 10 mL/100 g/min for even 5 to 10 minutes, infarction is inevitable and cytotoxic edema becomes marked. If perfusion is reduced but still present for 4 to 6 hours, there is loss of integrity of the blood-brain barrier, allowing a marked efflux of water and proteins into the extracellular space from the capillary bed. This phenomenon, termed vasogenic edema, is detected by both MRI and NCCT and typically reaches a peak in 3 to 5 days.

The natural history of a stroke varies, and current thinking holds that there are opportunities for effective interventions if the pathophysiology can be better characterized. ^4,11-13 Clinical progression of stroke occurs in as many as one third of cases, ¹⁴ and enlargement of the actual territory of the infarction is very common. One recent prospective study found that 46 of 49 untreated stroke patients demonstrated growth of infarct volume on scans performed within 24 hours of the event and on the second day after. ¹³ Other longitudinal studies ¹⁵ and animal models ¹⁶ support the theory that with an event, there is a core of infarcted tissue that is nonviable, surrounded by hypoperfused but still viable brain at risk for infarction by prolonged hypo-perfusion, failure of collateral circulation, secondary embolic events, further reduced perfusion secondary to parenchymal edema, or excitatory neurotransmitter-mediated metabolism alteration and oxygen-radical injury. The volume of nonviable tissue is strongly correlated with the risk of hemorrhagic transformation. ^17,18

The threatened tissue, or ischemic penumbra, is the target for treatment. Dedicated MRI techniques image this threatened tissue directly and changes of ischemia have been shown as early as 5 minutes after the onset of the insult using animal models. ¹⁹ Reperfusion treatments have proven benefit. ^2,6,20 Neuro-protective agents have yet to show benefit in humans.

Imaging techniques and requirements

CTA

Helical CT scanners must be used for CTA in order to image the cervical vessels through M3 branches prior to the venous phase, as traditional axial CT scanners are unable to cover the required territory of interest before the arterial bolus becomes venous. Following the acquisition of a routine NCCT, CTA is performed, followed by a CT perfusion scan after a minimum delay of 5 minutes. ^21-23

Typically, 75 to 150 mL of 300 mg I/mL or 370 mg I/mL iodinated (I) contrast is injected at rates of 3 to 4 mL/sec through an 18-gauge IV needle in an antecubital vein. ^21-24 Depending upon the likely cardiovascular status of the patient, standard delays between 18 and 24 seconds can be selected before initiating scan in a caudal-cranial direction from either the cervical vessels or the skull base through the Sylvian fissures 1.5 to 2.0 cm above the sella. A 10- to 15-mL timing bolus may be helpful if the patient is suspected of having an abnormal cardiac output that might delay the peak arterial contrast phase. Proprietary software packages vary among vendors regarding the specifics of a timing bolus technique. Beam collimations from 1 to 1.5 mm are used with a low pitch of 0.8 to 1.0 with 120 to 140 kV and maximum mA typically 250 to 300. The raw data is reconstructed with a low-noise algorithm with a small field of view (FOV) that includes the Circle of Willis and the lateral M3 branches, using 50% overlapping thin slices. The images should be reviewed on a workstation allowing multiplanar reformats, three-dimensional (3D) reconstructions with easily manipulated display windows and levels. Curved planar reformatting, maximum-intensity projection (MIP), volume-rendering (VR), and shaded surface display (SSD) have been investigated, ^23-25 showing an advantage for MIP when dense vessel calcifications are present, and for VR when viewing complex anatomy. ^23,25

CT perfusion

CT perfusion techniques are evolving, and many recent advances were presented at the 2001 annual meeting of the Radiological Society of North America (RSNA 2001). Eastwood et al ²¹ reported on a scanning protocol that performs a continuous cine in the transverse plane, with angulation of the gantry to minimize imaging artifact from dental amalgam. The continuous acquisition necessitates limiting the imaged territory to the greatest composite slice thickness the CT scanner can perform. Therefore a scanner with a single 1-cm detector could provide a single 10-mm slice perfusion study coverage, whereas a scanner with four 5-mm detectors could produce a total imaged territory of 2 cm. The imaged territory should be chosen to include the basal ganglia as well as cortical vascular territories of the anterior, middle, and posterior cerebral arteries bilaterally. Alternate slice locations may be chosen on the basis of clinical presentation and the preliminary findings of the angiographic scan. A total of 40 to 50 mL of contrast, a 25-cm FOV, and 512 * 512 pixel reconstruction are used. X-ray generation and contrast injection techniques vary. A high injection rate technique using 10 to 20 mL/sec would be imaged with a kV of 140 and 40 mA. ²⁶ A lower injection rate technique using 3 to 6 mL/sec, such as that recently validated by Eastwood et al and others, uses a lower energy technique of 80 kV and 200 mAs to improve tissue contrast while keeping contrast dose low. ^21,27,28 Higher rate techniques require large-bore venous access, such as 14-gauge IV needles, which can be difficult to place in the older patients, making the lower rate technique more feasible.

Uses, validation, and pitfalls

CTA

It is argued the value of noninvasive angiography should be equal to that of conventional angiography for diagnosis of large-vessel occlusion, without the 1% to 2% complication rate of stroke or arterial injury. ²³ The advantages of traditional angiography are: 1) it remains the gold standard for diagnosis and is itself evolving with the use of 3D rotational acquisition for single injection depiction of vascular anatomy; and 2) treatment of an identified large clot with catheter-administered intra-arterial thrombolytics is an immediately available option.

Conventional angiography has a proven and important role in the diagnosis of stroke. It can define the etiology of an ischemic event, directing both immediate and long-term treatment. ^24,29 Carotid dissection and involvement of the cerebral vasculature by aortic dissection and large-vessel high-grade stenosis/occlusion are readily shown, ^24,25,29,30 and ulcerated arterosclerotic plaque is well depicted. ^29-31 Third, the conclusions of the PROACT II study support the use of angiography to prove MCA occlusion for intra-arterial thrombolysis up to 6 hours after the onset of ictus ⁶ (figures 1 and 2).

There is validation in the literature comparing CTA with invasive methods on emergent patients. In one series of 44 patients, patients with abnormal findings on CTA were immediately taken for conventional angiography and a comparative accuracy of 99% for CTA was found. ²⁴ No patient with a normal CTA was found on clinical or imaging follow-up to have had a large-vessel occlusion. In another series of 42 patients, all 22 patients later shown to have large-vessel events were correctly identified by CTA. ²³ Multiple studies on nonemergent patients comparing CTA with digital subtraction angiography (DSA) and magnetic resonance angiography (MRA) have shown near 100% accuracy for occlusions. The literature disagrees, however, on CTA's ability to differentiate between stenoses of 50% to 69% and 70% to 99%. ^25,29-31

A complete overview of the challenges of CTA is beyond the scope of this article; however, the following examples illustrate common problems and how they can be addressed.

One pitfall identified in the literature is the potential for a false positive diagnosis of internal carotid artery (ICA) occlusion due to slow flow proximal to a stenosis or occlusion of a large branch vessel when using an early arterial phase bolus technique. ³⁰ A brief delay of 2 to 3 minutes, long enough to review initial images at the scanner monitor and to initiate a second scan, may aid in these cases. A second pitfall to avoid is injecting the contrast into a left arm vein. If the patient has a compromised cardiovascular status, the minimal relative impediment to passage of contrast through the brachiocephalic vein can cause relux into subcutaneous, muscular, and epidural plexus veins, degrading the study. This can be lessened or avoided altogether by a right arm injection.

Another pitfall is that although imaging vessel calcification is an advantage of CT, beam hardening artifacts remain difficulties, especially dense medial calcifications found in diabetic patients. MRA has been considered to have a decided advantage in these cases, ³¹ however, using VR and MIP displays for CTA of the posterior fossa for vasospasm secondary to subarachnoid hemorrhage has shown promising results. ³² Continuing efforts to address the obscuration caused by bone, aneurysm clips, dental amalgam, and other surgical metal are being made. A complicating factor is that often functions labeled "VR" and "MIP" can mean different renderings with different proprietary software configurations, and the radiologist's experience with the specific software package and workstation will prove important in the diagnostic value of the study.

CT perfusion

CT perfusion is based on tracer kinetic theory analysis using a single-compartment model of first-pass of a bolus of iodinated contrast material through the cerebral vasculature. The single compartment is the vascular system, and it is assumed there is no extravasation or stasis of contrast or alteration of the studied organ caused by the contrast agent. As the contrast bolus passes through the arterioles, capillary bed, and venules of the brain, the brain parenchyma becomes homogeneously and diffusely more X-ray attenuating. The entry and exit of the bolus can be seen directly as opacification of the major arteries and veins, respectively.

The distribution of an IV contrast bolus leaving the heart and entering the brain should follow a gamma-variate function, and a curve can be readily generated showing an increase, peak, and return to baseline of the X-ray attenuation by the brain as a function of time (figure 3). From this curve, peak enhancement, time to peak, and width at half-height--each proportional to blood flow--can be calculated readily. Perfusion protocols generate a contrast-enhancement curve for each pixel within the scanned slices using research ³³ or commercially available software packages. ^21,34 Regions-of-interest (ROIs) can be hand-drawn over one vessel for arterial input and over another for a venous input, typically a branch of the MCA and the superior sagittal sinus, respectively. Taken altogether, this data is then analyzed to produce values for three parameters: mean transit time (mTT), cerebral blood flow (CBF), and cerebral blood volume (CBV). Mean transit time can be defined either by the central volume principle as the ratio of CBV to CBF. Cerebral blood volume can be calculated as either proportional to the peak enhancement of a pixel, or from the central volume principle using the other two values. These values can be displayed as color images for interpretation. Interoperator reliability for generation of these results by experienced neuroradiologists has proven acceptable. ^21,35

The data are analyzed by one of two algorithms, the maximum slope ^34,36 or deconvolution method. ^21,33 The difference between methods lies in the determination of CBF. The maximum slope method assumes no venous contrast and calculates CBF by finding the slope of the tissue-concentration curve versus the integral of the arterial-concentration curve. This method requires very high injection rates of 10 to 20 mL/sec in order to obtain the needed arterial contrast concentration within the typical minimum transit time of 4.5 to 6.5 seconds into the venous system. ³⁷ The deconvolution technique compares the shape and height of the time-attenuation curve of each pixel with the shape and height of the arterial and venous time-attenuation curves to determine CBF. Venous contrast is accounted for, and therefore contrast injection rates of 3 to 6 mL/sec can be used. The resulting perfusion maps are noisier but diagnostically equivalent. ³⁸

A review of the literature finds that several small series of acute cases have been published. Three studies included 22 patients, ³⁹ 20 patients, ⁴⁰ and 32 patients ⁴¹ imaged within 6 hours of onset of symptoms, and one included 70 patients who were imaged within 12 hours of onset. ⁴² The reported sensitivities ranged from 89% ⁴¹ to 95% ³⁹ and specificities from 98% ⁴² to 100% ³⁹ for the detection of stroke.

MTT, CBV, and CBF

While it is agreed that tissue death is a direct result of time subjected to blood flow below a viability threshold, CBF has not proven to be a single-parameter predictor of tissue outcome.

The most sensitive and least specific indicator of ischemia is the mTT. ^21,34,40,42 With a stroke, this measure tends to be abnormal over the greatest area of brain, and likely indicates the infarct, the penumbra, and adjacent brain that is not at risk of infarction. An mTT below the threshold of 6 seconds likely reflects delayed blood flow by stenosed or collateral, circuitous routes. ²¹ This threshold agrees well with findings by MRI (figure 4). ¹³

CT perfusion-derived CBF has been validated in animals and humans in comparison with Xe CT, ³³ MRI, ⁴³ and SPECT, ⁴¹ although the issue of quantification remains, and currently ¹⁵ O labeled-water PET is the gold standard for determining absolute CBF values. Quantification is also an issue regarding MRI perfusion studies, and, to date, all MRI perfusion studies have used relative measures. Koenig et al ³⁴ quantified flow by CT perfusion as relative to the contralateral hemisphere--relative cerebral blood flow (rCBF) and relative cerebral blood volume (rCBV)--while others give unit values. These values are readily comparable and the results between groups substantially agree. Eastwood et al ²¹ showed strokes detected by NCCT have mean CBF of 13.1 + 8.4 mL/100 g/min compared with values for the contralateral hemisphere of 31.6 + 12.4 mL/100 g/min (roughly 40%). Mayer et al ⁴² showed that all of the patients with rCBF values below 30% progressed to infarction, as did half of those whose values were between 30% and 60%.

An rCBF cutoff value of 48% for prediction of infarction if left untreated has been proposed by the authors of a third study in which the lower range of normal perfusion was 55 mL/100 g/min in the contralateral hemispheres. This value had an efficiency of 74.7% predicting the outcome of tissue. ³⁴ These findings correlate well with the widely accepted values of 15 to 25 mL/100 g/min for infarction if prolonged (figure 5).

Cerebral blood volume abnormalities can vary between hyperemia, mild oligemia, and profound hypovolemia. ^21,34,44 There is vasodilatory autoregulation in the setting of acute ischemia, and some have posited a protective effect of hypervolemia in the setting of reduced flow. ^21,44 However, as compensatory mechanisms fail, CBV falls, therefore it has been argued that a markedly decreased CBV is the most specific indication of infarction. ^21,34,42 Eastwood et al ²¹ propose a cutoff CBV of 1.5 mg/100 g (roughly half the CBV of controls), and showed a mean CBV of 0.9 ± 0.4 mg/100 g within areas found to be abnormal on NCCT. A second study proposed an rCBV cutoff of 60%, ³⁴ finding this to be the best discriminator between infarct and peri-infarct tissue, with an accuracy of 83.1% (figure 6).

The most exciting findings are from two studies by the same group. The lowest values for rCBV and rCBF in tissue that did not progress to infarction were 40% and 29%, respectively, and were found in a subset of patients who received thrombolysis within 6 hours of onset. ³⁶ An earlier study analyzed the CT perfusion studies of 38 patients with ischemic stroke showing rCBF values between 20% and 35%. ²⁰ Sixty-one percent of the areas of brain treated by intra-arterial thrombolysis survived, compared with a 25% survival of areas of brain with the same severity of ischemia treated only with heparin. ²⁰

It is not established whether CBF or CBV will be the most accurate measure of irreversible injury. Each clearly has considerable predictive value. More investigation will be needed to demonstrate conclusively the role each value should play.

Clinical scenarios

A patient suspected of stroke could be scanned with a comprehensive CT protocol, and might be found to have a small area of the left MCA with rCBV and rCBF values below 40% and 30%, respectively. This confirms the diagnosis of stroke and excludes a number of frequently encountered clinical mimics, including migraines, central nervous system trauma, seizures, and hypoglycemia. Less severe derangements of rCBV, rCBF, and mTT of the entire MCA territory would indicate the viable tissue still at risk. If the CTA scan showed an MCA occlusion, the patient could be confidently taken for intra-arterial catheter-directed thrombolysis with a reasonable hope for substantial benefit at an acceptably low risk for a complicating hemorrhage. If, on the other hand, more than half the MCA territory showed rCBF and rCBV values below these cut-offs, this patient could be excluded from thrombolysis despite a normal NCCT.

Given tools such as these for the accurate and rapid evaluation of perfusion, it will be possible to gather data that more clearly define what volumes of infarct or what severity of derangements are actually found in patients who hemorrhage. Thus far, recommendations are based on large studies using the most elementary information imaging has to offer, blind, as it were, to what CT perfusion and CTA can safely and rapidly show.

Pitfalls

One pitfall is technical. Lee et al ⁴⁵ showed that the deconvolution algorithm must have an arterial input with contrast enhancement prior to tissue enhancement. If the selected artery is on the affected side distal to the occlusion, the results will be erroneous if the uninvolved contralateral hemisphere enhances before the affected vessel. ⁴⁵

A second shortcoming may soon be remedied by advances in hardware. Most published perfusion imaging techniques cover 1 or 2 adjacent centimeters of brain, usually through slices that include the territories of all major intracranial vessels. This may lead to false-negative findings due to limited coverage. ⁴¹ One group has used a table-toggling technique to cover a greater area; however, temporal resolution was sacrificed. ⁴⁶ One manufacturer is currently pioneering the develolpment of a large flat panel detector system of 256 contiguous 0.5 mm detectors, covering over 12 cm in the z-axis. This would allow continuous cine acquisition of data through the entire brain.

A more fundamental pitfall regards time. CBF values may suggest tissue viability of regions that have been ischemic for too long for recovery, and it has not yet been shown that this would be reflected by CBV values. Therefore, in cases where onset of ictus is unknown, it may be difficult to differentiate the ischemic penumbra from inevitable infarction by CT perfusion alone, thus exposing a person with a large volume of dead tissue to the risk of hemorrhagic transformation by thrombolytic therapy.

Comparative advantages and disadvantages

Nuclear medicine

PET imaging using ¹⁵ O-labeled water is accepted as the gold standard for measuring perfusion, but availability of this short-lived cyclotron-produced tracer is a major impediment to its widespread use. SPECT imaging may be more feasible than PET; however, spatial resolution is poor in comparison with PET, CT, and MRI.

A significant limitation of nuclear studies is their complete inability to exclude hemorrhage. Time is the most critical issue in stroke, and a protocol that needs two studies before treatment would not be warrented without some considerable and as yet not-shown advantage.

Stable xenon CT

This technique uses X-ray attenuation by inhaled xenon to image hematogenous delivery of this highly lipophilic agent to the brain as a reliable measure of cerebral blood flow. Neither spatial resolution nor exclusion of hemorrhage would be problems with this technique; however, the only parameter produced by this study is a measure of cerebral blood flow. ³³ CBV and mTT measures are not calculable. Furthermore, this would require equipment for delivery of xenon to the patient, while CTA and perfusion studies can be performed with the hardware and software already in place in most imaging departments.

MRI

Although a detailed discussion of MR diffusion, perfusion, and spectroscopy techniques is beyond the scope of this article, a brief overview follows. Conventional MRI is relatively specific and sensitive for subacute stroke, principally by detection of vasogenic edema, but it has not been shown to be useful for the acute presentation. More advanced MRI techniques for diffusion, perfusion, and spectroscopy have been extensively investigated for use in stroke. In 1990, Reith et al ⁴ demonstrated abnormal diffusion within 5 minutes of insult using an animal model. Cytotoxic edema is detected reliably by diffusion imaging techniques that show reduced Brownian motion of water as it moves into the injured or dying neuron. A partial explanation is that cellular structures impede movement relative to the simpler interstitial spaces. Apparent diffusion coefficient (ADC) maps have shown sensitivities of 88% to 100% and specificities of 86% to 100% for ischemia in the clinical setting as early as 5 minutes after an ischemic event. ¹³ Attempts to demarcate the core infarct from the ischemic penumbra by diffusion abnormalities, however, have not yet been successful. Small studies suggest a reduction of ADC of approximately 20% may prove a reliable time-independent indicator. ⁴⁷

Less severe abnormalities of ADC are problematic, and similar to CBF, it is both the degree of abnormality and the duration of ischemia, often unknown, that determine reversibility. MRI perfusion is very similar to CT perfusion, using the same models and assumptions regarding first-pass imaging of intravascular gadolinium contrast. Currently, the concept of the mismatch between severe MR diffusion derangement and MR perfusion abnormality (most commonly the mTT) is being investigated to distinguish between infarct and penumbra.

MR spectroscopy does show promise in delineating viable from nonviable brain by measuring biochemical markers. ⁴ Lactic acid accumulates in the ischemic tissue because of increased reliance on anaerobic glycolysis or decreased clearance from the region. This has not proven reliable in showing the duration of ischemia. More interestingly, N-acetyl aspartate (NAA), one of the most abundant amino acids within the brain, is readily measurable, and initial findings show a progressive decline in NAA levels with the length of insult. ⁴⁸ This offers a possible measure of the surviving population of neurons within the ischemic tissue independent of information regarding the time of onset. This is not yet proven.

Some advantages of CT over MRI are substantial but probably temporary. The length of time needed for MRI studies is constantly being reduced by advances in fast imaging techniques, so this does not appear to be a fundamental obstacle. In addition, FLAIR sequences have been shown to be highly sensitive for hemorrhage, ⁴ but NCCT has the advantage of having a clearly defined role for stroke. It is listed as a requirement for treatment in the package insert for alteplase in order to exclude hemorrhage and is recommended by the American Heart Association Thrombolysis Practice Advisory Guidelines to exclude extensive infarct. ⁴⁹ Further, adding perfusion and angiographic protocols for CT does not significantly lengthen imaging time. The average time for the performance of emergent NCCT, CT perfusion, and CTA for stroke in one series of 73 patients was 12 minutes, including an appropriate delay. ²²

The major drawback of MR in the setting of stroke appears to be an issue of availability. One study showed that urgent MR evaluation of stroke patients in the emergency department occurred more than twice as long after symptom onset than evaluation by CT. ⁵ It is telling that, although MR diffusion and perfusion imaging has been investigated in the use of stroke for several years, to date there has not been one large prospective study initiated using these techniques as part of a standard protocol for emergent stroke. The widespread availability of CT scanning, with 24-hour staffing for emergent patients, in contrast to MRI, gives CT imaging an immense advantage in the evaluation of stroke. Although the demands for training CT technologists to perform high-quality CTA and CT perfusion studies may prove substantial, well-composed imaging recipes have proven manageable in many centers.

Summary

CTA and CT perfusion can provide information in a timely manner. Not every issue regarding the modality is settled. However, it is clear more can be offered to the patient than is now widely available without any change in equipment and staffing. Current treatment regimens may be improved by a better understanding of the individual pathophysiology presenting to the emergency department. A large prospective study using these techniques as standard protocols for the evaluation of stroke needs to be initiated.

Acknowledgments

The author would like to thank Dr. Arthur Fountain and Dr. James Eastwood for the use of the images featured in this article. I would also like to thank Dr. Fountain and Dr. Srinivasan Mukundun for their invaluable support in preparing this manuscript.