1279

Current Concepts of Cerebrovascular Disease and Stroke

Magnetic Resonance and ClinicalCerebrovascular Disease

An Update

Nabih M. Ramadan, MD, Rajeev Deveshwar, MD, and Steven R. Levine, MD

Nuclear magnetic resonance (NMR) has beenrecognized for well over 40 years.Although the first successful in vivo NMR

experiments date back to 1946-1948, it was not untilthe 1970s that the first experiments involving NMRspectroscopy1-3 and imaging4-8 of living systemswere reported in the literature. Since that time theclinical application of NMR has experienced expo-nential growth. Because of this rapid pace of devel-opment and the concurrent increase in hospital-based NMR facilities, it is timely to review theapplication of NMR to the diagnosis and investiga-tion of cerebrovascular disease.

Nuclear Magnetic ResonanceDetailed physics of the NMR experiment are

beyond the scope of this review. However, theparameters used to determine contrast in magneticresonance imaging and the basics of NMR spectros-copy are briefly described.

Magnetic Resonance Imaging (MRI)There are three parameters that determine con-

trast in NMR imaging: nuclear density, p, and twointrinsic NMR parameters called T, and T2. T, andT2 are relaxation time constants that can havecharacteristic values for various tissues. By exploit-ing certain timing aspects of the pulse sequence (acollection of radio frequency [RF] pulses deliveredin a specifically timed sequence) used in obtainingan image, one can enhance the relative contributionof p, Tl5 and T2 contrast. The time delays in the RFpulse sequence, which are adjusted to producevariable contrast, are called TR (time-to-repetition)and TE (time-to-echo). TR is related to the speedwith which the entire pulse sequence is delivered,usually on the order of seconds, and TE is related tothe time between the individual RF pulses withinthe RF pulse sequence, usually on the order of

From the Center for Cerebrovascular Disease Research, Depart-ment of Neurology, Henry Ford Hospital, Detroit, Michigan.

Supported in part by NIH Grant NS23393 and the AmericanHeart Association, Michigan Affiliate.

Reprinted from Current Concepts of Cerebrovascular Diseaseand Stroke 1989;24:13-18.

milliseconds. Tl5 T2, and p image contrast are pro-duced as follows:

T, contrast. The faster the pulse sequence isdelivered (i.e., the shorter the TR), the more T,contrast will be apparent in the image, and con-versely, the slower the pulse sequence is delivered(i.e., the longer the TR), the less Tj contrast. (It isassumed that in T! contrast images TE is kept asshort as possible to avoid any contribution from T2.)

T2 contrast. The longer the TE (i.e., the longerthe time between individual RF pulses in thesequence, not the rate at which the entire pulsesequence is delivered), the more T2 contrast will beapparent in the image. Short TEs result in little T2contrast. (Here it is assumed that in T2 contrastimages TR is kept reasonably long to avoid anycontribution from Tj.)

p contrast. Nuclear density contrast, p, is obtainedby minimizing T, and T2 effects with the use of ashort TE and a long TR.

MR images are not pure p images or pure T, or T2images but rather p or T, or T2 "weighted" images.For practical reasons, however, images which accen-tuate either p, Tl5 or T2 are simply referred to as pimages, T, images, and T2 images, respectively.

From the discussion above, it is obvious that itis possible to obtain a full spectrum of p, T,, and T2contrast in any one image. Indeed, this is the basisfor the extreme versatility in soft tissue contrastseen in MRI. T) contrast is also dependent on thefield strength of the magnet that is used to obtainthe image, and in general, the higher the fieldstrength, the more difficult it can be to generate T,contrast. However, this factor has not been verylimiting in practice.

Magnetic Resonance Spectroscopy (MRS)The RF signals received from the nuclei can be

processed and displayed as a spectrum, which usu-ally comprises several peaks representing the indi-vidual groups of nuclei that make up the variousmolecules contained in the sample. For instance, ifwe collect a spectrum of the phosphorus-containingcompounds in human brain, we can see at leastseven readily identifiable peaks: phosphocreatine,inorganic phosphate, phosphodiesters, phospho-

by guest on May 25, 2018

http://stroke.ahajournals.org/D

ownloaded from

1280 Stroke Vol 20, No 9, September 1989

monoesters, and the a, /?, and y groups of 5' adeno-sinetriphosphate. The most common nuclei studiedin vivo include 31P, 'H, 13C, and 19F. As in other fieldsof spectroscopy, measurements are made of relativemagnitude and position, width, shape, and time-dependent appearance of the peaks. Mobility, struc-ture/conformation, local environment (pH, metal ionbinding), kinetics and relative concentrations, com-partmentalization, and transport, flow, and diffusioncan be inferred from these measurements.

MRI in StrokeIschemic Cerebrovascular Disease

Experimental studies have demonstrated that isch-emic changes due to stroke can be detected on MRIwithin the first 4 hours of the ictus and as early as 1hour.9-11 In the early stages of ischemia, theincreased accumulation of tissue water results inprolongation of both "P, and T2. In early ischemiathese effects are best appreciated with T2 images, inwhich the ischemic region appears as high signalintensity. Later in the ischemic process the infarctis better appreciated with low (relative) signal inten-sity produced from T, images (T\ prolongationeffect). With time an infarct will appear as a lowsignal (dark) region on 1^ images and as a highsignal (bright) on T2 images.10 The appearance ofinfarcted tissue in the subacute stage can be appre-ciated on both MRI and CT. One advantage of MRI,however, is its superior image resolution, whichallows for the identification of relatively smallerregions of infarct (including "lacunar" type stroke)missed by CT.11-14 Additionally, in the subacute tochronic stages (beyond the edema stage and beforecavitation starts to appear), CT images of the strokemight appear isodense with surrounding tissue(pseudonormalization effect), whereas with MRI arim of hyperintensity on T2 images can frequentlybe seen surrounding a darker zone (the actualinfarct).11 This may reflect either persistent isch-emia (penumbra) or Wallerian degeneration15 andmay have future therapeutic implications. In thechronic stage the cavity formed will have an appear-ance similar to that of cerebrospinal fluid (CSF) in aclosed space, that is, hypointense (dark) on Tiimages and hyperintense (bright) on T2 images.Another advantage of MRI over CT is better resolu-tion in the posterior fossa so that brainstem infarctsmissed on CT can be readily detected by MRI.

CT abnormalities in patients with transient isch-emic attacks (TIA) have been found with variablefrequency ranging from 0-20%.16-18 MRI is abnor-mal in up to 80% of patients with TIA.13-14 This widevariation reflects the lack of correlation between the"ischemic" lesion on CT or MRI and the clinicalpicture, as well as the broad definition of TIA used(deficit lasting less than 24 hours). SubclassifyingTIAs as completed infarcts with transient symp-toms (CITS)16'17 versus true transient ischemic epi-sodes, in which the deficit generally lasts less than

1-3 hours,19 might result in less variability in theincidence of abnormalities seen on MRI. MRI maythus aid in providing a more exact diagnosis forfuture natural history and treatment studies of lacu-nar infarction or small vessel disease.

There is a surprising lack of MRI-neuropatho-logical data on stroke.15 Dewitt and colleagues havecorrelated both ischemic and hemorrhagic cere-brovascular disease with gross and histopathologi-cal examination. More studies of this type on high-field (1.5 T) magnets are needed.

Intracerebral HemorrhageA detailed description of the evolution of a blood

clot and its MR image characteristics has beencovered in several recent publications.20-23 A muchsimplified schema will be presented here. Acutely,red blood cells (RBC) are hemoconcentrated andform deoxyhemoglobin which results in hypoin-tense regions on T2 images. In the subacute stage,as methemoglobin starts forming around the fourthto seventh day and proceeding centripetally, hyper-intense signals initially appear on T, images andlater on T2 images (Figure 1). This observation isbelieved to be related to the lag of RBC lysis behindmethemoglobin formation. Therefore, the brightsignal initially seen on T, images because of themethemoglobin T, shortening effect will not becomplete in the center of the hemorrhage until thedeoxyhemoglobin is completely replaced with lysisof centrally located RBC. At the same time, twoperipheral zones are delineated on T2 images: amore medial hemosiderin hypointense rim (darkzone) and a more peripheral hyperintense signal(bright zone), reflecting edema formation.

In the chronic state a well-demarcated brightsignal is seen from the whole clot on T, image and T2image as methemoglobin fills both the periphery andcenter of the hematoma. The hypointense rim ofhemosiderin is well formed at this stage, and vari-able edema (bright signal) is well demarcated on T2images. The hyperintensity of the hematoma maypersist for as long as a year or more. Subsequently,the hematoma is replaced by CSF, and T! prolon-gation effects appear. Therefore, the image of an oldclot of more than 1 year of age would appearhypointense on T! images and hyperintense on T2images with a hypointense rim (hemosiderin) sur-rounding the cavity. The rim of hypointensity ofhemosiderin is useful in differentiating ischemicinfarct from intracerebral hemorrhage of more than1 year of age since in the former no hemosiderinsignal is appreciated. An exception would be hem-orrhagic infarcts which after 1 year can have thesame MR signal characteristics as intracerebralhemorrhage. At this stage MRI is more useful thanCT for detecting small brainstem hemorrhage whichcan be missed on CT because of bone artifact. Arecently developed technique using gradient echopulse sequences (beyond the scope of this review)

by guest on May 25, 2018

http://stroke.ahajournals.org/D

ownloaded from

Ramadan et al NMR and Clinical Cerebrovascular Disease 1281

FIGURE 1. Acute to subacute intraparenchymal hematoma in the right occipital lobe. (Left) T,-weighted axial MRI(TR=600 msec, TE=30 msec) demonstrating the isointense lesion of hemorrhage (deoxyhemoglobin) (arrow) surroundedby the hyperintense rim of methemoglobin (arrow head). Blood forming a CSF-blood layer is seen as a hypointense signal(deoxyhemoglobin effect) in the left lateral ventricle (open arrow). (Right) T2-weighted axial MRI (Tr=2800 msec, TE=90msec) demonstrating a central region of hypointensity (deoxyhemoglobin) (short arrow) surrounded by the hyperintenseperipheral rim of methemoglobin (arrow head) and the most peripheral, less hyperintense zone of edema (long arrow).The hypointense signal from the contralateral ventricle represents deoxyhemoglobin (open arrow) within the CSF.(Reprinted through courtesy of Suresh Patel, MD, Division of Neuroradiology, Department of Radiology, Henry FordHospital, Detroit, MI)

has been shown to improve the specificity in thediagnosis of intracerebral hemorrhage.24

Subarachnoid HemorrhageContrary to initial in vitro studies, which indi-

cated that acute subarachnoid hemorrhage might bemissed on MRI,25 acute hemorrhage can be detectedwith pulse sequences that are intermediate in con-trast characteristics (i.e., mixed TrT2 images) sothat CSF is isointense with brain parenchyma.26

With such pulse sequences (TE=80 msec, TR=2200msec), using a 0.15-T magnet (low-field), the blood-stained CSF causes a marked shortening of T[ and aslight shortening of T2, resulting in a hyperintensesignal that contrasts with the isointense signal ofnormal CSF. In the subacute stage, methemoglobinformation (as in intracerebral hemorrhage) will causemarked shortening of T, and prolongation of T2,giving the bright signal of blood-stained CSF onboth Tl and T2 images. In addition, MRI couldprove superior to CT in demonstrating the cause ofhemorrhage. In a recent study no aneurysm wasdemonstrated by CT in 25 of 30 cases, whereas in 14of these cases, MRI detected aneurysms in variouslocations.26 MRI also appears superior to CT indemonstrating small posterior inferior cerebellarartery aneurysms that are often missed on CT whenonly trace hemorrhage into the CSF occurs.

Subdural and Epidural HematomaThe evolution of the blood clot in both subdural

and epidural hematoma is similar to that of intra-

cerebral hemorrhage as described previously (Table1). Epidural hematoma can be differentiated fromsubdural hematoma by the MR appearance of thefibrous dura mater.27 With intermediate magneticfield strength the fibrous dura mater, which has aneven lower signal characteristic than the hematomaon T2 images (TE=60 msec, TR=2000 msec), willappear darker than the underlying or overlyinglesion (subdural versus epidural, respectively). If ahigher field strength is used, the appearance of bothacute subdural and epidural hematoma is the samebecause dural demarcation is lost. In the subacutestage the high signal of the hematoma is clearlydemarcated from the low signal of the dura regard-less of field strength.

Cerebral Venous ThrombosisCT reveals direct signs (delta sign, cord sign) of

cerebral venous thrombosis in only 33% of cases.28

MRI may be more sensitive, possibly replacingangiography in confirming the clinical suspicion ofvenous sinus thrombosis. Sagittal MR images willshow not only the actual thrombosed superior sag-ittal sinus29 but also the extent of the thrombus andthe degree of recanalization, which may have futuretherapeutic implications. Acutely, the expected"flow void" signal of the sinus (hypointensity) onT, images is lost because it appears bright; T2images will reveal a low signal relating to thedeoxyhemoglobin effect. Subacutely, the intralumi-nal clot evolves in a pattern similar to that of

by guest on May 25, 2018

http://stroke.ahajournals.org/D

ownloaded from

1282 Stroke Vol 20, No 9, September 1989

TABLE 1. Optimum MRI Contrast in Cerebrovascular Disease

Ischemic strokeICH/SDH/EDHSAHCVT (sinus appearance with

associated clot)

Acute(0-3 days)

T2

T,Mixed T, and T2

T, and T2t

Subacute(4-14 days)

T] and T2

T, and T2

T, and T2

T, and T2

Chronic(2 weeks-1 year)

T, and T2

T, and T2*T, and T2*T] and T2t

ICH, intracerebral hemorrhage; SDH, subdural hematoma; EDH, epidural hematoma; SAH, subarachnoidhemorrhage; CVT, cerebral venous thrombosis.

'Progressive loss of signal after 1 year on T] image.tDisappearance of the flow void signal on T] image.^Reappearance of flow void signal on T] and T2 image.

intracerebral hemorrhage. In the chronic phaserecanalized vessels will result in reappearance ofthe dark flow void signal on T, images. Indirectsigns of thrombosis (intracerebral hemorrhage anddiffuse edema with slit ventricles) do not differ incharacteristics whether they are associated withcerebral venous thrombosis or other conditions.

Vascular Malformations and DevelopmentalAnomalies of the Cerebral Vasculature

Vascular malformations, which may present asintracerebral or subarachnoid hemorrhage, are cat-egorized as follows30:

1. Occult vascular malformations, includingthrombosed arteriovenous malformations, cavern-ous angiomas, capillary telangiectasia, and venousangioma.31 MR is more sensitive than CT and is themodality of choice for detecting and characterizingthese malformations. They appear as mixed signalintensity on both T, and T2 images. Unique areas ofsignal void are present that represent calcification,hemosiderin, or both.

2. Arteriovenous malformations. The entangled,dilated vessels appear as serpiginous and punctateregions of signal void because of rapid flow andturbulence. Previous hemorrhage, cysts, and calci-fication may be evident.

3. Aneurysms are diagnosed by the location,morphology, and mixed signal intensities in a con-centric lamellar arrangement. Flow void (high veloc-ity signal loss) within the lumen, or increased signalintensity on T, and T2 images associated with throm-bus (methemoglobin), or both may be detectedwithin all or part of the aneurysm. In giant aneu-rysms (>2.5 cm diameter), complex signal intensi-ties result from a combination of flowing blood,turbulence, organizing thrombus, calcification, andhemosiderin deposition.

MR SpectroscopyStudies of clinical stroke with MRS are

limited.32-34 Metabolic abnormalities detected byMRS (reduced phosphocreatine and ATP, increasedinorganic phosphate, elevated lactate, and acidosis)in ischemic brain may precede structural alterationsdetected by CT or MRI.35 The metabolic data

obtained by MR spectroscopy may eventually leadto biochemical markers of clinical stroke prognosisand response to therapy since serial studies can beeasily performed.32'34'36-37 Other applications of NMRtechnology that are being developed include theability to measure cerebral blood flow,38 oxygenutilization,39 and glucose metabolism.40 This infor-mation will undoubtedly be important to clinicianswho manage patients with cerebrovascular diseaseand will emphasize the need to treat stroke patientsas soon as possible to minimize the extent ofischemic damage seen on MRI.

SummaryMRI is becoming the imaging modality of choice in

patients with ischemic cerebrovascular diseasealthough CT is still the test of choice to exclude acutehemorrhagic stroke. We have briefly reviewed char-acteristic features of ischemic and hemorrhagic cere-brovascular disease as well as vascular anomalies asseen on MRI. In time MRS should provide usefulnoninvasive metabolic data to complement the ana-tomical data in patients with cerebrovascular disease.

References1. Moon RB, Richards RE: Determination of intracellular pH

by 31P magnetic resonance. JBiol Chem 1973;248:7276-72782. Hoult DI, Busby SJW, Gadian DG, Radda GK, Richards

RE, Seeley PJ: Observation of tissue metabolites using 31Pnuclear magnetic resonance. Nature 1974;252:285-287

3. Ackerman JJH, Grove TH, Wong GG, Gadian DG, RaddaGK: Mapping of metabolites in whole animals by 31-P NMRusing surface coils. Nature 1980;283:167-170

4. Mansfield P, Maudsley AA: Planar and line-scan spin imag-ing by NMR. Proc XlXth Congress Ampere, Heidelberg,1976, pp 247-252

5. Andrew ER, Bottomley PA, Hinshaw WS, Holland GN,Moore WS, Simaroj C: NMR images by the multiple sensi-tive point method: Application to larger biological systems.Phys Med Biol 1977;22:971-974

6. Damadian R, Goldsmith M, Minkoff L: NMR in cancer:Fonar image of the live human body. Physiol Chem PhysMed NMR 1977, pp 97-108

7. Clow H, Young IR: Britain's brains produce first NMRscans. New Scientist 1978;80:588

8. Moore WS, Holland GN, Kreel L: The NMR CAT scanner—A new look at the brain. CT 1980;4:l-7

9. Brant-Zawadzki M, Pereira B, Weinstein P, Moore S, Kucha-rczyk W, Berry I, McNamara M, Derugin N: MRI of acuteexperimental ischemia in rats. AJNR 1986;7:7-11

by guest on May 25, 2018

http://stroke.ahajournals.org/D

ownloaded from

Ramadan et al NMR and Clinical Cerebrovascular Disease 1283

10. Unger EC, Gado MH, Fulling KF, Littlefield JL: Acutecerebral infarction in monkeys: AN experimental studyusing MR imaging. Radiology 1987;162:789-795

11. Brant-Zawadzki M, Solomon M, Newton TH, Weinstein P,Schmidley J, Norma D: Basic principles of magnetic reso-nance imaging in cerebral ischemia and initial clinical expe-rience. Neuroradiology 1985;27:517-520

12. Dewitt LD: Clinical use of nuclear magnetic resonanceimaging in stroke. Curr Cone Cerebrovas Dis/Stroke 1985;20:25-29

13. Kinkel PR, Kinkel WR, Jacobs L: Nuclear magnetic reso-nance imaging in patients with stroke. Semin Neurol 1986;6:43-52

14. Salgado ED, Weinstein M, Furlan AJ, Modic MJ, Beck GJ,Estes M, Awad I, Little JR: Proton magnetic resonanceimaging in ischemic cerebrovascular disease. Ann Neurol1986;20:502-507

15. Dewitt L, Kistler JP, Miller DC, Richardson EP Jr, Buo-nanno FS: NMR-neuropathologic correlation in stroke. Stroke1987;18:342-351

16. Waxman SO, Toole JF: Temporal profile resembling TIA inthe setting of cerebral infarction. Stroke 1983;14:433-437

17. Bogousslavsky J, Regli F: Cerebral infarction with transientsigns (CITS): Do TIAs correspond to small deep infarcts ininternal carotid artery occlusion? Stroke 1984;15:536-539

18. Calandre L, Gommara S, Bermejo F, Millan JM, DelPozo G:Clinical-CT correlations in TIA, RIND, and strokes withminimum residuum. Stroke 1984;15:663-666

19. Levy DE: How transient are transient ischemic attacks?Neurology 1988;39:674-677

20. Gomori JM, Grossman RI, Goldberg I, Zimmerman RA,Bilaniuk LT: Intracranial hematomas: Imaging by high-fieldMR. Radiology 1985;157:87-93

21. Dooms CG, Brant-Zawadzki M, Lemme-Plaghosl: MRI ofhemorrhagic brain lesions (abstract). AJNR 1985;6:464

22. Norman D: Vascular disease: Hemorrhage, in Brandt-Zawadski M (ed): Magnetic Resonance Imaging of theCentral Nervous System. New York, Raven Press, Publish-ers, 1987, chap 13, pp 209-212

23. Zimmerman RD, Heier LA, Snow RB, Liu DP, Kelly AB,Deck MDF: Acute intracranial hemorrhage: Intensity changeson sequential MR scans at 0.5T. AJNR 1988;9:47-57

24. Atlas SW, Mark AS, Grossman RI, Gomori JM: Intracranialhemorrhage: Gradient-echo MR imaging at 1.5 T. Radiology1988;168:803-807

25. Chackeres DW, Bryan RN: Acute subarachnoid hemor-rhage: In vitro comparison of magnetic resonance and com-puted tomography. AJNR 1986;7:223-228

26. Jenkins A, Hadley DM, Teasdall GM, Condon B, Macpher-son P, Patterson J: MR imaging of acute subarachnoidhemorrhage. J Neurosurg 1988;68:731-736

27. Bradley WG Jr: Pathophysiologic correlates of signal alter-ations, in Brandt-Zawadski M (ed): Magnetic ResonanceImaging of the Central Nervous System. New York, RavenPress, Publishers, 1987, chap 3, pp 40-41

28. Virapongse C, Cazenave C, Quisling R, Sarwar M, HunterS: The empty delta sign: Frequency and significance in the76 cases of dural sinus thrombosis. Radiology 1987;162:779-785

29. Macchi PJ, Grossman JM, Gomon JM, Goldberg HI, Zim-merman RA, Bilaniuk LT: High field MR imaging of cerebralvenous thrombosis. J Comput Assist Tomogr 1986;10:10-15

30. Rao KC, Lee SH: Cerebrovascular anomalies, in Stark D,Bradley WG (eds): Magnetic Resonance Imaging. St. Louis,CV Mosby Co, 1988, chap 23, pp 473-505

31. Rigamonti D, Hadley MN, Drayer BP, Johnson PC, Hoenig-Rizamonti R, Knight TT, Spetzler RF: Familial occurrenceof cerebral cavernous malformation. N Engl J Med 1988;319:343-347

32. Welch KMA, Gross B, Licht J, Levine SR, Glasberg M,Smith MB, Helpern JA, Bueri J, Gorell JM: Magneticresonance spectroscopy of neurological diseases, in AppelSH (ed): Current Neurology, Vol 8. Chicago, Year BookMedical Publishers, Inc, 1988, chap 9, pp 295-332

33. Bottomley BA, Drayer BP, Smith LS: Chronic adult cere-bral infarction studied by phosphorus NMR spectroscopy.Radiology 1986; 160:763-766

34. Levine SR, Welch KMA, Helpern JA, Chopp M, Brucej R,Selwa J, Smith MB: Prolonged deterioration of ischemicbrain energy metabolism and acidosis associated with hyper-glycemia: Human cerebral infarction studied by serial 31PNMR spectroscopy. Ann Neurol 1987;23:416-418

35. Brant-Zawadzki M, Weinstein P, Bartkowski H, MoseleyM: MR imaging and spectroscopy in clinical and experimen-tal cerebral ischemia: A review. AJR 1987;148:579-588

36. Radda GK: The use of NMR spectroscopy for the under-standing of disease. Science 1986;233:640-645

37. Petroff OAC, Pritchard JW, Ogino T, Shulman RG: Protonmagnetic resonance spectroscopic studies of agonal carbo-hydrate metabolism in rabbit brain. Neurology 1988;38:1569-1574

38. Ewing JR, Branch CA, Helpern JA, Smith MB, Butt SM,Welch KMA: Cerebral blood flow measured by NMR indi-cator dilution in cats. Stroke 1989;20:259-267

39. Eidelberg D, Johnson G, Barnes D, Tofts PS, Deply D,Plummer D, McDonald WI: I9F NMR imaging of bloodoxygenation in the brain. Magn Reson Med 1988;6:344-352

40. Nakada T, Kwee IL, Card PJ, Matwiyoff NA, Griffey BV,Griffey RH: Fluorine-19 NMR imaging of glucose metabo-lism. Magn Reson Med 1988;6:307-313

KEY WORDS • cerebrovascular disordersimaging • nuclear magnetic resonance

magnetic resonance

by guest on May 25, 2018

http://stroke.ahajournals.org/D

ownloaded from

N M Ramadan, R Deveshwar and S R LevineMagnetic resonance and clinical cerebrovascular disease. An update.

Print ISSN: 0039-2499. Online ISSN: 1524-4628 Copyright © 1989 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Stroke doi: 10.1161/01.STR.20.9.1279

1989;20:1279-1283Stroke. 

http://stroke.ahajournals.org/content/20/9/1279World Wide Web at:

The online version of this article, along with updated information and services, is located on the

  http://stroke.ahajournals.org//subscriptions/

is online at: Stroke Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer available in the

Permissions in the middle column of the Web page under Services. Further information about this process isOnce the online version of the published article for which permission is being requested is located, click Request

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Stroke Requests for permissions to reproduce figures, tables, or portions of articles originally published inPermissions:

by guest on May 25, 2018

http://stroke.ahajournals.org/D

ownloaded from