Perfusion magnetic resonance imaging of acute ischemic stroke

Perfusion Magnetic Resonance Imaging of Acute Ischemic Stroke

Pamela W. Schaefer, Javier M. Romero, P. Ellen Grant, Ona Wu, A. Gregory Sorensen, Walter Koroshetz, and R. Gilberto Gonz~lez

T HE FUNDAMENTAL physiological abnormality in ischemic stroke is a failure of

cerebral perfusion. Information on cerebral perfusion provided by magnetic resonance imaging (MRI) is thus of great potential value. By definition, perfusion means capillary-level blood flow or tissue blood flow. Two broad categories of techniques exist for assessing cerebral perfusion in acute ischemia with MRI. The first of these is dynamic susceptibility contrast imaging. The second and currently infrequently used technique is spin tagging (also called arterial spin labeling). Although spin tagging has the advantage of not needing any contrast injection, it currently has a lower signal to noise ratio than the contrast-based techniques. Furthermore, approaches that involve multislice imaging with spin tagging still need refinement.

DYNAMIC SUSCEPTIBILITY CONTRAST IMAGING

Perfusion MRI aims to investigate flow at the capillary level. Capillary flow measurements with dynamic susceptibility contrast imaging are based on the interaction between the contrast agent and the spins surrounding a capillary. In this approach, a paramagnetic substance, typically a gadolinium- based contrast agent, is injected. As the gadolinium travels through the blood vessels, there are signal changes caused by T1 and T2 effects of the contrast agent. If contrast injection is performed as a bolus with a relatively high rate of injection (5 mL/s), then the contrast agent is still relatively concentrated, and the T2 effects outweigh the T1 effects. The T2 effects of paramagnetic contrast agents include both T2 and T2* changes. As the paramagnetic substance passes through the capil-

From the Neuroradiology Division and Stroke Service, Mas- sachusetts General Hospital and Harvard Medical School, Boston, MA.

Address reprint requests to R. Gilberto Gonzdlez, MD, PhD, Neuroradiolegy, GRB 285, Fruit Street, Massachusetts General Hospital, Boston, MA 02114-2696.

Copyright 2002, Elsevier Science (USA). All rights reserved. 0037-198X/02/3703-0009535.00/0 doi:lO.lO53/sroe.2002.34569

lary bed, it modifies the local magnetic environ- ment causing a signal loss because of susceptibility or T2* effects. Imaging during the rapid first pass of the contrast agent through the brain can detect enough signal change to estimate the amount of contrast agent present. This concentration of contrast agent can then be related to the amount of blood volume and blood flow. By measuring the concentration-versus-time curve, when signal changes are summed over time or integrated, the net signal change is proportional to the blood volume.

Although viewing the raw imaging data can be helpful, the large number of images at each slice (typically 40-50) make review of individual images somewhat impractical. Instead, after the ac- quisition of dynamic susceptibility contrast imaging data, postprocessing approaches are typically used. Our technique computes the log of the signal change in each time point (relative to a baseline image set), which effectively converts the map of signal change versus time to a map of R2* versus time (R2* = l/T2*). Integrating the R2*-versus- time curve produces a map of recombinant cerebral blood volume (rCBV) in each voxel.

Efforts have also focused on creating maps of relative cerebral blood flow (rCBF). Signal-to- noise limitations often preclude reliable absolute measurements. Nevertheless, it is possible to gen- erate reliable rCBF maps. Generation of CBF maps requires deconvolution of the tissue signal-versus- time curve from the arterial input function, typically the ipsilateral middle cerebral artery (MCA) stem. With this technique, one can compute a residue function, which is a mathematical term that describes how much of the contrast agent remains in a given voxel at a given time after the injection. This function allows the estimation of the initial blood flow in a particular voxel and the rCBF. Maps of mean transit time can be calculated from the relationship of the central volume theorem MTT = CBV/CBF. The types of analysis that may performed on the dynamic contrast MRI data sets are shown in Figure 1.

TECHNICAL CONSIDERATIONS

MRI perfusion maps provide qualitative data regarding the cerebral circulation. Therefore, the

230 Seminars in Roentgenology, Vol 37, No 3 (July), 2002: pp 230-236

PERFUSION MRI 231

, TTP ' ~ - - - ~

\ \ 1 st moment

time

Fig 1. Concentration-time curve. Typical shape of a concentration-time curve is shown with the various parameters used for transit time estimation. Integrating the R2-versus- time curve produces a map of rCBV. Generation of CBF maps requires deconvolution of the tissue signal-versus-time curve from the arterial input function. Maps of mean transit time can be calculated from the relationship of the central volume theorem MTT = CBV/CBF. AT, arrival time; 1st moment, first moment (center of gravity); rTI'P, relative TTP. (Reprinted with permission, z )

term relative "r" is used to express the qualitative nature of these maps. Precise quantitation of both CBF and CBV values can sometimes be obtained with the deconvolution method, although this is not routinely used clinically, t Cardiac output variations, vascular occlusions, or variations in collateral circulation can make the maps difficult to interpret. A major arterial occlusion can cause broadening of AR2* because of collateral circulation and lead to underestimation of CBV and CBF.

The effect of the calculation methods and underlying vasculopathy in magnetic resonance perfusion-weighted imaging of acute cerebral infarction were evaluated by Yamada et al. 2 From a pool of 113 patients who had undergone PWI during the study period, a total of 12 patients with nonlacunar ischemic strokes who were scanned within 24 hours after onset of symptom were selected for the study. The patient population consisted of 6 patients who had extracranial internal carotid artery stenosis (>70%) and 6 individuals without stenosis. Seven different postprocessing methods were evaluated: first moment, ratio of area to peak, time to peak (TTP), relative TTP, arrival time, full width at half maximum, and deconvolution. Fol- low-up MRI or computed tomography images, which served as the gold standard, were used to determine the areas that evolved into infarcts.

Sensitivity and specificity of each transit time technique were calculated. Calculation methods

with high sensitivity were the first moment (sensitivity, 74%), TTP (sensitivity, 77%), and deconvolution (sensitivity, 81%-94%). Between the 2 groups with and without internal carotid artery stenosis, the specificity of most of the techniques was lower in the internal carotid artery stenosis group. The first moment and deconvolution methods maintained relatively high specificity even in the stenosis group. The effect of different process- ing methods is shown in Figures 2 and 3.

NONCONTRAST TECHNIQUES

Noncontrast perfusion imaging techniques are very similar to time-applied magnetic resonance angiography. Both approaches attempt to saturate stationary tissue and highlight flowing spins. Not only is contrast administration unnecessary in this approach, but, in theory, absolute quantification of CBF is possible. Unfortunately, as with the dynamic susceptibility contrast imaging techniques, the assumptions made in these equations often prove to be erroneous in very low blood flow states. Obtaining the most accurate information in

Fig 2. Comparison of PWI map types. An 80-year-old woman presented with acute onset of aphasia and confusion. (A) Initial DWl showed an area of abnormality involving the left basal ganglia, (B) which was later shown to be an infarct also involving the anterior temporal cortex on follow-up computed tomography scan. (C) PWl at the time of onset using the first moment, (D) area to peak, (E) time to peak, and (F) relative time to peak images showed an area of abnormal perfusion in the left basal ganglia, but the cortical involvement was missed with these techniques. [G] arrival time and [H] full width at half maximum). (I-L) The deconvolution method was the only calculation technique that showed an unequivocal area of cortical involvement in the anterior temporal MCA territory, (Reprinted with permission, z)

232 SCHAEFER ET AL

Fig 3. Comparison of PWl map types. An 80-year-old woman was admitted to the hospital for balloon occlusion of a right cavernous carotid aneurysm, The patient remained asymptomatic until she developed a left facial droop during the evening after the procedure. (A) Her initial DWI was negative for an infarct. (B) On the follow-up DWI in 24 hours, the lesion in the right parietooccipital region at the MCA/PCA watershed zone became apparent. (I-L) PWls using deconvolution methods revealed abnormality involving the right pos- terior MCA/posterier cerebral artery watershed, (C-H) whereas the other methods failed to show an unequivocal area of abnormality. Note that the extent of abnormality showed on the deconvolution methods differs, depending on the selected arterial input functions. (J) The image calculated by the AIF from the contralateral IVICA overestimates the area of infarction, (L) whereas the images with the AIF from the peri-infarction area showed the closest agreement wi th the fol low-up DWI. (Reprinted with permissionY)

Diffusion in Combination with Perfusion MRI in the Evaluation of Acute Stroke

The role of perfusion MRI in conjunction with DWI is not completely understood. The most important clinical impact may result from defining the ischemic penumbra, a region that is ischemic but still viable and may infarct if not treated. Consequently, much investigation is focused on strokes with a diffusion-perfusion mismatch (ie, strokes with a perfusion lesion larger than the diffusion lesion) (Fig 4). Proximal occlusions are much more likely to result in a diffusion-perfusion mismatch than distal or lacunar infarctions. Oper- ationally, the diffusion abnormality is thought to represent the ischemic core and the region characterized by normal diffusion, but abnormal perfusion is thought to represent the ischemic penumbra. Definition of the penumbra is complicated because of the multiple hemodynamic parameters that may be calculated from the perfusion MRI data.

A number of articles have focused on volumetric data. The majority of strokes increase in volume on DWI, with the peak volumetric measurements

low blood flow states associated with acute stroke is exactly the situation in which error is least tolerable. In the future, it is likely that spin- labeling techniques will become increasingly important.

Perfusion MRI Reliability

In general, perfusion images are less sensitive than diffusion-weighted imaging (DWI) in the detection of acute stroke with sensitivities for CBV, CBF, and MTT, ranging from 74% to 84%. Lesions missed on perfusion images include (1) lesions with small abnormalities on DWI, which are not detected because of the lower resolution of magnetic resonance perfusion images, and (2) lesions with early reperfusion. Specificities for perfusion images range from 96% to 100%. Occa- sional false-positive lesions occur when there is an isehemic, viable hypoperfused region, which re- c o v e r s . 3

Fig 4. Diffusion/perfusion mismatch in a left MCA stroke. The axial DWl images (b value of 1000 s/mm2; effective gradient of 14mT/m; TR/TE 6000/108; matrix 256 x 128; field of view 400 × 200 ram, slice thickness 6mm with 1-mm gap) show a left insular infarction. The axial cerebral blood volume (CBV) images (spin echo echo planar technique; 0.2 mmol/kg gadopentetate dimeglumine (Nlagnevist, Berlex Laboratories, Wayne, N J); 51 images per slice; TR/TE 1500/75; matrix 256 x 128; FOV 400 x 200 mm; slice thickness 6mm with 1 mm gap), demonstrate a defect similar in size to the DWl abnormality. The CBF and MTT images show much larger abnormal regions in the left MCA territory. Follow-up T2-weighted images show that the infarct is similar in size to the initial DWl and CBV abnormalities but is much smaller than the initial CBF and MTT abnormalities.

PERFUSION MRI 233

Fig 5. Lacunar infarct with diffusion-perfusion match. A 63-year-old patient with sudden episode of right-sided weak- ness. DWl shows a focus of restricted diffusion in the tail of the left caudate nucleus and adjacent corona radiata, most likely representing a lacunar infarct. CBV, CBF, and MTT show the same area of change in signal intensity. The foUow-up T2-weighted image shows no changes in the volume in the final ischemic tissue.

achieved at 2 to 3 days after ictus. 4-1o The initial DWI lesion volume correlates highly with final infarct volume, with lesions, on average, growing approximately 20%. 4,11 The initial CBV lesion volume is usually similar to the DWI lesion volume, and CBV also correlates highly with final infarct volume, with abnormalities, on average, growing approximately 20%. When there is a rare DWI-CBV mismatch, the DWI lesion volume still correlates highly with final infarct volume, but the predicted lesion growth is approximately 60%. The CBV also correlates highly with final infarct volume, with no predicted lesion growth. In other words, when there is a DWI-CBV mismatch, the DWI abnormality grows into the size of the CBV abnormality,

CBF and MTT correlate poorly with final infarct volume and on average overestimate final infarct volume. Many more strokes are characterized by a DWI-CBF or a DWI-MTT mismatch compared with a DWI-CBV mismatch. In our experience, a DWI-CBF or DWI-MTT mismatch shows regions with altered hemodynamics but is not predictive of increased lesion growth. However, others have reported that DWI-CBF and DWI-MTT mis- matches do predict increased lesion growth and the size of the mismatch is predictive of amount of lesion growth. I2-14

In small vessel infarctions (perforator infarctions and distal MCA infarctions), the initial per-

fusion and diffusion lesion volumes are usually similar (diffusion-perfusion match, Fig 5), and the diffusion lesion volume increases only slightly with time. A diffusion lesion larger than the perfusion lesion or a diffusion lesion without a perfusion abnormality usually occurs with early reperfusion. In this situation, the diffusion lesion usually does not change significantly over time.

In animals treated with neuroprotective agents after MCA occlusion, there is stabilization of the size of the early diffusion MRI lesion. In other words, the expected growth of the early lesion is not present. 15,16 The benefit of this therapy has not been convincingly shown in human trials.

More recently, much research has focused on defining diffusion and perfusion MRI parameter lesion ratios in infarct core, penumbra, which progresses to infarction and hypoperfused regions that remain viable. Most articles have shown that rCBF is the most useful parameter in distinguish- ing hypoperfused tissue that will progress to infarction from hypoperfused tissue, which will remain viable in patients not treated with thrombolysis. Reported rCBF ratios for core range from 0.12 to 0.44, for penumbra that progresses to infarction from 0.35 to 0.56, and for hypoperfused tissue that remains viable from 0.58 to 0.78. 3,7-2o

The variability in CBF ratios likely results from a number of different factors. Most importantly, the data obtained represent only a single time point in a dynamic process. One major factor is variability in timing of tissue reperfusion. Jones et a121 showed in monkeys that both severity and duration of CBF reduction up to 4 hours define an infarction threshold. The threshold for tissue infarction with reperfusion at 2 to 3 hours was 10 to12 mL/100 g/rain, whereas the threshold for tissue infarction with permanent occlusion was 17 to 18 mL/100 g/rain. Ueda et al, 22 in a study of patients treated with thrombolysis, showed that duration of ischemia affected the threshold for tissue viability for up to 5 hours. 22 Another factor is that normal average cerebral blood flow in human parenchyma varies greatly from 21.1 to 65.3 mL/100 g/min, depending on age and location in gray matter versus white mat ter9 27 Other factors include variability in methodologies, variability in initial and follow-up imaging times, and variability in postischemic tissue responses.

Low rCBV ratios are predictive of infarction. However, rCBV ratios for penumbral regions are

234 SCHAEFER ET AL

not significantly different. They range from 0.69 to 1.44 for penumbra that progresses to infarction and 0.94 to 1.29 for hypoperfused regions that remain viable. 3A7-2°,28 The finding of elevated rCBV in the ischemic penumbra is in accordance with positron- emission tomography studies that have shown that in the early stages of ischemia decreased cerebral perfusion pressure produces vasodilatation and an increase in the cerebral blood volume, which maintains constant cerebral blood flow and oxygen extraction fraction. 29 With further decreases in cerebral perfi~sion pressure, tile compensatory vasodilatation reaches a maximum, cerebral blood flow begins to fall, and the cerebral blood volume initially continues to rise and then falls as capillary beds collapse and the oxygen extraction fraction reaches its maximum. However, elevated rCBV is not predictive of tissue viability.

Some studies have shown no statistically significant differences in the MTT values between infarct core or the 2 penumbral regions, whereas others have shown differences between all 3 regions or between hypoperfused tissue that remains viable compared with infarct core and penumbra that progresses to infarction? ,17-2°,28

In our experience, apparent diffusion coefficient (ADC) and DWI ratios are significantly different between the core and the 2 penumbral regions, but the ratios for the 2 penumbral regions show more overlap compared with the ratios reported for CBF. Others report no statistically significant difference between the 2 penumbral regions. Reported ADC ratios for infarct core, penumbra that progresses to infarction, and hypoperfused tissue that remains viable are 0.56 to 0.63, 0.89 to 0.91, and 0.93 to 0.96, respectively. 17,1s Reported DWI ratios for infarct core, penumbra that progresses to infarction, and hypoperfused tissue that remains viable are 1.84 to 1.9, 1.13, and 1.08, respectively. 3,3°

The aforementioned approaches have focused on regions or volumes of tissue. Because there is heterogeneity in diffusion and perfusion parameters within ischemic tissue, Wu et aP 4 performed a voxel by voxel analysis of abnormalities on 6 maps (T2, ADC, DWI, rCBV, rCBF, and MTT) compared with follow-up T2-weighted images and developed thresholding and generalized linear model algorithms to predict tissue outcome (Fig 6). 14 They found that, at their optimal operating points, thresholding algorithms combining DWI and PWI provided 66% sensitivity and 83% spec-

Fig 6. Probablity maps. Predicted risk of infarction maps overlaid on the 2-month follow-up T2 FSE for a patient who presented with a right hemiparesis. The maps were gener- ated from the hyperacute data obtained at admission. Over- laid values are the probability tissue will go on to infarction, with blue areas at low risk of infarction and yellow areas at high risk of infarction. The first row shows the probability of tissue becoming infarcted. The second and third rows are the lower and upper 95% confidence limits, respectively, for the risk estimates shown in the first row. For clarity, voxels with <30% probability of becoming infarcted are not shown. The first column are the results of using only diffusion data as predictors (rT2 + rADC + rDWl), second column are the results of using only perfusion data as predictors (rCBF + rCBV + rMTT), and the third column are the results of combining all 6 input parameters (rT2 + rADC + rDWl + rCBF + rCBV + rMTT). The area at risk depicted by the third column correlates best with the 2-month follow-up lesion area shown at bottom right. (Adapted with permission. 14)

ificity and generalized linear model algorithms combining DWI and PWI provided 66% sensitivity and 84% specificity.

HEMORRHAGIC TRANSFORMATION

Because ischemic tissue treated with thrombo- lytic therapy has an increased incidence of hemorrhagic transformation, investigators have begun to evaluate diffusion and perfusion MRI parameters in tissue that hemorrhages compared with tissue that does not. Tong et al, 3~ in an analysis of 17 patients presenting within 8 hours of acute ischemic stroke onset, showed that ischemic tissue that undergoes hemorrhagic transformation has a significantly greater percentage of pixels with low ADCs compared with ischemic tissue that does not develop hemorrhage. 3~ Forty-seven percent of pixels in tissue that developed hemorrhage had ADCs less than 550 × 10 -6 mm2/sec compared with 19% of tissues that did not hemorrhage. In another article, 32 they reported that there was a persistent perfusion defect (time to peak maps) in 83% of

PERFUSION MRI 235

hemorrhagic ischemic lesions compared with 30% of nonhemorrhagic ischemic lesions. In 12 patients with acute intracerebral hemorrhages, Kid- well et al33 showed that 6 of 12 patients had a rim of perihematoma-decreased ADC values that

correlated with poor clinical and radiographic outcome. In 5 of 6 patients who underwent perfusion imaging with calculation of time to peak, there was diffuse ipsilateral hemispheric hypoper- fusionY

REFERENCES

1. Ostergaard L, Weisskoff RM, Chesler DA, et al: High resolution measurement of cerebral blood flow using intravas- cular tracer bolus passages. Part I: Mathematical approach and statistical analysis. Magn Reson Med 36:715-725, 1996

2. Yamada K, Wu O, Gonzalez RG, et al: Magnetic resonance perfusion-weighted imaging of acute cerebral infarction: Effect of the calculation methods and underlying vasculopathy. Stroke 33:87-94, 2002

3. Schaefer P, Hunter,G, Hamberg, L, et al: Predicting cerebral ischemic infarct volume with diffusion and perfusion MRI. AJNR (in press)

4. Baird AE, Benfield A, Schlaug G, et al: Enlargement of human cerebral ischemic lesion volumes measured by diffusion-weighted magnetic resonance imaging. Ann Neurol 41: 581-589, 1997

5. Kucharczyk J, Mintorovitch J, Asgari HS, et al: Diffusion/ perfusion MR imaging of acute cerebral ischemia. Magn Resort Med 19:311-315, 1991

6. Moseley ME, Kucharczyk J, Mintorovitch J, et al: Diffu- sion-weighted MR imaging of acute stroke: correlation with T2-weighted and magnetic susceptibility-enhanced MR imaging in cats. Am J Neuroradiol 11:423-429, 1990

7. Mintorovitch J, Moseley ME, Chileuitt L, et al: Compar- ison of diffusion- and T2-weighted MR1 for the early detection of cerebral ischemia and reperfusion in rats. Magn Reson Med 18:39-50, 1991

8. Schwamm LH, Koroshetz WJ, Sorensen AG, et al: Time course of lesion development in patients with acute stroke: serial diffusion- and hemodynamic-weighted magnetic resonance imaging. Stroke 29:2268-22676, 1998

9. Rosen BR, Belliveau JW, Vevea JM, et al: Perfusion imaging with NMR contrast agents. Magn Reson Med 14:249- 265, 1990

10. Rosen BR, Belliveau JW, Buchbinder BR, et al: Contrast agents and cerebral hemodynamics. Magn Reson Med 19:285- 292, 1991

11. Quast MJ, Huang NC, Hillman GR, et al: The evolution of acute stroke recorded by multimodal magnetic resonance imaging. Magn Resort Imaging 11:465-471, 1993

12. Barber PA, Darby DG, Desmond PM, et al: Prediction of stroke outcome with echoplanar perfusion- and diffusion- weighted MRI. Neurology 51:418-426, 1998

13. Karonen JO, Vanninen RL, Liu Y, et al: Combined diffusion and perfusion MRI with correlation to single-photon emission CT in acute ischemic stroke. Ischemic penumbra predicts infarct growth. Stroke 30:1583-1590, 1999

14. Wu O, Koroshetz WJ, Ostergaard L, et al: Predicting tissue outcome in acute human cerebral ischemia using combined diffusion- and perfusion-weighted MR imaging. Stroke 32:933-942, 2001

15. Tatlisumak T, Carano RA, Takano K, et al: A novel endothelin antagonist, A-127722, attenuates ischemic lesion

size in rats with temporary middle cerebral artery occlusion: a diffusion and perfusion MRI study. Stroke 29:850-857, 1998; discussion 857-858

16. Minematsu K, Fisher M, Li L, et al: Effects of a novel NMDA antagonist on experimental stroke rapidly and quanti- tatively assessed by diffusion-weighted MRI. Neurology 43:

397-403, 1993 17. Schlaug G, Benfield A, Baird AE, et al: The ischemic

penumbra: operationally defined by diffusion and perfusiou MRI. Neurology 53:1528-1537, 1999

18. Rohl L, Ostergaard L, Simonsen CZ, et al: Viability thresholds of ischemic penumbra of hyperacute stroke defined by perfusion-weighted MRI and apparent diffusion coefficient. Stroke 32:1140-1146, 2001

19. Liu Y, Karonen JO, Vanninen RL, et al: Cerebral hemodynamics in human acute ischemic stroke: A study with diffusion- and perfusion-weighted magnetic resonance imaging and SPECT. J Cereb Blood Flow Metab 20:910-920, 2000

20. Grandin CB, Duprez TP, Smith AM, et al: Usefulness of magnetic resonance-derived quantitative measurements of cerebral blood flow and volume in prediction of infarct growth in hyperacute stroke. Stroke 32:1147-i 153, 2001

21. Jones TH, Morawetz RB, Crowell RM, et al: Thresholds of focal cerebral ischemia in awake monkeys. J Neurosurg 54:773-782, 1981

22. Ueda T, Sakaki S, Yuh WT, et aI: Outcome in acute stroke with successful intra-arterial thrombolysis and predictive value of initial single-photon emission-computed tomography. J Cereb Blood Flow Metab 19:99-108, 1999

23. Lassen NA: Normal average value of cerebral blood flow in younger adults is 50 ml/100 g/min. J Cereb Blood Flow Metab 5:347-349, 1985

24. Marchal G, Beaudouin V, Rioux P, et al: Prolonged persistence of substantial volumes of potentially viable brain tissue after stroke: A correlative PET-CT study with voxel- based data analysis. Stroke 27:599-606, 1996

25. Furlan M, Marchal G, Viader F, et al: Spontaneous neurological recovery after stroke and the fate of the ischemic penumbra. Ann Nenrol 40:216-226, 1996

26. Frackowiak RS, Lenzi GL, Jones T, et al: Quantitative measurement of regional cerebral blood flow and oxygen metabolism in man using 150 and positron emission tomography: theory, procedure, and normal values. J Comput Assist Tomogr 4:727-736, 1980

27. Rempp KA, Brix G, Wenz F, et al: Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR imaging. Radiology 193:637- 641, 1994

28. Hatazawa J, Shimosegawa E, Toyoshima H, et al: Cerebral blood volume in acute brain infarction: A combined

236 SCHAEFER ET AL

study with dynamic susceptibility contrast MRI and 99mTc- HMPAO-SPECT. Stroke 30:800-806, 1999

29. Powers WJ: Cerebral hemodynamics in ischemic cere- brovascular disease. Ann Neurol 29:231-240, 1991

30. Lansberg MG, Thijs VN, O'Brien MW, et al: Evolution of apparent diffusion coefficient, diffusion-weighted, and T2- weighted signal intensity of acute stroke. Am J Neuroradiol 22:637-644, 2001

31. Tong DC, Adami A, Moseley ME, et al: Relationship between apparent diffusion coefficient and subsequent hereof-

rhagic transformation following acute ischemic stroke. Stroke 31:2378-2384, 2000

32. Tong DC, Adami A, Moseley ME, et al: Prediction of hemorrhagic transformation following acute stroke: role of diffusion- and perfusion-weighted magnetic resonance imaging. Arch Neurol 58:587-593, 2001

33. Kidwell CS, Saver JL, Mattiello J, et al: Diffusion- perfusion MR evaluation of perihematomal injury in hyperacute intracerebral hemorrhage. Neurology 57:161 l- l 617, 2001

Documents

Perfusion magnetic resonance imaging of acute ischemic stroke