This Article Abstract Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Welch, K.M.A. Articles by Levine, S. R. Search for Related Content PubMed PubMed Citation Articles by Welch, K.M.A. Articles by Levine, S. R.

(Stroke. 1995;26:1983-1989.)
© 1995 American Heart Association, Inc.

Articles

A Model to Predict the Histopathology of Human Stroke Using Diffusion and T2-Weighted Magnetic Resonance Imaging

K.M.A. Welch, MD; Joseph Windham, PhD; Robert A. Knight, PhD; Vijaya Nagesh, PhD; James W. Hugg, PhD; Mike Jacobs, PhD; Donald Peck, MS; Patty Booker, RN; Mary O. Dereski, PhD Steven R. Levine, MD

From the Department of Neurology, Center for Stroke Research (K.M.A.W., R.A.K., V.N., J.W.H., P.B., M.O.D., S.R.L.), and Department of Diagnostic Radiology (J.W., M.J., D.P.), Henry Ford Hospital and Health Sciences Center, Detroit, Mich/Case Western Reserve University, Cleveland, Ohio.

Correspondence to K.M.A. Welch, MD, Department of Neurology, K-11, Henry Ford Hospital and Health Sciences Center, 2799 W Grand Blvd, Detroit, MI 48202.

	Abstract

Top Abstract Introduction Model Subjects and Methods Results Discussion References

 
Background and Purpose We sought to identify MRI measures that have high probability in a short acquisition time to predict, at early time points after onset of ischemia, the eventual development of cerebral infarction in clinical patients who suffer occlusion of a cerebral artery.

Methods We developed an MR tissue signature model based on experimentally derived relationships of the apparent diffusion coefficient of water (ADC_w) and T₂ to ischemic brain tissue histopathology. In eight stroke patients we measured ADC_w and T₂ intensity using diffusion-weighted echo-planar imaging (DW-EPI). Tissue signature regions were defined, and theme maps of the ischemic focus at subacute time points after stroke onset were generated.

Results Five MR signatures were identified in human stroke foci: two that may predict either cell recovery or progression to necrosis, one that may mark the transition to cell necrosis, and two that may be markers of established cell necrosis.

Conclusions An MR tissue signature model of ischemic histopathology using ADC_w and T₂ can now be tested for its potential to predict reversible and identify irreversible cellular damage in human ischemic brain regions.

Key Words: cerebral ischemia, focal • magnetic resonance imaging • stroke, acute • stroke outcome

	Introduction

Top Abstract Introduction Model Subjects and Methods Results Discussion References

 
For the acute management of stroke it is desirable to have one measuring device, thereby shortening the duration of the early investigative process and permitting treatment at the earliest possible time. This device should be readily available, rapid, and noninvasive and provide the maximum number of measurements, including diagnostics. MR technology can observe anatomic structure, identify arterial occlusion, and measure cerebral blood flow, metabolism, and integrity of the blood-brain barrier. Currently, however, the technology is such that it occupies hours of study to achieve this complete information in one session. The challenge is to (1) reduce the time of the MR methods and (2) identify the most critically important measures for the diagnosis, staging, and prediction of outcome of stroke. In this study we have used both approaches.

We report the measurement of ADC_w in clinical stroke patients by DW-EPI and the measurement of T₂I by T₂-weighted MRI. A model of experimentally derived relationships of ADC_w and T₂ to histologically graded ischemic damage and the designated tissue signatures is described and used in the study of these patients. We used EIGENTOOL image analysis software and cluster analysis to generate tissue signature maps (theme maps) and to quantitate proportions of stroke regions with different tissue signature at subacute time points after stroke onset. We describe how we have identified these tissue signatures in human stroke and discuss the probability with which they may predict the eventual development of cerebral infarction.

	Model

Top Abstract Introduction Model Subjects and Methods Results Discussion References

 
The model of tissue signatures of ischemic histopathology is based on the sensitivity of diffusion imaging to ischemia¹ exploited in a series of animal experiments conducted in our laboratory during the past 3 years.^{2 3 4 5 6 7 8 9 10} We investigated the evolution and changes in ADC_w, T₂, and other conventional MR measures in the rat permanent middle cerebral artery occlusion model. Measurements of MR indexes other than ADC_w and T₂ provided no different information and therefore are not described. By noting the dynamic shifts in ADC_w and T₂ after ischemic onset and the relationships of each to the other, we have attributed tissue histopathology described below to individual MR signatures.

We assessed ischemic injury in anatomically defined zones by neuronal, neuropil, and astrocyte grading and neuronal counting using light microscopy. The results most cogent to the model are provided in References 3 through 5 and are summarized in Table 1. The data obtained in these experiments indicated a reduction in ADC_w with gradation of change in terms of time and degree depending on severity of ischemia.

View this table: [in this window] [in a new window]   Table 1. Histopathologic Scoring of Brain Tissue in Rat Middle Cerebral Artery Occlusion With MR Tissue Signature Attribution

The precise physical basis for this decline remains to be determined but, in brief, includes temperature effects, increased tortuosity of the extracellular diffusion paths, restriction by reduced cellular membrane permeability, shifts in water from extracellular to intracellular space, membrane depolarization, and cytotoxic edema.^{4 11 12 13 14} The decreased ADC_w is associated with a specific threshold level of regional cerebral blood flow, similar to that which is associated with the loss of membrane potential,¹⁵ and is also associated with breakdown of energy metabolism, acidosis, cellular ionic shifts, and decreased Na⁺,K⁺-ATPase activity.^{16 17} Elevation of T₂ seemed to follow a similar gradation of change that was, however, delayed compared with ADC_w.

In the most heavily injured ischemic brain regions, the value of ADC_w began to turn toward normal, coincident with the emergence of eosinophilic neurons and the beginning of neuronal necrosis. The subsequent rebound increase of ADC_w, therefore, probably indicates a loss of cell membrane integrity (ie, it allows the unrestricted movement of water molecules) and cell necrosis. An elevated T₂ distinguishes the "normal" ADC_w (measured during the transition from low to high values) in damaged tissue from the normal ADC_w in recovered tissue. High ADC_w and high T₂ together were associated with cellular necrosis. Even when T₂ subsequently declined toward normal values, the ADC_w remained elevated and a signature of cellular necrosis. A low ADC_w before T₂ elevation predicted cellular necrosis at 1 week with a high degree of probability.⁶ A low ADC_w at 1 hour, however, was not predictive.

The relationships of ADC_w and T₂ changes derived from groups of animals studied in Reference 3 are depicted diagrammatically in Fig 1. These changes are derived from observations in a core of the ischemic focus because in this region maximal ADC_w and T₂ changes are exhibited throughout the course of ischemia. Thus, we can assign six different MR signatures (A through F) to tissue histopathology ranging from normal to necrosis. This model of experimentally derived MR signatures of ischemic histopathology may be tested in vivo for the potential to predict and mark the various stages in the evolution of cellular damage. The provisions of the model are summarized in Table 2 and as follows: (1) A low ADC_w in acute stroke may be associated with the potential for cellular recovery or eventual cell necrosis dependent on time, magnitude of initial ischemia, and subsequent degrees of ischemia. Thus, measurement of ADC_w alone, at a single moment in time or as is often seen in the marginal zones of the ischemic focus, has an uncertain probability of predicting recovery or cell necrosis. (2) A low ADC_w with high T₂ as ischemia progresses may have a probability of predicting but is not a marker of eventual cell necrosis, although events leading to irreversible cell injury may or may not have commenced. (3) The transition from low to high ADC_w values probably marks the onset of irreversible cell necrosis. (4) High ADC_w and high or normal T₂ is a marker of established cell necrosis in the tissue.

View larger version (19K): [in this window] [in a new window]   Figure 1. Computer-generated diagram of the dynamic relationships of the ADCw and T2 in the central region of an ischemic focus based on experimental data obtained from the rat middle cerebral artery occlusion model provided in the text, in Table 1, and in References 3, 4, and 15. There are five critical shifts in relationship between ADCw and T2 as ischemia progresses over time, to which we have attributed tissue signatures B through F. The corresponding histopathology for these signatures is provided in Table 1.

View this table: [in this window] [in a new window]   Table 2. MR Measures With Tissue Signature Attribution and Their Significance

	Subjects and Methods

Top Abstract Introduction Model Subjects and Methods Results Discussion References

 
Patients
Eight patients (aged 64.4±15.6 years) were studied during the subacute stage of stroke from 24 to 116 hours after onset of symptoms. One patient was studied at 24 hours and again at 48 hours. Two patients were studied once after stroke at 36 hours, two at 48 hours, and two at 72 hours. One patient was studied first at 116 hours and again at 70 days. All patients suffered focal cortical ischemic stroke due to occlusion of a single intracranial large or branch artery supplying anterior or posterior brain cortex. All patients or immediate family members gave written informed consent to the procedure, which was approved by the Institutional Review Board of the Henry Ford Hospital and Health Science Center.

Techniques
Studies were conducted in a 3-T, 80-cm-bore magnet (patient access bore is approximately 56 cm in diameter) with actively shielded gradients of 20 mT/m (Magnex Scientific, Ltd) interfaced to an SMIS imaging and spectroscopy console (Surrey Medical Imaging Systems, Ltd). A quadrature high-pass birdcage head-imaging coil was used for signal acquisition at 128 MHz. Patient head movement was comfortably restricted, and ear plugs were used to control noise discomfort. The patient's pulse rate and peripheral venous PO₂ were monitored by a pulse oximeter with finger probe. Patients were continuously observed by video camera, and a nurse was present throughout the procedure.

The studies used a DW-EPI sequence. In brief, a sequence was designed to acquire axial images of the whole brain one slice at a time and to allow separate application of diffusion gradients in three orthogonal directions. For technical reasons related to pixel distortion, only the DWI images along the z axis, ie, slice direction, were used for ADC_w calculation. Control studies demonstrated that cardiac gating was unnecessary. Two trapezoidal diffusion gradients, equal in magnitude and duration, were applied in the vertical (y or anterior-posterior) direction, and phase-encoding gradients were applied in the horizontal (x or left-right) direction. For a 20-cm field of view and 200-kHz continuous readout sampling, a plateau readout gradient of 12 mT/m was required. Readout gradient trapezoids were produced with 320-microsecond ramps and 640-microsecond plateaus, resulting in 1.28-millisecond readout lobes and 82-millisecond total readout time. The spin echo was placed coincident with the zero-phase-encoded gradient echo. For a maximum diffusion gradient of b=600 s/mm², a spin-echo time of 90 milliseconds was used, and the center of k-space sampling was placed asymmetrically at the 17th gradient echo (ie, 16 negative and 48 nonnegative k-space scan lines).

Rapid (15 seconds) FLASH (fast low-angle shot) scout images interleaved in two orthogonal planes (sagittal and transaxial) were used to position patients with the anterior-posterior commissure line at isocenter. Shimming was then performed to achieve nonlocalized water line widths of 20 Hz (0.16 ppm at 3 T) or less. Acquisition of DW-EPI began less than 10 minutes after the patient first entered the magnet bore.

Sixteen contiguous 5-mm slices (1 slice per second) were acquired in an interleaved order to minimize magnetization transfer and slice cross-talk effects. Four diffusion gradient strengths were applied (b=10, 207, 414, and 621 s/mm²) separately in each primary orthogonal direction. Reference scans were acquired without phase-encoding gradients, which allowed correction of echo position and phase before Fourier transformation reconstruction, to minimize image ghosts. A total of 384 DW-EPI scans (reference and imagex3 [gradient directions]x4 [b values]x16 [slices]) were acquired in 6.4 minutes. The total examination took 45 to 60 minutes.

The reference DW-EPI scans were used to determine gradient echo positions and phases to correct the image scans. The samples digitized during readout gradient ramps were discarded, and the even gradient echoes were time-reversed. After phase correction, the asymmetrical k-space samples from (-16 to +47)xk were conjugate-filled to range symmetrically from (-47 to + 47)xk. The resulting 128x94 matrix was mildly Hanning filtered and zero-filled, then 128x128 images were reconstructed by two-dimensional Fourier transformation. Nominal image resolution was 1.6 mmx2.1 mmx5 mm, giving a 17-µL nominal voxel.

Image Processing and Analysis
For the T₂I we used the b=10 s/mm² images (almost purely T₂-weighted). ADC_w maps were constructed and combined with T₂I images from which cluster scattergrams were generated for subsequent cluster analysis¹⁸ (Fig 2). Boundaries of the clustering regions were chosen with the use of an iterative supervised procedure. We identified normal gray and white matter clusters using the hemisphere contralateral to the stroke in the image slice of interest. Upper and lower threshold values of the T₂ and ADC_w were chosen. This normal range of values was checked by creating a theme map of normal tissue by means of projection of the cluster back onto the contralateral hemisphere image. The threshold values for the normal tissue obtained in this way were used to determine the boundaries for normal T₂ and ADC_w for the whole slice scattergram and cluster analysis (box A, Fig 2). Once the boundaries of the normal tissue box were established, the boundaries of the other boxes were automatically established (boxes B through F, Fig 2). A count of all pixels in each box was taken together with the mean and SD of the values in each box cluster. Next, theme maps of the distribution of each tissue signature in the stroke focus and whole brain slice were constructed (Figs 2 and 3). The theme maps were used to establish whether signatures such as E (high ADC_w/high T₂ also characterizes normal CSF) were distributed in the stroke region or in a normal anatomic distribution. For example, the cluster in box E, Fig 2, was distributed to the normal CSF compartments (not shown in the figure) and not to the stroke focus.

View larger version (0K): [in this window] [in a new window]   Figure 2. Illustration of use of the computer program EIGENTOOL to develop theme maps of the tissue signatures in an ischemic focus obtained from patient 1. Top left map shows a clusterplot of a whole brain slice ADCw map and T2I image. Top right map shows how a normal range cluster is identified (box A), from which boxes B through F are drawn, each representing the tissue signatures derived as in Fig 1 and Table 1. Bottom theme map is generated from this cluster analysis and reveals how the pixel clusters in boxes B through D are distributed as well as their proportions in the ischemic focus.

View larger version (114K): [in this window] [in a new window]   Figure 3. Illustration of how the pixel clusters generated from the top right cluster analysis of Fig 2 are projected onto the T2-weighted image of patient 1. Note how signature B (low ADCw) is distributed in peripheral segments of the stroke focus and around the sylvian fissure. In this case the prominent signatures are B and C, indicating that the majority of tissue has not yet begun the progression to necrosis, although a small proportion of the tissue does show a transition to necrosis (signature D). The cluster in box E was distributed as normal CSF (not shown).

Adopting the ADC_w/T₂I range of combined normal gray and white matter as normal control is considered necessary because of difficulties in distinguishing white from gray regions in the infarct. Also, because of having to combine gray and white matter values, means±SD could not be used to define control values because they overlap. Accordingly, the low ADC_w pixels may be underestimated in gray matter to a small degree because the ADC_w of white matter is lower than that of gray. Conversely, the number of pixels of high ADC_w in white matter may be underestimated. This may mean that a "normal" ADC_w in white matter that leads to a tissue signature designation of D may represent cell necrosis rather than transition to necrosis. Nevertheless, both represent an irreversible signature.

For each patient, we calculated (1) ADC_w of normal tissue as the mean±SD of the normal range for gray/white matter based on the cluster analysis; (2) ADC_w mean±SD of each tissue signature; (3) number of pixels of each signature in a single slice ischemic focus; (4) total number of abnormal pixels, ie, the area of the lesion in a slice; and (5) the percentage of each abnormal signature in the lesion area, ie, when P represents the total number of abnormal pixels, the proportion of each tissue signature in the total stroke area can be calculated. For example, where E is the number of pixels of tissue signature equivalent to cell necrosis, E/Px100=%E.

Because our purpose is to describe the model, the data evaluation is descriptive and was not subjected to statistical analysis.

	Results

Top Abstract Introduction Model Subjects and Methods Results Discussion References

 
The stroke shown in Figs 2 and 3 was studied during the subacute time point and was within the middle cerebral artery distribution. Fig 2 shows the cluster analysis from a single brain slice ADC_w map and T₂I image obtained from patient 1. The ADC_w is plotted on the ordinate and T₂I on the abscissa. Tissue signature A (normal) represents the cluster of normal gray and white matter. From this, ROIs of tissue signatures B through F were constructed. Pixels from the stroke focus were distributed in one or more ROIs dependent on the stage and heterogeneity of the ischemic tissue types. In this example, three different tissue types were found at 36 hours (day 2) after stroke onset. The pixels of each tissue signature were counted, and the proportion calculated is given in Table 3. Fig 3 depicts how the pixels of differing tissue type were projected onto an ischemic focus. The map in Fig 3 was obtained by taking the pixels that clustered in ROIs B through D and segmenting them onto the whole brain slice. It can be seen that these tissue types were distributed heterogeneously in the ischemic lesion. The same image-fitting procedures were followed for the cluster of normal gray/white matter, which was confirmed as the normal range of ADC_w/T₂I by taking this cluster of pixels and projecting it back onto normal tissue. Note that by these means we avoided the use of control homologous contralateral regions and the use of an "average" ADC_w value. Furthermore, pixels clustering in box E projected to CSF but not to the stroke region, illustrating how this method can differentiate the high ADC_w/high T₂I signature of CSF from cellular necrosis in an infarct.

View this table: [in this window] [in a new window]   Table 3. Proportions and Number of Pixels of Tissue Signatures B Through F in Subacute Stroke and Mean±SD of ADCw in Pixels of Each Signature

The majority of pixels in the stroke example were type C (low ADC_w/high T₂I). Type D (normal ADC_w/high T₂I), the transitional signature to necrosis, was distributed in central regions of the stroke. Tissue signature B was distributed in marginal, "penumbral" regions, which also included regions around the sylvian fissure.

Table 3 shows the tissue signatures we obtained from the eight stroke patients. Most of the cases exhibited tissue signatures D through F, which are markers of ongoing (D) to established (E, F) necrosis and cell loss. The transitional phase from low ADC_w to high ADC_w (D) for the most part constituted a major percentage of the lesions. Furthermore, the mean ADC_w values in these regions were somewhat higher than the mean calculated from the normal range data, suggesting that they were increasing toward the upper threshold of normal.

One patient (case 1, Table 3) appeared to show a slower evolution of tissue signatures with high percentages of tissue types B and C predominating. Two patients were studied twice. Patient 2 was studied at 4 days and again 2 months later. At the earliest time point, the tissue signature of normal ADC_w/high T₂ (D) constituted 72% of the stroke, and E and F, ie, necrosis, constituted 28%. After 2 months, 85% of the infarct showed the signatures of cellular necrosis (E and F), supporting the transition of signature D toward necrosis, as postulated. Patient 3 was studied at 24 hours and again at 48 hours. At 24 hours, 14% of the lesion had a low ADC_w (signature B), and 86% had the signatures of necrosis (E and F). By 48 hours the proportion of tissue that was classified as B became D (10%), ie, it had begun the transition to necrosis, and the established necrosis (E and F) had increased to 90%. There was a high percentage of tissue type F in the stroke areas, often distributed around the rim of infarcts. Analysis of histograms taken across the infarct margin (data not shown) also showed that the elevated ADC_w extended well beyond the margin of T₂ decline.

	Discussion

Top Abstract Introduction Model Subjects and Methods Results Discussion References

 
There have been few published studies of DWI in human stroke. Chien et al¹⁹ investigated 15 patients with focal cerebral ischemia and infarction within 24 hours (n=3) and up to 4 years, some studied serially (n=4). They found an increase of the "average" ADC_w in ischemic lesions compared with normal control subjects at these subacute and chronic times of study, observing further increases as time progressed. They attributed this to vasogenic edema complicating the subacute strokes and to gliosis or encephalomalacia in the 2- to 4-year-old strokes. Warach et al²⁰ studied 32 ischemic stroke patients with different techniques, at times from 2 to 12 hours (n=12) and up to 12 days, in whom the ADC_w ratio, calculated as the ADC_w in homologous brain regions divided by that of the stroke region, was low acutely, reached a nadir at 24 hours, and then remained low. There was then no change in ADC_w up to 2 months, but chronic infarcts had a relative increase in ADC_w, although not until 4 months. The differences in the results of these two studies can probably be explained on technical grounds as well as by the use of different controls. Indeed, the control values of Warach et al²⁰ were higher than expected, although the use of ratios adjusted for this in part. Both studies had small numbers of patients, particularly at the early times of measurement. The techniques used were less advanced than are currently available so that, for example, movement artifact could not be easily corrected, and only two images were available for calculation of the ADC_w. Measuring an average ADC_w value from a single locus of an ischemic focus of well-known heterogeneity may add to the variability. In a recent subsequent publication of Warach et al,²¹ many of the technical limitations were overcome by using EPI for diffusion measurements, although as before only single loci were analyzed. In this study the ADC_w reduction was observed only up to 4 days and was significantly elevated by 1 month. This time course of ADC_w change more closely approximates our own observations, although disparity in the timing of ADC_w elevation remains. One explanation could be that we studied different stroke populations. The rapidity with which ischemic necrosis advances may vary with factors such as location, collateral vasocapacitance, and cause, eg, embolism versus thrombosis.²²

In clinical practice, stroke patients cannot be studied at the precise time of stroke onset or serially over prolonged time periods so that moment to moment shifts in ADC_w cannot be observed. Instead a "snapshot" of events is obtained at the earliest possible time and possibly on one or two other occasions during stroke evolution. Thus, based on experimental observations,³ at these less acute time points the ADC_w in tissue destined for necrosis may be low, normal, or high as the ADC_w shifts from low to high values, and there will be a heterogeneous distribution of ADC_w values throughout central and peripheral regions. Accordingly, there must be some measure to discriminate a "normal" value of ADC_w in its transition to high values in regions undergoing active cellular necrosis from that of normal ADC_w values in recovering tissue. For these reasons at least one complementary measure is needed to characterize the evolving ischemic histopathology. The biophysical mechanisms of T₂ change are better known than those of the ADC_w,²³ and the T₂ correlates of histopathology, although regionally and dynamically limited, are established.^{24 25 26} These points influenced us to model the dynamic relationships between T₂ and ADC_w for their potential to predict and mark ischemic histopathology in human stroke.

We have identified five MR signatures in human stroke foci, of which, on the basis of our experimental model, two may predict either cell recovery or progression to necrosis and were more often distributed in stroke margins, one may mark the transition to necrosis, and two may be markers of established necrosis. The ADC_w values in nonischemic tissue used as control were for the most part lower than reported with other techniques,^{19 20 21} with a fairly large SD in a few cases. This might be explained on the basis of the DW-EPI technique used to obtain the ADC_w or other technical differences in the methods used to acquire data published by others (see below). Also, control data were obtained from all gray/white matter pixels in the contralateral hemisphere of the brain slice as opposed to the average of one normal locus. Diffuse changes in ADC_w, unrelated to the presenting stroke but contributing biological variability, may be present. The individual variabilities in normal ADC_w, however, do not influence tissue signatures because they are derived from ratios in the same subject.

In our clinical studies a low ADC_w alone was more often distributed in margins of the stroke foci. A low ADC_w can be caused by a spectrum of cellular change from potentially reversible metabolic deficit to destined cell necrosis.⁶ Thus, in clinical practice additional measures may be needed to discriminate degrees of recoverability, especially at later times of study.

It remains to be determined whether a low ADC_w and high T₂ has histopathological correlates in human stroke of (1) vasogenic edema in ischemic regions destined for recovery, (2) eventual cell necrosis, or (3) both (dependent on stroke region). If this information was known, it would have important implications for human stroke assessment. First, when brain regions are identified with this signature, although the cascade of events leading to cell necrosis may have begun, there still may be some potential benefit from cytoprotective drugs working at later points in the cascade, perhaps in more peripheral zones of the ischemic focus. Second, cellular necrosis may not have occurred yet in these brain regions, so that perhaps there will be less compliance of these tissues to hemorrhagic conversion or massive enhancement of vasogenic edema after reperfusion therapy.

We found regions characterized by a high ADC_w and high T₂ in our patients that we believe reflect cellular necrosis, but we also found evidence of significant proportions of tissue signature with a high ADC_w and normal T₂. We offer as possible explanation that the elevated ADC_w represents cellular necrosis and that the T₂ value is either reduced by the paramagnetic influence of iron from deoxygenated hemoglobin deposited by petechial hemorrhage in the tissue or could reflect selective neuronal necrosis unaccompanied by vasogenic edema.³ In support of the latter, tissue maps have shown the projection of this tissue signature often localizing to the periphery of the infarct.

The significance of an available predictor and marker of cell necrosis cannot be overemphasized. T₂-weighted MRI has been unable to discriminate regions of necrosis from less involved tissue in a subacute ischemic focus, which is essential if MRI is to be used for other than diagnostic purposes in stroke patients. In combination with DWI, however, it opens the possibility of identifying tissue that may respond to cytoprotective therapy and perhaps extending the therapeutic window or opening it for those patients in whom the time of stroke onset cannot be determined with certainty. At a minimum, such tissue signatures might indicate whether the total lesion is beyond recovery. Extended studies of our tissue signature model in acute ischemia seem warranted, with serial observations to test the ability of the model to predict the eventual volume of infarction. Ideally the model should be tested against human autopsy material, but the practical limitations of rapidly accumulating such data are considerable. The model can be tested for predictability almost as well with the use of the T₂-weighted image of the chronic established infarct as the "gold standard," and we are proceeding to do so.

	Selected Abbreviations and Acronyms

 

ADCw = apparent diffusion coefficient of water CSF = cerebrospinal fluid DW-EPI = diffusion-weighted echo-planar imaging DWI = diffusion-weighted imaging ROIs = regions of interest T2I = T2-weighted signal intensity

	Acknowledgments

 
This study was supported in part by National Institutes of Health grant NS-23393 and the American Heart Association.

Received April 10, 1995; revision received August 18, 1995; accepted August 18, 1995.

	References

Top Abstract Introduction Model Subjects and Methods Results Discussion References

 

1. Moseley ME, Cohen Y, Mintorovitch J, Chileuitt L, Shimizu H, Kucharczyk J, Wendland MF, Weinstein PR. Early detection of regional cerebral ischemia in cats: comparison of diffusion- and T₂-weighted MRI and spectroscopy. Magn Reson Med. 1990;14:330-346. [Medline] [Order article via Infotrieve]
2. Knight RA, Ordidge RJ, Helpern JA, Chopp M, Rodolosi LC, Peck D. Temporal evolution of ischemic damage in rat brain measured by proton nuclear magnetic resonance imaging. Stroke. 1991;22:802-808. [Abstract/Free Full Text]
3. Knight RA, Dereski MO, Helpern JA, Ordidge RJ, Chopp M. MRI assessment of evolving focal cerebral ischemia: comparison with histopathology in rats. Stroke. 1994;25:1252-1262. [Abstract]
4. Helpern JA, Dereski MO, Knight RA, Ordidge RJ, Chopp M, Qing MS. Histological correlations of nuclear magnetic resonance imaging parameters in experimental cerebral ischemia. Magn Reson Imag. 1993;11:241-246. [Medline] [Order article via Infotrieve]
5. Dereski MO, Chopp M, Knight RA, Rodolosi LC, Garcia JH. The heterogeneous temporal evolution of focal ischemic neuronal damage in the rat. Acta Neuropathol (Berl).. 1993;85:327-333. [Medline] [Order article via Infotrieve]
6. Jiang Q, Zhang ZG, Chopp M, Helpern JA, Ordidge RJ, Garcia JH, Marchese BA, Qing ZX, Knight RA. Temporal evolution and spatial distribution of the diffusion constant of water in rat brain after transient middle cerebral artery occlusion. J Neurol Sci. 1993;120:123-130. [Medline] [Order article via Infotrieve]
7. Dereski MO, Chopp M, Knight RA, Chen H, Garcia JH. Focal cerebral ischemia in the rat: temporal profile of neutrophil responses. Neurosci Res Commun. 1992;11:179-186.
8. Garcia JH, Yoshida Y, Chen H, Li Y, Zhang ZG, Lian J, Chen S, Chopp M. Progression from ischemia injury to infarct following middle cerebral artery occlusion in the rat. Am J Pathol. 1993;142:623-635. [Abstract]
9. Chen H, Chopp M, Zhang ZG, Garcia JH. The effect of hypothermia on transient cerebral artery occlusion in the rat. J Cereb Blood Flow Metab. 1992;12:621-628. [Medline] [Order article via Infotrieve]
10. Peck DJ, Spickler EM, Knight RA, Hearshen DO, Windham JP. Analysis of the evolution of focal cerebral ischemia in the rat using the eigenimage filter. Magn Reson Med. 1992;26:259-273. [Medline] [Order article via Infotrieve]
11. Benveniste H, Hedlund LW, Johnson GA. Mechanism of detection of acute cerebral ischemia in rats by diffusion-weighted magnetic resonance microscopy. Stroke. 1992;23:756-754.
12. Moonen CT, Pekar J, de Vleeschouwer MH, van Gelderen P, van Zijl PC, DesPres D. Restricted and anisotropic displacement of water in healthy cat brain and in stroke studied by NMR diffusion imaging. Magn Reson Med. 1991;19:327-332. [Medline] [Order article via Infotrieve]
13. Tanner JE. Transient diffusion is a system partitioned by permeable barriers: application to NMR measurements with a pulses field gradient. J Chem Phys. 1978;69:1748-1754.
14. Cooper RL, Chang DB, Young AC, Martin CJ, Ancker-Johnson B. Restricted diffusion in biophysical systems. Biophys Acta. 1970;230:482-486.
15. Busza AL, Allen KL, Gadian DG, Crockard HA. Early changes demonstrated by diffusion-weighted MR imaging in experimental cerebral ischemia. Magn Reson Med. 1991;1:328.
16. Moseley ME, Brant-Zawadzki M, Berry I, Bartkowski H, Weinstein P, Germano I, Nishimura MC, Chew W, Hurd R, Levy R. Magnetic resonance imaging and 31-P and 1-H spectroscopy of experimental brain ischemia. Am J Neuroradiol. 1986;7:538-539.
17. Mintorovitch J, Baker LL, Yang GY, Shimizu H, Weinstein PR, Moseley ME, Kucharczyk J. Diffusion-weighted hyperintensity of early cerebral ischemia: correlation with brain water content and ATPase activity. Proc Soc Magn Reson Med. 1991;10:329.
18. Soltanian-Zadeh H, Windham JP, Peck DJ. MRI feature extraction using a linear transformation. Presented at (and published in Proceedings of) the SPIE Medical Imaging VII Conference; February 14-19, 1993; Newport Beach, Calif.
19. Chien D, Kwong KK, Gress DR, Buonanno FS, Buxton RB, Rosen BR. MR diffusion imaging of cerebral infarction in humans. AJNR Am J Neuroradiol.. 1992;13:1097-1102. [Abstract]
20. Warach S, Chien D, Li W, Ronthal M, Edelman RR. Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology. 1992;42:1717-1723. [Abstract/Free Full Text]
21. Warach S, Gaa J, Siewert B, Wielopolski P, Edelman RR. Acute human stroke studied by whole brain echo planar diffusion-weighted magnetic resonance imaging. Ann Neurol. 1995;2:231-241.
22. D'Olhaberriague LD, Welch KMA, Nagesh V, Gymnopoulous C, Boska M, Levine SR, Chopp M, Ewing JR. Diffusion-weighted and T₂ MRI can discriminate embolic from nonembolic stroke in humans. Stroke. In press. Abstract.
23. Mansfield P, Morris PG. Water in biological systems. In: Waugh JS, ed. NMR Imaging in Biomedicine. London, England: Academic Press, Inc; 1982:15-31.
24. Bose B, Jones SC, Lorig R, Friel HT, Weinstein M, Little JR. Evolving focal cerebral ischemia in cats: spatial correlation of nuclear magnetic resonance imaging, cerebral blood flow, tetrazolium staining, and histopathology. Stroke. 1988;19:28-37. [Abstract/Free Full Text]
25. Brant-Zawadzki M, Pereira B, Weinstein P, Moore S, Kucharczyk W, Berry J, McNamara M, Derugin N. MR imaging of acute experimental ischemia in cats. AJNR Am J Neuroradiol.. 1986;7:7-11. [Abstract]
26. DeWitt LD, Kistler JP, Miller DC, Richardson EP Jr, Buonanno FS. NMR-neuropathologic correlation in stroke. Stroke. 1987;2:342-351.

This article has been cited by other articles:

				A M BLAMIRE The technology of MRI -- the next 10 years? Br. J. Radiol., August 1, 2008; 81(968): 601 - 617. [Abstract] [Full Text] [PDF]

				S M. Maniega, M E Bastin, P A Armitage, A J Farrall, T K Carpenter, P J Hand, V Cvoro, C S Rivers, and J M Wardlaw Temporal evolution of water diffusion parameters is different in grey and white matter in human ischaemic stroke J. Neurol. Neurosurg. Psychiatry, December 1, 2004; 75(12): 1714 - 1718. [Abstract] [Full Text] [PDF]

				P. D. Mitsias, J. R. Ewing, M. Lu, M. M. Khalighi, M. Pasnoor, H. B. Ebadian, Q. Zhao, S. Santhakumar, M. A. Jacobs, N. Papamitsakis, et al. Multiparametric Iterative Self-Organizing MR Imaging Data Analysis Technique for Assessment of Tissue Viability in Acute Cerebral Ischemia AJNR Am. J. Neuroradiol., October 1, 2004; 25(9): 1499 - 1508. [Abstract] [Full Text] [PDF]

				Y. Ozsunar, P. E. Grant, T. A. G. M. Huisman, P. W. Schaefer, O. Wu, A. G. Sorensen, W. J. Koroshetz, and R. G. Gonzalez Evolution of Water Diffusion and Anisotropy in Hyperacute Stroke: Significant Correlation between Fractional Anisotropy and T2 AJNR Am. J. Neuroradiol., May 1, 2004; 25(5): 699 - 705. [Abstract] [Full Text] [PDF]

				J. L. Bykowski, L. L. Latour, and S. Warach More Accurate Identification of Reversible Ischemic Injury in Human Stroke by Cerebrospinal Fluid Suppressed Diffusion-Weighted Imaging Stroke, May 1, 2004; 35(5): 1100 - 1106. [Abstract] [Full Text] [PDF]

				J. Chen, S.-K. Song, W. Liu, M. McLean, J. S. Allen, J. Tan, S. A. Wickline, and X. Yu Remodeling of cardiac fiber structure after infarction in rats quantified with diffusion tensor MRI Am J Physiol Heart Circ Physiol, August 7, 2003; 285(3): H946 - H954. [Abstract] [Full Text] [PDF]

				P. D. Mitsias, M. A. Jacobs, R. Hammoud, M. Pasnoor, S. Santhakumar, N. I.H. Papamitsakis, H. Soltanian-Zadeh, M. Lu, M. Chopp, and S. C. Patel Multiparametric MRI ISODATA Ischemic Lesion Analysis: Correlation With the Clinical Neurological Deficit and Single-Parameter MRI Techniques Stroke, December 1, 2002; 33(12): 2839 - 2844. [Abstract] [Full Text] [PDF]

				J.M. Wardlaw, S.L. Keir, M.E. Bastin, P.A. Armitage, and A.K. Rana Is diffusion imaging appearance an independent predictor of outcome after ischemic stroke? Neurology, November 12, 2002; 59(9): 1381 - 1387. [Abstract] [Full Text] [PDF]

				C. Z. Simonsen, L. Rohl, P. Vestergaard-Poulsen, C. Gyldensted, G. Andersen, and L. Ostergaard Final Infarct Size after Acute Stroke: Prediction with Flow Heterogeneity Radiology, October 1, 2002; 225(1): 269 - 275. [Abstract] [Full Text] [PDF]

				R. C. McKinstry, J. H. Miller, A. Z. Snyder, A. Mathur, G. L. Schefft, C. R. Almli, J. S. Shimony, S. I. Shiran, and J. J. Neil A prospective, longitudinal diffusion tensor imaging study of brain injury in newborns Neurology, September 24, 2002; 59(6): 824 - 833. [Abstract] [Full Text] [PDF]

				M. E. Mullins, P. W. Schaefer, A. G. Sorensen, E. F. Halpern, H. Ay, J. He, W. J. Koroshetz, and R. G. Gonzalez CT and Conventional and Diffusion-weighted MR Imaging in Acute Stroke: Study in 691 Patients at Presentation to the Emergency Department Radiology, August 1, 2002; 224(2): 353 - 360. [Abstract] [Full Text] [PDF]

				W. A. Copen, L. H. Schwamm, R. G. Gonzalez, O. Wu, C. B. Harmath, P. W. Schaefer, W. J. Koroshetz, and A. G. Sorensen Ischemic Stroke: Effects of Etiology and Patient Age on the Time Course of the Core Apparent Diffusion Coefficient Radiology, October 1, 2001; 221(1): 27 - 34. [Abstract] [Full Text] [PDF]

				I.-J. Huang, C.-Y. Chen, H.-W. Chung, D.-C. Chang, C.-C. Lee, S.-C. Chin, and M. Liou Time Course of Cerebral Infarction in the Middle Cerebral Arterial Territory: Deep Watershed versus Territorial Subtypes on Diffusion-weighted MR Images Radiology, October 1, 2001; 221(1): 35 - 42. [Abstract] [Full Text] [PDF]

				S. Ishikawa, K. Yokoyama, T. Kuroiwa, and K. Makita Evolution of cerebral ischaemia induced by thromboembolism in rats detected by early sequential MR imaging Br. J. Anaesth., September 1, 2001; 87(3): 469 - 476. [Abstract] [Full Text] [PDF]

				T Murata, H Kimura, H Kado, M Omori, J Onizuka, T Takahashi, H Itoh, and Y Wada Neuronal damage in the interval form of CO poisoning determined by serial diffusion weighted magnetic resonance imaging plus 1H-magnetic resonance spectroscopy J. Neurol. Neurosurg. Psychiatry, August 1, 2001; 71(2): 250 - 253. [Abstract] [Full Text] [PDF]

				A. Rabinstein, S. Falcone, A. Forteza, and A. Forteza Diffusion-weighted Imaging Shows Cytotoxic and Vasogenic Edema in Eclampsia AJNR Am. J. Neuroradiol., June 1, 2001; 22(6): 1068 - 1070. [Abstract] [Full Text] [PDF]

				M. G. Lansberg, V. N. Thijs, M. W. O'Brien, J. O. Ali, A. J. de Crespigny, D. C. Tong, M. E. Moseley, and G. W. Albers Evolution of Apparent Diffusion Coefficient, Diffusion-weighted, and T2-weighted Signal Intensity of Acute Stroke AJNR Am. J. Neuroradiol., April 1, 2001; 22(4): 637 - 644. [Abstract] [Full Text]

				O. Wu, W. J. Koroshetz, L. Ostergaard, F. S. Buonanno, W. A. Copen, R. G. Gonzalez, G. Rordorf, B. R. Rosen, L. H. Schwamm, R. M. Weisskoff, et al. Predicting Tissue Outcome in Acute Human Cerebral Ischemia Using Combined Diffusion- and Perfusion-Weighted MR Imaging Stroke, April 1, 2001; 32(4): 933 - 942. [Abstract] [Full Text] [PDF]

				M. A. Jacobs, Z. G. Zhang, R. A. Knight, H. Soltanian-Zadeh, A. V. Goussev, D. J. Peck, and M. Chopp A Model for Multiparametric MRI Tissue Characterization in Experimental Cerebral Ischemia With Histological Validation in Rat : Part 1 Stroke, April 1, 2001; 32(4): 943 - 949. [Abstract] [Full Text] [PDF]

				M. A. Jacobs, P. Mitsias, H. Soltanian-Zadeh, S. Santhakumar, A. Ghanei, R. Hammond, D. J. Peck, M. Chopp, and S. Patel Multiparametric MRI Tissue Characterization in Clinical Stroke With Correlation to Clinical Outcome : Part 2 Stroke, April 1, 2001; 32(4): 950 - 957. [Abstract] [Full Text] [PDF]

				J. R. Meyer, A. Gutierrez, B. Mock, D. Hebron, J. M. Prager, M. T. Gorey, and D. Homer High-b-value Diffusion-weighted MR Imaging of Suspected Brain Infarction AJNR Am. J. Neuroradiol., November 1, 2000; 21(10): 1821 - 1829. [Abstract] [Full Text] [PDF]

				S. L. Keir and J. M. Wardlaw Systematic Review of Diffusion and Perfusion Imaging in Acute Ischemic Stroke Stroke, November 1, 2000; 31(11): 2723 - 2731. [Abstract] [Full Text] [PDF]