(Stroke. 2000;31:3008.)
© 2000 American Heart Association, Inc.
Original Contributions |
N-Acetylaspartate Distribution in Proton Spectroscopic Images of Ischemic Stroke
Relationship to Infarct Appearance on T2-Weighted Magnetic Resonance Imaging
From the Departments of Medical Physics (J.M. Wild, I.M.) and Clinical Neurosciences (J.M. Wardlaw, C.P.W.), University of Edinburgh, Western General Hospital, Edinburgh, UK; and In Vivo NMR Center, Department of Biomedical Engineering, University of Alberta, Edmonton, Canada (J.M. Wild, I.M.).
Correspondence to Dr J. Wardlaw, Department of Clinical Neurosciences, Western General Hospital, Crewe Rd, Edinburgh, EH4 2XU, UK. E-mail JMW{at}skull.dcn.ed.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Background and Purpose—It is generally considered that tissue that appears abnormal on T2 MRI is already infarcted and that any penumbra lies outside the T2-visible lesion. We investigated the distribution of infarcted tissue using proton spectroscopic MRI.
Methods—In patients with symptoms of acute hemispheric ischemic stroke, imaged within a maximum of 3 days of stroke, we explored the distribution of N-acetylaspartate (NAA), a marker of intact neurons, within and around the abnormal (hyperintense) areas on T2-weighted MR images, using proton spectroscopic MRI.
Results—In 11 patients, imaged 24 to 72 hours after stroke onset, there was little evidence of damaged neurons (reduced NAA) beyond the margins of hyperintensity on the T2 image. However, within the abnormal T2 area, there were statistically significant differences in the amount of NAA (ie, the proportion of intact neurons) between areas that were obviously abnormal on T2 (very hyperintense) and those that were only slightly abnormal (slightly hyperintense).
Conclusions—The extent and degree of hyperintensity of the T2-visible lesion directly reflect the amount of neuronal damage; lack of a T2-visible lesion would suggest predominantly intact neurons at the time of imaging. We hypothesize that once tissue damage has reached a critical (probably irreversible) level, the T2 image quickly becomes abnormal without any significant time lag between the pathological staging of the infarct and its visualization on T2. Further testing in a larger study with information on blood flow levels would be required to confirm this.
Key Words: cerebral infarction • magnetic resonance imaging • pathology • spectroscopy
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In acute ischemic stroke, the concept of the ischemic penumbra, an area of nonfunctioning but potentially recoverable tissue around the core of an infarct, although difficult to demonstrate, is now generally accepted.1 2 Considerable effort has been directed recently at trying to find ways to visualize the extent of the penumbra to improve patient selection for acute stroke treatments. Implicit in this search for direct visualization of the penumbra is the idea that the ischemic/infarcted tissue does not appear abnormal on conventional CT or T2 MRI until some time after the permanent damage has actually occurred. Thus, an infarct could occur (but not be visible) on T2 MRI (or CT) until hours later. Radiologists and neurologists are familiar with patients who have had a definite stroke but who have a normal CT or T2 MR scan, although equally there are patients with obvious visible infarction on CT or T2 MRI even when scanned very early after the stroke. Thus, it is possible that visibility of ischemic/infarcted tissue on T2 is not dependent on the time lag per se but simply on the tissue reaching a particular (critical) depth of ischemia, at which point the image becomes abnormal immediately.
The original experiments of Astrup et al3 showed a series of thresholds of cellular events through which the brain tissue passes as the blood flow falls. Blood flow of approximately 9 mL/100 g per minute does not support neuronal viability, and cells die "instantly" at or below this level. Blood flow between approximately 10 and 20 mL/100 g per minute keeps neurons viable but "shut down" for variable periods of time but not indefinitely; if blood flow is not restored, the neurons eventually die. Thus, neuronal death is a product of blood flow reduction and time.1
Therefore, if an ischemic/infarcted area does not become visible on T2 MRI (or CT) until the neurons are actually dead, patients who have symptoms of a definite stroke (but in whom the blood flow is >9 mL/100 g per minute in the ischemic area, eg, perhaps 15 mL/100 g per minute) might not show any visible lesion on T2 until such time as the neurons have been ischemic for so long that they die in sufficient numbers to produce a T2-visible lesion. Restoration of blood flow before the neurons had died might mean that the infarct would never become visible on T2. When an infarct first becomes visible on T2 imaging, it is only slightly hyperintense (pale white) compared with normal brain. Close observation of the T2-visible lesion often demonstrates that the degree of hyperintensity is not uniform within the boundaries of the visible lesion but rather is patchy. In humans it progresses to maximal hyperintensity (very white) over the first 3 to 5 days, concomitant with the time of maximum infarct swelling,4 although the time may be quicker in rats.5
In the present study we used proton MR spectroscopic imaging to determine the relationship between the likely degree of permanent neuronal damage (ie, cerebral infarction) and the visible lesion on T2. Intact neurons contain the metabolite N-acetylaspartate (NAA), whose methyl protons can readily be detected by 1H MR spectroscopic imaging.6 The amount of NAA detectable with 1H MR spectroscopic imaging rapidly falls as neurons die in a cat stroke model.7 In stroke patients (mostly studied days after the stroke), the spectroscopic signal from NAA is reduced in infarcted brain compared with healthy brain.8 9 10 11 12 13 14 15 16 17 18 Thus far, the majority of MR spectroscopic studies in humans have used single-voxel spectroscopy, in which a region of interest (of typical volume 8 cm3) is placed over the area of brain thought to be infarcted (as seen on T2-weighted imaging) and in the contralateral region for comparison. Therefore, this gives little information about the distribution of metabolites across the whole infarct, into the surrounding presumed-penumbral and then presumed-normal brain. 1H MR spectroscopic imaging is potentially more useful because it gives metabolic information from a much larger region of the brain with a spatial precision comparable to single-voxel methods. Previous 1H MR spectroscopic imaging studies of human stroke have found reduced NAA in the core of the infarct demonstrated on T2 imaging, similar to the results of single-voxel spectroscopy studies.8 16 17 18 Systematic analysis of all the voxels in a spectroscopic image (to get a fuller metabolic picture of an ischemic brain) has not been undertaken in previous human in vivo stroke studies. There have been experimental studies (in rat models of stroke) in which the spatial and temporal variations in NAA have been compared with changes in diffusion images,19 20 21 although unfortunately the small size of the rat brain precludes high spatial resolution regional analysis of small localized brain regions even in a dedicated animal magnet.
We wanted to know first whether there was evidence of any infarcted but not yet T2-visible tissue beyond the boundary of the T2-visible lesion, ie, significantly reduced NAA in areas where the T2 signal was normal.22 We also wanted to determine whether there was evidence of variable tissue damage corresponding with different hyperintensities within the T2-visible lesion. If the T2-visible lesion appearance lags behind the point of actual tissue damage, there should be little NAA in the center of the T2-visible lesion (because the neurons would have died already), and the NAA might be reduced beyond the margins of the T2-visible lesion in damaged areas that had not yet become visible on T2. If, however, there is no significant time lag between the point of tissue damage and the appearance of the lesion on T2, then the NAA level should correspond with the degree of T2 hyperintensity within the T2-visible lesion but be relatively normal beyond the T2 lesion boundary. Therefore, the present study was performed as a further detailed analysis of spectroscopic image data obtained as part of a larger study in patients with recent stroke.15 23
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Materials and Patients
Patients with symptoms of an extensive hemispheric stroke (ie, at least 2 of the following 3 deficits: hemiparesis, hemianopia, deficit in higher cognitive function), with no contraindications to MR, were included if they could be imaged within 3 days of onset of their symptoms. The patients were examined by a stroke physician, and their clinical data were prospectively entered into our hospital stroke registry. Patients underwent MRI as soon as possible after hospital admission with the use of a 1.5-T Siemens SP63 Magnetom clinical MR scanner. T1 midline sagittal localizer and T2 and proton density (PD) axial brain imaging were performed before spectroscopy. Patients with hemorrhagic lesions were not imaged further, but those with either an obvious or possible infarct on the T2/PD-weighted imaging underwent MR spectroscopic imaging. Ethical approval was granted by the Lothian Area Ethics of Medical Research Committee. The T2 sequence was a spin-echo axial acquisition of the whole brain, with the following parameters: repetition time, 3500 ms; echo time, 93 ms; slice thickness, 5 mm; field of view, 240; and number of excitations, 1.
Spectroscopic Imaging
The acquisition and data processing methods for spectroscopic
imaging have been described in detail previously.24 The
prescan protocols included shimming, calculation of the optimum
1H nuclear MR chemical shift selective
suppression (CHESS) water suppression voltage,25
and positioning of the point-resolved spectroscopy (PRESS)
localized26 inner volume excitation volume of interest.
Water-suppressed and water reference27 data sets were then
acquired with 16x16 phase encodings over a
240-mm2 field of view, giving cubic voxels of
15 mm (Figure 1
). The echo time was
135 ms, and the repetition time between successive phase encoding steps
was 1600 ms. The data were then processed offline with software written
in C24 to give single-slice images of proton metabolites
within the PRESS inner volume placed in the brain. The data processing
steps were briefly as follows: reading of the raw files, voxel shifting
of the spectroscopic imaging grid, 2-dimensional spatial fast Fourier
transformation, phase correction with the water reference free
induction decay,26 4-Hz line broadening, Hankel-Lanczos
singular value decomposition water removal,28
spectral fast Fourier transformation, and peak area calculation for all
voxels. In addition, the 16x16 spectroscopic image grid was
zero-filled to 32x32 (Figure 2). No
further smoothing of the spectroscopic imaging grid was performed
because excessive smoothing may introduce an erroneous picture of the
actual metabolite distribution. Finally, the spectroscopic images were
normalized for head coil (B1) inhomogeneity using phantom
data,24 and the radio frequency coil loading was accounted
for using the amplitude of the 90° radio frequency pulse
voltage.29 The resulting images represent the
relative concentration of the metabolite in system-dependent units,
which allow intrapatient and interpatient comparisons to be made.
|
|
Image Analysis
The MR images from patients were closely examined to ensure that
most of the T2-visible lesion was visible on both adjacent T2 slices.
The spectroscopy slice was 15 mm thick, whereas the T2 slice was
only 5 mm thick, and therefore it was important to ensure that
there was sufficient lesion on the T2 slices adjacent to the slice
corresponding to the center of the spectroscopy grid. Patients for whom
the T2-visible lesion overlapped by at least 75% on the images on
either side of the slice on which the spectroscopy grid had been
centered were analyzed further in the present study. The
75% overlap was confirmed by measuring the area of the lesion in the 3
contiguous T2 images with the use of the interactive region-of-interest
facility of the ANALYZE (Mayo Clinic) image processing package.
Fulfillment of this condition of 75% overlap on adjacent slices should
have minimized inaccuracies introduced by partial volume
effects.30
The distributions of NAA levels from the spectroscopic grid in and around the T2-visible lesion were determined as follows. Each spectroscopic voxel was divided into quarters (subvoxels) by the process of zero-filling (see Spectroscopic Imaging). Subvoxels containing cerebrospinal fluid alone were omitted, as any NAA signal would have been, because of partial volume effects. These subvoxels were then coded blind to spectroscopic data according to 2 separate methods. The 2 coding methods were applied blind to the NAA values, and the clinical data on 2 different days were separated by a period of several weeks by the same neuroradiologist. The levels of NAA in the coded voxels were then read directly from the raw spectroscopic image.
In the first coding method, each subvoxel was coded N, P, or A
according to whether it was located in an area of definitely normal
brain on the T2 image (N), in an area of possible T2 signal
hyperintensity (P), or in an area of definitely abnormal T2
hyperintense signal (A) (Figure 3, top
panel). The N, P, A coding scheme was used to determine whether the
intensity of the T2 signal abnormality (ie, very hyperintense or only
slightly hyperintense) related to the degree of neuronal death as
determined by spatially corresponding NAA measurements. If there was a
significant time lag between neuronal death and the appearance of the
lesion on T2, then the NAA values in voxels coded P should be very
similar to those coded A, whereas if T2 changes closely mirrored
neuronal death without any time lag, then there should be a significant
difference between the amount of NAA in P and A.
|
In the second coding method, the subvoxels were coded core, inner rim,
outer rim, or normal according to their position in the area of T2
abnormality. Subvoxels in the central area of the T2-weighted lesion
were classified as follows: core (C)=the main lesion area; inner rim
(I)=a rim 1 subvoxel thick delineating the inner margin of the edge of
the visible T2 lesion, ie, abnormal on T2 filled to at least 50% to
100% with abnormal-appearing tissue; outer rim (O)=a rim 1 subvoxel
thick on the outer edge of the T2-visible lesion that appeared normal
on T2 or was only filled to <50% with visibly abnormal tissue; and
normal (N)=subvoxels containing 100% normal-appearing tissue (remote
from the T2 lesion) (Figure 3
, bottom panel). In this scheme,
the core of the presumed infarct was taken to be any subvoxel overlying
brain with any degree of T2 hyperintensity except for the rim of
subvoxels around the edge of the T2 hyperintense area. The square shape
of the subvoxels necessitated the use of a definition for inner rim of
subvoxels filled to between 50% and 100% with T2 hyperintensity and
for outer rim of subvoxels filled to <50% with T2 hyperintensity to
account for the visible boundary of the lesion often cutting obliquely
across its edge. The C, I, O, N coding scheme was used to determine
whether the infarcted tissue actually extended beyond the edges of the
T2-visible lesion, ie, whether T2 normal brain was in fact
infarcted—as inferred from low NAA levels—beyond the edge of the
visible lesion.
Statistical Analysis
The mean NAA levels for subvoxels in each coding classification
were calculated from all of the spectroscopic imaging surrounding the
T2-visible lesion for each of the patients studied, ie, to give a mean
value for all voxels coded N, P, or A or C, I, O, or N in each
patient.
NAA levels measured from all the voxels (as categorized according to the 2 methods outlined above) were compared after log transformation between voxel categories, with the use of a 2-way ANOVA model to control for patient-specific and system-dependent effects.30 Post hoc pairwise comparisons were performed with the Bonferroni correction to allow for multiple testing. All statistical analyses were performed with SPSS for Windows 7.5.1 (SPSS Inc).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The T2 images of 11 patients (of a total of 17 patients who had spectroscopic imaging) showed an area of hyperintensity consistent with a recent infarct that was large enough to overlap by 75% the 2 images immediately adjacent to the slice on which the spectroscopic grid had been centered. The patients were imaged at a mean time of 43 hours after stroke onset (range, 24 to 72 hours).
Analysis of Subvoxels Categorized as Normal, Possibly
Abnormal, or Abnormal According to T2 Appearance
There were statistically significant differences between the 3
categories of voxels (Table 1 and
Figure 4, top panel). Both the overall
ANOVA result (F2,902=53.8, P<0.001)
and all pairwise comparisons (P<0.001) were highly
statistically significant. There were clear differences in the NAA
values between voxels coded as normal (highest NAA values), possibly
abnormal, and abnormal (lowest NAA values). Thus, a marked increase in
T2 signal indicated that there had been significantly greater neuronal
loss compared with areas where the T2 signal was less hyperintense
although still definitely abnormal and indicated that even in this
possibly abnormal area, there was significant neuronal loss compared
with normal brain.
|
|
Analysis of Subvoxels Categorized as Infarct Core, Inner
Rim on the Edge of the Lesion, Outer Rim on the Edge of the Lesion, or
Normal Brain Outside the Lesion According to T2 Appearance
There were statistically significant differences in the NAA values
between the categories (F3,1328=75.7,
P<0.001) (Figure 4, bottom panel and Table 2). After Bonferroni adjustment of the
standard errors, there were statistically significant differences
between core subvoxels and all others individually (P<0.001
in each case), between the inner rim subvoxels and both the outer rim
and normal subvoxels (P<0.001 and P<0.04,
respectively), but not between the outer rim and the normal subvoxels
(P=0.06). This indicates that there was unlikely to have
been any significant neuronal loss outside the edge of the T2-visible
lesion.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In this small group of patients with large cerebral infarcts, we have demonstrated that the degree of signal abnormality (hyperintensity) on T2 imaging is directly proportional to the severity of neuronal loss as determined by the NAA level: the greater the hyperintensity in a T2 image, the greater is the neuronal damage as measured by depleted NAA levels in that region of the image plane. Furthermore, there was little evidence of significant neuronal loss beyond the margins of the T2-visible lesion. The use of the alternative classification of core, inner rim, outer rim, normal allowed us to explore the possibility that there was neuronal damage beyond the edges of the T2-visible lesion that had not yet become visible on T2. This second classification demonstrated that although there was a stepwise increase in the NAA level from the core of the infarct across the edge of the T2-visible infarct to definitely normal brain, the only statistically significant increases were from core to inner edge and inner edge to outer edge, but not outer edge to normal brain. This indicates that the edge of the T2-visible lesion is likely to be the true boundary of the infarct, there being little evidence of much neuronal loss (reduced NAA) beyond that boundary in the present group of patients. This does not completely exclude the possibility of there being some penumbral tissue beyond the T2-visible lesion, particularly since we were not able to image many of these patients very early after stroke, but indicates that if there is any, it is likely to be very minor.
These findings suggest that the appearance on T2 reflects the actual "real time" stage of the infarct during its evolution, rather than the slowly developing appearance of tissue that died many hours previously. While the traditional concept of ischemic stroke is of a central dead area and a surrounding penumbra of viable tissue, this simple geographic model may not be correct or useful. The ischemic area is probably patchy, ie, more damaged areas mingled with less damaged areas, as mirrored by the patchiness of the hyperintensity of T2. The longer the ischemia persists, the fewer viable neurons and the more dead neurons are present and the more visible the lesion becomes on T2. There is thus a balance depending on depth and duration of impaired cerebral blood flow1 ; perhaps, therefore, if there is no visible infarct on T2 at the time of imaging, the majority of the neurons are still alive.
In the present group, imaged >12 hours after the stroke, it is possible that the infarct had become completely established and that any penumbral tissue had already either died or survived. The lack of lactate in most patients might support this hypothesis. We were unable to image patients that quickly, but others have found evidence of incompletely damaged neurons up to several days after the infarct and therefore, by implication, penumbral tissue.2 31 32
What are the sources of error or bias in the present study? The problem of time has been mentioned previously. The partial volume effect of inclusion of tissue of a different composition from that suggested by the T2 image in-plane with the spectroscopic grid30 could affect the measured NAA values. However, by studying large T2 lesions, which were as extensive on the images on either side of the image on which the spectroscopic grid was placed as on that in-plane image itself, these partial volume effects were minimized. Excessive smoothing of metabolite images might lead to an NAA voxel of abnormally high or low intensity surrounded by voxels of uniform intensity, giving the impression of a core region surrounded by a region of intermediate intensity. We have tried to avoid this source of error by avoiding excessive smoothing of the spectroscopic images. The quantitative normalization procedures24 described in Subjects and Methods removed any machine system–dependent effects. One factor not accounted for in our analysis was metabolite relaxation time changes. As the T2 of water changes in ischemia, so do the T2s of the proton metabolites.11 However, multiple repetition time and echo time acquisitions, required to measure metabolite T1s and T2s, would have required prohibitively long scan times, which would be unethical in such ill patients, and therefore quantitative measurements of T1 and T2 change were not possible. Since our patients were all imaged well before the time of maximal water shift into the infarct and associated swelling (5 days), and given the magnitude of difference in NAA that we have observed across the infarcts, we think it unlikely that water content of the infarct alone would have had a major effect on the measured NAA values.
Visualization of infarction on CT depends on the time lapse from stroke to scan and the severity of the stroke.33 Some patients (usually those with the most severe strokes) have visible infarction on CT even when scanned within the first 90 minutes.34 It is likely that increasing stroke severity and time lapse increase visible infarcts on T2 MRI as well, although this has not specifically been examined. In our previous spectroscopy work (mostly with single voxel), increasing stroke severity was associated with larger infarcts on T2 MRI and lower NAA; the lower the infarct NAA, the greater was the likelihood of a poor clinical outcome.15 While the conclusions of the present study are based on T2 MR images, they may also apply to visible infarction on CT and would explain why patients with a clinically severe stroke and a definite visible infarct appear to be less likely to benefit from thrombolytic treatment.35 Other imaging techniques, more sensitive to early infarction, such as diffusion MRI,36 may simply be detecting tissue damage at higher blood flow levels at an earlier stage in the ischemic cascade. We were unable to perform diffusion imaging in the present study because the non–echo-planar technology imaging time was far too long for these patients to tolerate. However, we have started such a study on our new machine. Our findings in the present study are supported by studies in a rat model of middle cerebral artery occlusion in which (after a brief fall) T2 values in the infarct core rose linearly with time after arterial occlusion.5 We did not measure T2 values in the present study because that would also have considerably prolonged imaging time. T2 measurement would be feasible on newer, faster MR machines and might improve infarct recognition and characterization of the degree of ischemic damage. Our present results are also supported by a study in stroke patients in which those with middle cerebral artery occlusion had significantly lower NAA values in the infarct than those with either a patent middle cerebral artery or only minor branch occlusion at the time of imaging.37
In summary, the results of this study support the notion that the amount or severity of irreversible neuronal loss is directly reflected in the appearance of the brain on T2 MRI at that particular moment. Furthermore, the less obvious the lesion on T2, the more likely it is that there are viable neurons within it for a given stroke severity. This result has important implications for acute ischemic stroke treatment because it means that if there is little visible infarction on T2 MRI (or possibly CT), but the patient clinically has a moderate to severe stroke, then there has been little neuronal death thus far and there is likely to be a substantial proportion of viable but functionally shut-down neurons. Patients with obvious visible infarction are likely to have mostly dead neurons, and those with faintly visible infarcts are somewhere in between. Therefore, our results suggest that stroke physicians should worry less that they might not be seeing extensively and irreversibly damaged brain that is not yet visible on T2 at the time when the patient was scanned. Rather, it would appear that there is unlikely to be any time lag between the time of neuron death and visualization; if no lesion is visible, but the patient has symptoms of a moderate to severe stroke, then the majority of neurons have probably not yet died. Clearly this finding needs to be confirmed in a larger group of patients imaged earlier after ischemic stroke and ideally with some measure of the level of blood flow.
|
Acknowledgments |
|---|
This work was undertaken as part of the Medical Research Council Clinical Research Initiative in Clinical Neurosciences (grant numbers R21931 and R21932), which provided the financial support of Drs Wild, Marshall, and Wardlaw. We would like to thank Jim Cannon, Senior Radiographer, Martin Connell, Medical Physics, and Dr David Signorini, statistician, University of Edinburgh, and the medical staff and nurses for their contributions to this project.
Received May 25, 2000; revision received August 22, 2000; accepted August 22, 2000.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
1. Baron JC. Mapping the ischaemic penumbra with PET: implications for acute stroke treatment. Cerebrovasc Dis. 1999;9:193–201.[Medline] [Order article via Infotrieve]
2.
Marchal G, Beaudouin V, Rioux P, de la Sayette V, Le
Doze F, Viader F, Derlon JM, Baron JC. Prolonged persistence of
substantial volumes of potentially viable brain tissue after stroke.
Stroke. 1996;27:599–606.
3.
Astrup J, Siesjo BK, Symon L. Thresholds in cerebral
ischemia: the ischemic penumbra. Stroke. 1981;12:723–725.
4. Wardlaw JM, Dennis MS, Lindley RI, Warlow CP, Sandercock PAG, Sellar RJ. Does early reperfusion of a cerebral infarct influence cerebral infarct swelling in the acute stage or the final clinical outcome? Cerebrovasc Dis. 1993;3:86–93.
5. Calamante F, Lythgoe MR, Pell GS, Thomas DL, King MD, Busza AL, Sotak CH, Williams SR, Ordidge RJ, Gadian DG. Early changes in water diffusion, perfusion, T1, and T2 during focal cerebral ischemia in the rat studied at 8.5T. Magn Reson Med. 1999;41:479–485.[Medline] [Order article via Infotrieve]
6. Williams SR. In-vivo proton spectroscopy. In: Diehl P, Fluck E, Gunther H, Kosfeld R, Seelig J, eds. NMR Basic Principles and Progress. Berlin, Germany: Springer; 1992:56–71.
7. van der Toorn A, Verheul HB, van der Sprenkel J-WB, Tulleken CAF, Nicolay K. Changes in metabolites and tissue water status after focal ischemia in cat brain assessed with localized proton MR spectroscopy. Magn Reson Med. 1994;32:685–691.[Medline] [Order article via Infotrieve]
8.
Duijn JH, Matson GB, Maudsley AA, Hugg JW, Weiner MW.
Human brain infarction: proton MRS. Radiology. 1992;183:711–718.
9.
Petroff OAC, Graham GD, Blamire AM, Al-Rayess M,
Rothman DL, Fayad PB, Brass LM, Shulman RG, Prichard JW. Spectroscopic
imaging of stroke in humans: histopathology correlates of spectral
change. Neurology. 1992;42:1349–1354.
10. Christiansen P, Toft B, Gideon P, Danielsen ER, Ring P, Henriksen O. MR-visible water content in human brain: a proton MRS study. Magn Reson Imaging. 1994;12:1237–1244.[Medline] [Order article via Infotrieve]
11. Gideon P, Henriksen O. In vivo relaxation of N-acetyl-aspartate, creatine plus phosphocreatine, and choline containing compounds during the course of brain infarction: a proton MRS study. Magn Reson Imaging. 1992;10:983–988.[Medline] [Order article via Infotrieve]
12. Gideon P, Sperling B, Arlien-Soborg P, Olsen TS, Henriksen O. Long-term follow-up of cerebral infarction patients with proton magnetic resonance spectroscopy. Stroke. 1994;25:967–973.[Abstract]
13.
Gideon P, Henriksen O, Sperling B, Christiansen P,
Olsen TS, Jorgensen HS, Arlien-Soborg P. Early time course of
N-acetylaspartate, creatine and phosphocreatine, and
compounds containing choline in the brain after acute stroke: a proton
magnetic resonance spectroscopy study. Stroke. 1992;23:1566–1572.
14.
Saunders DE, Howe FA, van den Boogaart A, McLean MA,
Griffiths JR, Brown MM. Continuing ischemic damage after acute
middle cerebral artery infarction in humans demonstrated by short-echo
proton spectroscopy. Stroke. 1996;26:1007–1013.
15.
Wardlaw JM, Marshall I, Wild J, Dennis MS, Cannon J,
Lewis SC. Studies of acute ischemic stroke with proton magnetic
resonance spectroscopy. Stroke. 1998;29:1618–1624.
16. Barker PB, Gillard JH, van Zijl PCM, Oppenheimer SM, Bryan RN. Proton magnetic resonance spectroscopic imaging in acute stroke: identification of the ischaemic penumbra. In: Program and abstracts of the 2nd Scientific Meeting of The Society of Magnetic Resonance; August 6–12, 1994; Berkeley, Calif. Abstract 186.
17. Gillard JH, Barker PB, van Zijl PC, Bryan RN, Oppenheimer SM. Proton MR spectroscopy in acute middle cerebral artery stroke. AJNR Am J Neuroradiol. 1996;17:873–886.[Abstract]
18.
Lanfermann H, Kugel H, Heindel W, Herholz K, Heiss WD,
Lackner K. Metabolic changes in acute and subacute
cerebral infarctions: findings at proton MR spectroscopic imaging.
Radiology. 1995;196:203–210.
19. Norris DG, Hoehn-Berlage M, Dreher W, Kohno K, Busch E, Schmitz B. Characterization of middle cerebral artery occlusion infarct development in the rat using fast nuclear magnetic resonance proton spectroscopic imaging and diffusion-weighted imaging. J Cereb Blood Flow Metab. 1998;18:749–757.[Medline] [Order article via Infotrieve]
20. Dreher W, Kühn B, Gyngell ML, Busch E, Niendorf T, Hossmann KA, Leibfritz D. Temporal and regional changes during focal ischaemia in rat brain studied by proton spectroscopic imaging and quantitative diffusion NMR imaging. Magn Reson Med. 1998;39:878–888.[Medline] [Order article via Infotrieve]
21. Higuchi T, Graham SH, Fernandez EJ, Rooney WD, Gaspary HL, Weiner MW, Maudsley AA. Effects of severe global ischaemia on N-acetylaspartate and other metabolites in the rat brain. Magn Reson Med. 1997;37:851–857.[Medline] [Order article via Infotrieve]
22. Prichard JW. The ischemic penumbra in stroke: prospects for analysis by nuclear magnetic resonance spectroscopy. In: Waxman SG, ed. Molecular and Cellular Approaches to the Treatment of Neurological Disease. New York, NY: Raven Press; 1993.
23. Wild JM, Wardlaw JM, Marshall I, Signorini DF, Cannon J. NAA distribution in 1H spectroscopic images of stroke patients: metabolite evidence of a "penumbra." MAGMA. 1998;16:1. Abstract 577.
24. Wild JM, Marshall I. Normalisation of metabolite images in 1H NMR spectroscopic imaging. Magn Reson Imaging. 1997;15:1057–1066.[Medline] [Order article via Infotrieve]
25. Haase A, Frahm J, Hänicke W, Matthaei D. 1H NMR chemical shift (CHESS) imaging. Phys Med Biol. 1985;30:431–444.
26. Bottomley PA. Spatial localisation in NMR spectroscopy in vivo. Ann N Y Acad Sci. 1987;508:333–348.[Medline] [Order article via Infotrieve]
27. Ordidge RJ, Cresshull ID. The correction of transient B0 field shifts allowing for the application of pulsed gradient by phase correction in the time domain. J Magn Reson. 1986;69:151–155.
28. van den Boogart A, van Ormondt D, Pijnappel WWF, de Beer R, Ala-Korpela M. Removal of the water resonance from 1H magnetic resonance spectra. In: Mathematics in Signal Processing III. London, UK: Oxford University Press; 1992:175–195.
29. Danielsen ER, Michaelis T, Ross BD. Three methods of calibration in quantitative proton MR spectroscopy. J Magn Reson. 1995;6:287–291.
30. Blamire AM, Graham GD, Rothman DL, Prichard JW. Proton spectroscopy of human stroke: assessment of transverse relaxation times and partial volume effects in single voxel steam MRS. Magn Reson Imag. 1994;12:1227–1235.[Medline] [Order article via Infotrieve]
31.
Baron JC, von Kummer R, del Zoppo GJ. Treatment of
acute ischemic stroke: challenging the concept of a rigid and
universal time window. Stroke. 1995;26:2219–2222.
32. Heiss WD, Huber M, Fink GR, Herholz K, Pietrzyk U, Wagner R, Wienhard K. Progressive derangement of periinfarct viable tissue in ischemic stroke. J Cereb Blood Flow Metab. 1992;12:193–203.[Medline] [Order article via Infotrieve]
33.
Wardlaw JM, Lewis SC, Dennis MS, Counsell C, McDowall
M. Is visible infarction on computed tomography associated with an
adverse prognosis in acute ischemic stroke? Stroke. 1998;29:1315–1319.
34.
The National Institute of Neurological Disorders and
Stroke rt-PA Stroke Study Group. Tissue plasminogen
activator for acute ischemic stroke. N
Engl J Med. 1995;333:1581–1587.
35. Berrouschot J, Barthel H, von Kummer R, Koester J, Hesse S, Knapp WH, Schneider D. Thrombolysis with rt-PA: a SPECT study. Stroke. 1999;30:267. Abstract.
36. Warach S, Dashe JF, Edelman RR. Clinical outcome in ischaemic stroke predicted by early diffusion-weighted and perfusion magnetic resonance imaging: a preliminary analysis. J Cereb Blood Flow Metab. 1996;16:53–59.[Medline] [Order article via Infotrieve]
37. Kang D-W, Roh J-K, Lee Y-S, Song IC, Bae H-J, Yoon B-W, Chang KH. Impact of MCA flow patterns on ischemic injury in MCA territory infarction: 1H-magnetic resonance spectroscopy study. Stroke. 1999;30:262. Abstract.
This article has been cited by other articles:
 |
|
 |
|
  S. Munoz Maniega, V. Cvoro, F. M. Chappell, P. A. Armitage, I. Marshall, M. E. Bastin, and J. M. Wardlaw Changes in NAA and lactate following ischemic stroke: A serial MR spectroscopic imaging study Neurology, December 9, 2008; 71(24): 1993 - 1999. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J M Wardlaw Recent developments in imaging of stroke Imaging, October 1, 2002; 14(5): 409 - 419. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||