| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Stroke. 2001;32:950.)
© 2001 American Heart Association, Inc.
Original Contributions |
Multiparametric MRI Tissue Characterization in Clinical Stroke With Correlation to Clinical Outcome
Part 2
From the Departments of Neurology (M.A.J., P.M., S.S., M.C.), Radiology, Medical Image Analysis Research (M.A.J., H.S-Z., A.G., R.H., D.J.P), and Neuroradiology (S.P.), Henry Ford Health Sciences Center, Detroit, Mich; Department of Electrical and Computer Engineering, University of Tehran (Iran) (H.S-Z.); Department of Physics, Oakland University, Rochester, Mich (M.A.J., M.C.); and Department of Radiology, The Johns Hopkins University School of Medicine, Baltimore, Md (M.A.J.).
Correspondence to Michael Chopp, PhD, Department of Neurology, Center for Stroke Research, Henry Ford Hospital E&R 3056, 2799 W Grand Blvd, Detroit, MI 48202. E-mail chopp{at}neuro.hfh.edu
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionAppendix 1References |
|---|
Background and Purpose—Multiparametric MRI generates different zones within the lesion that may reflect heterogeneity of tissue damage in cerebral ischemia. This study presents the application of a novel model of tissue characterization based on an angular separation between tissues obtained with the use of an objective (unsupervised) computer segmentation algorithm implementing a modified version of the Iterative Self-Organizing Data Analysis Technique (ISODATA). We test the utility of this model to identify ischemic tissue in clinical stroke.
Methods—MR
parameters diffusion-, T2-, and T1-weighted imaging (DWI,
T2WI, and T1WI, respectively) were obtained from 10 patients at 3 time
points (30 studies) after stroke: acute (
12 hours), subacute (3
to 5 days), and chronic (3 months). The National Institutes of Health
Stroke Scale (NIHSS) was measured, and volumes were obtained from the
ISODATA, DWI, and T2WI maps on patients at each time
point.
Results—The acute (12
hours) multiparametric ISODATA volume was significantly
correlated with the acute (12 hours) DWI
(r=0.96,
P<0.05; n=10) and chronic (3
months) T2WI volume (r=0.69,
P<0.05; n=10). The
ISODATA-defined tissue regions exhibited MR indices consistent
with ischemic and/or infarcted tissue at each time point. The
acute (12 hours) multiparametric ISODATA volumes were
significantly correlated
(r=0.82,
P<0.009; n=10) with the
final NIHSS score. In comparison, the acute (12 hours) DWI volumes
were less correlated (r=0.77,
P<0.05; n=10) and T2WI volume
(12h) exhibited a marginal correlation
(r=0.66,
P<0.05; n=10) with the final
NIHSS score.
Conclusions—The integrated ISODATA approach to tissue segmentation and classification discriminated abnormal from normal tissue at each time point. The ISODATA volume was significantly correlated with the current MR standards used in the clinical setting and the 3-month clinical status of the patient.
Key Words: cerebral ischemia, focal • diagnostic imaging • diffusion imaging • magnetic resonance imaging • signal processing, computer assisted, ISODATA • stroke, acute • stroke classification
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1References |
|---|
Changes in the state of tissue water reflect physiological alterations that can be visualized by MRI. For example, diffusion-weighted imaging (DWI) generates images based on a quantitative assessment of the random brownian movement of water protons (diffusion) within tissues.1 2 3 4 5 6 Cytotoxic edema, secondary to energy depletion, and the loss of ionic homeostasis at the cellular level are believed to be major contributing factors to the changes seen on DWI during acute stroke.2 7 Clinical studies have shown that acute lesion volumes visualized on DWI scans in stroke patients have a strong correlation with final infarct volumes and clinical neurological outcomes.8
T2- and T1-weighted imaging (T2WI and T1WI, respectively) are part of the standard protocol for stroke imaging. A typical T2WI sequence obtained after stroke includes a proton density–weighted image and a T2WI. At later times after stroke, T2WI is a marker of vasogenic edema and is considered the diagnostic "gold standard" in the clinical setting for identification of cerebral infarction.9 10 11 12 Similarly, T1WI reflects vasogenic edema within the ischemic brain.13 14 15
In clinical practice, DWI, T2WI, and T1WI are often acquired during acute stroke, and integration of these MRI data may provide complementary information about the status of the tissue. Therefore, the introduction of an objective computerized segmentation of the MRI would assist in the identification and classification of brain tissue. Specifically, in this study we present unsupervised segmentation of stroke using the Iterative Self-Organizing Data Analysis Technique (ISODATA) method of postprocessing analysis.16 17 18 Objective computerized segmentation of multiparametric MRI has not been used to identify and classify ischemic tissue damage over time in the clinical setting after stroke.
This study presents the application of a novel model of
tissue characterization using the angular separation of tissue
signature vectors from the segmentation of multiparametric
MRI.17 19 This
model was validated in the experimental setting of cerebral
ischemia.17 19
We test the utility of this approach to identify ischemic
tissue in clinical stroke. To this end, segmented volumes of
ischemic tissue damage were compared with quantitative volumes
of the DWI, apparent diffusion coefficient of water (ADC), and T2WI
maps from acute (12 hours) to chronic (3 months) times after stroke
and compared with the neurological outcome of the patient as defined by
the National Institutes of Health Stroke Scale (NIHSS) at each time
point.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1References |
|---|
Clinical Subjects
All patients (n=10; 30 total studies) underwent a standard stroke MRI at 3 different time points: acute (0 to 12 hours), subacute (72 to 168 hours), and chronic (3 months). The acute stroke protocol consists of sagittal T1, axial multi-spin-echo T2WI, pregadolinium and postgadolinium T1WI, DWI, and circle of Willis 3-dimensional phase contrast MR angiography. All MRIs were performed on a 1.5-T General Electric Signa MRI unit. This protocol takes approximately 35 minutes to complete. The MRI parameters were as follows: T1 sagittal image (repetition time/echo time [TR/TE]=600/14 ms), axial T2WI (TR/TE=2500/30, 60, 90, 120 ms; number of excitations [NEX]=1), and T1WI 3-dimensional inversion recovery (TR/TE/inversion time=450/250/500 ms; NEX=2), with field of view=23x23 cm; matrix=256x192; and slice thickness=6 mm. Axial DWI (TR/TE=10 000/101 ms; b value=1000, 600, 300, and 0 s/mm2; field of view=46x23; matrix=128x128; slice thickness=6 mm/no slice gap; NEX=1; on 3 orthogonal axes) was obtained. Throughout the acquisition of the images, the patient was clinically monitored with the neurology service in attendance. A neuroradiologist (S.P.) blinded to the ISODATA segmentation interpreted all MRI examinations. In addition, the clinical status of each patient was estimated by the NIHSS at each time point (P.M., S.S.). All studies were approved by the Human Institutional Review Committee at our facility. Signed informed consent forms were obtained for each study from patients or appropriate family member or legal guardian.
MR Image Preprocessing and
Analysis
MR image analysis was performed with a SUN
UltraSPARC2 workstation (Sun Microsystems Inc). MRI data were processed
with Eigentool image analysis
software.20 21 22 23 24
Eigentool has a comprehensive set of functions for displaying,
restoring, enhancing, and analyzing images. The complete toolbox
consists of image analysis algorithms and advanced functions
such as morphological image operators, registration, and warping
methods.25
After reconstruction, preprocessing was performed on the MR data. Preprocessing consisted of subimaging and noise reduction. Subimaging of the intracranial volume was done with the use of thresholding and morphological operations to segment the image background, skull, and scalp from brain tissue.26 Finally, the images were restored with the use of a nonlinear restoration filter that reduces white noise while preserving edges and partial volume effects.27 Maps of the trace ADC and T2 were created for each time point using a least-squares fit from the slope of the signal intensity on a pixel-by-pixel basis.
Coregistration and Warping
Coregistration and warping of the MR data were
accomplished by a previously reported 2-step
methodology.25 28
This method consists of a modified head and hat surface-based
registration algorithm that registers the temporal MRI to a reference
MRI (acute T2WI), followed by nonlinear thin plate spline warping by
deformable contours to compensate for distortions between the T2WI and
DWI. In this study, for registration and warping, T2 was used as the
head data set with the other temporal MRI as the hat data set. After
coregistration, warping was performed on the DWI to exactly match the
reference
T2WI.28
MRI Data Analysis
The ISODATA Technique
The ISODATA technique is an unsupervised segmentation
method related to the K-means algorithm with additional splitting and
merging steps that allow for the adjustment of cluster centers and
their number.29 The ability
to adjust the number of clusters is the main advantage of the ISODATA
method because it requires no initial training or a priori
knowledge of the exact number of clusters (tissue classes) before
segmentation. The modified ISODATA algorithm consists of 4 main steps
that are summarized as follows: (1) Clustering parameters
are put into the program. (2) MR data are partitioned into random
clusters. (3) The pixels are grouped into clusters the lie closest to
each other, as defined by the intra-Euclidean distance (the clusters
and cluster centers are vectors). Inter-Euclidean distances are
calculated between pixel vectors and cluster centers. (4) Splitting and
merging of the clusters are performed on the basis of intra- and
inter-Euclidean distances computed. Steps 3 and 4 are repeated until
the algorithm converges or reaches the maximum number of iterations.
For a complete description, refer to the companion article (part 1) in
this issue of
Stroke.19
The MRI data set used in the ISODATA model consists of 2
T2WI (TE=30, 90 ms), 1 T1WI, and 2 DWI
(b=600, 1000
s/mm2). These MR images were selected
because each parameter provides different contrast between
tissue types during the evolution of stroke, which leads to better
separation of the tissue classes and increases the likelihood of tissue
class discrimination. The selected MR data set allows correlation with
animal experiments conducted in our
laboratory.17 By integrating
these parameters into a 5-dimensional (number of MR images)
feature space, the volume of tissue damage can be determined by
computer-assisted diagnosis independently of
time.17 18 A
representative MR data set used in the ISODATA model is
shown in
Figure 1
. Typical signature vectors used on the MR data set
are demonstrated in
Figure 2.
|
|
where Sijk is the k-th element of the ij-th pixel vector, ie, the mean gray level of the ij-th pixel in the k-th image in the image sequence for each cluster, and T is the transpose of the matrix.
Volume and Quantitative MRI
Measurements
Lesion volumes from each patient were obtained from a
DWI (b=1000
s/mm2) and a T2 map. The volume measurements
were accomplished by placing a region of interest (ROI) within the area
of signal intensity abnormality on the DWI and T2 map. Statistics (mean
and SD) were obtained from the gray level values within the ROI. The
DWI and T2 images were thresholded at the 95% CI to determine the area
of signal abnormality. The parametric outline of increased
signal intensity from the DWI was confirmed by a neuroradiologist
(S.P.) blinded to the ISODATA segmentation. These DWI volumes were
overlaid onto the ADC map to obtain quantitative values. Further
editing was done to remove any sulci or cerebrospinal fluid present
underneath the ROI that would have increased the ADC value.
Volume measurements similar to those described above were performed on
the T2 map. Volume analysis was performed on the
multiparametric ISODATA segmented ischemic regions
using the regions of abnormal and normal tissue defined by angular
separation between the different tissue
types.17 Total lesion
volumes were measured by summing the number of pixels from each region
in each slice defined by the ISODATA angle model within the
ischemic lesion and multiplying by the slice
thickness.
Comparisons between the ISODATA, acute (12 hours) DWI, and
chronic (3 months) T2 map volumes were performed. Correlation between
the ISODATA volume and clinical outcome, as defined by the NIHSS, was
obtained. The NIHSS evaluation was completed (P.M., S.S.) before the
ISODATA segmentation by investigators blinded to the ISODATA
segmentation. Intrareliability and interreliability measurements with
the use of ISODATA have been previously
reported.16 30
In addition, the defined ISODATA ROIs of ischemic regions used for volume measurements were overlaid onto the ADC and T2 maps to obtain quantitative measurements. Similar ROI analysis was performed for normal tissue by reflecting the ROI around the vertical axis. The MR estimates of ADC and T2 from the ischemic lesion were normalized to the homologous, noninvolved contralateral hemisphere and expressed as ratios of ADC (ADCr) and T2 (T2r), respectively.
Statistical Analysis
Paired t
tests were used to determine statistical significance between the
ISODATA volumes and the DWI and T2WI volumes. Linear regression
analysis was performed to test the correlation between the
volume measurements of ISODATA, DWI, and T2 maps. All
parametric map values are presented as mean±SD.
Statistical significance was assigned for
P<0.05.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1References |
|---|
The multiparametric ISODATA identified different regions of tissue damage at each time point after stroke. Figures 3
and 4 demonstrate the
multiparametric ISODATA segmentation of ischemic tissue
damage and corresponding DWI
(b=1000
s/mm2) and ADC maps, along with the T2 maps
at different times after stroke in representative
patients. In
Figure 3, a 63-year-old woman presented
acutely (12 hours) with a right hemispheric infarct extending from
the frontal to the occipital lobe, and she had no therapeutic
intervention. The multiparametric ISODATA segmented several
different regions of ischemic tissue damage and distinguished
an old infarction on the left frontal lobe near the
inferior frontal gyrus. These different regions of tissue
damage were not visualized on the DWI, ADC, and T2 maps. However, the
old infarct was clearly visualized on the ADC and T2 maps as a
hyperintense signal in the inferior frontal gyrus region.
Within 3 to 5 days, the ischemic region of tissue damage grew
to encompass most of the right hemisphere, including the basal ganglia.
The multiparametric ISODATA segmentation showed recruitment of
new areas of ischemic damage as distinct tissue classes. These
new areas of infarction were seen on the MR images; however, the
different characteristic tissue classes were not clearly demonstrated.
At the chronic time point (3 months), the ischemic tissue
damage was noted on all the MR and ISODATA images. The
multiparametric ISODATA approach clearly segmented the lesion
into different components, which were not obvious from the DWI, ADC,
and T2 maps.
|
|
In all patients, the average angular separation of abnormal
and normal tissue was 9.0±3.0° (range, 4.2° to 12.5°) at the
acute phase of stroke (12 hours after ictus). The angular separation
between abnormal and normal tissue continued to increase to
13.2±2.4° (range, 9.6° to 17.8°) at subacute time points and
to 13.1±2.9° (range, 10.7° to 19.5°) at chronic time points. The
ISODATA segmentation demonstrated recruitment of new and different
tissue classes within the volume of ischemic tissue damage, as
shown in
Figures 3
and 4 at all time points.
The patients’ demographic data, time to MRI, angular
measurements, infarct volumes, clinical information, and NIHSS scores
at each time point are summarized in
Table 1. The ISODATA-segmented ischemic tissue
volumes had significant correlation with the MRI-defined volumes at
each time point. The acute (12 hours) multiparametric ISODATA
volume exhibited excellent correlation
(r=0.96,
P<0.05; n=10)
with the acute DWI volume. Moreover, the early (12 hours) ISODATA
volume demonstrated excellent correlation
(r=0.69,
P<0.05; n=10) with the 3-month
infarct volume defined by the T2 map. The chronic (3 months) ISODATA
volume exhibited significant correlation with the 3-month volume
defined by the T2 map (r=0.96,
P<0.05; n=10).
|
|
MR indices of ADCr and
T2r are shown in
Table 2. Acutely (12 hours), the average
ADCr was decreased (0.73;
P<0.05; n=10) in all patients,
with a corresponding increased T2r (1.17;
P<0.05; n=10). Within 3 to 5
days after stroke, the mean ADCr was decreased
(0.72; P<0.05; n=10) in all
patients; however, in 1 patient the ADCr was
increased over unity (1.12). The average T2r
values continued to increase (1.48;
P<0.05; n=10). At 3 months,
both the mean ADCr (1.84;
P<0.05; n=10) and
T2r (1.70;
P<0.05; n=10) were elevated
above normal. The temporal evolutions of ADCr
and T2r values from the ISODATA volumes were
consistent with MR indices of ischemic tissue damage at
each time point.
|
Table 3 summarizes comparisons between the ISODATA,
DWI, and T2WI volumes with the NIHSS score at each time point. The
acute (12 hours) multiparametric ISODATA volume had superior
correlation (r=0.82,
P<0.009; n=10) with
the 3-month NIHSS score, outperforming both the acute (12 hours) DWI
and T2 volumes at this time point. The acute (12 hours) DWI volume
demonstrated good correlation
(r=0.77,
P<0.05; n=10); however, the
acute (12 hours) T2 volume exhibited marginal correlation
(r=0.66,
P<0.05; n=10) with the 3-month
NIHSS score.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1References |
|---|
In this report, the primary finding was that the volume of ischemic tissue damage defined by the integrated ISODATA angle approach was highly correlated with the final clinical status of the patient at each time point. The ISODATA angle model–defined volumes of damaged tissue outperformed both the DWI- and T2WI-defined volumes at all time points in correlation with the patients’ functional outcome after stroke, supporting previous findings in experimental animal models of stroke.18 This integrated model of ischemic tissue damage may be useful in determining the volume of tissue damage in clinical stroke. This is the first study to use the integrated ISODATA model in a clinical setting over a temporal course of stroke and demonstrates that ISODATA may be useful in defining the volume of tissue damage, which is highly correlated with the initial and final clinical status of the patients. Moreover, this report demonstrates that computer-assisted unsupervised segmentation using ISODATA of multiparametric MRI can detect heterogeneous zones within ischemic lesions during the evolution of clinical stroke. These heterogeneous regions were not clearly identified on the DWI, ADC map, and/or T2 map. In addition, the ISODATA model can distinguish between white and gray matter and cerebrospinal fluid while segmenting partial volume effects as distinct classes in feature space.16
The average angular separations between abnormal and normal
tissue types were similar to the experimental results in a rat model of
cerebral ischemia.19
In those reports, there was a clear tissue classification into normal
and different gradations of abnormal tissue characteristics with the
use of the angle model; however, the interpretation of tissue
classifications in animals should be tempered with the knowledge that
the progression of ischemic tissue damage varies over species.
This initial clinical study demonstrates that the angular separation
between abnormal and normal tissue is useful in defining the volume of
tissue damage and that the volume of tissue damage is highly correlated
with the initial (12 hours) and final (3 months) clinical status of
the patients.
The MR parameters used in this study were based
on our experimental results in a model of cerebral
ischemia.19 In these
studies it was demonstrated that the combined MR data set consisting of
T2WI, T1WI, and DWI used in the ISODATA segmentation was highly
correlated with the histological status of the tissue.
Since histological specimens are not routinely
available in the clinical setting, we selected the functional outcome
of the patient, as defined by the NIHSS score, to test the ISODATA
model. The heterogeneous tissue regions segmented by the
multiparametric ISODATA methodology exhibited the
characteristics of evolving cerebral ischemia at each time
point.3 5 6
Acutely (12 hours), the average ADCr was
decreased in all the patients studied, with a corresponding increased
T2r. However, there were
heterogeneous regions containing both increased and
decreased ADCr throughout the ischemic
volume, confirming previous
findings.30 Moreover,
between 3 and 5 days, 1 patient exhibited hypernormalized
ADCr, suggesting that progression of
ischemic damage is highly variable and depends on several
factors, such as location, duration, and mechanism of the
ischemic event.31 At
3 months, both ADCr and
T2r were elevated in the ISODATA-defined
volumes. Nonetheless, the multiparametric ISODATA segmentation
discriminated abnormal from normal tissue and detected their
heterogeneous distribution and revealed that in human
stroke, the concept of an infarcted core surrounded by potentially
viable ischemic tissue may not be valid
(Figures 3 and 4).
The multiparametric ISODATA overcomes several
limitations of our previously reported tissue signature
model,6 19 32 33
as clearly described in part 1 of this
report.19 The integrated
ISODATA provides an objective classification of the tissue clusters
independent of time, with the defined volumes consisting of multiple
tissue classes of ischemic tissue being significantly
correlated with NIHSS score at each time point. Tong et
al8 reported a correlation
(r=0.67,
P=0.03) between acute DWI
volumes and NIHSS scores at 24 hours in 10 patients who were imaged at
hyperacute time points (<6.5 hours). The present study found a
similar result (r=0.72,
P<0.009) at 3 to 5 days.
However, to the best of our knowledge, our report for the first time
extends the correlation of clinical outcome determined by NIHSS score
with DWI, T2WI, and the integrated ISODATA volumes to several time
points after stroke. The acute (12 hours) multiparametric
ISODATA volume had excellent correlation
(r=0.82,
P<0.009; n=10) with
the final 3-month NIHSS score, outperforming both the DWI and T2
volumes at each time point, which supported previous experimental
results.18 This study
demonstrates the advantages of using multiparametric MRI data
to identify ischemic tissue and to correlate the lesion with
the clinical status of the patient. Our data also support the concept
that a single MRI parameter is not sufficient to
characterize the status of the
tissue.6 32 In
addition, the power of the ISODATA methodology to discern different
regions of tissue damage within the lesion area not visible on
conventional MRI is illustrated. ISODATA has the advantage of utilizing
all the discriminating information implicitly available within the data
for segmentation.
Bernarding et al34 recently applied a supervised histogram-based methodology to multiparametric MR data to characterize ischemic tissue in 10 stroke patients at different time points ranging from 1 day to 4 months using DWI, ADC, T2WI, and T1WI in various combinations; longitudinal studies were not performed. Determination of normal tissue was accomplished interactively by drawing regions around surrounding clusters with the use of visualization and windowing techniques. They reported that at acute times after stroke, DWI should be included in the MR data set if T2WI or T1WI did not show any signal intensity changes. In addition, T1WI was needed to assist in the segmentation of normal tissue and, if hemorrhage was present in the brain, in the segmentation of ischemic tissue. This observation of the need to include DWI at early time points and T1WI for normal tissue differentiation is consistent with the findings of our studies.19 Taken together, their findings and our study demonstrate that multiparametric MRI and computer-assisted image processing provide a means for characterizing stroke in clinical applications.
Investigations are ongoing to incorporate other MRI parameters, such as perfusion-weighted imaging, into ISODATA. Recent reports have suggested that the combination of DWI and perfusion-weighted imaging may be useful in defining "tissue at risk."4 8 35 36 37 The addition of perfusion-weighted imaging into the ISODATA model may increase the segmentation of abnormal and normal tissue, including tissue at risk.
Conclusion
We have demonstrated that integration of
multiparametric MRI data in the ISODATA model can segment
ischemic tissue from normal tissue and that this segmentation
is highly correlated with the clinical status of the
patient.
|
Appendix 1 |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1References |
|---|
ISODATA Parameter Selection for Patients With Clinical Ischemia
For clinical stroke studies, the parameters are initialized for ISODATA as follows. We set the number of initial clusters (K) to K=15 and the desired clusters to K=5 (K=5 was selected because there are at least 5 types of tissue clusters after the brain is subimaged: white matter, gray matter, cerebrospinal fluid, and at least 2 clusters for abnormal tissue.) Note that this number (K=5 for desired clusters) will be automatically adjusted by ISODATA if the algorithm determines that the number of desired clusters is inadequate to represent the structure of the data. Other parameters are as follows: the minimum number of pixels in a cluster:
N=5; the SD of
white matter: S (in human brain, white matter
is the predominant tissue); the Euclidean distance between normal
tissue in brain: C, ie, white and gray
matter; the maximum number of cluster pairs to lump together in 1
iteration: L=1; the maximum number of iterations: I=100; and the
cluster center constant:
j=0.25
jmax, where
jmax is the maximum element for each SD
vector. See part 119 for
complete details of
ISODATA.
|
Acknowledgments |
|---|
This work was supported in part by National Institute of Neurological Disorders and Stroke grant PO1-NS23393 and National Institutes of Health grant CA61263. Dr Joe P. Windham was involved in the initial planning of these studies, and his untimely death is mourned by all his colleagues. The authors recognize the programming efforts of Lucie Bower and Thomas Brusca.
Received August 14, 2000; revision received November 28, 2000; accepted December 20, 2000.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1References |
|---|
1. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval Jeantet M. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology. 1986;161:401–407.
2. Moseley ME, Kucharczyk J, Mintorovitch J, Cohen Y, Kurhanewicz J, Derugin N, Asgari H, Norman D. Diffusion-weighted MR imaging of acute stroke: correlation with T2-weighted and magnetic susceptibility-enhanced MR imaging in cats. AJNR Am J Neuroradiol. 1990;11:423–429.[Abstract]
3. Baird AE, Warach S. Magnetic resonance imaging of acute stroke. J Cereb Blood Flow Metab. 1998;18:583–609.[Medline] [Order article via Infotrieve]
4.
Warach S, Li W,
Ronthal M, Edelman RR. Acute cerebral ischemia: evaluation with
dynamic contrast-enhanced MR imaging and MR angiography.
Radiology. 1992;182:41–47.
5.
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.
6.
Welch KMA, Windham
JP, Knight RA, Nagesh V, Hugg JW, Jacobs MA, Peck D, Booker P, Dereski
MO, Levine SR. A model to predict the histopathology of human stroke
using diffusion and T2-weighted magnetic resonance imaging.
Stroke. 1995;26:1983–1989.
7.
Benveniste H,
Hedlund L, Johnson G. Mechanism of detection of acute cerebral
ischemia in rats by diffusion-weighted magnetic resonance
microscopy. Stroke. 1992;23:746–754.
8.
Tong DC, Yenari MA,
Albers GW, O’Brien M, Marks MP, Moseley ME. Correlation of perfusion-
and diffusion-weighted MRI with NIHSS score in acute (<6.5 hour)
ischemic stroke.
Neurology. 1998;50:864–870.
9. Yuh WT, Crain MR, Loes DJ, Greene GM, Ryals TJ, Sato Y. MR imaging of cerebral ischemia: findings in the first 24 hours. AJNR Am J Neuroradiol. 1991;12:621–629.[Abstract]
10. Allegrini PR, Sauer D. Application of magnetic resonance imaging to the measurement of neurodegeneration in rat brain: MRI data correlate strongly with histology and enzymatic analysis. Magn Reson Imaging. 1992;10:773–778.[Medline] [Order article via Infotrieve]
11. Quast MJ, Huang NC, Hillman GR, Kent TA. The evolution of acute stroke recorded by multimodal magnetic resonance imaging. Magn Reson Imaging. 1993;11:465–471.[Medline] [Order article via Infotrieve]
12. Sauer D, Allegrini P, Cosenti A, Pataki A, Amacker H, Fagg G. Characterization of the cerebroprotective efficacy of the competitive NMDA receptor antagonist CGP40116 in a rat model of focal cerebral ischemia: an in vivo magnetic resonance imaging study. J Cereb Blood Flow Metab. 1993;13:595–602.[Medline] [Order article via Infotrieve]
13.
Go KG, Edzes HT.
Water in brain edema: observations by the pulsed nuclear magnetic
resonance technique. Arch
Neurol. 1975;32:462–465.
14. Kato H, Kogure K, Ohtomo H, Tobita M, Matsui S, Yamamoto E, Kohno H. Correlations between proton nuclear magnetic resonance imaging and retrospective histochemical images in experimental cerebral infarction. J Cereb Blood Flow Metab. 1985;5:267–274.[Medline] [Order article via Infotrieve]
15. Kato H, Kogure K, Ohtomo H, Izumiyama M, Tobita M, Matsui S, Yamamoto E, Kohno H, Ikebe Y, Watanabe T. Characterization of experimental ischemic brain edema utilizing proton nuclear magnetic resonance imaging. J Cereb Blood Flow Metab. 1986;6:212–221.[Medline] [Order article via Infotrieve]
16. Soltanian-Zadeh H, Windham JP, Robbins L. Semi-supervised segmentation of MRI stroke studies. Proc SPIE. 1997;3034:437–448.
17. Jacobs MA, Zheng ZG, Knight RA, Soltanian-Zadeth H, Mitsias P, Patel S, Jaing Q, Goussev AV, Windham JP, Chopp M. Multiparameter MRI tissue staging in experimental cerebral ischemia in rat. Stroke. 2000;31:346.
18. Jacobs MA, Knight RA, Soltanian-Zadeth H, Zheng ZG, Goussev AV, Peck DJ, Windham JP, Chopp M. Unsupervised segmentation of multiparameter MRI in experimental cerebral ischemia with comparison to T2, diffusion, and ADC MRI parameters and histopathological validation. J Magn Reson Imaging. 2000;11:425–437.[Medline] [Order article via Infotrieve]
19.
Jacobs MA, Zheng
ZG, Knight RA, Soltanian-Zadeth H, Goussev AV, Peck DJ, Chopp M. A
model for multiparametric MRI tissue characterization in
experimental cerebral ischemia with
histological validation in rat: part 1.
Stroke. 2000;32:943–949.
20. Windham JP, Abd-Allah MA, Reimann DA, Froelich JW, Haggar AM. Eigenimage filtering in MR imaging. J Comput Assist Tomogr. 1988;12:1–9.[Medline] [Order article via Infotrieve]
21. Soltanian-Zadeh H, Windham JP, Jenkins JM. Error propagation in eigenimage filtering. IEEE Trans Med Imaging. 1990;9:405–420.[Medline] [Order article via Infotrieve]
22. Soltanian-Zadeh H, Windham JP, Peck DJ, Yagle AE. A comparative analysis of several transformations for enhancement and segmentation of magnetic resonance image scene sequence. IEEE Trans Med Imaging. 1992;11:302–318.[Medline] [Order article via Infotrieve]
23. Peck DJ, Windham JP, Emery LL, Soltanian-Zadeh H, Hearshen DO, Mikkelsen T. Cerebral tumor volume calculations using planimetric and eigenimage analysis. Med Phys. 1996;23:2035–2042.[Medline] [Order article via Infotrieve]
24. Jacobs MA, Knight RA, Windham JP, Zheng ZG, Soltanian-Zadeth H, Goussev AV, Peck DJ, Chopp M. Identification of cerebral ischemic lesions in rat using eigenimage filtered MRI. Brain Res. 1999;1/2:83–94.
25. Jacobs MA, Windham JP, Soltanian-Zadeth H, Peck DJ, Knight RA. Registration and warping of magnetic resonance images to histological sections. Med Phys. 1999;26:1568–1578.[Medline] [Order article via Infotrieve]
26. Soltanian-Zadeh H, Windham JP. A multiresolution approach for contour extraction from brain images. Med Phys. 1997;24:1844–1853.[Medline] [Order article via Infotrieve]
27. Soltanian-Zadeh H, Windham JP, Yagle AE. A multidimensional nonlinear edge-preserving filter for magnetic resonance image restoration. IEEE Trans Image Proc. 1995;4:147–161.
28. Ghanei A, Soltanian-Zadeh H, Jacobs MA, Patel S. Boundary-based warping of brain MR images. J Magn Reson Imaging. 2000;12:412–423.
29. Ball GH, Hall DJ. A clustering technique for summarizing multivariate data. Behav Sci. 1967;12:153–155.[Medline] [Order article via Infotrieve]
30.
Nagesh V, Welch
KMA, Windham JP, Patel S, Levine SR, Hearshen D, Peck D, Robbins K, D.
Olhaberriague L, Soltanian-Zadeh H, Boska MD. Time course of ADCw
changes in ischemic stroke: beyond the human eye.
Stroke.. 1998;29:1778–1782.
31. Warach S, Boska M, Welch KMA. Pitfalls and potential of clinical diffusion-weighted MR imaging in acute stroke. Stroke. 1997;28:481–482.
32. Jiang Q, Chopp M, Zhang ZG, Knight RA, Jacobs MA, Windham JP, Peck D, Ewing JR, Welch KMA. The temporal evolution of MRI tissue signatures after transient middle cerebral artery occlusion in rat. J Neurol Sci. 1997;145:15–23.[Medline] [Order article via Infotrieve]
33. Jacobs MA, Windham JP, Knight R, Welch KMA, Peck D, Jiang Q, Chopp M. Methodology for MRI characterization of tissue histopathology during the evolution of cerebral ischemia in rat using feature space representation. J Cereb Blood Flow Metab. 1995;15(suppl 1):S125. Abstract.
34. Bernarding J, Braun J, Hohmann J, Mansmann U, Hoehn-Berlage M, Stapf C, Wolf KJ, Tolxdorff T. Histogram-based characterization of healthy and ischemic brain tissues using multiparametric MR imaging including apparent diffusion coefficient maps and relaxometry. Magn Reson Med. 2000;43:52–61.[Medline] [Order article via Infotrieve]
35.
Sorensen AG,
Buonanno FS, Gonzalez RG, Schwamm LH, Lev MH, Huang-Hellinger FR, Reese
TG, Weisskoff RM, Davis TL, Suwanwela N, Can U, Moreira JA, Copen WA,
Look RB, Finklestein SP, Rosen BR, Koroshetz WJ. Hyperacute
stroke: evaluation with combined multisection diffusion-weighted and
hemodynamically weighted echo-planar MR imaging.
Radiology. 1996;199:391–401.
36.
Albers GW.
Expanding the window for thrombolytic therapy in acute
stroke: the potential role of acute MRI for patient selection.
Stroke. 1999;30:2230–2237.
37. Kluytmans M, van Everdingen KJ, Kappelle LJ, Ramos LM, Viergever MA, van der Grond J. Prognostic value of perfusion- and diffusion-weighted MR imaging in first 3 days of stroke. Eur Radiol. 2000;10:1434–1441.[Medline] [Order article via Infotrieve]
This article has been cited by other articles:
 |
|
 |
|
  K. Y. Jonsdottir, L. Ostergaard, and K. Mouridsen Predicting Tissue Outcome From Acute Stroke Magnetic Resonance Imaging: Improving Model Performance by Optimal Sampling of Training Data Stroke, September 1, 2009; 40(9): 3006 - 3011. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. P. Adams Jr, G. del Zoppo, M. J. Alberts, D. L. Bhatt, L. Brass, A. Furlan, R. L. Grubb, R. T. Higashida, E. C. Jauch, C. Kidwell, et al. Guidelines for the Early Management of Adults With Ischemic Stroke: A Guideline From the American Heart Association/American Stroke Association Stroke Council, Clinical Cardiology Council, Cardiovascular Radiology and Intervention Council, and the Atherosclerotic Peripheral Vascular Disease and Quality of Care Outcomes in Research Interdisciplinary Working Groups: The American Academy of Neurology affirms the value of this guideline as an educational tool for neurologists. Circulation, May 22, 2007; 115(20): e478 - e534. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. P. Adams Jr, G. del Zoppo, M. J. Alberts, D. L. Bhatt, L. Brass, A. Furlan, R. L. Grubb, R. T. Higashida, E. C. Jauch, C. Kidwell, et al. Guidelines for the Early Management of Adults With Ischemic Stroke: A Guideline From the American Heart Association/ American Stroke Association Stroke Council, Clinical Cardiology Council, Cardiovascular Radiology and Intervention Council, and the Atherosclerotic Peripheral Vascular Disease and Quality of Care Outcomes in Research Interdisciplinary Working Groups: The American Academy of Neurology affirms the value of this guideline as an educational tool for neurologists Stroke, May 1, 2007; 38(5): 1655 - 1711. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. K. Schiemanck, G. Kwakkel, M. W. M. Post, and A. J. H. Prevo Predictive Value of Ischemic Lesion Volume Assessed With Magnetic Resonance Imaging for Neurological Deficits and Functional Outcome Poststroke: A Critical Review of the Literature Neurorehabil Neural Repair, December 1, 2006; 20(4): 492 - 502. [Abstract] [PDF]
|
|
|
|
 |
|
 O. Wu, S. Christensen, N. Hjort, R. M. Dijkhuizen, T. Kucinski, J. Fiehler, G. Thomalla, J. Rother, and L. Ostergaard Characterizing physiological heterogeneity of infarction risk in acute human ischaemic stroke using MRI Brain, September 1, 2006; 129(9): 2384 - 2393. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Jacobs, E. H. Herskovits, and H. S. Kim Uterine Fibroids: Diffusion-weighted MR Imaging for Monitoring Therapy with Focused Ultrasound Surgery--Preliminary Study Radiology, July 1, 2005; 236(1): 196 - 203. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. J. Rugg-Gunn, P. A. Boulby, M. R. Symms, G. J. Barker, and J. S. Duncan Whole-brain T2 mapping demonstrates occult abnormalities in focal epilepsy Neurology, January 25, 2005; 64(2): 318 - 325. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. S. Kidwell, J. R. Alger, and J. L. Saver Evolving Paradigms in Neuroimaging of the Ischemic Penumbra Stroke, November 1, 2004; 35(11_suppl_1): 2662 - 2665. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. J. Burton, R. A. Kenny, J. O'Brien, S. Stephens, M. Bradbury, E. Rowan, R. Kalaria, M. Firbank, K. Wesnes, and C. Ballard White Matter Hyperintensities Are Associated With Impairment of Memory, Attention, and Global Cognitive Performance in Older Stroke Patients Stroke, June 1, 2004; 35(6): 1270 - 1275. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. S. Kidwell, J. R. Alger, and J. L. Saver Beyond Mismatch: Evolving Paradigms in Imaging the Ischemic Penumbra With Multimodal Magnetic Resonance Imaging Stroke, November 1, 2003; 34(11): 2729 - 2735. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. C. Shih, J. L. Saver, J. R. Alger, S. Starkman, M. C. Leary, F. Vinuela, G. Duckwiler, Y. P. Gobin, R. Jahan, J. P. Villablanca, et al. Perfusion-Weighted Magnetic Resonance Imaging Thresholds Identifying Core, Irreversibly Infarcted Tissue Stroke, June 1, 2003; 34(6): 1425 - 1430. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Fisher Recommendations for Advancing Development of Acute Stroke Therapies: Stroke Therapy Academic Industry Roundtable 3 Stroke, June 1, 2003; 34(6): 1539 - 1546. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. P. Adams Jr, R. J. Adams, T. Brott, G. J. del Zoppo, A. Furlan, L. B. Goldstein, R. L. Grubb, R. Higashida, C. Kidwell, T. G. Kwiatkowski, et al. Guidelines for the Early Management of Patients With Ischemic Stroke: A Scientific Statement From the Stroke Council of the American Stroke Association Stroke, April 1, 2003; 34(4): 1056 - 1083. [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]
|
|
|
|
|
|
S. Warach Tissue Viability Thresholds in Acute Stroke: The 4-Factor Model Stroke, November 1, 2001; 32(11): 2460 - 2461. [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]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||