Download Comparison of Computed Tomographic and Magnetic Resonance

Comparison of Computed Tomographic and Magnetic
Resonance Perfusion Measurements in Acute
Ischemic Stroke
Back-to-Back Quantitative Analysis
Longting Lin, MMed; Andrew Bivard, PhD; Christopher R. Levi, MD; Mark W. Parsons, MD, PhD
Downloaded from http://stroke.ahajournals.org/ by guest on June 18, 2017
Background and Purpose—Magnetic resonance perfusion (MRP) and computed tomographic perfusion (CTP) are being
increasingly applied in acute stroke trials and clinical practice, yet the comparability of their perfusion values is not well
validated. The aim of this study was to validate the comparability of CTP and MRP measures.
Methods—A 3-step approach was used. Step 1 was a derivation step, where we analyzed 45 patients with acute ischemic
stroke who had both CTP and MRP performed within 2 hours of each other and within 9 hours of stroke onset. In this
step, we derived the optimal perfusion map with the least difference between MRP and CTP. In step 2, the optimal map
was validated on whole-brain perfusion data of 15 patients. Step 3 was to apply the optimal perfusion map to define crossmodality reperfusion from acute CTP to 24-hour MRP in 45 patients and, in turn, to assess how accurately this predicted
3-month clinical outcome.
Results—Among 8 different perfusion maps included in this study, time to peak of the residual function (Tmax) was the
only one with a nonsignificant difference between CTP and MRP in delineating perfusion defects. This was validated
on whole-brain perfusion data, showing high concordance of Tmax between the 2 modalities (concordance correlation
coefficient of Lin, >0.91); the best concordance was at 6 s. At Tmax>6 s threshold, MRP and CTP reached substantial
agreement in mismatch classification (κ >0.61). Cross-modality reperfusion calculated by Tmax>6 s strongly predicted
good functional outcome at 3 months (area under the curve, 0.979; P<0.05).
Conclusions—MRP and CTP can be used interchangeably if one uses Tmax measurement. (Stroke. 2014;45:1727-1732.)
Key Words: perfusion imaging ◼ reperfusion ◼ stroke
M
agnetic resonance perfusion (MRP) imaging has
been used in several past and current clinical trials.1,2
The role of MRP is to triage patients for reperfusion treatment by quantifying the acute ischemic lesion and classifying mismatch pattern. More recently, computed tomographic
perfusion (CTP) has been applied in a similar manner.3,4 In
the interest of timely recruitment, several studies now include
both MRP and CTP in their protocols.5,6 However, in terms
of triaging patients, it is unclear whether the use of the same
criteria is appropriate across the 2 imaging modalities.
In clinical practice, MRP and CTP are also being increasingly
used not only to triage patients for reperfusion therapy but also
to assess its success.7 In many centers, patients with stroke often
receive CTP pretreatment because it is more rapidly accessible.
Then after treatment, MRI (including MRP) is often the preferred
option because it avoids additional radiation and provides maximum pathophysiologic information (eg, stroke mechanism).
Thus, success of reperfusion therapy is assessed by comparing
perfusion change across the 2 modalities. Such assessment of
cross-modality reperfusion seems a common-sense approach for
clinicians, but its accuracy has not been studied.
To triage patients for reperfusion therapy or to assess its
success in the above manner, MRP and CTP are assumed to
be interchangeable. We previously showed that MRP and CTP
were comparable,8 but this was restricted to 1 single perfusion
threshold with limited brain coverage data. The aim of the
current study is to compare the 2 modalities comprehensively
across a broad range of thresholds and on whole-brain data.
Ultimately, we aimed to determine whether MRP and CTP
measurement had strong enough agreement to be used interchangeably in triaging patients for acute reperfusion therapy
and in assessing subsequent reperfusion.
Methods
Study Design
The study comprised 3 steps. Step 1 was a derivation step, where a
wide range of perfusion parameters and thresholds were analyzed to
identify the perfusion measure with the least volumetric difference
Received March 10, 2014; final revision received April 12, 2014; accepted April 14, 2014.
From the Stroke Research Program, Hunter Medical Research Institute (L.L., A.B., C.R.L., M.W.P.) and Department of Neurology, John Hunter Hospital
(C.R.L., M.W.P.), The University of Newcastle, Newcastle, New South Wales, Australia.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.
114.005419/-/DC1.
Correspondence to Mark W. Parsons, MD, PhD, Department of Neurology, John Hunter Hospital, The University of Newcastle, Locked Bag No 1,
Hunter Region Mail Centre, Newcastle, New South Wales 2310, Australia. E-mail [email protected]
© 2014 American Heart Association, Inc.
Stroke is available at http://stroke.ahajournals.org
DOI: 10.1161/STROKEAHA.114.005419
1727
1728 Stroke June 2014
between CTP and MRP. Steps 2 and 3 were validation steps on
whole-brain data. Step 2 tested the agreement of acute CTP and acute
MRP in detecting mismatch pattern to select patients for treatment.
Step 3 tested the accuracy of reperfusion from acute CTP to 24-hour
MRP in predicting clinical outcome.
Patients
Downloaded from http://stroke.ahajournals.org/ by guest on June 18, 2017
For this study, we retrospectively included patients with acute hemispheric ischemic stroke presenting to John Hunter Hospital between
2005 and 2013 who were imaged within 9 hours of symptom onset and who had an acute symptomatic intracranial occlusion on
CT angiography. The sites of occlusion were intracranial internal
carotid artery, middle cerebral artery, posterior cerebral artery, or
anterior cerebral artery. For the step 1 and 2 analyses, we used data
from patients who had CTP and MRP performed within 2 hours of
each other and both within 9 hours of stroke onset. For step 1, we
restricted patient data to those presenting between 2005 and 2010,
when limited-slice CTP was performed. For step 2 analysis, we restricted patient data to those presenting between 2010 and 2013,
when whole-brain CTP was performed (in addition to MRP within
2 hours of CTP and both within 9 hours of stroke onset). For the step
3 analysis, we used patients imaged with whole-brain CTP within 9
hours of stroke onset between 2010 and 2013, but in addition, they
had to have had follow-up MRP performed at 24 to 48 hours after
CTP. The data collection and analysis were approved by the institutional ethics committee.
Perfusion Image Acquisition
CTP images were acquired on 2 scanners. The first was 16-slice
Philips scanner (Philips Mx8000; Philips, Cleveland, OH). To obtain
reasonable brain coverage, 2 adjacent series were performed, each
with slice coverage of 24 mm. The second scanner was a 320-slice
Toshiba (Toshiba Aquilion ONE; Toshiba, Tokyo, Japan). This scanner has an axial coverage of 160 mm, obtaining whole-brain coverage
with a single acquisition. MRP was performed on a 1.5-Tesla scanner
(Siemens Avanto, Erlangen, Germany) with a dynamic susceptibility
contrast technique and gradient-echo echo-planar imaging technique.
Additional image acquisition details are provided in the online-only
Data Supplement.
Perfusion Image Postprocessing
CTP and MRP data were processed by the same commercial software
MIStar (Apollo Medical Imaging Technology, Melbourne, Victoria,
Australia). With this software, various perfusion measures can be
generated by deconvolution models.9,10 The measures include cerebral
blood flow (CBF), cerebral blood volume (CBV), mean transit time
(MTT), time to peak of the residual function (Tmax), and delay time to
peak of the residual function (DT).
In this study, for each imaging modality, 2 common deconvolution
models were applied. Those were delay- and dispersion-corrected
singular value decomposition (SVD)11,12 and standard SVD. From delay- and dispersion-corrected SVD, we generated 4 perfusion maps
labeled as CBF1, CBV1, MTT1, and DT; from standard SVD, we
generated another set of 4 parametric maps labeled as CBF2, CBV2,
MTT2, and Tmax (Figure I in the online-only Data Supplement).
Imaging and Statistical Analysis
Step 1: Comparison of MRP and CTP on 2 PerfusionMeasurement Levels
By setting perfusion thresholds, regions with hypoperfusion were
delineated on acute CTP and acute MRP separately. Hypoperfusion
refers to the ischemic area with low CBF/CBV and prolonged
MTT/DT/Tmax. To include most of the definitions for ischemia, a
wide range of thresholds was covered in this study (Figure I in
the online-only Data Supplement). For CBF1, CBF2, CBV1, and
CBV2, the threshold ranged from 95% to 5%, with 5% decrements;
for MTT1 and MTT2, the threshold range was from 105% to 195%,
with 5% increment; for DT and Tmax, it was from 0.5 to 9.5 s, with
increment of 0.5 s. The threshold levels were defined relative to the
mean tissue perfusion value of the unaffected hemisphere.
Then, difference of the 2 imaging modalities (MRP–CTP) was
calculated in terms of hypoperfusion volume measured by the same
threshold setting. To compare the result of the 2 imaging modalities
accurately, MRP was coregistered to CTP with a 3-dimensional orientation to adjust for anatomic variation between the 2 modalities.
One slice (12-mm thickness, at the level of the largest perfusion defect) was chosen to calculate the volumetric difference between the
CTP and MRP lesions.
For statistical analysis, a multilevel random-effects model was
used. In the model, the volumetric difference was treated as the dependent variable; the perfusion map was set as the level 1 random
effect and the threshold as the level 2 random effect. The model was
estimated by maximum-likelihood estimation.
Step 2: Intermodality Agreement
In this step, agreement of acute MRP and CTP was validated on wholebrain data. The cross-modality agreement included 2 aspects: (1) measuring the acute perfusion lesion and (2) classifying mismatch pattern.
To calculate ischemic lesion volume, the perfusion measure with the
least cross-modality difference (from step 1) was applied. To classify
mismatch pattern, 2 criteria were used. The first was the EXtending the
time for Thrombolysis in Emergency Neurological Deficits (EXTEND)
trial criteria6: mismatch ratio >1.2, mismatch volume >10 mL, and infarct core <70 mL; the second was the Diffusion and Perfusion Imaging
Evaluation for Understanding Stroke Evolution Study 2 (DEFUSE 2)
trial criteria13: mismatch ratio >1.8, mismatch volume >15 mL, infarct
core <70 mL, and Tmax 10 s <100 mL. Mismatch ratio=ischemic lesion volume/infarct core volume; mismatch volume=ischemic lesion
volume–infarct core volume; infarct core was defined by CBF <30% of
the contralateral hemisphere on CTP14 and by the diffusion-weighted
imaging lesion on MRI.15 Statistically, agreement of MRP and CTP in
measuring the acute perfusion lesion was quantified by concordance
correlation coefficient (CCC) of Lin. In terms of classifying patients
with mismatch pattern, the agreement was quantified by κ coefficient.
Step 3: Cross-Modality Reperfusion Index
Reperfusion was assessed by the change of perfusion status from
acute CTP to 24-hour MRP. To quantify the cross-modality reperfusion, a reperfusion index was introduced: reperfusion index=(acute
CTPvol–24-hour MRPvol)/acute CTPvol. CTPvol and MRPvol were the
perfusion volumes calculated by the optimal cross-modality measurement derived from step 1 (ie, the measurement with the least difference between CTP and MRP).
The cross-modality reperfusion index was then used as the independent variable in logistic regression to predict good clinical outcome at 90 days (modified Rankin Scale, 0–2). Afterward, receiver
operating characteristic analysis was performed to examine how well
cross-modality reperfusion predicted clinical outcome. All statistical
analyses were done using STATA 11.0, and a significance level of
0.05 was set.
Results
Patients
For step 1, 57 patients were eligible. However, 7 cases were
excluded because of severe motion artifact of perfusion
images, and 5 cases were excluded because of vessel recanalization between CT and MR. For step 2, 22 patients were
included, of which 7 patients were excluded for the above reasons. For step 3, 47 patients were eligible, but 2 were excluded
because of substantial motion artifact of perfusion images.
In summary, image data of 45 patients were used in step 1,
15 patients were used in step 2, and 45 cases were used in step
3. Demographics of patients are shown in Table 1.
Lin et al MRP vs CTP 1729
Table 1. Patient Demographics
Patients in Step 1 Analysis
Patients in Step 2 Analysis
Median
Median
IQR
IQR
Patients in Step 3 Analysis
Median
IQR
Age, y
75
18
70
16.5
71
17
Baseline NIHSS
17
5
14
7
12
6
Time to CTP, h
3.13
0.4
3.33
3.54
Time to MRP, h
3.67
0.77
4.33
2.58
29.65
3
12.83
2.08
CTP to MRP, h
0.5
0.17
1.23
0.72
26.5
13.17
CTP indicates computed tomographic perfusion; IQR, interquartile range; MRP, magnetic resonance perfusion; and NIHSS, National
Institutes of Health Stroke Scale.
Downloaded from http://stroke.ahajournals.org/ by guest on June 18, 2017
Step 1: Deriving the Map With the Least Difference
Between CTP and MRP
For each imaging modality, 6840 perfusion regions were
derived. Results of the random-effects model showed that
there was a significant variance between perfusion maps (mean
difference of MRP and CTP=1.11 mL, SD among perfusion
maps=3.84 mL with 95% confidence interval, 2.34–6.28).
However, within each parameter, the variance between thresholds was small (SD among thresholds=0.24 mL, with 95%
confidence interval, 0.26–1.07). Thus, the relationship of CTP
and MRP varied significantly between perfusion parameters
but was consistent among thresholds within each parameter.
We then further assessed the exact difference between CTP
and MRP for each respective perfusion map (Table 2). For
CBF and CBV, MRP volume was significantly smaller than
the respective CTP volume (P<0.01); for MTT and DT, MRP
volume was significantly bigger than its CTP comparator
(P<0.01). The only perfusion map with no significant volume
difference between the 2 modalities was Tmax (P=0.1333). Tmax
was, therefore, identified as the perfusion parameter to take
forward to the next 2 steps.
Step 2: Intermodality Agreement of Measuring Ischemic
Lesion With Tmax
From Tmax, 3 definitions were chosen to define the ischemic
lesion: Tmax>2 s, Tmax>4 s, and Tmax>6 s.16–18 For each threshold,
whole-brain lesion volume was extremely close between MRP
and CTP (Figures 1 and 2; Table I in the online-only Data
Table 2. Difference of Hypoperfusion Regions Between MRP
and CTP on Each Map
MRP–CTP
Difference, mL
95% Confidence
Interval
P Value
CBF1
–2.22
–2.82 to –1.66
<0.001
CBF2
–2.30
–3.05 to –1.60
<0.001
CBV1
–1.75
–2.18 to –1.34
<0.001
CBV2
–2.11
–2.51 to –1.70
<0.001
MTT1
7.80
7.03 to 8.65
<0.001
MTT2
6.41
5.60 to 7.27
<0.001
DT
2.81
2.33 to 3.28
<0.001
Tmax
0.67
–0.20 to 1.54
Perfusion Maps
0.133*
CBF indicates cerebral blood flow; CBV, cerebral blood volume; CTP, computed
tomographic perfusion; DT, delay time to peak of the residual function; MRP,
magnetic resonance perfusion; MTT, mean transit time; and Tmax, time to peak
of the residual function.
*Difference of MRP and CTP is nonsignificant on Tmax map.
Supplement). The definition with best cross-modality concordance was Tmax>6 s, with CCC of 0.93 (P<0.01). Setting the
threshold of Tmax>6 s, CTP resulted in a mean acute perfusion
lesion volume of 102.61±78.3 mL and MRP of 102.91±75.81
mL (P>0.05).
Tmax>6 s was then used as the definition for the extent of the
ischemic lesion for both imaging modalities to classify mismatch. In selection of patients with mismatch, CTP and MRP
reached substantial agreement (κ >0.61).19 With the EXTEND
criteria, 3 of 15 patients differed in classification across
modality (κ=0.613; P<0.01); with DEFUSE, there were 2 of
15 misclassified (κ=0.746; P<0.01). More details could be
found in Table II in the online-only Data Supplement.
Step 3: Cross-Modality Reperfusion Index With Tmax
Measurement
Increasing cross-modality reperfusion from acute CTP to
24-hour MRP at the Tmax>6 s threshold was associated with
improving clinical outcome (Figure 3A; Table III in the onlineonly Data Supplement). With each 1% increase in reperfusion,
the odds of good clinical outcome (modified Rankin Scale,
0–2) increased by 1.1 (95% confidence interval, 1.04–1.16;
P<0.05). More importantly, receiver operating characteristic
results (Figure 3B) suggested that the prediction of good clinical outcome by reperfusion was highly sensitive and specific
with an area under the curve value of 0.98 (95% confidence
interval, 0.95–1). A cross-modality Tmax reperfusion index
>77% predicted a good clinical outcome with 94.12% sensitivity and 93.33% specificity.
Discussion
Overall, the difference in acute perfusion lesions between
MRP and CTP modalities was not dramatic. However, we
did observe some variance across different perfusion maps.
Among the 8 different perfusion maps assessed in this study,
Tmax was clearly the best performer, with minimal difference
between MRP and CTP.
The finding adds evidence to current data that Tmax has superior performance to other perfusion maps in stroke. Previous
MRP studies16,17 showed that Tmax=4 to 6 s was the most accurate definition of the acute perfusion lesion. The MRP–Tmax
definition was also validated by showing the highest correlation to the ischemic reference on positron emission tomography (CBF <20 mL/100 g per minute), and notably Tmax was
more accurate than MRP–CBF.20 The current study found that
Tmax=4 to 6 s measurement had good agreement between MRP
1730 Stroke June 2014
Downloaded from http://stroke.ahajournals.org/ by guest on June 18, 2017
Figure 1. Concordance of the acute ischemic lesion between magnetic resonance perfusion (MRP) and computed tomographic perfusion
(CTP). A, Distribution of whole-brain lesion volume measured by time to peak of the residual function (Tmax) >2 s, >4 s, and >6 s separately. B, Concordance of MRP and CTP with Tmax>2 s measurement (CCC=0.89; slope of reduced major axis, 0.97). C, Concordance of
MRP and CTP with Tmax>4 s measurement (CCC=0.91; slope of reduced major axis, 0.97). D, Concordance of MRP and CTP with Tmax>6 s
measurement (CCC=0.93; slope of reduced major axis, 1.03). For all measurements, lesion distribution is similar between MRP and CTP,
and slope of reduced major axis is close to 1 (line of perfect concordance).
and CTP (CCC >0.91) and was superior to other perfusion
maps and thresholds. This finding allows more confidence in
generalizing the ischemic definition from MRP to CTP and
is a step forward in the quest standardizing perfusion lesion
assessment in acute stroke.
Our results have major significance for the application of
perfusion imaging in current and future acute stroke trials,
particularly in multicenter trials, where both modalities are
used for patient triage.5,6 There has been concern in past trials that there existed inconsistency in patient selection across
the 2 modalities.5 In the current study, we have shown that
there is a strong agreement (CCC >0.61) between MRP and
CTP in patient selection based on mismatch pattern, provided
our validated ischemic lesion thresholds are used. These are
Tmax>6 s/CBF <30% for penumbra/core mismatch on CTP
and Tmax>6 s/diffusion-weighted imaging lesion for the mismatch on MRP. This will maximize the likelihood that the
same treatment decision is made by MRP and by CTP.
For the first time, our study has validated the cross-modality
assessment of reperfusion. Reperfusion is a potent predictor
of early recovery and later clinical outcomes after stroke21
and also a powerful biological marker of acute treatment efficacy.2,4,6 Growing evidence22,23 suggests that reperfusion is a
stronger predictor of patient outcome than recanalization. Past
validation of reperfusion has been limited to the use of single
imaging modality. That is, it was calculated either by acute
CTP and follow-up CTP or by acute MRP and follow-up MRP.
This omits a common clinical scenario, whereby reperfusion
success is examined after acute reperfusion therapy (particularly intravenous tissue-type plasminogen activator), using
pretreatment CTP, and post-treatment MRP is performed to
minimize radiation dose. We found that cross-modality reperfusion, with Tmax>6 s measurement, had extremely high sensitivity and specificity in predicting good 3-month outcome.
This agrees with a recent CTP-only study23 that Tmax measurement of reperfusion, when compared with other perfusion
measurements, best predicted clinical outcome after stroke.
In summary, MRP and CTP are interchangeable if Tmax is
used as the perfusion threshold. We use the term interchangeable because MRP and CTP reached a high degree of agreement (CCC >0.9 or κ >0.61). This does not mean that they
are in perfect concordance because their technical differences
Lin et al MRP vs CTP 1731
Downloaded from http://stroke.ahajournals.org/ by guest on June 18, 2017
Figure 2. A case with good agreement of acute ischemic lesion
between magnetic resonance perfusion (MRP) and computed
tomographic perfusion (CTP). An 85-year-old man presented with
left middle cerebral artery syndrome (National Institutes of Health
Stroke Scale, 11). CTP and MRP were performed successively,
190 and 231 minutes after stroke onset. With time to peak of the
residual function (Tmax) >2 s setting, CTP and MRP detect the
ischemic lesion volume of 187.41 and 179.86 mL separately; with
Tmax>4 s setting, the volumes are 130.51 and 134.66 mL separately; with Tmax>6 s setting, the volumes are 92.45 and 100.37 mL
separately.
are unavoidable.24 Furthermore, the interchangeability of Tmax
measurement on MRP and CTP is contingent on the following 2 conditions being met. First, the same software is used
to process MRP and CTP. This does not necessarily mean
that it must be the software used in this study (MIStar). In
our previous study,8 an in-house software also resulted in a
good cross-modality agreement on Tmax>6 s. Second, the
Tmax map should be generated by standard SVD. Although
delay-corrected SVD has been reported to generate accurate
perfusion values to define infarct core and penumbra,25,26 the
correction might lead to different results on MRP versus CTP.
In our study, and in previous reports,27,28 Tmax values changed
significantly on CTP, but not on MRP, with delay correction.
The differential effect of delay correlation on the 2 imaging
modalities explains why delay time (delayed Tmax) had slightly
worse cross-modality agreement in this study than did nondelay corrected Tmax.
This is not the first study comparing MRP and concurrent CTP. However, previous studies were limited with only
1-threshold comparison29,30 and on limited slice coverage.8,31–35
Comparatively, our current study has 2 major advantages.
First, in the derivation step, we systematically compared the
2 imaging modalities with multilevel perfusion information,
including 19 threshold levels and 8 parameter levels. Second,
we further validated our findings in a new data set using wholebrain perfusion information of both CTP and MRP. This is the
first time such comparison has been done on a whole-brain
spatial level.
There are limitations in this study. First, MRP and CTP are
assumed to be concurrent, but there was a short interval time
between the 2. Changes might have happened between modalities. We excluded patients with any evidence of vessel recanalization between modalities, but we might have included some
with partial reperfusion or infarct growth between modalities.
Notably, these changes would underestimate the agreement
between MRP and CTP. Second, we were limited by relatively
small patient numbers with whole-brain CTP data because the
technology is relatively new. The small sample size in step 2
may, of course, have limited our ability to find better agreement between the modalities. Third, patients with stroke had
exclusively hemispheric ischemia with relatively large lesion
volumes. Additional study is needed in a broader range of
patients with stroke.
Conclusions
This study provides evidence for researchers and clinicians
that MRP and CTP can be used interchangeably if one uses
Figure 3. Cross-modality reperfusion and clinical outcome. Reperfusion index is calculated by the change of time to peak of the residual
function (Tmax) abnormality from acute computed tomographic perfusion (CTP) to follow-up magnetic resonance perfusion (MRP). In the
scatter plot of modified Rankin Scale (mRS) and the reperfusion index (A), we observe a trend that mRS decreases steadily as reperfusion
increases from 0% to 100%. Difference of the reperfusion index is significant between bad clinical outcome (mRS, 3–6) and good clinical
outcome (mRS, 0–2). For receiver operator characteristic (ROC) curve of the reperfusion index in prediction good clinical outcome (B),
area under curve is high (AUC, 0.9790), which is close to perfect prediction (AUC, 1). The cut-off point of reperfusion index with best prediction is 77%, which has the value of 0.9412 for sensitivity and 0.0667 for 1-specificity.
1732 Stroke June 2014
Tmax measurement. It also adds considerably to the evidence
that CTP, with whole-brain coverage, is comparable with MRI
in acute stroke imaging.
Sources of Funding
The study was funded by a National Health and Medical Research
Council partnership project grant (ID: 1013719).
Disclosures
Dr Parsons has research support from National Health and Medical
Research Council (NHMRC) Partnership Project Grant, ID: 1013719.
References
Downloaded from http://stroke.ahajournals.org/ by guest on June 18, 2017
1. Albers GW, Thijs VN, Wechsler L, Kemp S, Schlaug G, Skalabrin E, et
al. Magnetic resonance imaging profiles predict clinical response to early
reperfusion: The diffusion and perfusion imaging evaluation for understanding stroke evolution (DEFUSE) study. Ann Neurol. 2006;60:508–517.
2. Davis SM, Donnan GA, Parsons MW, Levi C, Butcher KS, Peeters A, et
al; EPITHET Investigators. Effects of alteplase beyond 3 h after stroke
in the Echoplanar Imaging Thrombolytic Evaluation Trial (EPITHET):
a placebo-controlled randomised trial. Lancet Neurol. 2008;7:299–309.
3. Parsons MW, Miteff F, Bateman GA, Spratt N, Loiselle A, Attia J, et
al. Acute ischemic stroke: imaging-guided tenecteplase treatment in an
extended time window. Neurology. 2009;72:915–921.
4. Parsons M, Spratt N, Bivard A, Campbell B, Chung K, Miteff F, et al.
A randomized trial of tenecteplase versus alteplase for acute ischemic
stroke. N Engl J Med. 2012;366:1099–1107.
5. Hacke W, Furlan AJ, Al-Rawi Y, Davalos A, Fiebach JB, Gruber F, et
al. Intravenous desmoteplase in patients with acute ischaemic stroke
selected by MRI perfusion-diffusion weighted imaging or perfusion CT
(DIAS-2): a prospective, randomised, double-blind, placebo-controlled
study. Lancet Neurol. 2009;8:141–150.
6. Ma H, Parsons MW, Christensen S, Campbell BC, Churilov L, Connelly
A, et al; EXTEND Investigators. A multicentre, randomized, doubleblinded, placebo-controlled Phase III study to investigate EXtending the
time for Thrombolysis in Emergency Neurological Deficits (EXTEND).
Int J Stroke. 2012;7:74–80.
7. Campbell BC, Tu HT, Christensen S, Desmond PM, Levi CR, Bladin CF,
et al. Assessing response to stroke thrombolysis: validation of 24-hour
multimodal magnetic resonance imaging. Arch Neurol. 2012;69:46–50.
8. Campbell BC, Christensen S, Levi CR, Desmond PM, Donnan GA,
Davis SM, et al. Comparison of computed tomography perfusion and
magnetic resonance imaging perfusion-diffusion mismatch in ischemic
stroke. Stroke. 2012;43:2648–2653.
9. Bivard A, Spratt N, Levi C, Parsons M. Perfusion computer tomography: imaging and clinical validation in acute ischaemic stroke. Brain.
2011;134(pt 11):3408–3416.
10. Bivard A, Levi C, Spratt N, Parsons M. Perfusion CT in acute stroke: a comprehensive analysis of infarct and penumbra. Radiology. 2013;267:543–550.
11. Yang Q. Method and system for obtaining improved data perfusion
measurements. PCT international application PCT/AU2004/000821.
Published November 10, 2005.
12. Yang Q. Method and system for mapping tissue status of acute stroke.
PCT international application PCT/IB2009/007446. Published May 20,
2010.
13. Lansberg MG, Straka M, Kemp S, Mlynash M, Wechsler LR, Jovin
TG, et al; DEFUSE 2 Study Investigators. MRI profile and response to
endovascular reperfusion after stroke (DEFUSE 2): a prospective cohort
study. Lancet Neurol. 2012;11:860–867.
14. Campbell BC, Christensen S, Levi CR, Desmond PM, Donnan GA,
Davis SM, et al. Cerebral blood flow is the optimal CT perfusion parameter for assessing infarct core. Stroke. 2011;42:3435–3440.
15. Campbell BC, Purushotham A, Christensen S, Desmond PM, Nagakane
Y, Parsons MW, et al; EPITHET–DEFUSE Investigators. The infarct
core is well represented by the acute diffusion lesion: sustained reversal
is infrequent. J Cereb Blood Flow Metab. 2012;32:50–56.
16. Olivot JM, Mlynash M, Zaharchuk G, Straka M, Bammer R, Schwartz
N, et al. Perfusion MRI (Tmax and MTT) correlation with xenon CT
cerebral blood flow in stroke patients. Neurology. 2009;72:1140–1145.
17. Olivot JM, Mlynash M, Thijs VN, Kemp S, Lansberg MG, Wechsler L,
et al. Optimal Tmax threshold for predicting penumbral tissue in acute
stroke. Stroke. 2009;40:469–475.
18. Takasawa M, Jones PS, Guadagno JV, Christensen S, Fryer TD, Harding
S, et al. How reliable is perfusion MR in acute stroke? Validation and
determination of the penumbra threshold against quantitative PET.
Stroke. 2008;39:870–877.
19. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174.
20. Zaro-Weber O, Moeller-Hartmann W, Heiss WD, Sobesky J. Maps of
time to maximum and time to peak for mismatch definition in clinical
stroke studies validated with positron emission tomography. Stroke.
2010;41:2817–2821.
21. Liebeskind DS. Recanalization and reperfusion in acute ischemic stroke.
F1000 Med Rep. 2010;2:71.
22. Soares BP, Tong E, Hom J, Cheng SC, Bredno J, Boussel L, et al.
Reperfusion is a more accurate predictor of follow-up infarct volume
than recanalization: a proof of concept using CT in acute ischemic stroke
patients. Stroke. 2010;41:e34–e40.
23. Eilaghi A, Brooks J, d’Esterre C, Zhang L, Swartz RH, Lee TY, et al.
Reperfusion is a stronger predictor of good clinical outcome than recanalization in ischemic stroke. Radiology. 2013;269:240–248.
24. Konstas AA, Goldmakher GV, Lee TY, Lev MH. Theoretic basis and
technical implementations of CT perfusion in acute ischemic stroke, part
1: theoretic basis. AJNR Am J Neuroradiol. 2009;30:662–668.
25. Calamante F, Gadian DG, Connelly A. Delay and dispersion effects in
dynamic susceptibility contrast MRI: simulations using singular value
decomposition. Magn Reson Med. 2000;44:466–473.
26. Ibaraki M, Shimosegawa E, Toyoshima H, Takahashi K, Miura S, Kanno I.
Tracer delay correction of cerebral blood flow with dynamic susceptibility
contrast-enhanced MRI. J Cereb Blood Flow Metab. 2005;25:378–390.
27. Kudo K, Christensen S, Sasaki M, Østergaard L, Shirato H, Ogasawara
K, et al; Stroke Imaging Repository (STIR) Investigators. Accuracy and
reliability assessment of CT and MR perfusion analysis software using a
digital phantom. Radiology. 2013;267:201–211.
28. Kudo K, Sasaki M, Østergaard L, Christensen S, Uwano I, Suzuki M, et
al. Susceptibility of Tmax to tracer delay on perfusion analysis: quantitative evaluation of various deconvolution algorithms using digital phantoms. J Cereb Blood Flow Metab. 2011;31:908–912.
29. Schaefer PW, Barak ER, Kamalian S, Gharai LR, Schwamm L, Gonzalez
RG, et al. Quantitative assessment of core/penumbra mismatch in acute
stroke: CT and MR perfusion imaging are strongly correlated when sufficient brain volume is imaged. Stroke. 2008;39:2986–2992.
30. Schramm P, Schellinger PD, Klotz E, Kallenberg K, Fiebach JB, Külkens
S, et al. Comparison of perfusion computed tomography and computed
tomography angiography source images with perfusion-weighted imaging and diffusion-weighted imaging in patients with acute stroke of less
than 6 hours’ duration. Stroke. 2004;35:1652–1658.
31. Wintermark M, Meuli R, Browaeys P, Reichhart M, Bogousslavsky J,
Schnyder P, et al. Comparison of CT perfusion and angiography and MRI in
selecting stroke patients for acute treatment. Neurology. 2007;68:694–697.
32.Wintermark M, Reichhart M, Cuisenaire O, Maeder P, Thiran JP,
Schnyder P, et al. Comparison of admission perfusion computed tomography and qualitative diffusion- and perfusion-weighted magnetic resonance imaging in acute stroke patients. Stroke. 2002;33:2025–2031.
33. Eastwood JD, Lev MH, Wintermark M, Fitzek C, Barboriak DP, Delong
DM, et al. Correlation of early dynamic CT perfusion imaging with
whole-brain MR diffusion and perfusion imaging in acute hemispheric
stroke. AJNR Am J Neuroradiol. 2003;24:1869–1875.
34. Ziegelitz D, Starck G, Mikkelsen IK, Tullberg M, Edsbagge M, Wikkelsö
C, et al. Absolute quantification of cerebral blood flow in neurologically
normal volunteers: dynamic-susceptibility contrast MRI-perfusion compared with computed tomography (CT)-perfusion. Magn Reson Med.
2009;62:56–65.
35. Alawneh JA, Jones PS, Mikkelsen IK, Cho TH, Siemonsen S, Mouridsen
K, et al. Infarction of ‘non-core-non-penumbral’ tissue after stroke: multivariate modelling of clinical impact. Brain. 2011;134(pt 6):1765–1776.
Comparison of Computed Tomographic and Magnetic Resonance Perfusion
Measurements in Acute Ischemic Stroke: Back-to-Back Quantitative Analysis
Longting Lin, Andrew Bivard, Christopher R. Levi and Mark W. Parsons
Downloaded from http://stroke.ahajournals.org/ by guest on June 18, 2017
Stroke. 2014;45:1727-1732; originally published online May 13, 2014;
doi: 10.1161/STROKEAHA.114.005419
Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 2014 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/45/6/1727
Data Supplement (unedited) at:
http://stroke.ahajournals.org/content/suppl/2014/05/13/STROKEAHA.114.005419.DC1
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click
Request Permissions in the middle column of the Web page under Services. Further information about this
process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Stroke is online at:
http://stroke.ahajournals.org//subscriptions/
SUPPLEMENTAL MATERIAL
Supplemental Methods
Patient
Patient data were retrospectively analysed in four steps. Firstly, we identified from the stroke
patient database of John Hunter Hospital from 2005 to 2013, and included patients with
multi-modal CT (NCCT+CTA+CTP) done within 9 hour of stroke onset. All patients had an
acute neurological deficit consistent with ischaemic stroke, and received intravenous t-PA if
they were clinically eligible (otherwise they received standard care). From this group we
then selected patients with an acute symptomatic intracranial occlusion on CTA. The sites of
occlusion were intracranial internal carotid artery (ICA), middle cerebral artery (MCA),
posterior cerebral artery (PCA) or anterior cerebral artery (ACA). From this group, for the
analysis of cross-modality agreement, we selected patients who had acute MR
(DWI+MRP+MRA) done within two hours of CTP. Patients with evidence of any
recanalization between acute CTA and acute MRA were excluded from this analysis. For the
analysis of cross-modality reperfusion, we selected patients who had MR performed 24-48
hours after CTP.
CTP acquisition
For patients admitted before 2010, CTP images were acquired on 16-slice Philips scanner
(Philips Mx8000; Philips, Cleveland, OH, USA). The image scanning commenced 4 seconds
after intravenous injection (40 ml, injected at 5ml/s) of non-ionic iodinated contrast (Ultravist
370; Bayer HealthCare, Berlin, Germany). It lasted for 60 seconds, with acquisition of one
image per second per slice. The acquisition parameters were 90 kV, 170 mA, Field of View
(FOV) 210×210mm, and Matrix 512×512. The 60-second series had brain coverage of 24mm,
which consisted two adjacent 12mm axial slices.
For patients admitted from 2010 to 2013, CTP images were derived from 320-slice Toshiba
scanner (Toshiba Aquilion ONE; Toshiba, Tokyo, Japan).The scanner had an axial coverage
of 160mm, generating 320 slices with the thickness of 0.5mm by one gantry rotation. Image
scanning started 7 seconds after intravenous injection (40 ml, injected at 6ml/s) of non-ionic
iodinated contrast (Ultravist 370; Bayer HealthCare, Berlin, Germany). It lasted for 60
seconds, acquiring 19 images per slice. The whole acquisition process was divided into three
phases: phase one, 1 image was generated (80kV, 310mA) as a mask for all subsequent
volumes; phase two, 13 images were generated at rate of one image per second (80kV,
150/300mA); phase three, 5 images were generated at rate of one image per five seconds
(80kV, 150mA). For all three phases, FOV was 220×220mm, and Maxtrix was 512×512.
MRP acquisition
MRP was performed on a 1.5-Tesla scanner (Siemens Avanto, Erlangen, Germany). It was
using dynamic susceptibility contrast technique and gradient-echo echo-planar imaging
technique (GE-EPI). Following a bolus of gadolinium contrast (Magnevist; Bayer HealthCare,
Berlin, Germany) into the antecubital vein (0.2mmol/kg, injected at speed of 5ml/s),
perfusion images were obtained with acquisition parameters as follows: TR 1400 ms, TE 30
ms, Flip 90, FOV 230×230mm, and Matrix 128×128. The scanning lasted for 60 seconds,
resulting in 40 images per slice. A total 19 slices were obtained with the thickness of 5mm,
and with a slice gap of 1.5mm.
1 Image post-processing
MRP and CTP data were processed by the same commercial software MIStar (Apollo
Medical Imaging Technology, Melbourne, VIC., Australia). Details of MIStar can be found
in the following links http://www.apollomit.com/. In summary, the software runs on all sorts
of consoles from different manufacturers, and provides a broad choice of processing
algorithm.
In this study, same post-processing protocol was used for MRP and CTP to improve the
comparability of the two modalities. To be more specific: (1) the same deconvolution model
was set to generate perfusion maps; (2) the anterior cerebral artery was selected as the arterial
input function (AIF) for both modalities; and (3) the same approaches were applied to remove
noise from perfusion maps. Regarding noise removal two approaches were taken; firstly, the
software automatically detected voxels without clear contrast peak and these voxels were
allocated with zero perfusion value of CBF and maximum perfusion value of MTT/Tmax/DT;
secondly, Gaussian smoothing was applied manually (sigma=0.5, kernel size =3×3).
Noticeably, there are still technical differences between the two modalities that cannot be
altered, such as the process of converting the time-intensity curve to a time-contrast curve. In
this process, different mathematical models were applied, since CT detects tissue density
which has a linear relationship with contrast concentration, whereas MR signal- intensity
relationship with contrast concentration is nonlinear.
Supplemental Figures
Supplementary Figure I. Data structure of perfusion modality. Each imaging modality is postprocessed by two mathematical models, delayed SVD (dSVD) and standard SVD (sSVD),
resulting in 8 perfusion maps. For each map, 19 thresholds are set to derive corresponding
perfusion defect regions. Comparison of MRP and CTP is based on perfusion region of the
same threshold of the same map.
2 Supplemental Tables
Supplementary Table I. Agreement of MRP and CTP on measuring whole-brain lesion with
Tmax
Tmax>2s
Tmax> 4s
Tmax >6s
MRP volume
(ml)
Mean ± SD
175.58 ± 112
136.69 ± 95.29
102.91 ± 75.81
CTP volume
(ml)
Mean ± SD
171.09 ± 108.61
131.91 ± 92.37
102.61 ± 78.3
Concordance correlation coefficient
(CCC)
Estimation 95% confidence interval
0.89
0.78
1.00
0.91
0.82
1.00
0.93
0.86
1.00
Supplementary Table II. Patient classification by mismatch pattern
EXTEND criteria
MR classification
Mismatch pattern (-)
Mismatch pattern (+)
MR classification
Mismatch pattern (-)
Mismatch pattern (+)
CTP classification
Mismatch pattern (-)
Mismatch pattern (+)
5
2
DEFFUSE-2 criteria
1
8
CTP classification
Mismatch pattern (-)
Mismatch pattern (+)
6
1
1
8
Supplementary Table III . Cross-modality reperfusion and clinical outcome
Modified Rankin Score
(mRS)
0
1
2
3
4
5
6
Patient number (N)
11
10
7
4
4
5
4
3 Reperfusion index (Median
±IQR)
100% ± 9.82%
100% ± 1.89%
95.27% ± 26.9%
71.29% ± 41.23%
28.58% ± 20.75%
26.86% ± 11.45%
0% ± 7.36%