AJR 2003; 180:3-12
© American Roentgen Ray Society
Perfusion CT with Iodinated Contrast Material
James D. Eastwood1,
Michael H. Lev2 and
James M. Provenzale1
1 Department of Radiology, Division of Neuroradiology, Duke University Medical
Center, Box 3808, Durham, NC, 27710-3808.
2 Department of Radiology, Division of Neuroradiology, Massachusetts General
Hospital, 55 Fruit St., Boston, MA 02114.
Received March 19, 2002;
accepted after revision June 24, 2002.
Address correspondence to J. D. Eastwood.
Introduction
The role of neuroimaging in the evaluation of acute stroke has changed
dramatically in the past decade. Previously, neuroimaging has been used in
this setting to provide anatomic imaging that indicated the presence or
absence of findings of acute cerebral ischemia and excluded lesions that
produce symptoms or signs that can mimic those of stroke, such as hemorrhage
and neoplasms. More recently, the introduction of an effective therapy for
acute stroke (recombinant tissue plasminogen activator) has changed the goals
of neuroimaging from providing solely anatomic information to providing
physiologic information that could help to determine which patients might
benefit from therapy. In particular, emphasis has increasingly been placed on
the differentiation of tissue (infarct core) that has already, irreversibly,
undergone infarction from ischemic tissue at immediate risk of infarction
(ischemic penumbra) in the absence of therapy.
Presently, IV-administered recombinant tissue plasminogen activator is the
only agent that has received approval from the United States Food and Drug
Administration for treatment of acute stroke. Studies have shown that this
therapy must be administered within the first 3 hr after symptom onset to
provide benefit to patients as now selected
[1]. Intraarterial thrombolysis
has also shown promise as a therapy for stroke if administered within 6 hr of
symptom onset, but its routine clinical use has not yet been approved, and
additional large trials are ongoing
[2,
3]. The treatment windows for
both of these thrombolytic therapies have mandated that imaging be performed
more quickly—and provide more information—than had been previously
required.
Clinical Imaging Techniques for Evaluation of Cerebral Ischemia
A number of techniques are now available on a clinical basis for evaluation
of patients with hyperacute stroke (an infarct < 6 hr old). Routine
unenhanced CT, the most widely available imaging technique, remains the
standard initial imaging examination for these patients because of its
convenience and its high sensitivity for the detection of intracranial
hemorrhage, which represents an absolute contraindication to thrombolytic
therapy. In many cases of hyperacute stroke, unenhanced CT can provide
information supportive of the diagnosis of evolving infarction (e.g., the
hyperdense artery sign, indicating arterial thrombus), even when changes in
the cerebral tissues, such as low attenuation and mass effect, are not
directly shown. Unfortunately, routine unenhanced CT has substantially lower
sensitivity for stroke detection than does diffusion-weighted MR imaging and
provides solely anatomic—and not physiologic—information
[4].
Xenon-enhanced CT is a technique that has proven valuable as a method for
noninvasive measurement of cerebral blood flow in ischemic cerebrovascular
disease [5,
6]. Xenon-enhanced CT offers
physiologic information using a technique that is not dependent on the choice
of a reference artery (an important consideration in CT perfusion studies
outlined later in this article)
[7]. Only a few institutions
have the technology to perform xenon-enhanced CT. Other limitations include
the difficulties some patients experience with xenon inhalation (e.g.,
decreased respiratory rate and headache), the fact that xenon itself can
influence cerebral blood flow, and at the time of this writing, lack of a
commercial supplier of medical xenon
[8,
9].
MR imaging has a number of advantages compared with routine CT, including
greater sensitivity for stroke detection and the capability to provide
physiologic information in the form of hemodynamic parameters that are
important in understanding areas at risk of undergoing infarction. When MR
angiography is also performed, determination of large vessel patency is also
possible. Despite these advantages, MR imaging remains underused in the
hyperacute stroke setting, most likely in part because of the delay incurred
by performing CT first to exclude hemorrhage and because emergency MR imaging
remains logistically difficult at many institutions, even during day-time
hours of operation.
A CT technique that uses equipment routinely available in most hospitals
(unlike xenon-enhanced CT) and that could provide the hemodynamic information
needed for making treatment decisions would be a valuable asset because it
would obviate moving the patient to another imaging device for physiologic
information. Rapid assessment of acute stroke patients with dynamic perfusion
CT combines the advantages of routine CT (speed, availability, and sensitivity
to hemorrhage) with a means to obtain physiologic information that is similar
to that available from xenon-enhanced CT and perfusion MR imaging. When used
in conjunction with CT angiography, dynamic perfusion CT also provides
information about cerebral artery patency. The purpose of this article is to
provide an overview of the emerging technology of perfusion CT with IV
infusion of iodinated contrast material. We discuss and compare two major
methods of performing perfusion CT—whole-brain perfused-blood-volume
perfusion CT, and first pass, single-detector dynamic perfusion CT—and
contrast these methods with other clinically available acute stroke perfusion
imaging methods.
Perfused-Blood-Volume Perfusion CT
Perfused-blood-volume perfusion CT is a rapid, high-resolution, and
convenient method for obtaining cerebral blood volume—weighted images of
the entire brain parenchyma
[10,11,12].
This technique involves obtaining an unenhanced CT scan followed by CT
angiography of the brain to simultaneously provide important information about
both arterial patency and tissue perfusion during the infusion of a single
bolus of iodinated contrast material. Perfused-blood-volume perfusion CT
combined with CT angiography is intended to be a simple and easy-to-implement
method to acquire important information that is useful in the rapid triage of
acute stroke patients. Most of the data required to make urgent thrombolysis
decisions can be obtained online, directly at the scanner console, with a
minimum of postprocessing.
Data Acquisition for Combined CT Angiography and
Perfused-Blood-Volume Perfusion CT
The combined CT angiography and perfused-blood-volume perfusion CT protocol
calls for a single 90-120 mL bolus of contrast material to be infused IV at
2.5-3 mL/sec, beginning 25 sec before the onset of scanning. These injection
parameters are intended to provide an approximate steady-state plasma
concentration of iodinated contrast material during image acquisition. Using a
helical multidetector CT scanner, two continuous data acquisitions, both using
a helical pitch of 3:1, are obtained. Using a helical pitch of 3:1 reduces the
anatomic distortion artifacts, such as those often seen in the posterior fossa
of the brain, that are often revealed on helical acquisitions obtained using
higher pitch [13]. The first
acquisition group (the acquisition that will result in data for the
intracranial CT angiogram and perfused-blood-volume scan) extends from the
skull base to the brain vertex. Suggested parameters include 3.75 mm/sec table
speed, 170 mAs, and 140 kV. The second acquisition group (the acquisition that
will result in data for the extracranial CT angiogram) extends from the aortic
arch to the skull base, using 7.5 mm/sec table speed, 250 mAs, and 140 kV
[14].
Scanning the brain (instead of the neck) during the first acquisition is
advantageous because if the study must be stopped prematurely for any reason,
the most clinically critical data for thrombolysis triage—the presence
or absence of thrombus in the middle cerebral artery—is acquired first.
The first acquisition is also collected during the peak phase of intravascular
contrast concentration (before the termination of infusion of contrast
material). This timing results in peak vascular density during the
intracranial portion of the examination, an advantage for detection of
thrombus in intracranial arteries.
Scanning the neck and aortic arch during the second acquisition helps to
decrease the severity of high-attenuation streak artifacts that can arise from
the presence of undiluted contrast material in the subclavian veins at the
thoracic inlet. Axial source images from multidetector CT are retrospectively
reformatted by the technologist into the thinnest possible slices (1.25-mm
thickness for the first acquisition group and 2.5-mm thickness for the second
acquisition group). This reformatting minimizes stairstep or zipper artifacts
that can be seen during visual assessment of three-dimensional data
constructed with slices that are too thick.
Evaluation of CT Angiography
Although most of the critical data required for thrombolysis triage can
typically be obtained from direct inspection of the thin reformatted axial CT
source images at the CT console, small peripheral middle cerebral artery
thrombi (mid and distal M2 branches) are more sensitively detected using a
maximum-density-projection reconstructed image viewed from a perspective
superior to the circle of Willis
[12]. This
maximum-density-projection image can routinely be created in less than 1 min
using most imaging workstations and scanner consoles. More detailed assessment
of the combined CT angiography and perfused-blood-volume CT perfusion data
sets can be performed offline at an imaging workstation. This analysis
typically includes construction of two-dimensional curved reformatted
projections of the bilateral common and internal carotid arteries and
vertebral arteries to screen for the presence of stenosis, and
maximum-density-projection and multiplanar volume-reformatted images of the
intracranial vessels [12]
(Fig.
1A,1B).
At our institutions, all image review is performed at a digital workstation
using operator-optimized window width and center-level display settings
[15].
View larger version (92K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1A. —54-year-old man with sudden onset of neck pain and right
hemiparesis. Superior maximum-density-projection image created from
whole-brain perfused-blood-volume CT angiography source images shows acute
left mid-M2 middle cerebral artery branch occlusion (arrow). Thrombus
was difficult to detect on source images alone. This reconstruction was
created at CT console in less than 1 min.
|
|
View larger version (119K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1B. —54-year-old man with sudden onset of neck pain and right
hemiparesis. Coronal curved reformatted image shows left internal carotid
artery skull base dissection (arrow), which likely represents source
of middle cerebral artery thromboembolus.
|
|
Using the CT angiography protocol described, one achieves a complete
evaluation of the cerebral arteries, including imaging of the great vessel
origins, carotid bifurcations, and circle of Willis vessels. In addition to
providing information relevant to acute treatment, the described CT-based
protocol also provides information traditionally obtained later during the
hospital admission by imaging methods other than CT. This CT angiography
protocol for acute stroke may obviate other frequently ordered tests, such as
carotid sonography, MR angiography, or conventional catheter arteriography.
The overall accuracy of CT angiography for detecting thromboses and stenoses
of large intracranial and extracranial arteries is in the range of 95-99%
[12,
16,17,18].
Visual Assessment of Perfused-Blood-Volume CT Perfusion Data
In clinical practice, when time is of the essence, we rely on the speed and
convenience of reviewing the unsubtracted, whole-brain, contrast-enhanced
axial CT angiography source images directly. We visually compare these images
to the unenhanced CT images of the brain for a quick estimate of regional
cerebral blood volume [19].
With few exceptions (such as in patients with early complete recanalization of
an occluded parent artery), the presence of a hypodense lesion on the
contrast-enhanced CT angiogram indicates irreversible infarction—despite
normal findings on an unenhanced CT scan
[11] (Fig.
2A,2B,2C,2D).
At Massachusetts General Hospital, patients are typically triaged to
thrombolysis if they present in an appropriate time window, their unenhanced
CT scans have no evidence of hemorrhage, and their CT angiography source
images (< one third of the middle cerebral artery territory) have small
hypoenhancing defects. However, patients with large hypoenhancing defects on
CT angiogram source images (> one third of the middle cerebral artery
territory) are typically excluded from thrombolysis.
View larger version (126K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2B. —69-year-old woman 2 hr after onset of aphasia and right-sided
weakness. Contrast-enhanced CT angiography axial source image acquired at same
time as A reveals large hypoenhancing area conforming to left middle
cerebral artery territory.
|
|
View larger version (121K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2C. —69-year-old woman 2 hr after onset of aphasia and right-sided
weakness. Diffusion-weighted MR image acquired 45 min after A and
B shows hyperintensity in left middle cerebral artery territory.
|
|
View larger version (137K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2D. —69-year-old woman 2 hr after onset of aphasia and right-sided
weakness. Unenhanced CT scan obtained for follow-up 34 hr after symptom onset
shows hypodensity corresponding to left middle cerebral artery territory. We
have found that final infarct volume on follow-up scans correlates well with
perfused-blood-volume perfusion CT lesion volume.
|
|
Subtracted Perfused-Blood-Volume Perfusion CT
Subtraction of unenhanced CT data from axial CT angiography source images
can be performed to obtain quantitative information about cerebral blood
volume [20]. The theory
underlying whole-brain perfused-blood-volume imaging is actually a special
case of the same multicompartmental tracer kinetic model that is used for
computing quantitative first-pass CT and MR imaging perfusion parameters
[10,13,21].
The resulting subtracted perfusion images are termed
"perfused-blood-volume maps" because the degree of contrast
enhancement detected in the brain parenchyma depends not only on the true
cerebral blood volume, but also on the quantity of contrast agent that is able
to reach that tissue during the actual duration of image acquisition. These
subtracted perfusion maps can be constructed off-line using proprietary
software. Such images are an important research tool from which we seek to
learn more about the changes in cerebral blood flow encountered during acute
stroke. Cerebral blood volume—weighted images are likely to have
important prognostic value for acute embolic stroke patients in predicting
both the minimum final infarct size and clinical outcome
[11]. Perfused-blood-volume CT
perfusion images have also shown promise for predicting which patients are
likely to experience cerebral hemorrhage after thrombolytic therapy (Swap C et
al., presented at the International Conference on Stroke and Cerebral
Circulation, February 2002). In one preliminary investigation, 80% of patients
with acute middle cerebral artery stroke who were treated with thrombolysis
and who had a greater than 8 H difference between normal and maximally
ischemic hypodense tissue on perfused-blood-volume CT perfusion progressed to
hemorrhage, whereas 80% of those with a less than 8 H difference, did not
(Swap C et al., ICSCC meeting, February 2002).
Published work also suggests that study of regional cerebral blood volume,
including the perfused-blood-volume CT perfusion images discussed in this
article, may provide information that is similar to that provided by
diffusion-weighted MR imaging with regard to irreversibly infarcted ischemic
brain parenchyma
[22,23].
Unlike the dynamic first-pass quantitative CT perfusion technique, however,
perfused-blood-volume images do not provide information about cerebral blood
flow or mean transit time. Cerebral blood flow and mean transit time are
important parameters for assessment of the so-called ischemic
penumbra—ischemic but potentially viable brain tissue at risk for
irreversible infarction if blood flow is not soon restored
[24,25].
Single-Detector Dynamic Perfusion CT
Like perfused-blood-volume perfusion CT, dynamic perfusion CT can be
combined with CT angiography of the head and neck. However, unlike
perfused-blood-volume perfusion CT, CT angiography is performed as a separate
step either before or after the dynamic CT perfusion acquisition. As opposed
to the whole-brain coverage provided by perfused-blood-volume CT perfusion
scans, dynamic CT perfusion scans are acquired at one location and without
table motion. Information about the passage of a bolus of infused contrast
material through the tissue can thus be tracked serially in time. It is from
this information that maps of a number of different perfusion parameters can
be computed on basis of the relative heights and shapes of the
time—attenuation curves within the brain tissue
[26,27,28,29,30].
For CT perfusion scans performed on a standard, single-detector, CT scanner, a
single 5-mm-thick or one 10-mm-thick slice is typically imaged. Studies
performed on multidetector CT scanners can have greater numbers of slices
(e.g., two 10-mm-thick or four 5-mm-thick slices). Using 10-mm slices results
in perfusion maps with somewhat greater signal-to-noise ratios than those
produced by use of 5-mm slices. Nevertheless, we typically use 5-mm instead of
10-mm slices for reasons that are discussed in the section on analysis of CT
perfusion images.
The location (level) chosen for dynamic CT perfusion scanning can be
selected at the time of examination, although for patients with acute stroke,
we typically choose a transverse slice through the level of the basal ganglia
(Fig.
3A,3B,3C,3D,3E,3F,3G)
because this level contains representative territories supplied by the
anterior, posterior, and middle cerebral arteries, and therefore offers the
opportunity to find abnormalities in each of these major vascular territories.
The present limitation of anatomic coverage to a 10- or 20-mm section
represents the main limitation of the dynamic perfusion CT method. One novel
strategy designed to increase anatomic coverage on dynamic CT perfusion
scanning was described by Roberts et al.
[31]. The authors described a
method that introduced scanner table motion during scanning so as to increase
the number of levels scanned during the same infusion of contrast material.
This method results in a temporal resolution of approximately 5 sec, from
which maps of relative perfusion parameter values can be obtained. At our
institutions, however, we have chosen not to use table motion to increase
anatomic coverage because we want to maintain the highest possible temporal
resolution for greatest accuracy. When needed, we increase anatomic coverage
by repeating the scanning at a second location (e.g., a more cephalad location
above the lateral ventricles) using a second bolus of contrast material
[23,32]
(Fig. 4). We think that future
development of larger multidetector CT scanners with greater arrays of
elements may help to address the issue of anatomic coverage without loss of
temporal resolution. In all patients, we attempt to choose CT scanning
locations and angles that avoid direct irradiation of the lens of the eye.
View larger version (120K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3A. —44-year-old man with onset of left hemiparesis 5 hr before
imaging. Single transverse CT source image from cine dynamic perfusion CT data
set (80 kVp, 200 mAs) shows substantial region of hypoattenuation in right
hemisphere. Anterior cerebral artery (green circle marked with
white arrow) and superior sagittal sinus (red circle marked
with black arrow) were chosen to serve as reference artery and
reference vein, respectively.
|
|
View larger version (174K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3B. —44-year-old man with onset of left hemiparesis 5 hr before
imaging. CT scan shows magnified arterial input region of interest (ROI)
placed on anterior cerebral artery. Small ROIs on order of 5-7 mm2
are typically used to reduce volume-averaging effects.
|
|
View larger version (184K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3C. —44-year-old man with onset of left hemiparesis 5 hr before
imaging. CT scan shows magnified venous outflow ROI in central part of
superior sagittal sinus. Venous outflow ROI represents "pure
blood" ROI to deconvolution analysis algorithm and is used to help
correct remainder of image for effects of volume averaging with nonvascular
tissue.
|
|
View larger version (102K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3E. —44-year-old man with onset of left hemiparesis 5 hr before
imaging. CT scan shows dynamic perfusion CT cerebral blood volume. Blue region
in right hemisphere represents tissue with cerebral blood volume less than 1.5
mL/100 g. Green regions represent blood volume of 1.5-3.0 mL/100 g and red
regions represent blood volume of greater than 3.0 mL/100 g.
|
|
View larger version (107K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3F. —44-year-old man with onset of left hemiparesis 5 hr before
imaging. CT scan shows cerebral blood flow. Blue regions in right hemisphere
represent tissue with cerebral blood flow less than 10 mL/100 g per minute.
Green regions represent blood flow of 10-20 mL/100 g per minute and red
regions represent blood flow of greater than 20 mL/100 g per minute.
|
|
View larger version (102K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3G. —44-year-old man with onset of left hemiparesis 5 hr before
imaging. CT scan shows mean transit time. Extensive red regions on right
represent tissue with mean transit time greater than 6 sec. Green regions
represent transit time of 3-6 sec and blue regions represent transit time of
less than 3 sec.
|
|
View larger version (119K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4. —74-year-old man with suspected stroke. Lateral scout image
from CT examination shows two levels chosen for single-detector perfusion CT.
First level (large arrows) corresponds to level of basal ganglia.
Second level (small arrows) is located more cephalad and helps to
increase coverage.
|
|
Most institutions that perform dynamic perfusion CT have chosen to use an
X-ray generation technique of 80 kVp with 200 mA. This is essentially the
technique advocated by Wintermark et al.
[33], who have shown that
using 80 kVp is preferable to using 120 kVp for increasing image contrast and
substantially decreasing patient radiation dose for a given current.
Wintermark et al. estimated that the dose delivered by a CT perfusion study
performed at 80 kVp is less than 300 mGy, whereas the dose delivered by a CT
perfusion study performed at 120 kVp can be more than 800 mGy
[33]. The choice of these
amperage and kilovoltage settings is not without controversy, however, because
the patient's absorbed dose is also higher at 80 kV than at 120 or 140 kV, and
some centers select alternative scanning settings
[13].
Although it is technically possible to perform perfusion CT using
nonhelical CT scanners (discontinuous scanning), temporal resolution less than
2 sec is not routinely possible with these scanners, which represents a
problem because low temporal resolution could result in maps of decreased
quality and accuracy. Because high temporal resolution (interscan interval of
1 sec) is desirable for increased accuracy, a continuous (cine) scanning
technique should be used, if possible, for dynamic CT perfusion studies. A
minimum gantry speed of 1 sec per revolution, which results in an intrinsic
temporal resolution of one "frame" per second, is suggested for
routine use. However, in an effort to enhance accuracy even further, we
routinely reconstruct the original cine data (acquired using 1-sec revolutions
of the scanner gantry) at half-second intervals. We typically reconstruct CT
perfusion data using overlapping segments, each of 1 sec in duration, centered
at half-second intervals. The software to perform this extra step is available
on some CT scanners.
A typical protocol for dynamic CT perfusion calls for a scanning duration
of 45 sec; scanning begins 5 sec after the start of contrast material
infusion. This 5-sec preparation delay helps to limit total patient radiation
dose but does not result in loss of useful data because delay is always
encountered in the passage of the bolus from the systemic venous circulation
to the cerebral arteries. In our experience with younger patients (those with
excellent cardiac output), a delay of more than 5 sec in the onset of scanning
could result in loss of important data regarding early portions of the
arterial and tissue enhancement versus time curves, because the time from
onset of infusion to the time that contrast material first reaches the
cerebral arteries can be as short as 8 sec. A total of 40-50 mL of iodinated
contrast material is typically infused in an antecubital vein at a rate of at
least 4 mL/sec. Contrast material having a density of at least 300 mg/dL is
used, although we often use contrast material of 370 mg/dL density. The total
infusion time is kept to 10 sec or less to maintain a compact bolus. Evidence
exists that increasing the infusion rate (and decreasing the injection time)
can help to increase the quality of the resultant perfusion maps (Cook AJ II
et al., presented at the Radiological Society of North America meeting,
November 2001). However, there appears to be no advantage in using infusion
rates higher than approximately 10 mL/sec
[34]. For infusion rates of 4
mL/sec, we have found that either 18- or 20-gauge antecubital catheters can be
safely used.
Dynamic Perfusion CT Analysis
A number of algorithms are available for processing perfusion CT data
[26,
29,
34]. This article focuses
mainly on one method, deconvolution analysis, which has advantages over other
methods of analysis. One advantage of deconvolution analysis is that, unlike
some other algorithms, it does not impose on the computation potentially
unrealistic assumptions about venous outflow that may affect the accuracy of
quantification [26]. A related
practical advantage of deconvolution analysis is that IV infusion rates lower
than 10 mL/sec may be used. Lower infusion rates permit the use of IV
catheters smaller than the 14- to 16-gauge catheters required for studies that
use high (10-20 mL/sec) infusion rates
[35,
36]. Using lower infusion
rates may be safer than using high infusion rates.
A detailed discussion of the mathematics of deconvolution analysis is
beyond the scope of this article. The interested reader is referred to a
number of excellent reviews
[37,38,39,40,41].
Briefly and in simplified terms, deconvolution analysis is a mathematic
technique in which a specified arterial contrast concentration versus time
curve (the "arterial input function") and the tissue contrast
concentration versus time curves (the "tissue residue function")
are combined to determine an idealized tissue concentration versus time curve
that would result if the entire bolus of contrast could be delivered
instantaneously to the tissue bed of interest. Additionally, examination of
the venous contrast concentration versus time curve assists in the
normalization of the resulting perfusion parameters by helping to correct for
partial volume averaging effects. For this reason, analysis of perfusion CT
data using the deconvolution method requires the selection of two small
regions of interest (ROIs) reflecting representative time—attenuation
curves for the arterial input function and venous outflow function
[26,
39]. The optimal choices for
both the arterial and venous ROIs are large vessels that have courses nearly
perpendicular to the transverse plane of section used in CT studies. Our usual
approach is to select either one of the two (unaffected) anterior cerebral
arteries or the unaffected middle cerebral artery as the reference artery; we
typically select the superior sagittal sinus as the reference vein (Fig.
3A,3B,3C,3D,3E,3F,3G).
An important function of the venous outflow time—attenuation curve is to
correct for volume averaging effects (by providing to the algorithm a
reference equal to pure blood). This function of the venous outflow curve is
important for providing accurate perfusion parameter values. It is essential,
therefore, to minimize partial volume effects both within the plane of section
and through the plane of section. It is for this reason that we prefer
5-mm-thick slices to 10-mm-thick slices, despite the slightly lower
signal-to-noise ratios seen in 5-mm slices. We also use very small ROIs (on
the order of 6 mm2) placed in the center portion of the sagittal
sinus (Fig.
3A,3B,3C,3D,3E,3F,3G).
Some controversy exists concerning the best strategy for choosing an
appropriate arterial input function ROI in the setting of an acute embolic
stroke with occlusion of a major cerebral artery
[37,40].
The concern is that of obtaining accurate quantifiable values for the cerebral
blood flow and mean transit time parameters. Based on the deconvolution model,
the arterial input ROI should ideally be chosen from the most proximal large
feeding vessel directly supplying the tissue of interest. Because, for
practical reasons, it is not always possible to select such an ROI, there are
limits to the accuracy of any cerebral blood flow or mean transit time
calculation based on firstpass dynamics. In practice, however, such
considerations are seldom clinically significant. One possible approach for
constructing maps in the presence of a large proximal artery occlusion is to
select two arterial input functions (and hence two ROIs): the first (from a
normal artery) to be used for assessment of the nonischemic tissues, and the
other (from the obstructed artery) to be used for assessment of the ischemic
tissues [13]. Although this
approach has some mathematic benefit, it may be impractical for clinicians
trying to assess an individual patient with acute stroke. A simpler approach
is to choose a single patent artery from the unaffected side as the reference
artery. The results obtained from different observers using the same
unaffected artery have been shown to be reproducible (Sanelli PC et al.,
presented at the Radiological Society of North America meeting, November
2001). Additionally, prior work indicates that perfusion maps derived using
various unaffected arteries (e.g., unaffected middle cerebral artery versus
unaffected anterior cerebral artery) provide similar results (Lee T et al.,
presented at the Radiological Society of North America meeting, November
2001). The potential problem with using a single unaffected artery as the
reference artery for the whole brain is that values of cerebral blood flow on
the affected side may be underestimated, and mean transit time values on the
affected side may be overestimated. Further work in this area is needed to
determine the optimal arterial ROI selection strategy for assessment of
patients with various forms of cerebrovascular disease.
Visual Assessment of Dynamic Perfusion CT Scans in Acute Stroke
In the setting of stroke, it is important that the imaging methods used
have high interobserver and intraobserver reproducibility. To date, few
studies have provided guidance regarding the optimal way to assess the extent
of perfusion abnormality on perfusion CT scans. A study by Eastwood et al.
[36] of patients with acute
middle cerebral artery stroke used a method of establishing thresholds of
abnormality for three perfusion parameters: cerebral blood flow, cerebral
blood volume, and mean transit time. For each parameter, values thought to be
consistent with severe ischemia were chosen to represent abnormal levels. In
that study, intraobserver variation in determination of extent of cerebral
blood flow abnormality was less than 10%. That study also reported a high
degree of interobserver reproducibility as evidenced by strong correlation
between extents of abnormality determined by two observers
[36].
In the study by Eastwood et al.
[36], cerebral blood flow
values ranging between 0 and 10 mL/100 g per minute were considered to be
those consistent with severe ischemia. Cerebral blood volume values of 0-1.5
mL/100 g and mean transit time values above 6 sec were chosen to represent
abnormal regions (Fig.
5A,5B,5C,5D,5E).
A study by Nabavi et al. [26]
in an animal model that used correlation with final infarct extent has shown
that one may determine the optimal range of perfusion parameter values to
correlate with final infarct size. Ideally, a future study in humans using
follow-up imaging to determine the final infarct extent could help to show the
optimal ranges of values of cerebral blood flow, cerebral blood volume, and
mean transit time for prediction of final infarct size.
View larger version (91K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5B. —32-year-old woman with onset of left arm and face weakness 6
hr before scanning. Dynamic perfusion CT cerebral blood flow map shows
decreased blood flow in right insula and lateral basal ganglia (thin
arrow) and temporal lobe (thick arrows). Decreased blood flow in
range of 0-100 mL/100 g per minute is represented by blue.
|
|
View larger version (100K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5C. —32-year-old woman with onset of left arm and face weakness 6
hr before scanning. Mean transit time map calculated from same data as
B shows much more extensive region of abnormality. Increased mean
transit time greater than 6 sec is represented by red.
|
|
View larger version (17K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5D. —32-year-old woman with onset of left arm and face weakness 6
hr before scanning. Tissue time—attenuation curves. Iodinated contrast
material is slower to enter (thin arrow) and slower to wash out
(thick arrow) of right middle cerebral artery territory compared with
normal left middle cerebral artery territory. Inspection of such curves is not
usually done in clinical situation. Instead, computer algorithm computes
tissue values pixel by pixel to create perfusion maps such as those in
B.
|
|
View larger version (119K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5E. —32-year-old woman with onset of left arm and face weakness 6
hr before scanning. Unenhanced CT scan obtained 3 days after admission to
hospital shows final size of infarction in basal ganglia and small part of
temporal lobe (arrow).
|
|
Use of Perfusion CT to Evaluate Chronic Cerebrovascular Disease
Patients with symptoms thought to be due to chronic hemodynamic
insufficiency represent a challenging clinical problem
[42]. The goal of therapy is
to prevent further symptoms, but the best therapy for an individual patient
(such as anticoagulant therapy or intervention to increase cerebral blood
flow) is frequently not known with certainty. Perfusion imaging represents one
strategy used to select patients for possible intervention
[43,
44]. Previous studies have
shown that cerebral perfusion imaging coupled with a pharmacologic challenge
can provide valuable information about regional blood flow and reactivity to
acetazolamide
[43,44,45].
In particular, patients with decreased blood flow and decreased response to
acetazolamide are thought to be those patients most likely to benefit from
interventions designed to increase blood flow
[43,44].
Although a number of different methods of perfusion imaging may be used to
evaluate patients with cerebrovascular insufficiency, many of these previously
reported methods (such as positron emission tomography and xenon-enhanced CT)
may be unavailable outside of large institutions. The use of dynamic perfusion
CT for evaluation of chronic ischemia has now been reported by several authors
[32,
46,
47]. In particular, a recently
published study by Furukawa et al.
[46] has shown that dynamic
perfusion CT with acetazolamide challenge compares favorably with
xenon-enhanced CT with acetazolamide challenge for this purpose. The authors
of that study concluded that dynamic perfusion CT is a useful alternative to
xenon-enhanced CT for the evaluation of patients with acetazolamide
challenge.
Use of Perfusion CT for Cerebral Neoplasms
Although perfusion imaging has mainly been used for assessment of ischemia,
perfusion imaging has recently been applied to other disease processes. For
example, perfusion imaging of brain tumors has been shown to be helpful for
preoperative assessment of tumor grade and for differentiating enhancement due
to tumor from that of radiation necrosis
[48,49,50,51].
In particular, the parameters of cerebral blood volume and permeability
surface flow (a measure of how "leaky" tumor capillaries are)
correlate well with tumor grade
[48,49,50].
MR imaging is ideally suited for assessment of brain tumors because of high
tissue contrast and multiplanar capabilities. However, perfusion MR imaging
demands rapid scanner gradients and MR-compatible contrast material injection
devices that may not be available in every department. Furthermore, some
patients (such as those with pacemakers or aneurysm clips) cannot undergo MR
imaging. In cases like these, an alternative method for measuring parameters
such as cerebral blood volume and permeability surface flow would be
valuable.
In a study of a rabbit model of brain tumor, Cenic et al.
[52] described the utility of
perfusion CT for assessment of brain tumor hemodynamics, including
measurements of cerebral blood flow, cerebral blood volume, and vascular
permeability. In that study, each of these three parameters was found to be
increased in enhancing tumor—findings that were similar to those of
prior MR studies in humans
[52]. Nabavi et al.
[47] reported increased
cerebral blood flow and cerebral blood volume values within enhancing tumor in
a patient studied with perfusion CT. In another case we describe in a paper
that is presently submitted for publication, we found increased cerebral blood
flow, cerebral blood volume, and vascular permeability in an enhancing
glioblastoma multiforme tumor (Eastwood JD, et al., unpublished data). When an
algorithm that permits calculation of permeability of blood vessels to
contrast material is used, values of cerebral blood flow and cerebral blood
volume that are corrected for the effect of this leakage of contrast material
can be computed [52]. Another
possible advantage of perfusion CT for assessment of tumors may be that it is
possible to compute absolute values of perfusion parameters. MR methods of
perfusion imaging provide relative, rather than absolute, values of perfusion
parameters. Dynamic perfusion CT with correction for vascular permeability is
also being developed for tumor imaging outside the central nervous system (Lee
T-Y, personal communication).
Conclusion
Perfusion CT with the infusion of iodinated contrast material is a rapid
and widely available method of cerebral perfusion imaging. For assessment of
patients with acute stroke, perfusion CT may be combined with unenhanced CT
and CT angiography to offer rapid assessment of anatomy, vascular patency, and
regional hemodynamics. A successful and comprehensive CT-based protocol may
ultimately help to direct therapy for acute stroke. Perfusion CT may also be
an important alternative to MR perfusion imaging when MR imaging is
unavailable or contraindicated.
References
- Hacke W, Kaste M, Fieschi C, et al. Intravenous thrombolysis with
recombinant tissue plasminogen activator for acute hemispheric stroke: the
European Cooperative Acute Stroke Study (ECASS). JAMA
1995;274:1017
-1025[Abstract]
- Furlan A, Higashida R, Wechsler L, et al. Intra-arterial
prourokinase for acute ischemic stroke. The PROACT II study: a randomized
controlled trial—Prolyse in acute cerebral thromboembolism.
JAMA
1999;282:2003
-2011[Abstract/Free Full Text]
- del Zoppo GJ, Higashida RT, Furlan AJ, Pessin MS, Rowley HA, Gent
M. PROACT: a phase II randomized trial of recombinant pro-urokinase by direct
arterial delivery in acute middle cerebral artery stroke. PROACT
Investigators. Prolyse in acute cerebral thromboembolism.
Stroke
1998;29:4
-11[Abstract/Free Full Text]
- Gonzalez R, Schaefer P, Buonanno F, et al. Diffusion-weighted MR
imaging: diagnostic accuracy in patients imaged within 6 hours of stroke
symptom onset. Radiology
1999;210:155
-162[Abstract/Free Full Text]
- Kauffman AM, Firlik AD, Fukui MB, Wechsler LR, Jungies CA, Yonas H.
Ischemic core and penumbra in human stroke. Stroke
1999;30:93
-99[Abstract/Free Full Text]
- Firlik AD, Kaufmann AM, Wechsler LR, Firlik KS, Fukui MB, Yonas H.
Quantitative cerebral blood flow determinations in acute ischemic stroke:
relationship to computed tomography and angiography.
Stroke
1997;28:2208
-2213[Abstract/Free Full Text]
- Wintermark M, Thiran J-P, Maeder P, Schnyder P, Meuli R.
Simultaneous measurement of regional blood flow by perfusion CT and stable
xenon CT: a validation study. AJNR
2001;22:905
-914[Abstract/Free Full Text]
- Latchaw RE, Yonas H, Pentheny SL, Gur D. Adverse reactions to
xenon-enhanced CT cerebral blood flow determination.
Radiology
1987;163:251
-254[Abstract/Free Full Text]
- Horn P, Vajkoczy P, Thome C, Muench E, Schilling L, Schmiedek P.
Xenon-induced flow activation in patients with cerebral insult who undergo
xenon-enhanced CT blood flow studies. AJNR
2001;22:1543
-1549[Abstract/Free Full Text]
- Hunter GJ, Hamberg LM, Ponzo JA, et al. Assessment of cerebral
perfusion and arterial anatomy in hyperacute stroke with three-dimensional
functional CT: early clinical results. AJNR
1998;19:29
-37[Abstract]
- Lev MH, Segal AZ, Farkas J, et al. Utility of perfusion-weighted CT
imaging in acute middle cerebral artery stroke treated with intra-arterial
thrombolysis: prediction of final infarct volume and clinical outcome.
Stroke
2001;32:2021
-2028[Abstract/Free Full Text]
- Lev MH, Farkas J, Rodriguez VR, et al. CT angiography in the rapid
triage of patients with hyperacute stroke to intraarterial thrombolysis:
accuracy in the detection of large vessel thrombus. J Comput Assist
Tomogr 2001;25:520
-528[Medline]
- Lev M, Gonzalez R. CT angiography and CT perfusion imaging. In:
Mazziotta J, Toga A, eds. Brain mapping: the methods,
2nd ed. San Diego: Academic Press, 2002:427
-484
- Lev MH, Nichols SJ. Computed tomographic angiography and computed
tomographic perfusion imaging of hyperacute stroke. Top Magn Reson
Imaging 2000;11:273
-287[Medline]
- Lev M, Farkas J, Gemmete J, et al. Acute stroke: improved
nonenhanced CT detection—benefits of soft-copy interpretation by using
variable window width and center level settings.
Radiology
1999;213:150
-155[Abstract/Free Full Text]
- Wildermuth S, Knauth M, Brandt T, Winter R, Sartor K, Hacke W. Role
of CT angiography in patient selection for thrombolytic therapy in acute
hemispheric stroke. Stroke
1998;29:935
-938[Abstract/Free Full Text]
- Shrier D, Tanaka H, Numaguchi Y, Konno S, Patel U, Shibata D. CT
angiography in the evaluation of acute stroke. AJNR
1997;18:1011
-1020[Abstract]
- Cumming MJ, Morrow IM. Carotid artery stenosis: a prospective
comparison of CT angiography and conventional angiography.
AJR
1994;163:517
-523[Abstract/Free Full Text]
- Bove P, Lev M, Chaves T, et al. CT perfusion imaging improves
infarct conspicuity in hyperacute stroke. In: Hachinski V, ed. Proceedings
of the 26th International Conference on Stroke and Cerebral Circulation.Stroke 2001;32:325
-b
- Alpert NM, Berdichevsky D, Levin Z, Thangaraj V, Gonzalez G, Lev
MH. Performance evaluation of an automated system for registration and
postprocessing of CT scans. J Comput Assist Tomogr
2001;25:747
-752[Medline]
- Hamberg LM, Hunter GJ, Kierstead D, Lo EH, Gilberto Gonzalez R,
Wolf GL. Measurement of cerebral blood volume with subtraction
three-dimensional functional CT. AJNR
1996;17:1861
-1869[Abstract]
- Berzin T, Lev M, Goodman D, et al. CT perfusion imaging versus MR
diffusion weighted imaging: prediction of final infarct size in hyperacute
stroke. In: Hachinski V, ed. Proceedings of the 26th International
Conference on Stroke and Cerebral Circulation. Stroke
2001;32:317
- Wintermark M, Reichhart M, Thiran J, et al. Prognostic accuracy of
cerebral blood flow measurement by perfusion computed tomography, at the time
of emergency room admission, in acute stroke patients. Ann
Neurol 2002;51:417
-432[Medline]
- Heiss WD, Kracht L, Grond M, et al. Early [(11)C] flumazenil/H(2)O
positron emission tomography predicts irreversible ischemic cortical damage in
stroke patients receiving acute thrombolytic therapy.
Stroke
2000;31:366
-369
Top
Introduction
Clinical Imaging Techniques for...
