This copy is for personal use only. To order printed copies, contact reprints@rsna.

org
1717

Brain Imaging
CT for Treatment Selection in
Acute Ischemic Stroke: A Code
Stroke Primer
Christopher A. Potter, MD
Achala S.Vagal, MD CT is the primary imaging modality used for selecting appropriate
Mayank Goyal, MD, FRCPC treatment in patients with acute stroke. Awareness of the typical
Diego B. Nunez, MD, MPH findings, pearls, and pitfalls of CT image interpretation is therefore
Thabele M. Leslie-Mazwi, MD critical for radiologists, stroke neurologists, and emergency depart-
Michael H. Lev, MD ment providers to make accurate and timely decisions regarding
both (a) immediate treatment with intravenous tissue plasminogen
Abbreviations: ACA = anterior cerebral artery, activator up to 4.5 hours after a stroke at primary stroke centers
ASPECTS = Alberta Stroke Program Early CT and (b) transfer of patients with large-vessel occlusion (LVO) at
Score, CBV = cerebral blood volume, DWI =
diffusion-weighted imaging, EVT = endovascu- CT angiography to comprehensive stroke centers for endovascu-
lar thrombectomy, GWD = gray-white matter lar thrombectomy (EVT) up to 24 hours after a stroke. Since the
differentiation, ICA = internal carotid artery,
ICH = intracranial hemorrhage, LVO = large- DAWN and DEFUSE 3 trials demonstrated the efficacy of EVT
vessel occlusion, MCA = middle cerebral artery, up to 24 hours after last seen well, CT angiography has become the
MIP = maximum intensity projection, tPA = tis-
sue plasminogen activator
operational standard for rapid accurate identification of intracranial
LVO. A systematic approach to CT angiographic image interpre-
RadioGraphics 2019; 39:1717–1738
tation is necessary and useful for rapid triage, and understand-
https://doi.org/10.1148/rg.2019190142 ing common stroke syndromes can help speed vessel evaluation.
Content Codes: Moreover, when diffusion-weighted MRI is unavailable, multiphase
From the Department of Radiology, Brigham CT angiography of collateral vessels and source-image assessment
and Women’s Hospital, 75 Francis St, Boston, or perfusion CT can be used to help estimate core infarct volume.
MA 02115 (C.A.P., D.B.N.); Department of
Radiology, University of Cincinnati, Cincin-
Both have the potential to allow distinction of patients likely to ben-
nati, Ohio (A.S.V.); Department of Diagnostic efit from EVT from those unlikely to benefit. This article reviews
Imaging, University of Calgary, Calgary, AB, CT-based workup of ischemic stroke for making tPA and EVT
Canada (M.G.); and Department of Radiology,
Massachusetts General Hospital, Boston, Mass treatment decisions and focuses on practical skills, interpretation
(T.M.L.M., M.H.L.). Presented as an educa- challenges, mimics, and pitfalls.
tion exhibit at the 2018 RSNA Annual Meeting.
Received February 28, 2019; revision requested ©
RSNA, 2019 • radiographics.rsna.org
June 13 and received July 12; accepted July 25.
For this journal-based SA-CME activity, the au-
thors A.S.V., M.G., and M.H.L. have provided
disclosures (see end of article); all other authors,
the editor, and the reviewers have disclosed no
relevant relationships. Address correspondence
Introduction
to C.A.P. (e-mail: [email protected]). CT is the first-line imaging modality used in neurologic emergen-
©
RSNA, 2019
cies owing to its speed, accurate depiction of acute intracranial
disease, and availability (1). The critical role of nonenhanced CT
for stroke evaluation began in 1996, when the U.S. Food and Drug
SA-CME Learning Objectives
Administration (FDA) approved intravenous tissue plasminogen
After completing this journal-based SA-CME activator (tPA) for clot thrombolysis (2). Nonenhanced CT should
activity, participants will be able to:
be performed rapidly in patients with signs and symptoms of acute
■■Describe CT findings that are early in-
dicators of ischemia and infarction. stroke to exclude intracranial hemorrhage (ICH) and identify large
■■Compare stroke imaging patterns that (ie, >100 mL or more than one-third of a brain territory at risk)
are favorable or unfavorable for treat- well-established infarcts. However, the revised 2018 American Heart
ment. Association (AHA) guidelines state that the extent or severity of
■■Recognize
pitfalls in interpretation of the hypoattenuation seen at CT should not be used as a criterion
nonenhanced CT images. for withholding tPA owing to insufficient evidence (3). Early signs
See rsna.org/learning-center-rg. of proximal middle cerebral artery (MCA) large-vessel occlusive
infarction seen at nonenhanced CT include loss of gray-white matter
differentiation (GWD) at the insula, basal ganglia, and caudate head
as well as sulcal effacement (4–7).
1718  October Special Issue 2019 radiographics.rsna.org

fusion-weighted images, CT angiography source

Teaching Points images as [length 3 width 3 height]/2) (10).
■■ Since the DAWN (Clinical Mismatch in the Triage of Wake Up Diffusion-weighted imaging (DWI) remains
and Late Presenting Strokes Undergoing Neurointervention
with Trevo) and DEFUSE 3 (Endovascular Therapy Following
the unequivocal reference standard used to assess
Imaging Evaluation for Ischemic Stroke) trials showed the ef- core infarct volume (3,11). When MRI is un-
ficacy of endovascular thrombectomy (EVT) up to 24 hours available, multimodal CT assessment is helpful.
after the onset of stroke, CT angiography has become the Perfusion CT or multiphase CT angiography col-
operational standard for rapid accurate identification of intra- lateral and source image assessment are emerging
cranial large-vessel occlusion (LVO).
as important tools for treating stroke in eligible
■■ To expedite the decision to transfer a patient to a stroke center
patients, especially those with delayed presenta-
that is capable of performing EVT, all patients should undergo
CT angiography immediately after undergoing nonenhanced tion (>6 hours from symptom onset) or stroke of
CT without being removed from the CT scanner. unknown onset (from time last seen well) (3).
■■ The insula, caudate heads, and basal ganglia show early find- This review focuses exclusively on the current
ings of proximal MCA thrombosis at nonenhanced CT and role of CT in management of acute stroke. We
should be carefully evaluated. present an overview of CT-based workup in eval-
■■ Knowledge of common stroke syndromes is part of an effi- uation of acute ischemic stroke that emphasizes
cient evaluation in conjunction with interpreting CT angio- practical considerations in nonenhanced CT, CT
grams and identifying vessel occlusions.
angiography, and perfusion CT, including key
■■ Potential acute stroke mimics include intracranial masses due aspects of interpretation, pitfalls, and challenges.
to primary or metastatic malignancy, cerebritis, cerebral ab-
scess, and posterior reversible encephalopathy syndrome
(PRES). Imaging Rationale and
Code Stroke Workflow
Although an in-depth discussion of the patho-
physiology of acute ischemic stroke is beyond
Since the DAWN (Clinical Mismatch in the the scope of this review, the concepts of ischemic
Triage of Wake Up and Late Presenting Strokes core and penumbra are essential to understand-
Undergoing Neurointervention with Trevo) (8) ing the role of the different components of imag-
and DEFUSE 3 (Endovascular Therapy Fol- ing evaluation. In ischemic stroke, arterial occlu-
lowing Imaging Evaluation for Ischemic Stroke) sion due to embolism or less commonly in situ
(9) trials showed the efficacy of endovascular thrombosis leads to a cascade of cellular events.
thrombectomy (EVT) up to 24 hours after the These events cause local dysfunction and ulti-
onset of stroke, CT angiography has become the mately cell death, leading to development of an
operational standard for rapid accurate iden- infarct core, which is defined as brain tissue likely
tification of intracranial large-vessel occlusion to die despite immediate reperfusion. The infarct
(LVO). The proximal intracranial vessels are core is surrounded by the ischemic penumbra,
evaluated to identify LVOs suitable for catheter the ischemic tissue at risk for infarction that may
thrombectomy, whereas the cervical vessels are be salvaged with timely reperfusion (12,13). In
evaluated for the presence of dissection, critical the setting of a proximal MCA or ICA LVO, if
stenoses, or other vascular variants or abnor- there is persistent insufficient tissue reperfusion
malities that may complicate EVT. as time passes, there is continued core infarct
At primary stroke centers, nonenhanced CT growth with penumbral loss (12).
is essential in the decision-making process by The rate of penumbral loss in untreated stroke
helping rule out ICH and large well-established does not appear to be directly time dependent
infarct. CT angiography helps identify proximal but appears to depend primarily on the quality
LVOs in patients with acute MCA or intracranial and maintenance of collateral flow (9,10). Re-
internal carotid artery (ICA) syndromes, aiding gardless of whether stroke patients are classified
in the decision of whether to transfer a patient as rapid or slow progressors, failure of collateral
to a comprehensive stroke center accredited to flow with associated penumbral loss will continue
perform EVT. up to 24 or even 48 hours after stroke onset in
However, some measure of core infarct volume the absence of LVO recanalization, necessitating
is required before proceeding with EVT to deter- imaging approaches for selecting patients that
mine both ICH risk and the likelihood of treat- may benefit from EVT (13). Accurate identifica-
ment benefit (5). Small cores (<50–70 mL or in a tion of the infarct core has therefore been a target
symmetric collateral pattern) are considered favor- of intense scientific investigation.
able for EVT treatment, and large cores (>100 mL In determining a patient’s eligibility for treat-
or in a collateral pattern typical of malignancy) are ment of acute stroke, imaging must rapidly answer
considered unfavorable to treat. Core volume can the following three essential questions: (a) Is
be estimated from cross-sectional images (ie, dif- there an ICH seen at nonenhanced CT that is a
RG  •  Volume 39  Number 6 Potter et al  1719

contraindication to intravenous tPA or EVT, or mum acquired section thickness of 2 mm or less

is there a large well-established hypoattenuating and a table pitch no greater than 2:1 (15). Refor-
infarct? (b) Is there a proximal LVO seen at CT matting coronal and sagittal images is routine at
angiography that can be treated with EVT? (c) Is many institutions, although not universal.
there a large core infarct seen at DWI, at CT angi- Dual-energy CT is available with different
ography of collateral vessels or on CT angiographic system architectures, the most popular being
source images, or at perfusion CT that is a relative dual-source and fast kilovolt peak–switching
contraindication to intravenous tPA or EVT? systems. Because dual-energy CT acquires data
Rapidly acquiring and interpreting nonen- using high and low x-ray energy spectra, materi-
hanced CT images of a patient suspected of hav- als can be characterized on the basis of their x-ray
ing acute stroke is critical to expedite administra- absorption characteristics and contrast can be
tion of intravenous tPA. Optimally, evaluation of optimized. Using the data from the low and high
the nonenhanced CT images of a patient desig- kilovolt peak x-ray spectra, dual-energy CT post-
nated with “code stroke” occurs at the scanner processing can generate virtual monochromatic
console at the time of acquisition and findings are images that display the expected attenuation at a
directly communicated to the stroke team. single virtual x-ray spectrum, which is not pos-
To expedite the decision to transfer a patient sible in conventional nonenhanced CT using a
to a stroke center that is capable of performing single polychromatic spectrum. Spectral absorp-
EVT, all patients should undergo CT angiogra- tion curves can be plotted at different energy
phy immediately after undergoing nonenhanced levels and can be used to optimize the contrast of
CT without being removed from the CT scanner. GWD on images (16,17).
Performing CT angiography in this setting
should never delay the administration of intra- Image Review
venous tPA. If a pharmacist is present as part of
the code stroke team, imaging and medication Initial Image Review.—During the initial imag-
preparation may occur simultaneously. Once the ing evaluation in a patient with stroke, it must be
nonenhanced CT images have been reviewed and determined rapidly whether there is ICH or a large,
the decision has been made to administer tPA, CT well-established, hypoattenuating territorial infarct.
angiography may be performed during the several Revised 2018 American Heart Association
minutes needed to mix and prepare the tPA. guidelines recommend that at least 50% of stroke
Efficient imaging workflow is essential. Ideally, treatment candidates undergo nonenhanced CT
a code stroke page should be initiated as soon imaging within 25 minutes of arrival (3). Initial
as the patient is identified, which may be during rapid nonenhanced CT evaluation within 4.5
transportation or at presentation to the emer- hours from onset in patients without other con-
gency department. The page notifies the stroke traindications to intravenous tPA administration
team, radiologists, and CT and MRI technolo- should focus on identification of a large territorial
gists about an incoming patient. This allows infarct and exclusion of ICH, which is an absolute
the technologists to accommodate the patient contraindication to intravenous tPA treatment (3).
as quickly as possible in the scanner, minimiz- These results should be communicated im-
ing the door-to-imaging time to no more than mediately after imaging to enable rapid intrave-
the 25 minutes recommended by the American nous tPA administration, which should never be
Heart Association’s Get With The Guidelines delayed by CT angiography for EVT triage. Cur-
program (3,14), which can be seen at https:// rently, guidelines for treatment with intravenous
www.heart.org/en/professional/quality-improvement/ tPA within 4.5 hours and EVT within 6 hours
get-with-the-guidelines/get-with-the-guidelines-stroke/ are based only on nonenhanced CT relative and
get-with-the-guidelines-stroke-overview. absolute exclusion criteria (3,14).
The sensitivity and specificity of nonenhanced
Nonenhanced CT CT for depiction of ICH is estimated to be high
with modern CT scanners, likely exceeding 95%–
Technique 98% depending on the patient cohort and level
There has been rapid development of CT technol- of training of the radiologists (18). Indeed, recent
ogy over the past decades, enabling rapid imaging work published in Nature Biomedical Engineering
with a relatively low dose of radiation. Current has suggested that even subtle hemorrhages can
American College of Radiology recommendations be sensitively depicted with artificial intelligence
for CT of the brain include a reconstructed axial platforms, with accuracy approaching that of
section thickness of no more than 5 mm using experienced subspecialty-trained neuroradiolo-
overlapping or contiguous sections. Imaging times gists and exceeding that of less experienced non–
should be no greater than 2 seconds, with a mini- subspecialty-trained readers (19).
1720  October Special Issue 2019 radiographics.rsna.org

Nonenhanced CT is much less sensitive about ASPECTS can be found at the educational
compared with DWI for depiction of both the website www.aspectsinstroke.com.
cytotoxic edema and low cerebral blood volume Identification of vessel occlusion is limited at
(CBV) that accompany early ischemia (13). De- nonenhanced CT, although an “attenuating vessel
spite this, the widespread availability and speed of sign” is highly specific when it is present. The at-
CT make it the most practical first-line imaging tenuating vessel sign due to thrombus in the M1
strategy at most institutions, as it can be per- segment of the MCA has relatively low sensitivity
formed in minutes and requires no prescreening of 17%–52% but high specificity of 95% (24,25).
or patient exclusion. Considered an equivalent of the attenuat-
The sensitivity and specificity for acute infarc- ing vessel sign in the M1 segment, the MCA
tion at nonenhanced CT likely depend on the “dot” sign in the proximal M2 vessels is often
duration, infarct size, and degree of ischemia. associated with early insular GWD loss (26). At
Sensitivity and specificity for depiction of early imaging, the dot sign appears in the vessels in the
ischemic change are also likely better for the lateral sylvian fissure adjacent to the insula wall.
anterior circulation than the posterior circulation, The mixed images obtained with dual-energy
primarily owing to artifact at the skull base from CT are produced by blending both high and low
thick surrounding bone, although this has not kilovolt peak datasets. Dual-energy CT can be
been definitively demonstrated. used in the initial examination similar to the way
Several areas of the brain including the insula, conventional nonenhanced CT is used. Monochro-
basal ganglia, and caudate head are early indi- matic imaging is not widely used in this context,
cators of acute infarct owing to their vascular since image reconstruction is not typically auto-
anatomy in relation to the most common patterns mated and it is too time-consuming. Nevertheless,
of proximal MCA LVO distribution (4,5,20). optimization of GWD has been reported to occur
Special attention should therefore be directed to at a virtual kilo–electron voltage of 66–75 (17).
these regions during initial image review (Fig 1).
Well-established frank ischemia seen at CT in Tip or Pearl for Image Review.—A methodical
these regions as an area of low attenuation with search pattern is important because the imaging
loss of GWD is typically attributed to vasogenic findings in acute ischemic stroke can be subtle on
edema from blood-brain barrier breakdown. This nonenhanced CT images.
imaging finding indicates irreversible infarction
in the clinical setting of an acute MCA occlusion Gray-White Matter Differentiation.—Loss of
and is not attributable to cytotoxic edema from GWD is the imaging hallmark of acute infarct at
membrane pump failure and low CBV with poor nonenhanced CT. Use of a narrow stroke window
collateral flow (21). Cytotoxic edema is a more while performing soft-tissue sequences has been
subtle hyperacute finding and is typically attribut- studied, and recommended values have been
able to membrane pump failure and low CBV. reported (window width, 8 HU; window level, 32
The Alberta Stroke Program Early CT Score HU) (27).
(ASPECTS) was developed to establish a repro- Evaluation of coronal and sagittal images is
ducible scoring system for early ischemic changes essential because signs of sulcal hemorrhage on
at nonenhanced CT due to MCA stroke, since axial images are subtle and can easily be missed.
estimation of one-third of the MCA territory can The imaging findings of acute infarct can be
be challenging in practice (22). difficult to notice in the following locations: the
In the ASPECTS system, the brain is sepa- extreme vertex owing to volume averaging; the
rated into 10 discrete labeled areas of the deep inferior temporal lobes, where the gray-white
and superficial gray matter on each side. Start- matter junction is oriented axially; the occipital
ing with a maximum of 10 points on each side, lobes owing to frequent artifact caused by the ir-
the patient’s score is decreased by one point for regular contours of the skull; and the deep gray
each area of the brain with early ischemic change matter, particularly the caudate heads (Fig 3)
(Fig 2). As it is a topographic scoring strategy, it (4). The insula, caudate heads, and basal ganglia
requires only axial images for implementation. show early findings of proximal MCA thrombo-
The nonenhanced CT ASPECTS is often sis at nonenhanced CT and should be carefully
used by stroke neurologists to identify the extent evaluated.
of early ischemic change. Patients with AS- However, nonenhanced CT has low sensitiv-
PECTS of less than 7 have been shown to have ity for depiction of hyperacute and early acute
a lower chance of independent recovery after a hypoattenuating ischemic changes (28). Even
stroke (22). Automated quantification of AS- large territorial infarcts may not be seen until the
PECTS using machine learning is being imple- subacute phase. An acute territorial infarct visible
mented at some sites (23). More information at nonenhanced CT that is greater than one-
RG  •  Volume 39  Number 6 Potter et al  1721

Figure 1.  Left MCA infarction. (a) Axial nonenhanced CT image shows subtle hy-
poattenuation of the gray matter of the left insula in an “insular ribbon” configura-
tion (arrow), resulting in loss of GWD. In the setting of persistent proximal MCA
occlusion, the insular ribbon sign is an early indicator of a proximal MCA infarct
and is predictive of core infarct growth due to penumbral loss (15). More extensive
loss of GWD is seen throughout the left frontal lobe. (b) Nonenhanced CT image
obtained at follow-up demonstrates well-established left insular and extensive left
MCA hypoattenuation due to vasogenic edema, indicating infarction. (c) Nonen-
hanced CT image depicts loss of GWD in the right caudate head (arrow) and basal
ganglia, which are also important early indicators of infarction. (d) Nonenhanced
CT image obtained at follow-up shows the established right caudate and anterior
frontal lobe infarcts.

third of the MCA territory has been considered successful recanalization. This is not an indica-
a relative contraindication for administration of tion to withhold intravenous tPA according to
intravenous tPA, but the revised 2018 American the revised 2018 American Heart Association
Heart Association guidelines state that the extent guidelines (3,30).
or severity of the hypoattenuation should not be In addition to reviewing the nonenhanced
used as a criterion for withholding intravenous CT images for acute infarct, it is important
tPA owing to lack of sufficient evidence (3). to remember that mimics of acute stroke are
common in the emergent setting. A full routine
Sulci.—Focal sulcal effacement is also an impor- evaluation of the nonenhanced CT images is
tant early secondary sign of acute ischemia and essential to identify alternative diagnoses such as
can help identify subtle acute infarcts. intracranial masses, infection, and hemorrhage,
all of which can cause focal and nonspecific
Vessels.—The hyperattenuating vessel sign can be neurologic findings.
seen in the intracranial ICA, M1 and M2 seg-
ments of the MCA, and A1 and A2 segments of Pitfalls of Image Review.—Challenges in evalu-
the anterior cerebral artery (ACA) as well as the ation of nonenhanced CT images in stroke are
basilar and vertebral arteries (29) (Fig 4). This often caused by artifact from patient motion that
sign typically indicates the need for intervention. can be associated with alterations in conscious-
Although CT angiography should always be per- ness as well as from streak artifact from bone or
formed for definitive diagnosis in LVO, identifica- hardware (26). Skull base artifact at the petrous
tion of a hyperattenuating MCA sign greater than apices or posterior fossa often limits the visibility
8 mm in length also indicates that administration of subtle hypoattenuation in acute infarct (Fig
of intravenous tPA alone is not likely to achieve 3). Skull base streak artifact can also mimic the
1722  October Special Issue 2019 radiographics.rsna.org

Figure 2.  Regions of the brain

used to calculate ASPECTS.
Axial nonenhanced CT im-
ages show the labeled MCA
territories that are given one
point each for a score out of 10
points by using the ASPECTS
system. C = caudate, I = insula,
IC = internal capsule, L = lenti-
form nucleus, M1–M6 = corti-
cal regions.

hyperattenuating MCA sign. Comparison with the ography should also never be delayed by remov-
contralateral side will typically help resolve any ing the patient from the CT table for additional
ambiguity. clinical evaluation or to review the nonenhanced
Coronal and sagittal reformations can improve CT images at the scanner.
visualization and help one distinguish artifact, in- An 18–20-gauge needle should be confirmed
farct, and hemorrhage in some cases, especially at before the patient arrives, as power injection
locations such as the sylvian fissure, where axial of contrast material is required. The right arm
volume averaging can mimic GWD loss. In situ- should be used to limit streak artifact from hyper-
ations where axial images are tilted off axis, coro- attenuating contrast material crossing the mid-
nal and sagittal reformations can display areas of line, since the flow from the left brachiocephalic
nonanatomic linear hyperattenuation that cross vein enters the superior vena cava at an average
through different brain regions and structures. rate of 4–5 mL/sec in adults.
Evaluation of the basilar artery for acute Administration of iodinated contrast material
thrombus is another example where coronal and without first testing renal function in candidates
sagittal reformations can be helpful for distin- for EVT is acceptable, as the likelihood of contrast
guishing artifact from thrombus. material–induced nephropathy has been shown to
be very low relative to the potential for brain injury
CT Angiography from stroke (3,31). Correct timing of the injec-
tion of contrast material can be achieved by using
Technique a test bolus to calculate the imaging delay or by
Intravenous tPA administration should never be using semiautomated or automated triggering after
delayed to perform advanced CT, and CT angi- injection of the full dose of contrast material.
RG  •  Volume 39  Number 6 Potter et al  1723

Figure 3.  Potential pitfalls that can cause missed infarcts at nonenhanced CT. (a) Small focus of GWD loss at the right frontal
vertex (arrow), which could be missed owing to atypical location just deep to the skull, especially with thick (≥5-mm) axial sections.
(b) Left inferior cerebellar GWD loss (arrow) might also be overlooked owing to skull base beam-hardening artifact in this region.
(c) Subsequent diffusion-weighted image shows the evolving infarct. (d–f) The presence of nonstroke disease can also distract the
reader owing to satisfaction of search or anchoring bias, as well as nonspecific findings. (d, e) In another patient, low attenuation
due to vasogenic edema from a left temporal meningioma (white arrow in d) distracts from and overlaps with hypoattenuation from
an acute left anterior frontal infarct (black arrow in d). (f) Diffusion-weighted image shows the acute left anterior frontal infarct.
(g, h) Similarly, encephalomalacia from prior infarcts makes evaluation for new infarction difficult. (g) In a patient with old right
MCA–distribution encephalomalacia, CT image shows new left frontal GWD loss (arrow) due to an acute left MCA infarct, which
might be considered “age-indeterminate.” (h) Subsequent MR image shows the acute left MCA infarct.

Images should be acquired at a maximum erage should begin at the origins of the cervical
thickness of 1.5 mm and reconstructed in sec- vessels at the aortic arch and extend through the
tions overlapping by 50% or less. Anatomic cov- vertex (32). Multiplanar and maximum intensity
1724  October Special Issue 2019 radiographics.rsna.org

Figure 4.  Attenuating vessel sign. This is an early hyperacute ischemic finding at nonenhanced CT that can be seen before changes
of vasogenic edema are visible and allows anticipation of the presence of a clot at CT angiography for both anterior and posterior
circulation strokes (24). (a) A hyperattenuating MCA sign (arrow) is seen in the M1 segment. (b) Diffusion-weighted image shows
a large left MCA infarct. (c) Thrombus in an M2 branch vessel can create an MCA dot sign, seen in the left anterior sylvian fissure
(arrow). (d) Subsequent nonenhanced CT image shows a left MCA infarct in the territory of the left anterior temporal artery. (e) An
MCA dot sign is seen in another patient in the posterior sylvian fissure (arrow). (f) Follow-up diffusion-weighted image shows the
extent of the infarct, which corresponds to the occlusion of the inferior MCA M2 division.

projection (MIP) reformations can be acquired at (Fig 5) (34). Automated delineation of LVOs is
orthogonal planes for optimal evaluation, de- under investigation as well, which may help to
pending on the institution. speed detection, particularly in busy centers.
CT angiography can be performed as a single, CT angiographic data can be acquired at dual-
delayed, or multiphasic study. Multiphasic CT energy CT as well, which has the potential to pro-
angiography begins with an arterial imaging vide robust three-dimensional vessel reconstruction
sequence spanning from the aortic arch to the using bone subtraction techniques (16,35,36). The
vertex. The next two imaging sequences are lengthy postprocessing time has largely limited its
performed after a single contrast material bolus clinical application. In addition, iodine maps are
and are timed to depict the peak and late venous reconstructions that can be used to show attenuat-
phases in the head (33). The images acquired ing contrast material throughout an image. The
with multiphasic CT angiography can be rapidly attenuation can also be subtracted from an image to
and automatically reformatted into sequential create a virtual nonenhanced CT image (37).
axial MIP images that can be used for collateral
flow assessment and EVT decision making (33). Image Review
Four-dimensional CT angiography is per-
formed at some centers. This is a dynamic CT Tips or Pearls for Image Review.—Patients
angiographic technique in which bone-subtracted suspected of having LVO should undergo CT
MIP images can be viewed like digital subtraction angiography to expedite treatment with EVT.
angiograms are viewed over time. The images are Although the entire intracranial vasculature is
used to evaluate collateral filling or in a condensed depicted well at current-generation CT angi-
three-dimensional MIP image of all time points ography, occlusions of the intracranial ICA and
RG  •  Volume 39  Number 6 Potter et al  1725

Figure 5.  Four-dimensional CT angiography. Dynamic four-dimensional CT angiography data can be

reconstructed into time-resolved MIP reconstructions that appear similar to conventional catheter digital
subtraction angiograms. The earliest images show the arteries (A), followed by increased venous filling
over time (B, C) and dense opacification of the venous sinuses without visible arterial enhancement (D).

proximal MCA segments should be the focus at eral flow is an important determinant of the rate
imaging, since these vessels are the targets for of infarct progression (12).
EVT (3). The skull base and neck vessels are Multiple groups and studies have described
largely orthogonal to the axial imaging plane, and systems that help define collateral flow, includ-
their patency can be assessed quickly by viewing ing ASPECTS, the Miteff and Maas systems,
the axial CT angiography source images at the and the modified Tan scale (39–41). There is no
scanner console at the time of acquisition. consensus on which standardized scoring system
Overlapping thick-section MIP images of is optimal for collateral grading.
the circle of Willis in axial, coronal, and sagittal Several groups consider a malignant collateral
planes can help expedite identification of LVOs pattern at delayed phase CT to be characterized
(3-cm section thickness at overlapping inter- by contrast material failing to reach more than
vals of 5 mm). The more distal branches of the 50% of the intracranial vessels and tissue bed
circle of Willis can be difficult to interpret on of the at-risk MCA territory (42). Conversely,
thin-section CT angiography source images, and a good collateral pattern can be defined as a
using thick-section reconstructions increases their symmetric or nearly symmetric leptomeningeal
visibility. CT technologists can produce MIP flow when comparing the hemisphere affected
reformations at the scanner console in less than 1 by ischemia with the contralateral hemisphere.
minute (28,38). In the setting of an intracranial ICA or proxi-
Collateral flow assessment of the intracranial mal MCA occlusion, infarct core growth and
circulation at CT angiography can be used to penumbral loss progress more rapidly in pa-
help assess the potential risks (ie, ICH) versus tients with malignant collateral patterns (ie, fast
benefits (ie, limited infarct growth and improved progressors), whereas infarct core growth and
clinical outcomes) of EVT. The quality of collat- penumbral loss progress more slowly in patients
1726  October Special Issue 2019 radiographics.rsna.org

with good collateral patterns (ie, slow progres- M2 division is located anteriorly in the sylvian
sors) (43,44). fissure and primarily supplies the superior
ASPECTS is one of the best-known collateral frontal-parietal regions. A receptive aphasia
score systems and compares the filling of the (Wernicke aphasia) in the dominant hemisphere
intracranial arteries at all three phases of mul- or visual field problems from disruption of the
tiphase CT angiography. When there are no or visual tracts suggest occlusion of the inferior M2
just a few vessels visible in the ischemic territory division of the MCA (Fig 8b). The inferior M2
at any phase, the collateral score is given a 0 or division is in the posterior sylvian fissure and
1, indicating poor collaterals. If there is normal supplies the temporal-parietal regions.
symmetric vessel filling, or a delay of a single A combination of these symptoms suggests oc-
phase with similar extent of vessel filling, collater- clusion more proximally at the MCA bifurcation
als are given a score of 5 or 4, respectively, indi- (Fig 8c). Complete hemiplegia and hemisensory
cating good collaterals. Intermediate collaterals loss suggest a proximal M1 segment MCA occlu-
with scores of 2 or 3 demonstrate a delay of two sion, which includes occlusion of the lenticulostri-
phases with either a similar or decreased extent of ate vessels that supply the internal capsule (Fig 8d).
perfused vessels (Fig 6) (45). Isolated focal arm, hand, and face weakness
Modified by the time of stroke onset, the de- suggests a more distal occlusion, usually in an
gree of collateral flow is an important indicator of M3 or M4 branch. These occlusions are depicted
the potential efficacy of EVT. Malignant or poor well on sagittal thick-section MIP images.
collateral flow is associated with a low probabil- Complete hemiplegia and hemisensory loss
ity of penumbral salvage or clinical benefit from can also be caused by occlusion of the intracra-
EVT unless thrombectomy is performed very nial ICA. Dense monocular blindness can also
quickly after onset of stroke (33,46) (Fig 7). occur and is due to an occlusion of the ophthal-
Complete assessment of the intracranial mic artery. However, occlusion of the ICA may
vasculature can seem daunting. MIP and multi- be more difficult to detect. The symptoms can
planar reformations are indispensable for efficient range from subtle to severe, and there may be no
assessment. Coronal reformations rapidly depict symptoms if the circle of Willis is intact and there
the proximal MCA, ACA, and terminal ICA, is good contralateral flow (47).
while sagittal reformations demonstrate the ACA Contralateral lower extremity weakness and
and distal MCA vessels. Evaluating the overall sensory loss suggest occlusion of the ACA (Fig
symmetry of vessels can help identify regions of 8e). Occlusion of the A1 segment of the ACA is
relative hypoperfusion, particularly in the cere- an easily overlooked cause of infarct in the ante-
bral convexities. Assessing hypoperfused regions rior lentiform nuclei and caudate head.
can help differentiate stroke from mimics such as Visual changes are the most frequent sign
seizure and directs the reader’s attention to likely of posterior cerebral arterial occlusion. These
areas of vessel occlusion. changes typically manifest as homonymous
Knowledge of common stroke syndromes is hemianopsia owing to infarction of the occipital
part of an efficient evaluation in conjunction visual cortex.
with interpreting CT angiograms and identify- Profound alteration of consciousness or cranial
ing vessel occlusions. Although a patient’s signs nerve deficits suggest brainstem involvement and
and symptoms may not be available at the time of basilar thrombosis. Readers should always check
their imaging orders, discussing the results of the for the presence of these conditions, and they are
initial nonenhanced CT with referring provid- usually clearly depicted at CT angiography.
ers often offers the opportunity to obtain a brief Although symptoms of ataxia, vertigo, and
clinical history that can help focus the CT an- dizziness are nonspecific, they should prompt
giography assessment. Knowledge of symptoms careful evaluation of the posterior circulation,
such as laterality, acuity of onset, and specific including the posterior inferior cerebellar artery
functional involvement (eg, aphasia, hemiparesis, (PICA) (Fig 8f) and anterior inferior cerebellar
hemisensory deficit, and hemineglect) is essential artery (AICA). The distribution of vessels supply-
to correlate specific imaging findings with the ing the cerebellum is highly variable. The AICA
acute manifestation of stroke. and PICA develop in a reciprocal relationship,
For example, contralateral arm and face and a single artery or side could be dominant. If
weakness with or without a sensory disturbance, the ipsilateral PICA and AICA are not visible at
which might also include an expressive aphasia imaging, this may indicate vascular occlusion and
(Broca aphasia) in the dominant hemisphere prompts careful evaluation (47).
(typically the left hemisphere), suggests occlu- The degree and location of vessel narrowing
sion of the superior M2 division or anterior or partial obstruction should be included in the
temporal MCA branches (Fig 8a). The superior final interpretation. Intracranial arterial stenosis
RG  •  Volume 39  Number 6 Potter et al  1727

Figure 6.  (a) Good collateral flow. A and B, Initial nonenhanced CT images show some patchy GWD loss in the left anterior frontal
lobe. C, Axial MIP CT angiogram shows occlusion at the left MCA bifurcation. E–G, MIP images of the three sequential CT angiogra-
phy phases show that the extent of the visualized vessels on the affected side has a single phase delay, consistent with good collater-
als. D and H, Diffusion-weighted images after treatment with intravenous tPa and EVT demonstrate that only a small focus of acute
infarct remains. (b) Poor collateral flow. A and B, Initial nonenhanced CT images depict no loss of GWD. C, Axial MIP image from CT
angiography shows occlusion of distal right M1. E–G, MIP images of the three sequential CT angiography phases show few opacified
vessels on the affected side, consistent with poor collaterals. The patient was treated with intravenous tPA and EVT. D and H, Follow-
up nonenhanced CT images show a large right MCA infarct that developed after treatment.
1728  October Special Issue 2019 radiographics.rsna.org

Figure 7.  Collateral flow at multiphase CT angiography with right-sided M1 occlusion. (a) ColorViz software summation images—
with color-coded early arterial (red), late arterial or venous (green), and delayed venous (blue) phases—show predominantly green
vessels (arrowheads) ipsilateral to the right MCA occlusion (arrow), indicating symmetric but delayed collaterals relative to the
predominantly red vessels in the unaffected hemisphere and suggesting good pial arterial filling collateral flow. (b) Corresponding
conventional multiphase CT angiograms. The patient received intravenous tPA. (c) Follow-up diffusion-weighted images 24 hours
after ictus show hyperintense signal consistent with acute infarction in the caudate head and lentiform nucleus only with sparing of
the cortex, where good collateral flow was present.
RG  •  Volume 39  Number 6 Potter et al  1729

Figure 8.  Common stroke syndromes. (a) Axial CT angiogram (left) shows the left superior M2 vessel (arrow). Along with the ante-
rior temporal artery, the M2 segment perfuses the left frontal lobe and frontal operculum (purple region on nonenhanced CT image
[right]). The frontal operculum contains the Broca area in the dominant hemisphere, and infarction may cause a nonfluent aphasia.
(b) Axial CT angiogram (left) shows the left inferior M2 vessel (arrow), which typically perfuses the inferior parietal and temporal
lobes (yellow region on nonenhanced CT image [right]). Infarction in this region in the dominant hemisphere could cause a fluent
(Wernicke) aphasia. (c) Axial CT angiogram (left) shows the left MCA bifurcation (arrow). Occlusion can cause a pattern similar to
occlusion of both M2 vessels but without significant involvement of the deep gray matter structures (red region on nonenhanced CT
image [right]). (d) Axial CT angiogram (left) shows the proximal left M1 segment of the MCA (arrow). Occlusion in this segment typi-
cally causes hemiplegia and hemisensory loss owing to involvement of the cortex, basal ganglia, and internal capsule (blue region on
nonenhanced CT image [right]). (e) Axial CT angiogram (left) shows the proximal left A2 segment of the ACA (arrow). The vascular
territory of the ACA is parafalcine, and occlusion can cause contralateral lower extremity weakness and sensory loss (orange region
on nonenhanced CT image [right]). (f) Axial CT angiogram (left) shows the left posterior inferior cerebellar artery (PICA) (arrow). The
PICA often supplies the inferior medial cerebellum (blue region on nonenhanced CT image [right]), and infarction can cause a variety
of symptoms, including ataxia.

may be clinically significant and can create falsely of this review. Thrombus density, distance to the
delayed arrival time values on perfusion CT im- terminal ICA, and length longer than 8 mm have
ages (48–50). However, management of proximal been associated with worse functional outcome
intracranial vascular stenoses is beyond the scope but are not used as criteria for EVT (51).
1730  October Special Issue 2019 radiographics.rsna.org

Figure 9.  Potential false-negative interpretation of a posterior cerebral artery (PCA) occlusion in a patient with new-onset vision
changes. (a) Axial nonenhanced CT image shows mild hypoattenuation with loss of GWD in the medial right occipital lobe (arrow).
(b) CT angiogram shows a patent proximal right PCA and a more distal and possibly contiguous enhanced vascular structure lateral
to the midbrain and pons (arrow). (c) Axial thin-section source image from CT angiography shows that the distal structure is the basal
vein of Rosenthal (arrow), a common confounder when assessing the patency of the PCA, which is occluded in this case.

It is important to evaluate initial CT angio- stroke can be indispensable, particularly in con-

grams of the neck at the scanner console to iden- firmation of chronic vessel occlusions.
tify areas of critical stenosis, dissection, or vessel Late-phase CT angiography can lead to exten-
tortuosity. These findings can help inform the sive venous attenuation that complicates interpre-
plan for EVT. CT angiograms can depict clini- tation by introducing distracting vessels and veins
cally relevant severe ICA stenosis at the cervical that can be confused with patent arteries. When
ICA origin (>70% narrowing), although heavy venous contamination is present, it is essential to
circumferential calcifications can lead to overesti- carefully trace the arteries, particularly the poste-
mation of the degree of narrowing. rior cerebral arteries (PCAs) to avoid mistaking a
Atherosclerotic disease of carotid origin and patent basal vein of Rosenthal for a patent distal
embolus associated with untreated atrial fibril- PCA segment (Fig 9).
lation are some of the most common causes of Hypoattenuation on CT angiography source
stroke in adults (52). Spontaneous arterial dissec- images has been studied as a correlate of low
tion is more common in patients younger than 40 CBV in brain regions. CT angiography may be
years or after acute vascular trauma. superior to nonenhanced CT alone for identifica-
Multiplanar and MIP reconstructions aid in tion of parenchymal ischemic change (28). Stud-
rapid screening of patency and vessel contour of ies have also shown that CT angiography source
the cervical carotid arteries as well as in dem- images and diffusion-weighted images are more
onstrating underlying vascular abnormalities, accurate at depicting core infarct volume than
such as fibromuscular dysplasia. However, MIP nonenhanced CT images (28,53).
images may poorly depict small intimal flaps in Like perfusion CT, CT angiography is flow and
the carotid arteries, and evaluation of the verte- timing dependent and demonstrates tissue viabil-
bral arteries is complicated by the bony foramina. ity on the basis of hemodynamic parameters. For
Evaluation of the cervical arteries in multiple this reason, CT angiography source images have
planes is important, as pseudoaneurysms and dis- the potential to overestimate core infarct volume
sections may be better visualized in the coronal in regions of poor but not critical hypoperfusion.
and sagittal planes than on axial images. This overestimation can lead to inaccurate patient
At perfusion CT, it is important to identify selection and underuse of EVT in patients who
extracranial narrowing that decreases the rate might benefit from treatment (28,54,55).
of intracranial flow. This can cause an abnormal Collateral vessels can be viewed at single-
appearance at perfusion CT and lead to misinter- phase, dual-phase, or multiphase CT angiogra-
pretation of imaging findings, as discussed in the phy. Images must be acquired and reviewed by
next section (48, 49). using at least arterial and delayed phases. The
vessels can then be classified as good or poor by
Pitfalls of Image Review.—CT angiography can comparing the attenuation to that of the contra-
be challenging owing to the complexity of the lateral uninvolved hemisphere (33).
anatomy, the number of images obtained, and In regions of ischemia caused by long-segment
patient cooperation. Images obtained before a proximal vascular occlusion, there may be a de-
RG  •  Volume 39  Number 6 Potter et al  1731

Figure 10.  Poor collateral flow with early ischemia in a hospital visitor who collapsed after abrupt onset of left hemiparesis. (a) Non-
enhanced CT image less than 30 minutes after stroke shows a hyperattenuating MCA sign (arrow) but is otherwise normal without
subtle or well-established ischemic hypoattenuation, not surprising given the hyperacute manifestation. (b) Coronal thick-section
MIP CT angiogram shows a T occlusion of the right ICA terminus involving the proximal M1 and A1 segments (arrow). (c) Axial
thick-section MIP image of collateral vessels obtained in the arterial phase shows absent enhancement in over 50% of the territory at
risk, suggestive of a malignant pattern with both vascular and parenchymal window and level settings. Given the hyperacute mani-
festation, normal nonenhanced CT findings, and immediate availability of the interventional team in a patient with a poor collateral
pattern likely to progress to a large infarction, the decision was made to perform rapid EVT after thrombolysis with intravenous tPA.
(d) Diffusion-weighted images obtained after removal of a right M1 thrombus show only trace final infarct in the right MCA territory.

layed arrival time of contrast material with rela- Perfusion CT

tively maintained collateral flow. In this context,
reviewing arterial single-phase MIP images may Technique
be misleading. Dynamic or first-pass perfusion CT is performed
For example, a region with intermediate col- by sequentially imaging a defined section of tis-
laterals distal to a long-segment occlusion might sue after a single high-flow bolus of contrast ma-
be labeled incorrectly as a poor or malignant terial is administered. The same section is imaged
pattern if the images are obtained too early in the multiple times in cine mode as the bolus passes
arterial phase. This may inappropriately exclude to track the degree of attenuation at both the tis-
some patients from receiving treatment with EVT sue and arterial levels as a function of time.
(56,57) (Fig 10). Older CT scanners with narrow detectors pro-
The ASPECTS system partially addresses this vided craniocaudal coverage of a limited section
pitfall by using multiphase CT angiography. Col- of the brain only. Modern scanners with helical
lateral flow is categorized as good, intermediate, capability and broader z-direction detectors can
or poor on the basis of interhemispheric compari- perform whole-brain perfusion CT.
sons corrected for contrast material arrival time. As with CT angiography, an 18- or 20-gauge
Other advantages of multiphase CT angiog- peripheral needle is preferred to achieve optimal
raphy include low cost of implementation and contrast agent flow rates with a saline chaser of
relative insensitivity to patient motion (33). at least 15–20 mL. A minimum contrast material
1732  October Special Issue 2019 radiographics.rsna.org

Figure 11.  Perfusion CT findings favorable for EVT in an 85-year-old woman with atrial fibrillation who had received intravenous tPA
for right MCA syndrome. CT angiography showed a right M1 occlusion extending to the inferior M2 division (not shown). CT perfu-
sion images 10 hours after last known normal show a small 15-mL region of critical cerebral blood flow (CBF) reduction (CBF < 30%
[purple]) surrounded by a larger 75-mL region of delayed transit time (time to maximum of the tissue residue function [Tmax] > 6 sec-
onds [green]) for a Tmax-CBF mismatch ratio of 5. On the basis of these favorable perfusion CT findings, the patient was referred for EVT.

injection of 40 mL with a minimum rate of 4 mL/ quality perfusion CT images, it is also important
sec by using power injection is recommended. A to assess for patient motion, adequate attenuation
higher rate of injection forms a tighter bolus and of the contrast material bolus, and sufficiently
helps improve hemodynamic maps. long acquisition time to avoid truncation of the
Technique should be optimized to the particu- tissue and vessel time-attenuation curves (36,63).
lar scanner, but 70–90 kVp and 100–200 mAs
are recommended to keep the total radiation dose Image Review
as low as reasonably achievable. In cine mode,
one image per second should be acquired over a Tips and Pearls for Image Review.—Using the
period of at least 50–60 seconds. Perfusion CT time-attenuation curves generated from initial
can be performed concurrently with or separately ROI selection, color-coded perfusion parameter
from CT angiography (58,60–62). maps can be generated automatically or semiau-
Automated or semiautomated postprocess- tomatically. Parameters demonstrated on these
ing of the CT perfusion data generates multiple maps include CBF, time to maximum of the
perfusion maps (38,58,59). Perfusion parameter tissue residue function (Tmax), CBV, and mean
calculations are performed using deconvolu- transit time. Only CBF and Tmax have been widely
tion approaches. The methodologies of different studied in recent randomized clinical trials.
deconvolution algorithms are beyond the scope of For example, in the CT arms of the DAWN
this review (63). and DEFUSE 3 trials, automated perfusion
Because attenuation change is linearly associ- CT software estimated core infarct volume on
ated with the concentration of iodinated contrast the basis of a less than 30% threshold for CBF
material in a region, absolute values of perfusion reduction and penumbral volume on the basis of
parameters can be calculated. Cerebral blood a threshold greater than 6 seconds for prolonga-
flow (CBF) and time to maximum enhance- tion of Tmax (8,9). Mismatch ratios and ischemic
ment are among the most accurate values for use volumes can be calculated, displayed, and used
in acute stroke evaluation (64,65). Automated for clinical trial enrollment, depending on the
software platforms can generate qualitative and specific eligibility criteria (Figs 11, 12).
quantitative maps of ischemic lesion volumes The accuracy of perfusion CT for helping
from perfusion CT data with automatically distinguish large (>100 mL) from small (<50–70
selected arterial and venous inputs. The arterial mL) core infarct volumes in EVT selection has
inflow region of interest (ROI) is usually located been studied compared with a DWI reference
in the A2 segment of the ACA or the M2 seg- standard (55,66–68). Use of perfusion CT has
ment of the MCA. The venous ROI is selected been increasing since its successful use in helping
over the dural venous confluence. determine eligibility for enrollment in subsets of
Even small differences in vessel ROI selection patients in the recent late-window DAWN and
can result in substantial differences on the maps DEFUSE 3 trials (8,9).
and in reported tissue values, so assessing the im- Perfusion CT is grouped with MRI in the
aging input quality is important. To ensure high- updated 2018 American Heart Association–
RG  •  Volume 39  Number 6 Potter et al  1733

Figure 12.  Perfusion CT mismatch findings unfavorable for EVT in a 54-year-old

woman with a history of coronary artery disease and hypertension who awoke with
right hemispheric symptoms. (a) Axial CT angiogram shows complete occlusion of the
right ICA and distal M1 and M2 (arrow). (b) Perfusion CT images show a large 105-mL
region of critical cerebral blood flow (CBF) reduction (CBF < 30% [purple]) surrounded
by a larger 155-mL region of delayed transit time (Tmax > 6 sec [green]) for a Tmax-CBF
mismatch ratio of 1.5. The patient was not a candidate for intravenous tPA; on the basis
of these unfavorable findings, EVT was not performed.

American Stroke Association guidelines for early when imaging and other eligibility criteria from
management of patients with acute ischemic RCTs [randomized controlled trials] showing
stroke, which states, “in selected patients with benefit are being strictly applied in selecting pa-
AIS [acute ischemic stroke] within 6 to 24 hours tients for mechanical thrombectomy” (3, p e59).
of last known normal who have LVO in the an- However, in the recent Hermes meta-analysis
terior circulation, obtaining CTP [CT perfusion of early-window EVT trials and in a subgroup
imaging], DW-MRI [diffusion-weighted MRI], or analysis of the DEFUSE 3 late-window trial
MRI perfusion is recommended to aid in patient that were presented at the 2018 American Heart
selection for mechanical thrombectomy, but only Association International Stroke Conference,
1734  October Special Issue 2019 radiographics.rsna.org

DWI was found to be more efficient than perfu- overestimate infarct core volume, since perfusion
sion CT when used to help select candidates for parameter values may be underestimated in voxels
EVT. DWI was found to have higher odds ratios with incomplete bolus tracking (77-79).
for selecting patients more likely to experience Poor signal-to-noise ratio can result when
clinical improvement and functional indepen- insufficient contrast material reaches the imaging
dence (69). voxel. This can be caused by factors such as poor
Given this, it is not surprising that there is cardiac output associated with atrial fibrillation
currently little consensus on the optimal imag- (commonly associated with acute embolic stroke)
ing strategy for late-window EVT selection. Use or unsuspected venous or arterial stenoses delay-
of DWI, CT angiography of collateral ves- ing the arrival of contrast material at the circle of
sels, and perfusion CT for core infarct volume Willis (48,63,49).
assessment varies widely between different Moreover, thresholds for determining core and
comprehensive stroke centers and in different penumbra can vary between vendors and be-
clinical situations (8,9,66,70). tween postprocessing platforms (68,69). Thresh-
Additionally, recent trials have suggested that olds for determining irreversible ischemia likely
advanced CT and MRI can be used to extend vary with time after stroke, quality of collateral
the time window for intravenous tPA adminis- flow, and ischemic preconditioning (11,55).
tration (71,72). The recent EXTEND (Extend- Perfusion CT thresholds can also be unreliable
ing the Time for Thrombolysis in Emergency in the presence of old infarcts, partial reperfu-
Neurological Deficits) trial results have shown sion, or hyperemia associated with compensatory
that patients benefit from treatment with intra- vasodilatation.
venous tPA at 9 hours or after a wake-up stroke
when they are selected using perfusion CT or Stroke Mimics
perfusion MRI (73). Potential acute stroke mimics include intracranial
Tenecteplase is a newer intravenous tPA that masses due to primary or metastatic malignancy,
is currently being investigated in late-window cerebritis, cerebral abscess, and posterior revers-
clinical trials. When perfusion CT is used for ible encephalopathy syndrome (PRES). Most
treatment selection, tenecteplase may provide of these pathologic processes cause vasogenic
stronger benefits up to 6 hours after symptom edema and accentuation of—rather than loss
onset (74). of—GWD. Low-grade gliomas and other hypoat-
tenuating lesions without surrounding vasogenic
Pitfalls of Image Review.—Potential pitfalls edema may require use of MRI for definitive
of perfusion CT include motion artifact, poor diagnosis (Fig 13).
signal-to-noise ratio from a suboptimal con- Increased gyriform enhancement with mass ef-
trast material bolus, faulty arterial and venous fect in subacute infarction may be indistinguish-
input functions, and truncation of the tissue and able from high-grade glioma. Metastases, cereb-
vascular time-attenuation flow curves from a ritis, abscess, demyelinating disease, and PRES
shortened acquisition time. can all mimic subacute infarction by blurring or
It is important to identify the degree of effacing GWD owing to mass effect and displace-
motion to decide whether the images are of ment (Fig 13).
adequate quality. Moderate-to-severe motion ar- Contrast material staining after EVT can
tifact occurs frequently, and one study reported mimic acute hemorrhage. Dual-energy CT iodine
that it occurred in 25% of patients (75). maps and virtual nonenhanced CT images can
Although automated perfusion software gen- help confirm that intracranial hyperattenuation is
erates volumetric data for core and penumbra, it due to iodine leakage through a nonintact blood-
is important to be aware of its potential techni- brain barrier (16,79).
cal and clinical pitfalls. The time-attenuation Perfusion CT can also be useful in stroke dif-
curves should be evaluated to ensure appropri- ferential diagnosis by helping distinguish wedge-
ate location of the ROI in a vessel that runs shaped perfusion deficits caused by arterial
nearly perpendicular to the imaging plane to occlusive emboli from other flow derangements,
avoid volume averaging with the vessel wall and such as those caused by seizure, hypoglycemia, or
surrounding structures. Slow flow in a stenotic hyperglycemia (49).
major vessel selected for the arterial input func-
tion (eg, chronic carotid occlusion) will result in Conclusion
inaccurate relative parameter values. CT evaluation of acute ischemic stroke is robust,
The effect of truncated time-attenuation curves rapid, and widely available. The DAWN and
has also been a concern (68,69,79). Perfusion DEFUSE 3 trials proved that EVT results in
CT protocols less than 60 seconds in length can improved outcomes and functional independence
RG  •  Volume 39  Number 6 Potter et al  1735

Figure 13.  Stroke mimics caused by nonspecific hypoattenuating lesions at nonenhanced CT. (a) Nonenhanced CT image in a
58-year-old man with confusion shows cortical effacement from a well-circumscribed rounded area of hypoattenuation with loss
of GWD in the medial right parietal lobe (arrow). (b) Subsequent FLAIR (fluid-attenuated inversion-recovery) image shows that the
lesion is hyperintense. DWI showed no diffusion restriction in the lesion, which was confirmed at biopsy to be a low-grade glioma.
(c) Admission nonenhanced CT image in another patient shows an irregular lesion with patchy hypoattenuation (arrow). (d) Contrast-
enhanced T1-weighted MR image shows serpentine irregular surrounding enhancement. These findings are suggestive of a subacute
evolving infarct, but the lesion was proved to represent a left frontal glioblastoma multiforme. (e) Nonenhanced CT image in another
patient shows subcortical and gray-white matter junction left frontal lobe hypoattenuation (arrow), which is suggestive of vasogenic
edema but nonspecific. (f) Diffusion-weighted image shows diffusion restriction. Along with peripheral enhancement on postcontrast
T1-weighted images (not shown), these findings helped confirm the final diagnosis of cerebral abscess. (g) Nonenhanced CT image
in a young patient shows loss of GWD in the right insula and external capsule (arrow). (h) FLAIR image shows additional involvement
of the right mesial temporal lobe, amygdala, and hypothalamic regions. The final diagnosis was herpes encephalitis.
1736  October Special Issue 2019 radiographics.rsna.org

even up to 24 hours after a stroke. Because of 8. Nogueira RG, Jadhav AP, Haussen DC, et al. Thrombec-
tomy 6 to 24 Hours after Stroke with a Mismatch between
this, nonenhanced CT and CT angiography have Deficit and Infarct. N Engl J Med 2018;378(1):11–21.
become the community standards for selecting 9. Albers GW, Marks MP, Kemp S, et al. Thrombectomy for
patients for intravenous tPA and EVT in both Stroke at 6 to 16 Hours with Selection by Perfusion Imaging.
N Engl J Med 2018;378(8):708–718.
early and late time windows. 10. Sims JR, Gharai LR, Schaefer PW, et al. ABC/2 for rapid
Although DWI is the reference standard for clinical estimate of infarct, perfusion, and mismatch volumes.
determining core infarct lesion volume, CT an- Neurology 2009;72(24):2104–2110.
11. Schellinger PD, Bryan RN, Caplan LR, et al. Evidence-
giography and perfusion CT may be used when based guideline: the role of diffusion and perfusion MRI
DWI is not available. These uses of CT may help for the diagnosis of acute ischemic stroke: report of the
distinguish patients with large infarct cores—who Therapeutics and Technology Assessment Subcommit-
tee of the American Academy of Neurology. Neurology
are least likely to benefit from and most likely to 2010;75(2):177–185. [Published correction appears in
be harmed by EVT—from patients with small Neurology 2010;75(10):938.] https://doi.org/10.1212/
infarct cores. WNL.0b013e3181e7c9dd.
12. González RG. Imaging-guided acute ischemic stroke therapy:
Awareness of common findings, pearls, and from “time is brain” to “physiology is brain.” AJNR Am J
pitfalls of multimodal stroke CT evaluation and Neuroradiol 2006;27(4):728–735.
interpretation has therefore become essential for 13. Kanekar SG, Zacharia T, Roller R. Imaging of stroke. II.
Pathophysiology at the molecular and cellular levels and
providers caring for patients with acute stroke. corresponding imaging changes. AJR Am J Roentgenol
2012;198(1):63–74.
Disclosures of Conflicts of Interest.—A.S.V. Activities related 14. Kelly AG, Hellkamp AS, Olson D, Smith EE, Schwamm
to the present article: disclosed no relevant relationships. Activi- LH. Predictors of rapid brain imaging in acute stroke:
ties not related to the present article: institution received grants analysis of the Get with the Guidelines—Stroke program.
from Cerenovus, ENDOLOW trial, GE Healthcare, Imaging Stroke 2012;43(5):1279–1284.
Core Lab, and Johnson & Johnson. Other activities: disclosed 15. ACR-ASNR-SPR Practice Parameter for the Performance
no relevant relationships. M.G. Activities related to the present of Computed Tomography (CT) of the Brain. https://www.
article: disclosed no relevant relationships. Activities not related acr.org/-/media/ACR/Files/Practice-Parameters/CT-Brain.
to the present article: author received payment from Medtronic, pdf. Revised 2015. Accessed April 2, 2019.
Mentice, Microvention, and Stryker for consulting; institution 16. Potter CA, Sodickson AD. Dual-Energy CT in Emergency
received grants from Medtronic and Stryker; author received Neuroimaging: added Value and Novel Applications. Ra-
payment from GE Healthcare for a licensing agreement. Other dioGraphics 2016;36(7):2186–2198.
activities: disclosed no relevant relationships. M.H.L. Activi- 17. Pomerantz SR, Kamalian S, Zhang D, et al. Virtual
ties related to the present article: disclosed no relevant relation- monochromatic reconstruction of dual-energy unenhanced
ships. Activities not related to the present article: received grant head CT at 65-75 keV maximizes image quality com-
from GE Healthcare; consultant to Takeda Pharmaceuticals pared with conventional polychromatic CT. Radiology
and GE Healthcare; institution received research funding from 2013;266(1):318–325.
GE Healthcare; institution received research software license 18. Perry JJ, Stiell IG, Sivilotti ML, et al. Sensitivity of computed
from Siemens Healthineers. Other activities: patents pending tomography performed within six hours of onset of headache
for portable electrical impedance spectroscopy device, artificial for diagnosis of subarachnoid haemorrhage: prospective
intelligence for use in CT of brain hemorrhage, and artifical cohort study. BMJ 2011;343:d4277.
intelligence to depict collaterals. 19. Lee H, Yune S, Mansouri M, et al. An explainable deep-
learning algorithm for the detection of acute intracranial
haemorrhage from small datasets. Nat Biomed Eng
References 2019;3(3):173–182.
1. Tomandl BF, Klotz E, Handschu R, et al. Comprehensive 20. Kamalian S, Kemmling A, Borgie RC, et al. Admission
imaging of ischemic stroke with multisection CT. Radio- insular infarction >25% is the strongest predictor of large
Graphics 2003;23(3):565–592. mismatch loss in proximal middle cerebral artery stroke.
2. National Institute of Neurological Disorders and Stroke rt-PA Stroke 2013;44(11):3084–3089.
Stroke Study Group. Tissue plasminogen activator for acute 21. Kucinski T. Unenhanced CT and acute stroke physiology.
ischemic stroke. N Engl J Med 1995;333(24):1581–1587. Neuroimaging Clin N Am 2005;15(2):397–407, xi–xii.
3. Powers WJ, Rabinstein AA, Ackerson T, et al. 2018 22. Pexman JH, Barber PA, Hill MD, et al. Use of the Alberta
Guidelines for the Early Management of Patients With Stroke Program Early CT Score (ASPECTS) for assess-
Acute Ischemic Stroke: A Guideline for Healthcare Profes- ing CT scans in patients with acute stroke. AJNR Am J
sionals From the American Heart Association/American Neuroradiol 2001;22(8):1534–1542.
Stroke Association. Stroke 2018;49(3):e46–e110. [Pub- 23. Kuang H, Najm M, Chakraborty D, et al. Automated AS-
lished corrections appear in Stroke 2018;49(3):e138 and PECTS on Noncontrast CT Scans in Patients with Acute
Stroke 2018;49(6):e233–e234.] https://doi.org/10.1161/ Ischemic Stroke Using Machine Learning. AJNR Am J
STR.0000000000000158. Neuroradiol 2019;40(1):33–38.
4. Tomura N, Uemura K, Inugami A, Fujita H, Higano 24. Leys D, Pruvo JP, Godefroy O, Rondepierre P, Leclerc X.
S, Shishido F. Early CT finding in cerebral infarction: Prevalence and significance of hyperdense middle cerebral
obscuration of the lentiform nucleus. Radiology 1988; artery in acute stroke. Stroke 1992;23(3):317–324.
168(2):463–467. 25. Mair G, Boyd EV, Chappell FM, et al. Sensitivity and speci-
5. Truwit CL, Barkovich AJ, Gean-Marton A, Hibri N, ficity of the hyperdense artery sign for arterial obstruction
Norman D. Loss of the insular ribbon: another early CT in acute ischemic stroke. Stroke 2015;46(1):102–107.
sign of acute middle cerebral artery infarction. Radiology 26. Barber PA, Demchuk AM, Hudon ME, Pexman JH, Hill MD,
1990;176(3):801–806. Buchan AM. Hyperdense sylvian fissure MCA “dot” sign: a
6. Dippel DW, Du Ry van Beest Holle M, van Kooten F, CT marker of acute ischemia. Stroke 2001;32(1):84–88.
Koudstaal PJ. The validity and reliability of signs of early 27. Lev MH, Farkas J, Gemmete JJ, et al. Acute stroke: im-
infarction on CT in acute ischaemic stroke. Neuroradiology proved nonenhanced CT detection—benefits of soft-copy
2000;42(9):629–633. interpretation by using variable window width and center
7. Wardlaw JM, Mielke O. Early signs of brain infarction at level settings. Radiology 1999;213(1):150–155.
CT: observer reliability and outcome after thrombolytic treat- 28. Camargo EC, Furie KL, Singhal AB, et al. Acute brain
ment— systematic review. Radiology 2005;235(2):444–453. infarct: detection and delineation with CT angiographic
RG  •  Volume 39  Number 6 Potter et al  1737

source images versus nonenhanced CT scans. Radiology ischemia: patterns and pitfalls. AJNR Am J Neuroradiol
2007;244(2):541–548. 2010;31(9):1552–1563.
29. Goldmakher GV, Camargo EC, Furie KL, et al. Hyperdense 50. Holmstedt CA, Turan TN, Chimowitz MI. Atherosclerotic
basilar artery sign on unenhanced CT predicts thrombus intracranial arterial stenosis: risk factors, diagnosis, and
and outcome in acute posterior circulation stroke. Stroke treatment. Lancet Neurol 2013;12(11):1106–1114.
2009;40(1):134–139. 51. Borst J, Berkhemer OA, Santos EMM, et al. Value of Throm-
30. Riedel CH, Jensen U, Rohr A, et al. Assessment of thrombus bus CT Characteristics in Patients with Acute Ischemic
in acute middle cerebral artery occlusion using thin-slice Stroke. AJNR Am J Neuroradiol 2017;38(9):1758–1764.
nonenhanced computed tomography reconstructions. Stroke 52. Benjamin EJ, Virani SS, Callaway CW, et al. Heart
2010;41(8):1659–1664. Disease and Stroke Statistics—2018 Update: A Report
31. Dittrich R, Akdeniz S, Kloska SP, et al. Low rate of From the American Heart Association. Circulation
contrast-induced nephropathy after CT perfusion and 2018;137(12):e67–e492. [Published correction appears in
CT angiography in acute stroke patients. J Neurol Circulation 2018;137(12):e493.] https://doi.org/10.1161/
2007;254(11):1491–1497. CIR.0000000000000558.
32. ACR-ASNR-SPR Practice Parameter for the Performance 53. Schramm P, Schellinger PD, Fiebach JB, et al. Comparison
and Interpretation of Cervicocerebral Computed Tomog- of CT and CT angiography source images with diffusion-
raphy Angiography (CTA) of the Brain. https://www.acr. weighted imaging in patients with acute stroke within 6
org/-/media/ACR/Files/Practice-Parameters/CervicoCer- hours after onset. Stroke 2002;33(10):2426–2432.
ebralCTA.pdf. Revised 2015. Accessed April 2, 2019. 54. Pulli B, Schaefer PW, Hakimelahi R, et al. Acute ischemic
33. Menon BK, d’Esterre CD, Qazi EM, et al. Multiphase CT stroke: infarct core estimation on CT angiography source
Angiography: a New Tool for the Imaging Triage of Patients images depends on CT angiography protocol. Radiology
with Acute Ischemic Stroke. Radiology 2015;275(2):510–520. 2012;262(2):593–604.
34. Kortman HG, Smit EJ, Oei MT, Manniesing R, Prokop 55. Schaefer PW, Souza L, Kamalian S, et al. Limited reli-
M, Meijer FJ. 4D-CTA in neurovascular disease: a review. ability of computed tomographic perfusion acute infarct
AJNR Am J Neuroradiol 2015;36(6):1026–1033. volume measurements compared with diffusion-weighted
35. Deng K, Liu C, Ma R, et al. Clinical evaluation of dual- imaging in anterior circulation stroke. Stroke 2015;46
energy bone removal in CT angiography of the head and (2):419–424.
neck: comparison with conventional bone-subtraction CT 56. Christoforidis GA, Mohammad Y, Kehagias D, Avutu
angiography. Clin Radiol 2009;64(5):534–541. B, Slivka AP. Angiographic assessment of pial collaterals
36. Buerke B, Puesken M, Wittkamp G, et al. Bone subtraction as a prognostic indicator following intra-arterial throm-
CTA for transcranial arteries: intra-individual comparison bolysis for acute ischemic stroke. AJNR Am J Neuroradiol
with standard CTA without bone subtraction and TOF- 2005;26(7):1789–1797.
MRA. Clin Radiol 2010;65(6):440–446. 57. Reid M, Famuyide AO, Forkert ND, et al. Accuracy and
37. McCollough CH, Leng S, Yu L, Fletcher JG. Dual- and Reliability of Multiphase CTA Perfusion for Identifying
Multi-Energy CT: principles, Technical Approaches, and Ischemic Core. Clin Neuroradiol 2018 Aug 21. [Epub ahead
Clinical Applications. Radiology 2015;276(3):637–653. of print] [Published correction appears in Clin Neuroradiol
38. de Lucas EM, Sánchez E, Gutiérrez A, et al. CT protocol 2019 Jun 3.] https://doi.org/10.1007/s00062-018-0717-x.
for acute stroke: tips and tricks for general radiologists. 58. Hoeffner EG, Case I, Jain R, et al. Cerebral perfusion
RadioGraphics 2008;28(6):1673–1687. CT: technique and clinical applications. Radiology
39. Yeo LL, Paliwal P, Teoh HL, et al. Assessment of in- 2004;231(3):632–644.
tracranial collaterals on CT angiography in anterior cir- 59. Srinivasan A, Goyal M, Al Azri F, Lum C. State-of-the-art
culation acute ischemic stroke. AJNR Am J Neuroradiol imaging of acute stroke. RadioGraphics 2006;26(suppl
2015;36(2):289–294. 1):S75–S95.
40. Maas MB, Lev MH, Ay H, et al. Collateral vessels on CT 60. Eastwood JD, Lev MH, Provenzale JM. Perfusion CT
angiography predict outcome in acute ischemic stroke. with iodinated contrast material. AJR Am J Roentgenol
Stroke 2009;40(9):3001–3005. 2003;180(1):3–12.
41. Miteff F, Levi CR, Bateman GA, Spratt N, McElduff P, 61. Christensen S, Lansberg MG. CT perfusion in acute stroke:
Parsons MW. The independent predictive utility of com- practical guidance for implementation in clinical practice. J
puted tomography angiographic collateral status in acute Cereb Blood Flow Metab 2018 Oct 22:271678X18805590
ischaemic stroke. Brain 2009;132(Pt 8):2231–2238. [Epub ahead of print].
42. Souza LC, Yoo AJ, Chaudhry ZA, et al. Malignant CTA 62. ACR-ASNR-SPR Practice Parameter for the Performance of
collateral profile is highly specific for large admission DWI Computed Tomography (CT) Perfusion in Neuroradiologic
infarct core and poor outcome in acute stroke. AJNR Am Imaging. https://www.acr.org/-/media/ACR/Files/Practice-
J Neuroradiol 2012;33(7):1331–1336. Parameters/ct-perfusion.pdf. Revised 2017. Accessed April
43. Goyal M, Menon BK, Almekhlafi MA, Demchuk A, Hill 2, 2019.
MD. The Need for Better Data on Patients with Acute Stroke 63. Konstas AA, Goldmakher GV, Lee TY, Lev MH. Theoretic
Who Are Not Treated Because of Unfavorable Imaging. basis and technical implementations of CT perfusion in acute
AJNR Am J Neuroradiol 2017;38(3):424–425. ischemic stroke. II. Technical implementations. AJNR Am
44. Rocha M, Jovin TG. Fast Versus Slow Progressors of Infarct J Neuroradiol 2009;30(5):885–892.
Growth in Large Vessel Occlusion Stroke: Clinical and 64. Bivard A, Levi C, Spratt N, Parsons M. Perfusion CT in
Research Implications. Stroke 2017;48(9):2621–2627. acute stroke: a comprehensive analysis of infarct and pen-
45. ASPECTS Collateral Score. http://www.aspectsinstroke. umbra. Radiology 2013;267(2):543–550.
com/collateral-scoring/. Accessed April 2, 2019. 65. Campbell BC, Mitchell PJ, Kleinig TJ, et al. Endovascular
46. Sun CH, Connelly K, Nogueira RG, et al. ASPECTS de- therapy for ischemic stroke with perfusion-imaging selection.
cay during inter-facility transfer predicts patient outcomes N Engl J Med 2015;372(11):1009–1018.
in endovascular reperfusion for ischemic stroke: a unique 66. Goyal M, Menon BK, Derdeyn CP. Perfusion imaging in
assessment of dynamic physiologic change over time. J acute ischemic stroke: let us improve the science before
Neurointerv Surg 2015;7(1):22–26. changing clinical practice. Radiology 2013;266(1):16–21.
47. Aminoff MJ, Josephson SA, eds. Aminoff’s Neurology and 67. Kamalian S, Kamalian S, Maas MB, et al. CT cerebral blood
General Medicine. 5th ed. London, England: Academic flow maps optimally correlate with admission diffusion-
Press/Elsevier, 2014. weighted imaging in acute stroke but thresholds vary by
48. Allmendinger AM, Tang ER, Lui YW, Spektor V. Imaging postprocessing platform. Stroke 2011;42(7):1923–1928.
of stroke. I. Perfusion CT: overview of imaging technique, 68. Kamalian S, Kamalian S, Konstas AA, et al. CT perfusion
interpretation pearls, and common pitfalls. AJR Am J mean transit time maps optimally distinguish benign olige-
Roentgenol 2012;198(1):52–62. mia from true “at-risk” ischemic penumbra, but thresholds
49. Lui YW, Tang ER, Allmendinger AM, Spektor V. vary by postprocessing technique. AJNR Am J Neuroradiol
Evaluation of CT perfusion in the setting of cerebral 2012;33(3):545–549.
1738  October Special Issue 2019 radiographics.rsna.org

69. Campbell BCV, Majoie CBLM, Albers GW, et al. Penumbral 75. Fahmi F, Beenen LF, Streekstra GJ, et al. Head move-
imaging and functional outcome in patients with anterior ment during CT brain perfusion acquisition of patients
circulation ischaemic stroke treated with endovascular throm- with suspected acute ischemic stroke. Eur J Radiol
bectomy versus medical therapy: a meta-analysis of individual 2013;82(12):2334–2341.
patient-level data. Lancet Neurol 2019;18(1):46–55. 76. Borst J, Marquering HA, Beenen LF, et al. Effect of ex-
70. Leslie-Mazwi TM, Hirsch JA, Falcone GJ, et al. Endovas- tended CT perfusion acquisition time on ischemic core
cular Stroke Treatment Outcomes After Patient Selection and penumbra volume estimation in patients with acute
Based on Magnetic Resonance Imaging and Clinical Criteria. ischemic stroke due to a large vessel occlusion. PLoS One
JAMA Neurol 2016;73(1):43–49. 2015;10(3):e0119409.
71. Thomalla G, Simonsen CZ, Boutitie F, et al. MRI-Guided 77. Copen WA, Deipolyi AR, Schaefer PW, Schwamm LH,
Thrombolysis for Stroke with Unknown Time of Onset. N González RG, Wu O. Exposing hidden truncation-related
Engl J Med 2018;379(7):611–622. errors in acute stroke perfusion imaging. AJNR Am J Neu-
72. Ma H, Campbell BCV, Parsons MW, et al. Thrombolysis roradiol 2015;36(4):638–645.
Guided by Perfusion Imaging up to 9 Hours after Onset of 78. Schaefer PW, Mui K, Kamalian S, Nogueira RG, Gon-
Stroke. N Engl J Med 2019;380(19):1795–1803. zalez RG, Lev MH. Avoiding “pseudo-reversibility” of
73. Schwamm LH, Wu O, Song SS, et al. Intravenous throm- CT-CBV infarct core lesions in acute stroke patients after
bolysis in unwitnessed stroke onset: MR WITNESS trial thrombolytic therapy: the need for algorithmically “delay-
results. Ann Neurol 2018;83(5):980–993. corrected” CT perfusion map postprocessing software.
74. Parsons M, Spratt N, Bivard A, et al. A randomized trial Stroke 2009;40(8):2875–2878.
of tenecteplase versus alteplase for acute ischemic stroke. 79. Phan CM, Yoo AJ, Hirsch JA, Nogueira RG, Gupta R.
N Engl J Med 2012;366(12):1099–1107. Differentiation of hemorrhage from iodinated contrast in
different intracranial compartments using dual-energy head
CT. AJNR Am J Neuroradiol 2012;33(6):1088–1094.

TM
This journal-based SA-CME activity has been approved for AMA PRA Category 1 Credit . See rsna.org/learning-center-rg.