Eur Radiol (2005) 15: 416–426

DOI 10.1007/s00330-004-2633-5 NEURO

A. Rovira
E. Grivé
Distribution territories and causative
A. Rovira
J. Alvarez-Sabin
mechanisms of ischemic stroke

Received: 30 June 2004

Accepted: 13 December 2004
Published online: 19 January 2005
# Springer-Verlag 2005
A. Rovira (*) . E. Grivé . A. Rovira .
J. Alvarez-Sabin
Hospital Vall d’Hebron, Unidad de
Resonancia Magnetica,
Passeig Vall d’Hebron 119-129,
08035 Barcelona, Spain
e-mail: alex.rovira@idi-cat.org
Abstract Ischemic stroke prognosis,
risk of recurrence, clinical assess-
ment, and treatment decisions are
influenced by stroke subtype (ana-
tomic distribution and causative
mechanism of infarction). Stroke
subtype diagnosis is better achieved
in the early phase of acute ischemia
with the use of multimodal MR
imaging. The pattern of brain lesions
as shown by brain MR imaging can
be classified according to a modified
Oxfordshire method, based on the
anatomic distribution of the infarcts
into six groups: (1) total anterior
circulation infarcts, (2) partial anterior
circulation infarcts, (3) posterior cir-
culation infarcts, (4) watershed in-
farcts, (5) centrum ovale infarcts, and
(6) lacunar infarcts. The subtype of
stroke according to its causative
mechanism is based on the TOAST
method, which classifies stroke into
five major etiologic groups: (1) large-
vessel atherosclerotic disease, (2)
small-vessel atherosclerotic disease,
(3) cardioembolic source, (4) other
determined etiologies, and (5) unde-
termined or multiple possible etiolo-
gies. The different MR imaging
patterns of acute ischemic brain
lesions visualized using diffusion-
weighted imaging and the pattern of
vessel involvement demonstrated with
MR angiography are essential factors
that can suggest the most likely
causative mechanism of infarction.
This information may have an impact
on decisions regarding therapy and
the performance of additional diag-
nostic tests.
Keywords Diffusion . MRI . Stroke

Introduction
Ischemic stroke prognosis, risk of recurrence, clinical as-
sessment, and treatment decisions are influenced by stroke
subtype. Nevertheless, treatment decisions are often made
before an extensive, time-consuming evaluation to identify
a likely diagnosis is completed. Therefore, early classifica-
tion of ischemic stroke subtype is of substantial practical
clinical value. The most widely used methods for stroke
subtype classification are the Oxfordshire Community
Stroke Project and the TOAST (Trial of ORG 10172 in
Acute Stroke Treatment) method.
In 1991 the Oxfordshire Community Stroke Project (OCSP)
proposed four subgroups of cerebral infarction (Table 1)
based solely on presenting signs and symptoms [1]. This
method has the ability to predict the prognosis and shows
good correlation with the underlying pathophysiology and
imaging findings on cranial computed tomography (CT)
[2].
The TOAST method is a set of guidelines developed for
prospectively classifying ischemic strokes into specific sub-
types, based mainly on the mechanism of infarction [3, 4].
Stroke patients are classified into five major etiologic/pa-
thophysiologic groups (Table 2).
With the widespread use of diffusion-weighted MR im-
aging (DWI) and MR angiography (MRA) in the acute
stage of ischemic stroke, accurate early diagnosis of ische-
mic stroke subtype can be better achieved [5]. This in-
formation can be helpful for establishing the most likely
 417

Table 1 Topographic clinical pattern of brain infarction (Oxford-
shire method)
1. Lacunar infarcts (LACI) Acute stroke that includes one of the
major recognized lacunar syndromes:
pure motor, sensory, or sensorimotor
strokes, ataxic hemiparesis, and dysar-
thria (clumsy hand syndrome)
2. Total anterior circula- Clinical syndrome in which there is
tion infarcts (TACI) ischemia in both the deep and superfi-
cial territories of the middle cerebral
artery (higher cerebral dysfunction such
as dysphasia, dyscalculia, visuospatial
disorder; homonymous visual field de-
fect; and ipsilateral motor and/or sen-
sory deficit of at least two areas of the
face, arm, and leg)
3. Partial anterior circula- Clinical syndrome that includes only
tion infarcts (PACI) two of the three components of the
TACI syndrome, with higher cerebral
dysfunction alone, or with more re-
stricted sensorimotor deficit than those
classified as LACI
4. Posterior circulation in- Clinical syndrome that includes ipsi-
fracts (POCI) lateral cranial nerve palsy with con-
tralateral motor and/or sensory deficit;
bilateral motor and/or sensory
deficit; disorder of conjugate eye
movement; cerebellar dysfunction
without ipsilateral long-tract deficit
(i.e., ataxic hemiparesis); or isolated
homonymous visual field defect
ditions that govern flow in leptomeningeal anastomoses
connecting the different arterial territories [6]. Despite this
variability, brain imaging can accurately locate an ischemic
stroke lesion in a specific vascular distribution in the ma-
jority of cases [7–10].
Arterial cerebral circulation can be divided into two
systems: (1) the leptomeningeal (also known as superficial
or pial) artery system; and (2) the perforating (or deep
perforating) artery system (Table 3).
Leptomeningeal artery system
The leptomeningeal arteries comprise the terminal branches
of the cerebral and cerebellar arteries, which penetrate the
cortex and subjacent white matter. Infarcts within the ter-
ritories irrigated by these arteries are often described as
territorial infarcts.
Territorial anterior infarcts
Territorial anterior infarcts, mostly related to the middle
cerebral artery (MCA) territory, can be divided into large
and limited types.
Large infarcts, defined as those covering at least two of
the three MCA territories (deep, superficial anterior, and
superficial posterior), show a high frequency of clinical
deterioration, a minimum chance of good outcome, and a
high mortality rate. Large MCA infarctions are associated
with cardioembolism, internal carotid artery (ICA) occlu-
sion, and ICA dissection. In patients with large infarcts
without ICA occlusion, the frequency of cardioembolic

mechanisms of ischemia and the risk of clinical progres-

sion, and for initiating the most appropriate therapy.
This review article is divided into three parts. The first Table 3 Topographic radiologic pattern of brain infarction
addresses the topographic patterns of brain infarction, the Leptomeningeal artery Territorial anterior circulation infarcts
second is devoted to the mechanisms implicated in the system Large
genesis of multiple synchronous acute brain infarcts, and Malignant
the third part reviews the various stroke categories on the Limited
basis of their causative mechanisms. Territorial posterior circulation
infarcts
Large territorial
Topographic pattern of brain infarcts Small or end zone infarcts
Brainstem infarcts
The boundaries between vascular distributions are deter- Centrum ovale infarcts
mined by anatomic variations and by hemodynamic con- Large
Small
Table 2 Stroke categories (TOAST method)
Watershed infarcts
1 Large-vessel disease Internal
2 Small-vessel disease Cortical anterior
3 Cardioembolism Cortical posterior
4 Other etiology Deep perforator artery Lacunar infarcts
5 Undetermined or multiple possible etiologies system
418
disease is clearly higher than in large infarcts with ICA
occlusion or limited infarcts without ICA occlusion [11].
Malignant MCA infarction refers to life-threatening (80%
associated mortality), complete or almost complete MCA
infarction. This type of infarction occurs in up to 10% of all
stroke patients. The main cause of death is severe post-is-
chemic brain edema leading to raised intracranial pressure.
The clinical course is uniform, with clinical deterioration
developing within the first 2 to 3 days after stroke. DWI in the
early phase of large territorial anterior infarction is an accurate
method for predicting malignant MCA infarction (lesion
volume >145 cm3) [12] in patients with persistent arterial
occlusion and signs of total anterior circulation infarction.
Early detection may be important since treatment for these
large infarcts, such as hypothermia and/or hemicraniectomy
can significantly reduce mortality [13].
Limited infarcts, covering only one of the three MCA
territories, show a very low frequency of clinical deteriora-
tion. The mechanism of infarction in these cases is either
cardioembolism or large-artery atherosclerosis in equal
numbers. The lower incidence of cardioembolism in limited
MCA infarcts as compared to large infarcts can be ex-
plained by the fact that cardiac thrombi are generally larger
than thrombi of the large vessels [2] (Fig. 1).
Fig. 1 Diffusion-weighted MR imaging in anterior choroidal artery
(AChA) infarcts. Right acute lacunar infarct limited to the AChA
territory with no significant stenosis in an ipsilateral large artery,
probably related to small-vessel disease (A). Acute right AChA ter-
ritory infarct associated with other ipsilateral internal carotid artery
infarcts due to internal carotid artery dissection (arrow) (B).
Territorial posterior infarcts
With the use of MR imaging, posterior circulation infarcts
can be diagnosed and their topography delineated with
high sensitivity. Infarcts are recognized in the territory of
the posterior inferior (PICA), anterior inferior (AICA), and
superior (SCA) cerebellar arteries and their branches, and
in the territory of the posterior cerebral arteries (PCA) [8].
Large-vessel atherosclerosis of the extracranial and in-
tracranial vertebrobasilar arteries, has been demonstrated
angiographically in more than two thirds of patients with
cerebellar infarction, and in situ branch artery disease in
almost one fifth of these patients. Proximal disease of the
vertebral artery is the most common feature in large-artery
disease, leading to PICA, SCA, and PCA infarcts. The
vessels most frequently affected by intraarterial embolism
are the intracranial vertebral artery (leading to PICA
infarcts) and the distal basilar artery (leading to SCA and
PCA infarcts). Thrombus originating from proximal ver-
tebral artery disease never occludes the intracranial verte-
bral artery, but it can affect the PICA, AICA, and SCA.
Therefore, vascular imaging in patients with atherothrom-
botic cerebellar infarcts must assess the proximal vertebral
artery (V1) [14].
Clinical deterioration in posterior circulation infarcts is
associated with severe brain atrophy (suggesting longstand-
ing hypoperfusion), and significant stenosis in the verte-
brobasilar arteries. This feature implies that large-vessel
disease plays an important role in clinical worsening in
posterior circulation infarcts [2].
Small cerebellar infarcts (<2 cm) are frequently recog-
nized with MR imaging. These small infarcts were thought
to affect the boundary zone between various territories
(nonterritorial infarcts), but in fact, they seem to be very
small territorial infarcts resulting from involvement of
small distal arteries. Therefore, small cerebellar infarcts
correspond to “end zone” infarcts, with embolic or local
arterial disease as the mechanism of infarction in the ma-
jority of cases; a low-flow hemodynamic state is the likely
cause of infarction in only a minority of patients.
Territorial and nonterritorial (small) cerebellar infarcts
are essentially the same, and it is likely that their extent
and location simply depend on the size of the embolus
causing the infarct [15, 16].
Brainstem infarcts may be related to large-vessel disease
(stenosis or occlusion) affecting the vertebral and basilar
arteries or their main branches. In this situation, brain MR
usually shows additional infarcts involving the territories
irrigated by the cerebellar and posterior cerebral arteries.
Centrum ovale infarcts
The centrum ovale is the central white matter of the cerebral
hemispheres, including the most superficial part of the
corona radiata and the long associated fasciculi. The cen-

trum ovale is supplied by long (2 to 5 cm) noninterdigitating
medullary arteries that perforate it and course toward the
upper part of the lateral ventricles. At the deeper part of
the corona radiata the medullary branches tend to form an
area of junction with the deep perforating branches of the
MCA and the AChA. Centrum ovale infarcts are those
limited to the territory of the medullary branches without
accompanying involvement of the cortex or deep perfo-
rator territory [17].
Centrum ovale infarcts can be large and small.
Large centrum ovale infarcts (>1.5 cm) affect more than
one medullary branch. A hemodynamic mechanism related
to severe ipsilateral internal carotid or MCA disease may
be the leading cause. In this situation, the infarct affects the
area of the internal border zone, between the deep per-
forators and superficial medullary territories of the MCA
(internal border zone infarcts). However artery-to-artery
embolism and cardioembolism cannot be ruled out in some
patients.
Small centrum ovale infarcts (<1.5 cm) involve only one
medullary branch. Small infarcts were thought to be related
to small-vessel disease involving the medullary branches,
in a manner similar to lacunar infarction. In fact, the neu-
rologic picture of small infarct of the centrum ovale is
consistent with a so-called lacunar syndrome, although the
motor or sensory distribution pattern is less often complete
(face, arm, and leg) than partial. However, recent studies
have shown that in a significant percentage of patients,
small centrum ovale infarcts are associated with large-
vessel and heart disease, and should be distinguished from
the more common lacunar infarcts [18]. Identification of
subsidiary small acute infarcts in addition to an acute small
infarct in the centrum ovale on DWI suggests an embolic
mechanism [17].
Watershed infarcts
Watershed infarcts (WIs) are ischemic lesions that occur at
the junction between two or three arterial territories and
Fig. 2 Watershed infarcts. Dif-
fusion-weighted MR images
showing the classical pattern of
anterior cortical (A), posterior
cortical (B) and internal
(C) watershed infarcts.
419
account for approximately 10% of ischemic strokes. The
pathogenesis of WIs is controversial. It may involve var-
ious mechanisms such as systemic hypotension, severe
arterial stenosis or ICA occlusion, microemboli, or a com-
bination of these. Watershed infarcts account for 72% of
delayed strokes in patients with ICA occlusion, but are
rarely the initial manifestation of ICA occlusion (5%) [19].
Recent data indicate that WIs are often explained by a
combination of two inter-related processes: hypoperfusion
and embolization. In fact, severe ICA occlusive disease and
cardiac surgery cause both embolization and decreased
brain perfusion. This decreased perfusion might alter blood
flow currents, encouraging microemboli to reach recipient
blood vessels with the least effective blood flow. Moreover,
microemboli that reach a border zone area with decreased
blood flow are difficult to wash out [20].
Two types of vascular border zone areas exist within the
cerebral hemispheres: the cortical and the internal. Cortical
border zone areas are located between the cortical supply
of the ACA and MCA (anterior cortical border zone), and
between the MCA and PCA (posterior cortical border
zone). Internal border zone areas are located between the
ACA (anterior cerebral artery), MCA, and PCA, and the
area supplied by the Heubner, lenticulostriate, and ACh
arteries (Fig. 2).
Purely anterior cortical WIs are very rare, as in most
cases they are associated with internal border zone in-
farcts. Posterior cortical WIs are frequently difficult to
differentiate from limited territorial infarcts affecting the
posterior division of the MCA. Embolism, not distal field
perfusion failure, is the predominant stroke mechanism in
this type of WI.
Internal border zone infarcts, commonly associated with
severe ICA stenosis, are larger than lacunar infarcts within
the vascular territory of the deep perforators. In some cases
it is difficult to distinguish internal border zone infarcts
from centrum ovale infarcts within the territory irrigated by
the medullary branches of the MCA. The presence of two or
more lesions, appearing as a chain of round infarcts along
the internal vascular border zone, suggests an internal bor-
420
der zone infarct. Lesions in the internal border zone are
mainly attributed to the effect of hemodynamic impairment
caused by severe stenosis or occlusion of the ICA or MCA.
Nevertheless, some studies have suggested an embolic me-
chanism for both cortical watershed and internal border
zone infarcts [21].
Watershed infarcts involving more than one of the
border zone areas in a single hemisphere are mostly related
to severe ICA stenosis or occlusion. Bilateral watershed
infarcts are typically related to a profound global reduction
in perfusion pressure (hypoxia, hypovolemia) or to diffuse
cerebral vessel disease (sickle-cell disease).
The perforating (or deep perforating) artery system
The perforating arteries arise from the arterial circle of
Willis, the AChA, and the basilar artery, perforating the
brain parenchyma as direct penetrators and supplying the di-
encephalon (thalamus, hypothalamus, subthalamus, and epi-
thalamus), basal ganglia, internal capsule, and brainstem [10].
Lacunar infarcts
The strict pathologic definition of a lacunar infarction (LI)
is a small (<1.5 cm in diameter), fluid-filled cavity rep-
resenting the healed stage of a small deep infarct, which
was likely due to occlusion of a single penetrator artery
arising from the large arteries of the circle of Willis or
from the basilar artery. The most frequently affected per-
forating arteries include the lenticulostriate, the thalamo-
perforate and the perforators arising from the AChA.
Thus, LI can be located deep within the cerebral he-
mispheric white matter, the upper two thirds of the basal
ganglia, the internal capsule, the thalamus or the parame-
dian and lateral regions of the brainstem. Brainstem LI are
mainly found in the paramedian region of the pons, which is
supplied by long arterioles (midline and anteromedial
perforators) arising from the basilar artery. The medulla and
midbrain are supplied by short arterioles, which are less
vulnerable to aging and hypertension, and as a consequence
LI are very uncommon in these locations.
The maximum size of 15 mm is probably true for LI in
the chronic stage. In fact, most of them have a diameter
<10 mm, but this dimension does not apply in the acute
phase, when cellular swelling and extracellular edema can
produce LI larger than 15 mm in diameter.
Most first-ever symptomatic LI are located in the area
supplied by the AChA [9, 22], while multiple, asymp-
tomatic LI are mainly located within the territory irrigated
by perforating arteries arising from the MCA or ACA.
Most LI are asymptomatic. Symptoms are related to size
(most lesions less than 200 μm in diameter are silent) and
location (LI in the posterior limb of the internal capsule or
in the pontine base are usually symptomatic, whereas
lesions in the basal ganglia tend to be silent).
Although most lacunar strokes appear to be a con-
sequence of small-vessel disease (atherosclerosis or lipo-
hyalinosis) [23], many of the other potential causes of
small-vessel occlusion rarely cause LI, such as infective or
immune vasculitis, artery-to-artery or cardiogenic embo-
lism, arterial dissection, and in situ thrombosis due to a
variety of hypercoagulable states [24]. This non–small-
vessel disease mechanism of LI is supported by the not
infrequent association of acute LI and superficial infarcts.
As a consequence, vascular imaging of the cervical and
intracranial arteries and a focussed search for a cardiac
source of the embolism is needed in first-ever lacunar
strokes. The detection of subsidiary infarctions in patients
presenting with a classic lacunar syndrome and often
diagnosed with intrinsic small-vessel disease (with a high
probability of excellent recovery, recommendation of anti-
platelet agents for secondary prevention, and low priority
placed on extensive cardiac or arterial testing for other
causes of stroke), should prompt the physician to search
for an underlying embolic source and tailor a secondary
stroke prevention strategy to treat the underlying cause.
Embolic infarcts can present as a clinically well-defined
lacunar syndrome. However, the concept of embolic LI is
dubious for two main reasons: the low likelihood of an
embolus entering a vascular territory that receives such a
small proportion of cerebral blood flow, and the sharp
angle of the penetrating vessels arising from the parent
vessel, which makes it more likely for an embolus to be
directed toward the leptomeningeal arteries. However, as
compared to a leptomeningeal territory, the lack of col-
lateral flow pathways in the deep perforating artery ter-
ritories may make them more susceptible to infarction on
entry of a very small embolus. In fact, it has been sug-
gested that a massive shower of emboli such as that oc-
curring during cardiac and aortic operations, in cholesterol
embolization from ulcerated atheroma, and in paradoxical
fat or air embolism can produce LI.
Acute multiple brain infarction
Synchronous acute multiple brain infarcts (AMBIs) can
have various mechanisms of origin, which are suggested by
their topographic pattern. With the use of DWI, the different
types of AMBI can be easily identified [25, 26]:
(1) Internal watershed infarcts: multiple rosary-like infarcts
within border zone areas. Unilateral watershed infarcts
are usually related to ICA disease, whereas bilateral
ones are typically related to a profound overall re-
duction in perfusion pressure or to diffuse cerebral vas-
cular disease.
(2) Multiple small cortical or subcortical infarcts within
the same arterial territory. This type of AMBI is attri-

buted to fragmentation of an embolus near its origin or
located within a major proximal intracranial artery.
Most of these infarcts are found within one cerebral
hemisphere in the anterior circulation.
(3) Multiple small infarcts attributable to multiple artery-
to-artery thromboembolic material (located in one or
more major arterial territories, depending on the arterial
anatomy). Sometimes these infarcts are located in two
hemispheres and in both the anterior and posterior cir-
culation. Acute multiple brain infarcts exclusively af-
fecting the posterior circulation are, in most cases,
related to large-artery atherosclerosis. The anatomic
variations that may explain AMBI include (1) posterior
communicating artery (PCoA) originating from the
ICA (fetal-type PCA), which explains the simulta-
neous infarcts in the anterior and posterior cerebral
artery territory (25% of cerebral hemispheres); (2)
PCoA patency (67% of anatomic dissections), explain-
ing multiple infarcts in both the anterior and posterior
circulation; and (3) AMBI in the territories of both
ACAs when a single artery supplies the two medial
aspects of the hemispheres, occurring in 18% of the
normal population.
(4) Multiple infarcts located in one or more major arterial
territories of the anterior and/or posterior circulation
produced from embolic sources proximal to the cer-
vical arteries (heart or aortic arch), not attributable to
anatomic variations of the cervicocranial arterial ves-
sels. The pattern of multiple, small and large bihemi-
spheric AMBIs is especially frequent in nonbacterial
thrombotic endocarditis (marantic endocarditis).
(5) Multiple simultaneous deep perforators. Most of these
cases are related to bilateral, simultaneous small-artery
occlusion.
Cardioembolism can be the mechanism of infarcts in
groups 2, 3, and 4, although it is much more frequent in
group 4. Nevertheless, bilateral carotid stenosis or occlu-
sion can also be associated with acute infarcts in both
cerebral hemispheres.
The main causes of AMBI in one hemisphere in the
anterior circulation are MCA and ICA disease (75%).
Although the factors that determine contemporary in-
farcts in small-vessel occlusion or severe bilateral large-
vessel disease are unknown (groups 3 and 5), it is believed
that erytrocytosis (elevated primary or secondary hemat-
ocrit) and increased serum fibrinogen, may be important
contributory factors [25]. In fact, these factors are signif-
icantly associated with bilateral cerebral infarction in
patients with large-artery atherosclerosis or small-artery
occlusion. Other possible explanations for AMBI located in
different vascular territories include bilateral or unilateral
watershed infarctions, or diffuse thrombotic or inflamma-
tory processes, such as thrombotic thrombocytopenic pur-
pura, granulomatous angiitis, and anticardiolipin syndrome,
421
which lead to multiple small-vessel occlusions within a
short period of time.
Stroke categories
The subtype of stroke according to its causative mech-
anism (TOAST) is based on risk factor profiles, clinical
features, and the results of diagnostic tests [3, 4].
The TOAST algorithm classifies patients with ischemic
stroke into five major etiologic and pathophysiologic
groups: large-vessel atherosclerotic disease, small-vessel
atherosclerotic disease, cardioembolism, other etiologies,
and undetermined or multiple possible etiologies.
Examining responses to acute treatment in each one of
these subgroups of stroke mechanisms is clinically im-
portant; therefore, highly accurate early stroke subtyping
is needed. The sensitivity and positive predictive value of
the initial TOAST diagnosis of large- and small-vessel
disease improves considerably with the combined use of
DWI and MRA techniques [5].
Large-vessel atherosclerotic disease
Large-vessel atherosclerosis represents about 13% of all
patients with a first-ever stroke.
Cortical or cerebellar lesions and brainstem or subcor-
tical hemispheric infarcts greater than 1.5 cm in diameter
on brain imaging are considered to be potential large-artery
disease strokes. Supportive evidence by vascular imaging
of more than 50% stenosis in an appropriate intracranial
or extracranial artery (presumably due to atherosclerosis)
is needed. Embolic (artery-to-artery) and hemodynamic
mechanisms are the cause of stroke in these patients. The
coexistence of both mechanisms has been postulated.
Five patterns of ischemic lesions can be differentiated in
patients with acute stroke and large-vessel disease [27, 28]
(Fig. 3).
Cortical territorial infarction
Cortical territorial infarcts are ischemic lesions involving
the cerebral cortex and subcortical structures in one or
more major cerebral artery territories. Almost half of the
patients with ICA occlusion have territorial infarction.
However about 20% of strokes in the territory of a high-
grade symptomatic ICA are cardioembolic or lacunar.
This pattern can be subclassified into three forms: (1)
limited MCA infarction in occlusions of a distal MCA
branch or the proximal MCA, associated with effective
collateral circulation; (2) large MCA infarction, frequently
related to large emboli that proximally occlude the MCA in
the absence of an efficient collateral system, or to oc-
clusions of the distal ICA with a partially effective collateral
422

Fig. 3 Different patterns of

cortical territorial infarcts within
the internal carotid artery (ICA)
territory demonstrated with dif-
fusion-weighted MR imaging
and MR angiography. A Limited
left middle cerebral artery
(MCA) infarction due to occlu-
sion of the proximal MCA.
B Complete right ICA infarction
due to occlusion of the ICA.
C Large left MCA infarction
due to occlusion of the MCA.
D Subcortical right MCA in-
farction due to MCA occlusion
with small fragmented subcorti-
cal and cortical subsidiary
infarcts. E Fragmented right
cortical MCA infarction due to
severe MCA stenosis (arrow).
F Small fragmented right in-
farctions in the left MCA terri-
tory due to severe stenosis at the
origin of the right ICA.

system; and (3) complete infarcts involving two major ICA
cerebral artery territories in distal occlusions of the ICA
without an effective collateral system.
Subcortical infarction
Subcortical infarcts are ischemic lesions in the territory
of the deep perforating branches arising from the distal
ICA or MCA trunk in proximal occlusions of the MCA
or ICA in the presence of patent collaterals. Additional
fragmented small cortical or subcortical infarctions can
be also identified.
Cortical territorial infarction with fragmentation
Large ischemic cortical lesions with additional smaller
cortical or subcortical lesions due to partial fragmentation
of an embolus fall into the category of cortical infarction
with fragmentation.
Fragmented infarction
Fragmented infarcts are defined as several small, dissem-
inated lesions sprinkled randomly in the cortical ICA re-
gions due to multiple emboli or the break-up of an embolus.
Fragmented infarctions are more common in moderate ICA
stenosis or in fragmentation of a lodged embolus. In the
latter case, there is often no evidence of stenosis or oc-
clusion in the intracranial arteries (resolved embolism).
Watershed infarction
The pathogenesis of watershed infarctions is controversial.
The leading mechanism is believed to be critical ICA
stenosis or occlusion, which may or may not be associated
with transient hypoperfusion. In fact, 75% of patients with
WI have high-grade ICA stenosis or occlusion associated
with hemodynamically significant heart disease, increased
hematocrit, or acute hypotension. Conversely, 50% of pa-
tients with high-grade or subtotal ICA stenosis have water-
shed infarcts. Atherosclerotic disease of the MCA may also
cause watershed infarcts. For this reason, in addition to the
extracranial vessel, cerebrovascular investigation in these
patients should include the large intracranial vessels.
Small-vessel disease (small-artery occlusion)
Small-vessel disease is the cause of about 25% of all first-
ever strokes. The most frequent pathologic events related to
small-vessel disease are atherosclerosis and lipohyalinosis
limited to the small penetrator vessels [23]. Chronic hyper-
423
tension seems to be the main etiology of these pathologic
events, but a variety of other conditions such as aging,
hypoperfusion, generalized atherosclerosis, diabetes, and
orthostatic hypotension can contribute to the brain micro-
circulation compromise. In fact, in a significant proportion
of patients, small-vessel disease is identified in normoten-
sive, nonelderly, nondiabetic individuals.
Atherosclerosis
Atherosclerosis of the small penetrator vessels or micro-
atheromatosis is the leading cause of small-artery disease.
Atheroma plaques are localized in the proximal perforat-
ing arteries (microatheroma), in the origin of the penetrator
artery (junctional atheroma), or in the parent artery on the
circle of Willis (mural atheroma). Mural atheroma is par-
ticularly frequent in the basilar artery, producing infarcts
limited to its pontine perforating branches.
The atheroma plaques result in occlusion of the proximal
segment of the large penetrating arteries (300–900 μm) and
usually lead to single, large, frequently symptomatic LI
[29]. This type of small-vessel disease, which is not so
strongly related to hypertension, seems to predominantly
involve the penetrator vessels arising from the anterior
choroidal artery. Radiologic studies in patients with a first-
ever lacunar stroke of this type commonly show no ad-
ditional asymptomatic infarcts or leukoaraiosis (Fig. 4a).
Lipohyalinosis
Lipohyalinosis, a vascular disease associated with long-
lasting, severe hypertension, is the second small-vessel
lesion of relevance in lacunar infarction. It is a destructive
lesion of the small penetrating arteries (<200 μm) that
leads to small LI, which are commonly asymptomatic and
located predominantly in the striatocapsule and thalamus.
Leukoaraiosis and old asymptomatic LI are commonly
seen on brain imaging in these patients (Fig. 4b).
The mechanism of infarction seems to be related to
occlusive thrombosis (perhaps exacerbated by a hyperco-
agulable state) or to non-occlusive poststenotic hypoper-
fusion. Patients with this type of small-vessel disease have
a better outcome and a smaller prevalence of large-vessel
cerebral disease and coronary disease than patients with
the atherosclerotic type.
Cardioembolism
Cardiogenic embolism is responsible for about 15–27% of
all first-ever strokes. The incidence is higher in patients
under 45 years old, primarily because of the lower inci-
dence of atherosclerotic disease in this age group. About
16% of ischemic strokes are associated with atrial fibril-
424

Table 4 Sources of risk for cardioembolism

Higher risk sources for Atrial fibrillation
cardioembolism Mural thrombus associated with acute
myocardial infarction
Prosthetic heart valve
Dilated cardiomyopathy
Bacterial endocarditis
Rheumatic mitral stenosis
Ascending aorta atheroma (≥4 mm in
size)
Intracardiac thrombus
Spontaneous left atrial echo contrast
Left ventricular aneurysm or large area
of dyskinesia
Nonbacterial (marantic) endocarditis
Lower risk sources for Sick sinus syndrome
cardioembolism Calcified aortic stenosis
Patent foramen ovale or atrial septal
defect
Atrial septal aneurysm
Mitral annulus calcification
Ventricular septal defect
Mitral valve prolapse

Fig. 4 Acute LI (Fast-Flair and diffusion-weighted MR images). Other etiologies

A Acute left internal capsule (territory of the anterior choroidal artery)
infarct with no additional infarcts or leukoaraiosis, probably related to
microatherosclerosis. B Acute left thalamic lacunar infarct associated
with small chronic LI and dilated Virchow-Robin spaces, suggesting
small-vessel atherosclerotic disease due to lipohyalinosis.
lation, and 10% are probably due to embolism from an
atrial appendage thrombus, with the remainder caused by
other stroke mechanisms [30]. Cerebral infarction in atrial
fibrillation tends to be large and severely disabling [31]. A
possible or probable diagnosis of cardioembolic stroke
requires the identification of at least one cardiac source for
an embolus (high-risk or medium-risk sources) (Table 4).
Evidence of a previous transient ischemic attack or stroke
in more than one vascular territory or systemic embolism
supports a clinical diagnosis of cardiogenic stroke.
On brain imaging, large cortical territorial infarction
should suggest a cardioembolic mechanism, particularly if
it is not associated with ICA occlusion. The median volume
of infarcts caused by cardiogenic embolism is more than
twice the median volume of infarcts caused by artery-to-
artery embolism [31].
Although it has been suggested that simultaneous acute
infarction indicates a cardioembolic mechanism, in the
majority of cases they are not caused by a proximal em-
bolism from the heart or aortic arch, but instead by artery-
to-artery embolism or fragmentation of a proximal artery
embolus.
Only 2% of all patients with a first-ever stroke fall into
the other etiology category. The lesion can have any size
or location. To classify a stroke under this category, car-
diac sources of embolism and large-artery atherosclerosis
should be excluded. These unusual mechanisms of stroke
include nonatherosclerotic, nonhypertensive vascular dis-
eases (Moya-Moya disease, craniocervical arterial dis-
section, and primary and systemic vasculitis), migraine,
hypercoagulable states, hematologic disorders, stroke after
catheter angiography, and sporadic or genetically deter-
mined small-vessel occlusion such as cerebral autosomal
dominant arteriopathy (CADASIL) and Fabry’s disease.
These rare forms of small-vessel occlusion cannot be dif-
ferentiated radiologically from the classical atherosclerotic
and hypertensive forms of small-vessel disease.
Undetermined etiology or multiple possible etiologies
Strokes classified as having undetermined or multiple possi-
ble etiologies must possess one of the following conditions:
(1) No cause found despite extensive assessment.
(2) Most likely cause could not be determined because
more than one plausible cause was found (e.g., atrial
fibrillation or lacunar infarct associated with >50%
symptomatic large vessel stenosis).
 425

This type of stroke represents about 35% of all patients
with a first-ever stroke. However, this percentage can vary
considerably, since in some cases an extensive diagnostic
evaluation is performed (ECG, Duplex, MRI/MRA, trans-
craneal Doppler ultrasound, transesophageal echocardio-
graphy, laboratory tests for coagulation factors, proteins C
and S, antithrombin III, and various autoantibodies), where-
as in others the evaluation is cursory.
Conclusion
The goal of imaging in the acute phase of ischemic stroke
is to identify the location and extension of the relevant
lesion and the presence of significant arterial stenosis or
occlusion. Multimodal MR imaging facilitates the achieve-
ment of these diagnostic goals, improving the accuracy of
early ischemic stroke subtype identification. The various
MRI patterns of acute brain ischemia (topography, size,
and multiplicity) visualized using DWI, the pattern of
vessel involvement demonstrated with MR angiography,
and the presence of previous ischemic lesions detected with
conventional MRI, are essential factors that can suggest the
most likely mechanisms of origin. This information may
have an impact on decisions regarding therapy and the
performance of additional diagnostic tests.

References
1. Bamford J, Sandercock P, Dennis M
et al (1991) Classification and natural
history of clinically identifiable sub-
types of cerebral infarction. Lancet
337:1521–1526
2. Tei H, Uchiyama S, Ohara K,
Kobayashi M, Uchiyama Y, Fukuzawa
M (2000) Deteriorating ischemic stroke
in 4 clinical categories classified by the
Oxfordshire Community Stroke Pro-
ject. Stroke 31:2049–2054
3. Adams HP, Bendixen BH, Kappelle LJ
et al (1993) Classification of subtype of
acute ischemic stroke: definitions for
use in a Multicenter Clinical Trial.
Stroke 24:35–41
4. Kolominsky-Rabas PL, Weber M,
Gefeller O, Neundoerfer B,
Heuschmann PU (2001) Epidemiol-
ogy of ischemic stroke subtypes ac-
cording to TOAST criteria: incidence,
recurrence, and long-term survival in
ischemic stroke subtypes: a population-
based study. Stroke 32:2735–2740
5. Lee LJ, Kidwell CS, Alger J et al
(2000) Impact on stroke subtype diag-
nosis of early diffusion-weighted mag-
netic resonance imaging and magnetic
resonance angiography. Stroke
31:1081–1089
6. van der Zwan A, Hillen B, Tulleken
CA, Dujovny M, Dragovic L (1992)
Variability of the territories of the major
cerebral arteries. J Neurosurg 77:927–
940
7. Tatu L, Moulin T, Bogousslavsky J,
Duvernoy H (1998) Arterial territories
of the human brain: cerebral hemi-
spheres. Neurology 50:1699–1708
8. Tatu L, Moulin T, Bogousslavsky J,
Duvernoy H (1996) Arterial territories
of human brain: brainstem and cere-
bellum. Neurology 47:1125–1135
9. Hupperts RM, Lodder J, Heuts-van
Raak EP, Kessels F (1994) Infarcts in
the anterior choroidal artery territory:
anatomical distribution, clinical syn-
dromes, presumed pathogenesis and
early outcome. Brain 117:825–834
10. Takahashi S, Suzuki M, Matsumoto K,
Ishii K, Higano S, Fukasawa H,
Sakamoto K (1994) Extent and location
of cerebral infarcts on multiplanar MR
images: correlation with distribution of
perforating arteries on cerebral
angiograms and on cadaveric
microangiograms. Am J Roentgenol
163:1215–1222
11. Heinsius T, Bogousslavsky J, Van
Melle G (1998) Large infarcts in the
middle cerebral artery territory: etiolo-
gy and outcome patterns. Neurology
50:341–350
12. Oppenheim C, Samson Y, Manai R,
Lalam T, Vandamme X, Crozier S,
Srour A, Cornu P, Dormont D,
Rancurel G, Marsault C (2000) Pre-
diction of malignant middle cerebral
artery infarction by diffusion-weighted
imaging. Stroke 31:2175–2181
13. Georgiadis D, Schwarz S, Aschoff A,
Schwab S (2002) Hemicraniectomy and
moderate hypothermia in patients with
severe ischemic stroke. Stroke
33:1584–1588
14. Min WK, Kim YS, Kim JY, Park SP,
Suh CK (1999) Atherothrombotic cer-
ebellar infarction: vascular lesion–MRI
correlation of 31 cases. Stroke
30:2376–2381
15. Amarenco P, Kase CS, Rosengart A,
Pessin MS, Bousser MG, Caplan LR
(1993) Very small (border zone) cere-
bellar infarcts: distribution, causes,
mechanisms and clinical features. Brain
116:161–186
16. Amarenco P, Levy C, Cohen A,
Touboul PJ, Roullet E, Bousser MG
(1994) Causes and mechanisms of
territorial and nonterritorial cerebellar
infarcts in 115 consecutive patients.
Stroke 25:105–112
17. Bogousslavsky J, Regli F (1992) Cen-
trum ovale infarcts: subcortical infarc-
tion in the superficial territory of the
middle cerebral artery. Neurology
42:1992–1998
18. Yonemura K, Kimura K, Minematsu K,
Uchino M, Yamaguchi T (2002) Small
centrum ovale infarcts on diffusion-
weighted magnetic resonance imaging.
Stroke 33:1541–1544
19. Bogousslavsky J, Regli F (1986)
Borderzone infarctions distal to internal
carotid artery occlusion: prognostic
implications. Ann Neurol 20:346–350
20. Belden JR, Caplan LR, Pessin MS,
Kwan E (1999) Mechanisms and clin-
ical features of posterior border-zone
infarcts. Neurology 53:1312–1318
21. Del Sette M, Eliasziw M, Streifler JY,
Hachinski VC, Fox AJ, Barnett HJ
(2000) Internal borderzone infarction: a
marker for severe stenosis in patients
with symptomatic internal carotid ar-
tery disease. For the North American
Symptomatic Carotid Endarterectomy
(NASCET) Group. Stroke 3:631–636
426
22. Boiten J, Lodder J, Kessels F (1993)
Two clinically distinct lacunar infarct
entities? A hypothesis. Stroke 24:652–
656
23. de Jong G, Kessels F, Lodder J (2002)
Two types of lacunar infarcts: further
arguments from a study on prognosis.
Stroke 33:2072–2076
24. Lodder J, Bamford JM, Sandercock PA,
Jones LN, Warlow CP (1990) Are
hypertension or cardiac embolism
likely causes of lacunar infarction?
Stroke 21:375–381
25. Roh JK, Kang DW, Lee SH, Yoon BW,
Chang KH (2000) Significance of acute
multiple brain infarction on diffusion-
weighted imaging. Stroke 31:688–694
26. Baird AE, Lovblad KO, Schlaug G,
Edelman RR, Warach S (2000) Multi-
ple acute stroke syndrome: marker of
embolic disease? Neurology 54:674–
678
27. Min WK, Park KK, Kim YS, Park HC,
Kim JY, Park SP, Suh CK (2000)
Atherothrombotic middle cerebral ar-
tery territory infarction: topographic
diversity with common occurrence of
concomitant small cortical and subcor-
tical infarcts. Stroke 31:2055–2061
28. Szabo K, Kern R, Gass A, Hirsch J,
Hennerici M (2001) Acute stroke
patterns in patients with internal carotid
artery disease: a diffusion-weighted
magnetic resonance imaging study.
Stroke 32:1323–1329
29. Arauz A, Murillo L, Cantu C,
Barinagarrementeria F, Higuera J
(2003) Prospective study of single and
multiple lacunar infarcts using mag-
netic resonance imaging: risk factors,
recurrence, and outcome in 175 con-
secutive cases. Stroke 34:2453–2458
30. Hart RG, Halperin JL (2001) Atrial
fibrillation and stroke: concepts and
controversies. Stroke 32:803–808
31. Timsit SG, Sacco RL, Mohr JP, Foulkes
MA, Tatemichi TK, Wolf PA, Price TR,
Hier DB (1993) Brain infarction se-
verity differs according to cardiac or
arterial embolic source. Neurology
43:728–733