Scott
W. Atlas,
MD
Vascular
Alexander
S. Mark,
#{149}
M
AGNETIC
resonance
ing can rapidly
vascular
flow
as high
when
sequential
Aneurysm,
was
intracranial,
the
Department
tal of the University
Spruce
St. Philadelphia,
RIG.),
of Radiology,
of Pennsylvania,
PA 19104
and the Department
Hospi-
3400
(S.W.A.,
of Radiology,
Uni-
versity
of California
at San Francisco
(A.S.M.,
E.K.F.).
From the 1987 RSNA
annual
meeting.
Received
April
13, 1988; revision
requested
May 18; revision
received
June 1; accepted
June
27. Address
reprint
requests
to S.W.A.
© RSNA,
1988
See also the article
by Atlas et al (pp 449-453)
and the editorial
by Bradley
(pp 574-575)
in
this issue.
and
define
the
Robert
vessel
in
seven
venous
AND
angioma
invasion,
both
spin-echo
(SE)
Images
time
(TR)
of 500-800
echo
time
(TE)
800/20)
and
with
GRE
images
a short
msec
obtained
were
repetition
and
of 20 msec
were
a short
(TR/TE
in all
500-
cases,
as
were long TR/short
TE (2,500-3,000/2030) and long TR/TE
(2,500-3,000/80)
sequences.
For 42 of 63 long TR SE images,
first-order
gradient
moment
nulling
was
employed
for constant-velocity-flow
compensation
(3),
which
has
been
our
stan-
dard head imaging
protocol
since it became available.
In all 63 cases, the sequential
section
acquisition
technique
gradient
recalled
state
(GRASS)
a TR of 150-200
msec,
and
a flip
variables
mize
the
intensity
depiction
and
the
nal
and
fluid
stationary
dient
moment
GRE
of two
section
gap,
images
were
to four
X 128
to opti-
flow
as high
of cerebrospi-
fluid
were
as low
in the
obtained
256
malformation
imgra-
two
without
SE
with
the
intersection
X 256
matrix,
a 20-cm
field of view.
The diagnosis
was arteniovenous
formation
(AVM)
in 23 cases, occult
brovascular
in-
a 3-5-mm
a 0-2.5-mm
or
other
obtained
technique.
excitations,
thickness,
a 256
the
61 of 63 cases,
GRE
with
first-order
nulling;
cases,
GRE images
this flow-compensation
use
selected
of blood
depiction
(1,2,4,5).
In
was performed
in
of 50#{176}-60#{176}.
These
were
tensity
aging
and
acquisition
(2) was employed,
msec,
a TE of 13-16
angle
imaging
in
five
occlusion,
in
neofive
cases,
and
on thrombosis
cases,
and
on
a combination
tomognaphic
findings
in
in
(CT),
SE MR.
16 cases.
The
compared
a retrospective,
with
the
17
of computed
and
SE
GRE
clinical
MR
images
MR images
nonblinded
fashion.
RESULTS
Malformation
METHODS
Sixty-three
patients,
aged 1 1 weeks
to
80 years, were examined
with MR imaging at 1.5 T (Signa;
General
Electric
Medical Systems,
Milwaukee).
In all patients,
obtained.
11 cases,
intracranial
cases,
aneurysm
in 12 cases.
These
diagnoses
were
based
on surgical
and
pathologic
findings
in
cases,
on angiognaphic
findings
in 30
were
limitaMR imlesions.
MD
Applications
MR Imaging’
cases,
clini-
I. Grossman,
#{149}
Arteriovenous
PATIENTS
Radiology
From
MD
mangioma)
plasm
in
acquisition
cal applications
and possible
tions
of gradient-echo
(GRE)
aging
in vascular
intracranial
steady
with
1988; 169:455-461
(MR)
imagdepict
intrasignal
intensity
section
to evaluate
17.73
#{149}
Arteniovenous
malformations,
cranial,
17.70
#{149}
Brain neoplasms,
MR studies,
10.36 #{149}
Cerebral
blood
vessels,
abnormalities,
17.70 #{149}
Cerebral
blood
vessels,
MR studies,
17.1214
#{149}
Magnetic
resonance
(MR), pulse sequences
I
K. Fram,
and gradient
refocusing
for echo
formation
are used
(1,2).
However,
the
role of this technique
in clinical
neurologic
imaging
has not yet been
detenmined.
The purpose
of this study
called
terms:
#{149}
Evan
Intracranial
Lesions:
of Gradient-Echo
To investigate
the role of the gradient-echo
(GRE) technique
in clinical
intracranial
magnetic
resonance
(MR) imaging,
63 patients
with a variety of vascular
intracranial
lesions
were examined
at 1.5 T with the use
of spin-echo
(SE) and GRE sequences.
In all cases, the sequential
section
acquisition
technique
called
gradient
recalled
acquisition
in the
steady
state (GRASS)
was employed;
a repetition
time of 150-200
msec,
an echo time of 13-16 msec, and a
flip angle
of 500_600
were used to
optimize
the depiction
of blood
flow as high intensity
and the depiction
of stationary
fluid as low intensity.
In 61 of 63 cases, gradient
moment
nulling
was utilized
to
compensate
for first-order
flow effects. Although
GRE images
rapidly
demonstrated
flow in vascular
intracranial
lesions
as high intensity,
the vascular
nature
of these lesions
was also clearly
evident
on SE images in most cases. In some cases,
GRE images
can be used to clarify
the vascular
nature
of a lesion
or to
characterize
a neoplasm.
Other
applications
include
the detection
of
vascular
thrombosis,
occult
vascular
malformations,
and hemorrhagic
complications
of vascular
lesions.
Index
MD
(cavernous
and
malcemehe-
In 21 of 23 cases
of AVM,
the lesions
were
clearly
depicted
by both
SE and GRE imaging
(Fig 1). In the
other
two cases,
the GRE images
were
superior
in defining
flow
in residual
patent
vessels
after
surgery
or proton-beam
therapy
(Fig 2). In all 23
AVMs,
the SE findings
consisted
of
serpentine
and mound
areas
of signal
void
(Fig 1). The GRE images
varied
in their
depiction
of these
lesions.
In
i4 of 23 cases,
the entire
AVM,
as
seen
on SE images,
was depicted
as
high
intensity
on GRE images
(Fig 1).
In seven
of 23 cases,
large
areas
of the
lesions
depicted
as signal
void
on SE
images
were
also seen
as signal
void
on GRE images
obtained
with
gradient moment
nulling
(Fig 3). In one
AVM,
a substantial
portion
of the lesion was obscured
by magnetic
susceptibility-melated
hypointensity
antifacts
due to its subfmontal
location
near
the interface
between
the brain
and the posterior
ethmoid
sinus
(Fig
4). In one AVM,
low intensity
was
seen
in most
of the lesion,
presumably because
no flow-compensation
gradients
were
employed;
in this
case,
the same
regions
were
high
intensity
on GRE images
obtained
with
the use of gradient
moment
nulling
(Fig 5). In six AVMs,
the GRE images
were
superior
in depicting
associated
hemorrhage,
seen
as regions
of
marked
hypointensity
(Fig 6).
455
T.
‘
J#{149}
Figure
1.
MR images
biguously
depicted
ages.
(c) Axial
GRE
images.
of AVM
as round
(150115)
with
ganglionic
and serpentine
image
obtained
and
a.
components.
Large
right
ganglionic
2.
demonstrate
MR images
linear
between
areas
hemosiderin
15) image
obtained
of residual
after
of hypointensity
from
with
AVM
prior
50#{176}
flip
hemorrhage
larger
(open
therapy.
regions
arrows)
Axial
of edema
and
short
TR/TE
(600/20)
or encephalomalacia
residual
patent
turbulent
flow.
vessel
MR images
(b) SE
of left insular
and
Radiology
is unam-
(b) SE imvoid
on SE
(closed
arrows)
long
TRITE
occipital
lobe.
can be made
(2,800/80)
(b) SE images
(c) Confident
distinction
only
GRE (150/
on axial
c.
left frontal
images
clearly
depict
AVM
as signal
void.
the lesion
as high
intensity
(closed
arrows),
large
nulling
was employed.
Persistence
of signal
void
fects may also have
contributed
to this appearance.
#{149}
(a) and
in left
b.
3.
AVM
(2,800/80)
as signal
angle.
a.
Figure
intraventnicular
C.
proton-beam
within
and
short
TR/TE
(600/20)
(a) and long
TRITE
high
intensity
in areas
of flow
depicted
b.
Figure
456
intraventricular
regions
of signal
void on axial
with
50#{176}
flip angle
demonstrates
AVM
with
Axial
short
TR/TE
(600/20)
(a) and
(c) Although
axial GRE (150/15)
image
obtained
with
50#{176}
flip
regions
of patency
remain
as signal
void (open
arrows),
even
was presumably
due to turbulence,
although
uncompensated
long
TR/TE
(2,800/80)
angle
depicts
some
areas
though
gradient
moment
higher-order
flow
motion
November
of
ef-
1988
7). In nine
Occult
Cerebrovascular
Malformation
In four of the 1 1 cases of occult
cerebrovascular
malformation,
all internal
signal
characteristics
seen on
SE images
were lost on the GRE images (ie, marked
hypointensity
was
present
throughout
the lesion)
(Fig
cases,
no evidence
of flow-
ing blood
within
the lesion
was seen
on GRE images.
In two cases, high intensity
was present
within
the lesion
on the GRE images,
which
did not
differentiate
ty clot.
pointense
GRE
flow
In
images,
cases,
two
lesions
which
from
Venous
high-intensi-
additional
were
identified
were
seen
on SE images;
these
lesions
sumably
represented
other
occult
rebrovasculam
malformations.
not
hyon
clearly
prece-
Angioma
In three
of the
venous
angiomas,
five patients
with
the GRE images
more
more
cleanly
and
specifically
defined
the nature
of the lesion,
when
compared
with the SE images
(Fig 8). In two cases, no additional
information
GRE
was
obtained
from
the
images.
Aneurysm
In one
of the
eurysms,
the
five
patent
men was more
the GRE image.
tensity
was
ages
b.
portions
with
portion
of the
anlu-
clearly
identifiable
In two cases, high
present
in the
patients
on
the
thrombosed
of the
GRE
and
aneurysm
on
in-
im-
patent
lumen,
thereby
making
it virtually
impossible to distinguish
a high-intensity
clot from blood
flow on the GRE sequence
alone
(Fig 9). In two cases,
there
was
no
information
imaging
difference
in diagnostic
obtained
techniques.
with
the
two
Neoplasm
In three
of the seven
cases in
which
theme was a neoplasm,
the GRE
images
were more specific
than the
SE images
in delineating
ence
of large
intratumoral
Figure 4. MR images of subfrontal
AVM. Sagittal
short TR/TE (600/20)
(a, b) and axial long
TR/TE (2,800/80)
(c) SE images clearly depict subfrontal
AVM as regions
of signal void
(closed
arrows).
(d) On axial GRE (150/15)
image obtained
with 50#{176}
flip angle, portions
of
the lesion are obscured
by hypointensity
(open arrows)
due to diamagnetic
susceptibility
gradient-induced
signal loss from nearby
air-brain
interface.
one of these
eas of signal
cases,
void
the presvessels.
In
linear
and focal amon SE images
were
thought
to represent
large
intratumoral
vessels
rather
than
calcification on the basis
of morphologic
characteristics,
but GRE images
showed
even
tensity
in
magnetic
more
these
profound
areas,
susceptibility-induced
nal loss (Fig 10).
found
to represent
ependymoma
another
patient
the
in an
low
which
of vessels.
GRE
In
fossa
a vascular
hemanbasis
of CT finddemonstrated
linear
mass,
Axial
were
at CT and surgery.
with a posterior
heterogeneous
within
sig-
These
areas
calcification
mass
thought
to be
gioblastoma
on the
ings,
the SE images
tive
hypoin-
indicating
intensity
was
images
GRE (150/
sugges-
failed
15) images
to
Figure
5.
tamed
draining
tensity
with
50#{176}
flip angle.
Enlarged
vessel
AVM
(arrows)
appears
as low inon image
obtained
without
gradient
ob-
moment
nulling
(a), even though
sequential
section
acquisition
gradient
refocusing
was
used. High intensity
is seen within
this yessel
ment
Volume
169
Number
2
on
image
nulling
obtained
with
gradient
mo-
(b).
Radiology
457
#{149}
show
and
high
intensity
an avascular
found
at surgery
Vessel
(Fig
steady-state
between
areas,
was
11).
Patency
In three
med
in these
astrocytoma
of the
for patency
12 patients
exam-
of vessels
at the
base
of the skull,
GRE MR imaging
clanfied the presence
of thrombosis
(Figs
12, 13). In two cases,
large-vessel
displacement
and compression
by mass
lesions
were
more
clearly
seen
on
GRE
images.
In two
cases,
high
the
situation,
signal
the
intensity
imaging,
difference
of in-
can
signal
be
movement
flowing
spins
and that of stationary
spins
is, to a great
extent,
related
to
the flip angle-the
larger
the flip angle, the greater
the difference
in signal intensities
(1,2,5).
2. The nonselectivity
of the refocusing
mechanism
with
GRE imaging
is also a factor.
In conventional
SE
void
at least
from
partially
rapid
flow
ascribed
of spins
out
of the
to
section
between
the excitation
and refocusing pulses,
both
of which
are section-
selective
(6,7). To emit a signal
in SE
imaging,
both pulses
must be expenienced
by the spin. In GRE imaging,
the refocusing
case,
reversal
mechanism
of the read-out
(in
this
gradi-
inten-
sity was present
in clotted
vessels,
which
appeared
as high
intensity
on
SE images,
thereby
making
flow
indistinguishable
from
clot on the basis
of GRE imaging
alone.
In five cases,
no additional
information
was provided
by the GRE sequence.
I
DISCUSSION
Rapid
MR
ed flip
angles
imaging
and
for echo
acquisition
blood
flow
as high
when
the
posed
images,
utilizing
gradient
can
signal
appropriate
ables
are used
the depiction
high
intensity
a.
(1,2,4,5).
of moving
on GRE
limit-
reversal
demonstrate
intensity
imaging
ular
(a) and
8, 9.
nulling
MR images
of intraventniculan
Intraventricular
long
TRITE
AVM
(2,800/80)
image
obtained
more
obvious
siderosis
is clearly
seen
(b) SE images.
on
with
GRE
from prior hemorrhage
as signal
(c) AVM
50#{176}
flip angle.
image,
indicates
void
on
is depicted
Marked
prior
axial
and intraventnicshort
as high
hypointensity
intraventnicular
TR/TE
intensity
lining
right
hemorrhage.
(600/20)
on axial
ventri-
van-
on SE
fea-
a.
Figure
b.
7.
MR images
of right
c.
panietal
occult
cerebrovascular
malformation.
Focal
high-in-
tensity
regions
are depicted
within
areas
of hemosidenin-related
hypointensity
on axial
short
TR/TE
(600/20)
(a) and long TRITE
(2,500/80)(b)
SE images.
(c) Axial
GRE (150/15)
image
obtained
with
50#{176}
flip angle
depicts
the lesion
entirely
as an area of hypointensity
with
no evidence
of intralesional
flow and without
internal
signal
characteristics.
8b.
Figures
6.
AVM.
GRE (150/15)
cle (arrows),
tunes
of GRE imaging:
1. With
the use of sequential
section acquisition
rather
than
an interleaved
multisection
technique,
theme
is inflow
of unsaturated
spins
into
every
section
(assuming
that at least
some
component
of flow
is present
along
the section-select
gradient).
As
a result,
each
section
is like an “entry
section”-that
is, flow-related
enhancement
(6,7) is observed
on every
section
in the acquisition
(as long
as
the TR is long
enough
to avoid
steady-state
effects).
In the non-
ment
Figure
The basis
of
spins
as
images,
as op-
to the signal
void seen
lies in three
important
8a.
b.
(8) MR images of subtle venous
angioma.
depicts
lesion as area of slight hyperintensity
9a.
(a) Coronal
long TR/short
due to misregistnation
9b.
TE (2,800/30)
of compensated
SE image obtained
with use of gradient
moflow (arrow)
in left frontal
deep white
matter.
(b) Coronal
GRE (150/15)
image
obtained
with
50#{176}
flip angle
demonstrates
more
obvious
and more
extensive
hyperintensity
(arrows)
in angioma,
unambiguously
indicating
flow.
(9) MR images
of partially
thrombosed
left middle
cerebral
artery
giant
aneurysm
depict
both
thrombosed
(1) and patent
(2) portions
of lumen.
(a) On axial short
TRITE
(600/20)
SE image,
thrombosed
portion
of lumen
appears
as
high
intensity,
clearly
distinguishable
from
patent
portion
of lumen,
which
is seen
as signal
void.
(b) Axial
GRE (150/15)
image
obtained
with
50#{176}
flip angle
depicts
both
portions
of the lumen
as areas
of high
intensity.
458
Radiology
#{149}
November
1988
Figure
10.
MR images
depicting
nal void
(arrows)
on axial short
(c) On axial GRE (150/15)
image
compatible
with
CT-documented
calcification
TR/TE
(600/20)
obtained
with
calcification,
in ependymoma.
Right
cerebellopontine
angle
mass contains
linear
and focal
(a) and long
TR/TE
(2,800/80)
(b) SE images,
compatible
with
flow
or dense
50#{176}
flip angle,
more
profound
hypointensity
(arrows)
in areas
of signal
void
not flow,
in patent
vessels.
regions
of sigcalcification.
on SE images
is
the convexities
(Figs
1, 3). More
subtle
vascular
lesions,
such
as small
venous
angiomas
(Fig 8), may become
more
conspicuous
on GRE im-
ages.
b.
Figure
showed
(600/20)
C.
11.
MR images
of avascular
astrocytoma.
marked
enhancement
consistent
with
(a) and long TR/TE
(2,500/80)
of hypointensity
(arrows),
possibly
(150/15)
image
obtained
with
50#{176}
flip
eas
played
hypointensity
cytoma
rather
on SE images
than
169
posterior
fossa mass
On axial short TR/TE
is consistent
with
operative
findings
focal
and irregular
ar(c) On axial
GRE
in regions
that dis-
of avascular
astro-
hemangioblastoma.
moment
nulling
for constant-velocity
Volume
large
(b) SE images,
lesion
contains
indicating
extensive
vascularity.
angle,
absence
of high
intensity
ent) is not section-selective.
Therefore,
spins
that originally
are in the
appropriate
section
for the application
of the excitation
radio-frequency
pulse
can still contribute
signal,
even
if they
have
moved
out of that plane
during
the intempulse
time
(TE/2),
assuming
that they
have
not left the
volume
subject
to the application
of
the read-out
gradient.
The nonselectivity
of refocusing
in GRE imaging
does
not result
in hypenintensity
of
incoming
spins;
it merely
gives
flowing spins
the same
signal
intensity
as
stationary
spins
(if all other
factors
are the same)
rather
than
signal
void.
3. Another
component
of GRE imaging
that contributes
to the high
intensity
displayed
by blood
flow
is the
introduction
of gradient
pulses
for
correction
of phase
changes
due to
flow.
We used
the technique
of gmadient
sate
On CT scans,
hemangioblastoma.
Number
(3)
2
to compenflow
only
(first-order
motion).
Rephasing
induced
by this method
of flow
compensation
cannot
result
in higher
signal intensity
for blood
flow
than
for
stationary
fluid
of the same
composition.
Extra
gradient
pulses
were
added to the section-select
and read-out
gradients
in our GRE acquisitions.
Our
results
indicate
that the value
of GRE imaging
for vascular
intracranial lesions
is limited
and often
depends
on the specific
lesion
in question.
AVMs
are usually
easily
diagnosable
on the basis
of the SE images
alone.
The enlarged
vascular
structunes
were
clearly
seen
as signal
void
in the vast majority
of cases
on short
TR/TE
and long
TR/TE
images.
The
high
intensity
of cemebrospinal
fluid
on long
TR/TE
SE images
often
was
very
useful
in highlighting
lesions
that contained
regions
of signal
void,
especially
when
the abnormal
vessels
were
intraventnicular
or located
over
In two
patients
who
had
under-
gone
proton-beam
radiation
therapy
for AVMs,
GRE images
allowed
confident
distinction
between
small
megions
of residual
patent
vascular
channels
and hypointensity
of penlesional
hemosidenin
seen
on SE images (Fig 2). Treated
AVMs
therefore
may
necessitate
adjunctive
GRE imaging
emphasizing
flow
for complete
noninvasive
evaluation.
The appearance
of large
tumor
yessels was more
specific
on GRE images.
In these
cases,
the initial
interpretation
of hypointensity
on SE images
as due to vascular
flow
void
was
verified
or disproved
on the basis
of
the GRE images.
For example,
in a
surgically
proved,
partially
calcified
ependymoma,
the sensitivity
of GRE
acquisition
to magnetic-susceptibility
differences
allowed
a confident
specific diagnosis
of intratumomal
calcification,
rather
than
blood
flow
in yessels, as the cause
of SE signal
voids
(Fig 10). In another
case, a surgically
proved
avascular
posterior
fossa
astrocytoma,
thought
initially
to represent
a hemangioblastoma
on the basis
of CT findings,
was depicted
as being
without
large
vessels
on the GRE image (Fig 11). Adjunctive
GRE imaging
of intracranial
neoplasms
for possible
vasculamity
therefore
seems
to be another
indication
for this technique,
both
as a means
of improving
the
specificity
of MR imaging
in brain
tumors
and perhaps
as a means
of assessing
the degree
of malignancy.
GRE
imaging
was
often
helpful
in
ascertaining
whether
ambiguous
intravascular
SE signal-intensity
pattemns represented
patent
or thrombosed
lumina
(Figs
12, 13). The diag-
Radiology
459
#{149}
a.
b.
Figure
12.
MR images
of thrombosed
right
internal
ies (closed
arrows)
all demonstrate
signal
void
ular vein
(open
arrow)
shows
moderate
intensity
ther slow
flow or thrombosis.
(c) High-intensity
(open
arrow)
can be differentiated
unambiguously
a.
on
vein.
Normal
left internal
jugular
13.
MR images
of thrombosed
basilar
artery
(600/20)
(a) and long TR/TE
(2,500/80)
(b) SE images
seen on both SE images.
(c) Axial GRE (150/15)
image
thrombosis
rather
than
slow
Radiology
and
left and
right
internal
carotid
arter-
Right
internal
(b), indicating
jugular
vein
jugei-
C.
(arrow).
could
Ambiguous
indicate
obtained
signal-intensity
relatively
with
slow
50#{176}
flip angle
pattern
seen
flow
or thrombosis.
depicts
definite
in basilar
Left
artery
pontine
low intensity
on axial
short
infarction
in basilar
TRITE
is clearly
artery,
indicating
flow.
nosis
of vascular
occlusion
and the
clarification
of patency
in partially
thrombosed
giant
aneurysms
were
often
possible
on these
images,
due
to the depiction
of blood
flow
as
high
intensity.
It has been
clearly
shown
that certain
SE sequences
may
depict
vascular
flow
as variable
intensities
under
certain
conditions,
mainly
due to the phenomena
of
flow-related
enhancement
and evenecho
rephasing
(6,7).
Furthermore,
the recent
implementation
of flowcompensation
techniques,
such
as
gradient
moment
nulling
(3), in moutine head
imaging
(on long
TR SE
images)
often
results
in high
intensity within
or adjacent
to patent
yessels.
There
are several
pitfalls
in evalu-
#{149}
vein
axial short
TR/TE
(600/20)
(a) and long TR/TE
(2,800/80)
(b) SE images.
on short
TRITE
image
(a) and increased
intensity
on long
TR/TE
image
patent
vessels
(closed
arrows)
and low-intensity
thrombosed
right
internal
on axial
GRE (150/15)
image
obtained
with
50#{176}
flip angle.
b.
Figure
460
C.
jugular
ating
nique.
blood
The
GRE
imaging
flow
major
with
this
limitation
to discriminate
GRE techof using
flow
from
thrombus
lies in cases
in which
a hypemintense
subacute
clot of methemoglobin
is present.
Methemoglobin appears
as high
intensity
on GRE
images
obtained
with
the use of imaging
variables
(relatively
large
flip
angle
and short
TE) selected
to demonstrate
flow
as high
intensity
(Fig
9), since the contrast
in these
images
is relatively
Ti dependent
(4). Furthemmore,
even
though
flow
may be
present,
GRE images
can still demonstrate
low intensity
in certain
cases.
When
GRE images
are obtained
without gradient
moment
nulling
(on
with
inadequate
compensation
for
higher-order
flow
effects),
flow
is
not consistently
tensity,
even
variables
bulent
sity
depicted
as high
in-
though
other
imaging
are unchanged
(Fig 5). Turflow is also seen as low inten-
due
to irretrievable
random
spin
dephasing
(8); in approximately
onethird
of the AVMs
in our series,
substantial
areas
of patency
were
depicted as low intensity
on GRE images,
presumably
due to turbulence
(Fig 3).
In-plane
flow
and extremely
slow
flow
(5,9)
can
theoretically
fest as low intensity
although
in practice
be mani-
on GRE images,
these
have
not
been problematic
in our experience.
Hemorrhagic
complications
of vascular lesions
are often
more obvious
on GRE
sitivity
breakdown
images.
of GRE
This
heightened
sen-
imaging
to bloodproducts
has been reportNovember
1988
this
Table
1
Uses of GRE Imaging
Intracranial
Lesions
in Vascular
Small
vascular
malformations
with
abnormal
flow
Venous
angiomas
Residual
AVMs
after therapy
Vascular
malformations
without
abnormal
flow (eg, cavernous
Neoplasm
tibility
(10)
pointensity
blood
flow from
blood
flow from
calcification
Differentiating
hemosiderin
Intravascular
ty.
thrombosis
subacute
(except
when
complications
specific
for
all are associated
on GRE images.
In
most
cases,
of vascular
of signal
with
however,
is due
investigators
to local
(10-12)
and
magnetic-susceptibility
differences
from paramagnetic
tionary
stages
of hemorrhage.
evoluThese
susceptibility
changes
result
in the
static
local
magnetic
field
pentunbations
that shorten
T2*, an effect
that
is compensated
for by the 180#{176}
radio-
frequency
pulse
used
Therefore,
hemorrhage
AVM on a hemorrhagic
in SE imaging.
from an
occult
cere-
brovasculam
malformation
seen
on GRE images
when
can
the
be
SE im-
ages fail to depict
the lesion
(Fig 6).
In one of our cases, however,
an
AVM was partly
obscured
on the
GRE image
by a diamagnetic
susceptibility
gradient-induced
hypointensity artifact
due to its proximity
to
air-brain
be noted
T2*
interfaces
that the
effects
seen
be associated
with
chronic
hemorrhage.
Volume
169
GRE
both
Number
images
acute and
Furthermore,
2
hy-
void.
6.
In a limited
tions
of vascular
lesions,
number
the
7.
on
SE images.
of
Haase
A, Frahm
FLASH
imaging:
AVMs
8.
#{149}
ing
flip
low
angle
pulses.
J Magn
9.
10.
11.
us-
1986; 67:258-266.
2.
Wehrli
F, Hecker
Prost
distinguishing
properties
called acquisition
(GRASS) (abstr).
can
1987;
3.
Pattany
Motion
(MAST)
Tomogr
J, Roberts
F. The
of gradient-re-
flow
imaging.
effects
in magnetic
AJR 1984;
143:1157-
Evans
A, Herfkens
R, Spritzer
CE, et al.
of turbulent
flow on MRI signal intensity
using
gradient
refocused
echoes.
In: Book of abstracts:
Society
of
Magnetic
Resonance
in Medicine
1987.
Vol 1. Berkeley,
Calif: Society
of Magnetic
Resonance
in Medicine,
1987; 354.
Fram E, Karis J, Evans
A, et al.
Fast imaging of CSF: the effect of CSF motion.
In:
Book of abstracts:
Society
of Magnetic
Resonance in Medicine
1987. Vol 1. Berkeley,
Calif: Society
of Magnetic
Resonance
in
Medicine,
1987; 314.
Edelman
RR, Johnson
K, Buxton
R, et al.
MR of hemorrhage:
new approach.
AJNR
1986; 7:751-756.
Mills TC, Ortendahl
DA, Hylton
NM,
Crooks
LE, Carlson
JW, Kaufman
L. Partial flip angle
MR imaging.
Radiology
1987;
12.
Reson
Blood
The effects
detec-
J, Matthaei
D, et al.
rapid NMR imaging
154:443-450.
L.
1166.
References
1.
Axel
resonance
tion of occult
cerebrovascular
malfonmations
and lange-vessel
thrombosis
(unless
the clot is in the high-intensity methemoglobin
stage),
and chamactenization
of intracranial
neoplasms
visualized
as heterogeneous
signal
loss
Buxton
RB, Edelman
RR, Rosen
BR, et al.
Contrast
in rapid
MR imaging:
Tl- and
T2-weighted
imaging.
J Comput
Assist
Tomogr 1987; 11:7-16.
Fram E, Hedlund
L, Dimick
R, Glover
G,
Herfkens
R. Parameters
determining
the
signal
of flowing
fluid in gradient
refocused
imaging:
flow velocity,
TR and flip
angle.
In: Book of abstracts:
Society
of
Magnetic
Resonance
in Medicine
1986.
Vol 1. Berkeley,
Calif: Society
of Magnetic
Resonance
in Medicine,
1986; 84-85.
Bradley
WG Jr. Waluch
V. Blood
flow:
magnetic
resonance
imaging.
Radiology
1985;
the vascular
naby the presence
with an equivocal
appearance
on the
SE examination,
the GRE technique
can clarify
the vascular
nature
of the
area in question.
Other
useful
applications
of this technique
include
the
diagnosis
of hemorrhagic
complica-
(Fig 4). It should
hypointensity
from
on
5.
SE images
cases
(Table
1), such
as treated
and small
vascular
malformations
ed by many
4.
in-
GRE images
can be
to successfully
demflow in vascular
intraas high signal
intensi-
cleanly
demonstrate
ture of these
lesions
clot is present)
Hemorrhagic
lesions
is not
In summary,
rapidly
obtained
onstrate
blood
cranial
lesions
hemangiomas)
characterization
Differentiating
appearance
tracellular
blood
by-products,
since
calcification
(13), turbulent
flow (8),
chemical
shift, and the boundary
of
regions
differing
in magnetic
suscep-
13.
162:531-539.
Atlas SW, Mark AS, Grossman
RI, Gomori
JM.
Intracranial
hemorrhage:
gradientecho MR imaging
at 1.5 T-comparison
with spin-echo
imaging
and clinical
applications.
Radiology
1988; 168:803-807.
Atlas SW, Grossman
RI, Hackney
DB, et al.
Calcified
intracranial
lesions:
detection
with gradient
echo acquisition
rapid
MR
imaging.
AJNR
1988; 9:253-259.
in the steady state
Magn Reson Imaging
5(suppl
1):l05.
PM, Phillips
JJ, Chiu LC, et al.
artifact
suppression
technique
for MR imaging.
J Comput
Assist
1987; 11:369-377.
Radiology
461
#{149}