Didier
Mathieu,
Catherine
MD
Beges,
#{149} Alain
MD
Focal Nodular Assessment TurboFLASH Twenty-two
patients
focal nodular proved with
hyperplasia pathologic
Rahmouni,
#{149} Pierre
with
25 cases
of
imaged with a TurboFLASH (fast low angle shot) sequence combined with bolus administration of gadolinium tetraazacyclododecanetetraacetic (DOTA), spin-echo (SE) T2-weighted sequences, and postcontrast Ti-
quantified;
the
acid
of the
central
scar were qualitatively SE T2-weighted images, were
hyperintense;
central
scar
intensity
tense
exhibited
associated
areas
analyzed. On all FNHs in two cases the a high signal with
corresponding
tissue within Unenhanced
the branches TurboFLASH
hypoin-
40-80
seconds
(GRE)
RADIENT-ECHO
injec-
magnetic proved
evaluation
of liver
However, the are sometimes
results limited
tumors
our
combined
(1-6).
of this technique because the
study
was
this
to study
characteristics of this trast material-enhanced MR
images
and
180:25-30
I From the Department of Radiology, H#{244}pital Henri Mondor, 51 Avenue du Mar#{233}chal de Lattie de Tassigny, 94010 Cr#{233}teil,France (D.M., AR., M.C.A., B.F., C.B., PG., J.J.M., NV.); and Siemens Medical Systems, Paris (B.F.). From the 1990 RSNA scientific assembly. Received December 21, 1990; revision requested January 23, 1991; revision received March 1; accepted March 13. Address reprint requests to D.M. C RSNA, 1991
In a project mors at high referred to
field
1-year
period
with
the same
since
implementation
sequence.
strength,
all the
our MR imaging underwent
We
have
cases and
drium cases;
prompted a palpable
patients.
of FNH were pain in the
These
right
lesions
were
patients
a used
fortu-
hypochon-
located
patients
dynamic
corn-
after
intravenous
of contrast lesion.
All
bo-
medium the
in
patients
to our institution underwent MR with a 1.5-T magnet (Magnetom Siemens). The study was performed
1. T2-weighted
images
were
obtained
with a TR of 2,2.00 msec and a TE of 45 and 90 msec (TRIFE = 2,200/45, 90), a matrix of 256 x 256, two acquisitions, and an 8-mm with
a gap
sections
with
of 0.8 mm
and
inter-
an acquisition
Artifacts
TurboFLASH
TB/FE
matrix of 128 mm (Fig 1).
re-
in the found
and
of the
follows:
discovered
discovery mass was
plane
2.
viewed the MR imaging characteristics of 25 cases of FNH proved with pathologic examination in 22 patients. These patients were 21 women (mean age, 34 years; range, 18-52 years), 19 of whom had used oral contraceptives, and one man who was 42 years old with no history of steroid use. Eight itously,
before
such
images
with a multisection covering the entire pended respiration nous injection. The
TurboFLASH
retrospectively
ob-
time of
as motion
and
flow were suppressed by means of flow and respiratory artifact obliteration with directed orthogonal pulses in 20 patients (9), with a presaturating band 2 cm thick close to the ventral subcutaneous fat.
hi-
that had been of the
with
puted tomography (CT) performed in other medical centers. In all cases, adjacent axial CT scans 8 or 10 mm thick were
leaved
examination
protocol
and
17 minutes.
unit during
mean
proved of samples
in all these
sonography
thickness
the
liver
were
The
4-12 cm). All
by means of the same protocol with a body coil and axial views in all patients. These successive images were obtained:
in compar-
different
and in the left
included
referred imaging SF 63;
METHODS
studying
PhD
tamed at surgery. Initial examination
the
ison with other previous reports, the features of FNH on spin-echo (SE) MR images in this large group of lesions proved with pathologic examination.
AND
of FNH examination
lus administration
lesion on conTurboFLASH
to discuss,
MD,
lobe in 12 patients
obtained
technique
twofold:
right
these cases pathologic
acquisition time is too long to perform a suitable hemokinetic study of these lesions (6). TurboFLASH (fast low angle shot) (Siemens Medical Systems, Paris) MR imaging techniques have recently been developed to obtain very fast acquisition of images (7,8). have
Falise,
MD
lobe in the other 10 patients. diameter was 6.8 cm (range,
contrast agents have imthe capability of MR imaging
PATIENTS
1991;
in
T2-
Index terms: Liver, focal nodular hyperplasia, 761.3119 #{149} Liver neoplasms, MR studies, 761.1214 #{149} Magnetic resonance (MR), pulse sequences #{149} Magnetic resonance (MR), rapid imaging #{149} Magnetic resonance (MR), tissue characterization Radiology
sequences
magnetic resonance (MR) imaging reduce measurement time by virtue of short repetition times (TRs). GRE sequences enhanced with para-
with the injection of paramagnetic contrast agent in the evaluation of 22 patients with focal nodular hyperplasia (FNH) of the liver. The purpose of
after
with
G
#{149} Beatrice
Vasile,
Liver:
We
scar.
lion. Both unenhanced and enhanced TurboFLASH sequences produced the best signal-difference-to-noise
ratios in comparison weighted images.
MD
#{149} Norbert
MD
of the images
central scar as a hypointense area within the lesion. After bolus injection, arterial enhancement of FNH was clearly seen, and in 10 of 25 lesions, enhancement within the scar
seen
Anglade,
Mollet,
to fibrous
always demonstrated the FNHs as hypointense and always depicted the
was
J.
#{149} Jean
in the
FNH-liver sigratios were
features
#{149} Marie-Christine
MD
Hyperplasia ofthe with Contrast-enhanced MR Imaging’
(FNH) study were
weighted sequences. nal-difference-to-noise
MD
Gheung,
other in seven
in the
3. A bolus
of 7/4, x 128, of 0.2
were
obtained
sequential acquisition liver without susand without intraveparameters were as a flip
angle
mLlkg
sous
bois,
(Laboratoire France)
of 10
of gadolinium
tetraa.zacyclododecanetetraacetic
(DOTA)
of 10#{176}, a
and a thickness
acid
Guerbet,
Aulnay
in a concentration
mol/L Gd-DOTA was injected antecubital vein, followed by sion of saline. With the same as in the previous paragraph, FLASH sequence was started
of 0.5
through an a rapid infuparameters this Turboin the plane
Abbreviations: DOTA = tetraazacyclododecanetetraacetic acid, FLASH = fast low angle shot, FNH = focal nodular hyperplasia, GRE gradient echo, SD/N = signal-difference-tonoise ratio, SE = spin echo, TE = echo time, TI = inversion time, TR = repetition time.
=
25
a.
c.
b.
f.
g.
d.
h.
e.
j.
i.
Figure 1. Images of FNH. (a) TurboFLASH image obtained with an inversion time (TI) of 100 msec (TRTIE,ffI = 7/4/100). Banding artifacts (arrow) and presence of high signal intensity in the aorta and inferior vena cava were caused by flow-related enhancement. The central scar is not seen. In a-g, the flip angle was 10#{176}. (b) TurboFLASH image (7/4/400). FNH is hypointense, with a good visualization of the central scar. (c) TurboFLASH image (7/41600). The FNH-liver signal-difference-to-noise ratio (SD/N) is decreased. (d) TurboFLASH image (7/411,000). Note the area of signal isointensity between FNH and the surrounding liver. (e) TurboFLASH image (7/41400) obtained 10 seconds after injection of GdDOTA. Hypervascularization of the lesion is seen with identification of the hepatic artery (arrow). (f) TurboFLASH image (7/4/400) obtained 18 seconds after injection of Gd-DOTA. Hypervascularization of the lesion is seen with identification of the right portal vein. (g) TurboFLASH image (7/41400) obtained 50 seconds after injection of Gd-DOTA. Note enhancement of the central scar (arrow). (h) Ti-weighted SE image (420/ 15). The lesion is isointense relative to the surrounding parenchyma. Note signal hypointensity of the central scar. (i) T2-weighted SE image (2,200/90). The lesion is hyperintense, with high signal intensity of the central scar. (j) Contrast-enhanced Ti-weighted SE image (420/15). Note enhancement of the central scar with a small area of signal hypointensity (arrow) that could not be explained at pathologic examination.
of the lesion immediately after injection and lasted 2 minutes. After the end of the bolus
injection,
every 4.
an
image
2 seconds for Then, 4 minutes
was
a total after
8
acquired
of 60 images. the bolus injec-
tion, a delayed contrast-enhanced weighted sequence was started
6 0
z
TIwith the
0
z
0
0
Cl)
CI)
following parameters: 256 x 256 matrix, and with an 8-mm thickness, and
interleaved
tion
time
sections
of 7 minutes
TurboFLASH
acquisition
an
Cl)
small
flip
very
small
provides nique,
described by that allows
angles.
With
such
TI relaxation
Haase very
TI
short
et al, short
place TE.
before
before To gener-
inverts one
the
short
im-
as the time and the beHence, the
relationship
the
TI and
of the TurboFLASH on the spin-lattice
of the tissue and is similar inversion recovery sequence. The shortest TR and TE
26
#{149} Radiology
times
3000
0
2000
1000
Inversion
( TIs
Times
3000
( TIs)
b.
a.
Figure 2. (a) Mean values (b ) Mean values of FNH-liver bottom
are
of liver-spleen SD/N versus
SD/N versus Tis in three
TIs in five healthy volunteers. cases of FNH. In a and b, numbers
at
milliseconds.
a
age acquisition. TI is defined between the inversion pulse ginning of the GRE sequence. between
Inversion
2000
and thus
contrast. In this techis also minimal, because
a 180#{176} pulse
magnetization
TRs,
occurs
dephasing takes with a very short
ate Tl contrast,
strength depends
2
1000
times
a poor T2 contrast
little T2* sampling
z
::
acquisi-
30 seconds.
(7,8). This ultrafast of images is obtained by means sequences with short TRs and
acquisition of GRE
spins
with
4
a
Technique
This sequence, a new technique
is data
TRII’E of 420/15, four acquisitions a 0.8-mm gap,
signal
sequence Ti relaxation
were for
fixed a 128
at 7 and x
128
gle
of 10#{176}; with appears. used, 128 lines msec.
signal
To and
obtain to study
We
respectively, chose
enhancement of the lesions, the SD/N between the liver with different TIs. With TIs
in an
300 msec, the from 300-600
on
system
tamed
and
700
a flip
an-
a smaller angle, a loss of When a TRItE of 7/4 is of data are acquired in 900
heavily Ti-weighted the paramagnetic
to that our
4 msec,
matrix.
SD/N msec,
dropped
was this for
poor. ratio TIs
images contrast we compared and the spleen shorter than Ranging was mainlonger
than
msec
(Fig
2a).
Because
we
wanted
to
study FNH of the liver, we established the same curve for the SD/N between FNH and liver. To determine the best contrast, we measured SD/Ns of three typical cases of FNH normal
(ie, central surrounding
examination).
This
for short TIs (less Hence, we chose of 400
msec,
which
scar and lesion with liver at pathologic contrast
was
than 600 msec) the compromise provided
trast of FNH and liver. The ratio of signal difference FNH and surrounding liver
good
obtained (Fig 2b). of a TI Ti
con-
between to noise was
July
1991
b.
c.
e.
f.
Figure 3. FNH of the left lobe. (a) Dynamic perintense signal corresponds to the central with low signal intensity of the scar (arrow).
CT scan shows scar within the (d) TurboFLASH
g.
hypervascular mass without central scar. (b) T2-weighted SE image (2,200/90). Hylesion. (c) TurboFLASH image (7/4/400, flip angle of 10#{176}) shows hypointense lesion image (7/4/400, flip angle of 10#{176}) obtained 12 seconds after injection of Gd-DOTA.
At the
arterial phase, tumor shows high enhancement. The scar remains hypointense. (e) TurboFLASH image (7/4/400, flip angle of 10#{176}) ob20 seconds after injection of Gd-DOTA. The scar remains hypointense relative to the tumor. (f) TurboFLASH image (7/4’400, flip angle 10#{176}) obtained 60 seconds after injection of Gd-DOTA. The scar (arrow) shows high signal intensity within the tumor. (g) Contrast-enhanced Ti-weighted SE image (420/15) shows high signal intensity within the scar.
tamed
computed
as follows:
tensity
of lesion
-
SD/N
signal
(signal
=
in-
intensity
different
of
liver)/standard deviation of background noise. Measurements of signal intensities
in the homogeneous the central scar were ages
by
interest
use
area of FNH outside obtained from im-
of operator-defined
encompassing
Noise
was
of interest
with
large
that encompassed
method
offers
regions
of interest
ences
regions
ventral
and are more
measurement
of noise
tient
respiratory
includes
the entire
average ventral
Second, to the
motion
pa-
artifacts
that may degrade image quality in the region of the liver. As a result, our noise measurement
reflects
both
incoherent
noise and coherent motion 12). MR images were thus evaluated for FNI-I signal
graded
tensity of surrounding TurboFLASH images, jection of paramagnetic FNH. quences
Student
were t test
characteristics
Volume
liver on SE and before and after contrast agent,
of the contrast
SD/Ns
for
the
enhancement
various
compared for paired of the
180
(10-
intensity and with the signal in-
in comparison
the peak
artifacts quantitatively
#{149} Number
pulse by means samples.
central
scar
1
inat
of seof a The
on
the
plain were
and
en-
also
cases
present
in the
lesions
were
SD/Ns
between
hyperintense,
FNH
with and
surround-
ing liver higher on the first echo at 45 msec than on the second echo at 90 msec (9.4 msec ± 3.2 [standard deviation] and 5.8 msec ± 3.1, respectively). of FNH
were
same
lobe
in three
patients
differ-
reproducible.
Twenty-five
were
First, large small
on
images
demonstrated with the different imaging techniques. Two cases of FNH
regions
to the patient, bands. This
two advantages:
and
RESULTS of
at least 25 pixels.
measured
image background out of the presaturated
SE images
hanced TurboFLASH qualitatively assessed.
of
The
differences
in the
of lesion-liver 2,200/45 and not statistically
absolute
values
SDINs on both SE SE 2,200/90 images were significant (Figs ii, 3b,
and were visualized on the same plane at MR imaging. One FNH, 1 cm in diameter, adjacent to a large 7-cm main FNH tumor, was diagnosed at pathologic examination only and was not found retrospectively with the different imaging techniques. Furthermore, three patients had associated hemangiomas, 3 and 4 cm in diameter, that were resected along with FNH. These hemangiomas had been previously diagnosed at sonography, dynamic CT, and MR imaging. Furthermore, at pathologic
4).
examination
changes in the large septal vessels and the presence of loose connective tissue within the branches of the scar without necrosis. These vascular changes were observed in two
in two
patients,
hemo-
siderosis without obvious cause was present within the surrounding liver resected with FNH. On SE T2-weighted images, all the
The
fled
central
with
peared
scar
these
was
hyperintense
(24 cases)
always
sequences
identi-
and
ap-
as a thin
or as a network
band
pattern
(one case). The scar was identified CT scans in 18 of 25 cases of FNHs
(72%).
However,
(10 and
7 cm
hypomntense were
noticed
in two
large
in diameter),
radiating on
on
lesions
associated
linear
areas
T2-weighted
SE im-
ages (Fig 4a, 4b). In these two cases, pathologic examination demonstrated prominent
obliterative
vascular
Radiology
#{149} 27
a.
b.
d.
c.
e.
f.
Figure 4. Large lesion of the right lobe in a patient with hemosiderosis at pathologic hypointense band (arrow) and hyperintense bands within the tumor. (b) T2-weighted bands within the tumor. (c) TurboFLASH image (7/4/400, flip angle of 10#{176}) shows slightly surrounding liver. Hypointense bands are present within the tumor. (d) TurboFLASH after bolus injection, shows hypervascular mass at the arterial phase. (e) TurboFLASH after bolus injection, shows persistence of hypointense bands within this hypervascular (420/20) shows high enhancement of the previously hypointense bands with a network
women
who
contraceptives
had for
been taking oral 8 and 10 years, re-
cases were then hyperintense on sequence (Fig 40. The high signal intensity of the scar was homogeneous in 23 cases of FNH, and in two two
this
spectively. On
unenhanced
TurboFLASH
im-
ages obtained with the parameters previously described, all the lesions were recognized as hypointense masses in comparison with the liver (Figs lb. 3c, 4c). The SD/N was equal to -14.7 ± 2.3. The scar was always demonstrated as a hypointense area within the tumor on these precontrast images (Figs lb. 3c). On
contrast-enhanced
cases
In our
Turbo-
to the
#{149} Radiology
because
of
study,
both
the
unenhanced
accumulation
lesions.
of
with
Because
facts
can
ages
of the
degrade
that
Recent methods have permit MR imaging of
upper
the
quality
abdomen,
motion of MR
shot
a modified
ery-type
of paramagnetic
contrast
agents
The
and
not et al and
or GRE systems used
ation single-shot
first consists
These
with
very
GRE short
allows
seTRs
For the
inversion
TurboFLASH,
of on
prepulses.
This
desnap-
available
(7,8).
of Ti-weighted
than
generally
called
technique
a 180#{176} prepulse images.
yet
TurboFLASH,
radio-frequency
less necessitates
Haase
are
ample,
times
technique technique,
FLASH
different
this
described imaging
Echo-planar
Another by
im-
characterization
tumors.
has
authors
been with
less.
hardware
scribed
quences
of hepatic
sequences
imaging
but
available.
anti-
numerous
or
permits
msec,
techniques have been developed to reduce these different artifacts. Breathholding with fast imaging and panticularly GRE sequences has been used for role
i second
specialized
respiratory
fast
by different
(1-6).
100
vanliver
these
emphasized
times
MR imaging has been used with ous degrees of success for studying
SE
associated
been
imaging
DISCUSSION
paramagnetic contrast agent was observed within the scar (Figs lj, 3g). Furthermore, the hypointense areas noticed on T2-weighted images in 28
heterogeneous
and enhanced TurboFLASH images produced the best SD/Ns (-14.7 ± 2.3 and 21.0 ± 5.3, respectively), in cornparison with the poor SD/N of the enhanced Ti-weighted images (4.3 ± 2.1). Comparison of unenhanced and enhanced TurboFLASH images with the SE T2-weighted images obtained with a TE of 45 msec (9.4 ± 3.2) showed significant differences (P < .01) (Table).
2 minutes
Ti-weighted images, all areas of FNH were hyperintense to the liver, with an SD/N equal to 4.3 ± 2.i. In all 25 cases of FNH, high signal intensity corresponding
it was
the presence of a hypointense area less than 0.5 cm in diameter (Fig lj). In these two cases, pathologic examination did not reveal peculiar features within the scar.
FLASH images, the peak of contrast enhancement was observed 10-24 seconds after bolus injection (Figs le, 3d, 3e, 4d, 4e). All the lesions were then hyperintense, with an SD/N between FNH and liver equal to 21.0 ± 5.3. In 10 cases, an enhancement was visualized within the scar 40-80 seconds after injection (Figs 1g. 3f). In the remaining 15 cases of FNH, the scar remained hypointense after bolus injection. On delayed contrast-enhanced
examination. (a) T2-weighted SE image (2,200/45) shows SE image (2,200/90) shows persistence of hypointense hypointense lesion in relation to hemosiderosis of the image (7/4/400, flip angle of 10#{176}), obtained 18 seconds image (7/4/400, flip angle of 10#{176}), obtained 24 seconds mass. (f) Delayed postcontrast Ti-weighted SE image pattern.
technique
ex-
cre-
recovof
acquiring
all
July
1991
lines
of data,
has
different
advantages:
on the
plain
TurboFLASH
multisection sequential acquisition and a short imaging time allowing a suitable
thin hypointense of lines within
study
for
i28
ment
(13). Another
dynamic
segmented
been
contrast
technique,
k-space
called
TurboFLASH,
has
by Edelman et al (14). This method spans k space in four segments of 32 steps each to generate
recently
enhance-
a 256
described
x 128-pixel
breathhold
image
of 12-14
method offers different comparison with the
FLASH technique ter tissue contrast,
during
seconds.
advantages single-shot
in Turbo-
: high resolution, and diminution
artifacts.
The
with this to buildup
technique are probably of unspoiled transverse
netization. technique
banding
a good
tumors
Ti
without
contrast
tion
by
time,
the
the
from
300
subsecond
an enhancement in
40-80
3f).
This
after
and/or
stant
visualization
or into
and
the
In the other is usually
cases
of FNH
at 0.6
frequency
T on Ti-weighted
always
appeared
high signal ter injection)
lowed
at the portal
phase,
described
slightly
hyperintense
one would of hyperplastic, hepatocytes study,
all
of
these
with
2,200/45
images
than
SE 2,200/90
significant
lesions
a higher
on
the
images,
statistical
on
difference
The reason
contrast quence
first echo unclear:
has
were
comparison with
an The
Volume
slightly
with SD/N
central
180
the
ratio scar
surrounding
was
#{149} Number
to 4.3 always
1
resolution
intensity
of the
of the scar
on Ti-weighted intensity on T2-
images
images
can be explained
of slow
blood
flow
contains fibrous the vessels, and (22). However,
characteristics can
be
observed
of these also
by within
tissue the bile MR vascular
in
been
ages
in
liver, ± 2.i.
identified
did
not
become
(19,23). carcinoma
hyperintense
In
on
long TRTFE SE images (24). On T2weighted images in two cases of FNH, we also observed the presence of hypointense areas corresponding to fibrous tissue and obliterative vascular changes at pathologic examination 4a, 4b). This finding was previously
mentioned
by of FNH on the tamed at 1.5 T correspond to FNH,
particularly
(Fig
Schiebler et al in one case long TRITE images ob(20). These findings could a complex appearance of in large
lesions,
(25).
with
a
vascular
inthe
malforma-
Finally,
on the
images, demonstrated
scar and
in the previ-
hypointense
areas
to the stagnation contrast agent fibrous
hy-
tissue.
of this
cor-
of the within The
the
incon-
enhancement
during contrast-enhanced TurboFLASH MR imaging can be explained by the relative short time (2 minutes) of the acquisition.
In a recent
letter,
Tham
et al
reported one case of FNH on Tiweighted images obtained 6 minutes after administration of gadopentetate dimeglumine ing septa
that demonstrated as structures with low
intensity
separating
high
signal
(26).
In
nodular
intensity
lobules
within
the
fact, as demonstrated the
contrast
radiatsignal
of
tumor
in our
agent
is present
within the scar delineating the lesion and not in the nodules with lower sig-
nal intensity In
DOTA
ij, 3g. 4f).
(Figs
conclusion,
enhancement
significantly
Gd-
with
improves
lesion-
liver contrast when dynamic scanning performed with this TurboFLASH se-
reported by Titelbaum et al, the scarlike area hypointense on short TRTFE im-
of this seAt pathologic
equal
CT in
no
of this higher
hyperintense
The
with
the central mentioned
study,
ence of proteinaceous fluids fibrolamellar hepatocellular
hy-
examinations of these resected tumors, no necrosis was observed despite their large size. On Ti-weighted images obtamed 4-12 minutes after bolus injection of paramagnetic contrast agent, the lesions
18 of 25
SE
second-
although
demonstrated. on the remains
were SD/N
in
in all le-
in
both hepatocellular carcinoma and giant hemangioma, which present with frequently inflammatory scar tissue marked by edema, hypercellularity, and the presscars
composed
in
on CT scans.
better
presence
imaging
but otherwise normal, (i5-i7,i9-2i). However,
perintense,
echo
with
a finding
in a lesion
the
the scar, which accompanying duct proliferation
or
lesion,
expect
to
present in FNH and high signal the
previous images
isointense
due
weighted
surroundinjection)
as described
as an
a
new CT scans. The low signal
of isointense
FNH and the seconds after
CT (18). In different FNH on T2-weighted
reports,
our
arterial
images
this study is higher than the 43.5% frequency in our previously described series of 23 cases of FNH (18). It is probably
seconds afphase, fol-
MR
study,
of this observation
bo-
lesions with
(10-24
by an area
signal between ing liver (30-50 dynamic
After
the
hyperintense
intensity at the
rapidly
was
images.
of Gd-DOTA,
and
images.
the scar
on
as in our
(72%)
thick-
reports,
analyzed
images
at
authors
the scar was identified MR
dis-
of six
obtained
of the
published
sequences
injection
interval
on CT scans,
which
three
the
Ti-weighted was always
responding paramagnetic
pat-
women
or the hepatocytes.
ously
in
et al have
the section
gap
better
nodules
network
Mattison
not mentioned
on
lus
into
a peculiar
in one case.
sions
0.35
the lesion
thromboand was cords, as
and Medline these findings
in two
postcontrast perintensity
T2-weighted images in our study, scar was always identified as a thin
dividing
of previous
of either
tion
fibrous of some
of oral contraceptive is known to stimulate
vessels
than
performed by Butch et al (15) by El Rahman et al (16) at 0.15 T on Ti-weighted images and in one of the six FNH lesions reported by Matisson et al (17) at a field strength of
history which
On the
(Figs
T and
observed
long take,
within
have
in only three cases to on inversion recovery
were
In the previous reports, this scar appeared hypointense relative to liver on Ti-weighted images and hyperintense on T2-weighted images (i5-i7,i9-21).
tinguished this scar in only cases of FNH on SE images 0.35 T (i7). However, these
been
Evidence
reported by Wanless It is noteworthy that
ig,
could be due to blood flow within scar contains nu-
enhancement
in the
thrombosis
sis was seen in 45% of lesions prominent in 22% with fibrous
of the (Figs
scarring
organized
arteries.
25 lesions
injection
and
growth
tern
had
10 of the
by dense
body
merous dilated, thin-walled vascular channels among dense and/or loose connective tissue and occurs as a result of vascular malformation in FNH (22).
ness
already reported our knowledge:
seconds
24 cases
studies
hypointensity
seen
the presence of slow the scar. This vascular
On unenhanced TurboFLASH images in our study, FNH always appears hypointense with a TI equal to 400 msec This
was
marked
as a
of enhanced
scar
the
performed.
i, 3).
By use
TurboFLASH,
to 700
acquisi-
dynamic
x 128 matrix.
band
liver
by the images.
the
enhanced
easily
related mag-
degrading
supported
Furthermore,
observed
in
artifacts
information
betof
however, demon-
for a TI ranging
msec
are
artifacts
In our study, of TurboFLASH
strates
a
This
image
band or as a network the lesion despite a
quence,
particularly
in cases
of FNH.
In
our study, the unenhanced FLASH sequence permitted visualization of the central
Turboconstant scar, which
was
in only
identified
on CT scans
is
72%
of the cases. With this sequence, the enhancement characteristics of this tumor are similar to those on dynamic CT scans.
The
delayed
weighted
sequence
all cases
in showing
enhanced
Ti-
appeared
useful
the
persistent
in en-
hancement of the scar, probably in relation to the stagnation of the paramagnetic contrast and/or in the
the scar.
agent fibrous
within tissue
the vessels present in
This
sequence permitted the in two cases in which hypointense bands, associated with hyperintense bands, were present on T2weighted images, because late enhancement of these areas was shown. Therefore, these MR imaging sequences (unenhanced, enhanced TurboFLASH, and delayed Ti-weighted sequences) diagnosis
of FNH
demonstrated
constant
patterns
when
findings
were
atypical
T2-weighted studies
images. that
include
helpful
present
However, assessment
tocellular adenomas and hepatocellular carcinomas
on
larger of hepa-
fibrolamellar with these Radiology
#{149} 29
a.
c.
b.
Figure image flip
Al. (a) Dynamic CT scan, (2,200/90) shows hyperintense
angle
obtained 15 seconds and homogeneous
sequences
will
be helpful
ADDENDUM
A 62-year-old
hepatitis
man with
B without
B
virus
chronic
replication
underwent
of the
MR
hep-
imaging
ac-
cording to our protocol for a focal liver mass. Previous sonography and dynamic CT demonstrated a well-limited tumor of the right lobe of the liver (Fig Ala). The
a-fetoprotein level was normal. T2weighted MR images showed a homogeneous and hyperintense lesion with a hyperintense seconds
central scar after intravenous
(Fig Aib). Ten injection of Gd-
DOTA, the lesion was markedly enhanced on a TurboFLASH image (Fig Aic). On the TurboFLASH image obtained 2 minutes after injection of contrast material, the enhancement
ever,
of the scar
was
a rim enhancement
also
seen
(Fig Aid).
weighted
the lesion
features
tence
of a rim enhancement finding
ries
as FNH was
of 25 FNHs,
not
was Ti-
exhibited
except
capsule
with
the persisin our
the patient
numerous
composed
Se-
under-
vessels,
of fibrous
and
tissue.
a true
Patho-
for must
is a well-known
of a capsule feature
cellular carcinoma and has been described in rare cases of hepatocellular adenomas
this
#{149} Radiology
This
this presence
In our
14.
on
emphasizes
be
This
misleading
2.
Yoshida
3.
4.
5.
6.
Edelman
RR,
1990;
8.
Siegel
JB, Singer
A, Matthaei 0, Bartowski Leibfritz D. Inversion recovery
Edelman
RR,
FRODO ology Stark
flow
DJ, Silver
a new and
method
of hepatic
metastases:
Crooks Magnetic
12.
Wehrli
LE, Arakawa resonance
20.
21.
22.
field strength. FW,
MacFall
JR. Clover
GH,
scars
in primary
liver
CT
5, et al.
tumors:
hyperplasia
of the liver.
MR fea-
J Comput
24.
25.
BA, nod-
Assist
AC, Gyi B.
HY, Burke
hepatocellular
J Comput Wanless
Hepatocellular
OR, Saul carcinoma:
Assist Tomogr IR, Medline A.
carcinoma:
SH.
nodular
hy-
ML,
Fibrolamel-
MR appearance. 1988; 12:588-591.
The
role of estrogens
as promoters of hepatic neoplasia. Lab Invest 19fi2; 46:313-320. Tham RTOTA, Holscher HC, Falke THM, Arndt JW, Lamers CB. Focal nodular hyperplasia of the liver: features on Gd-DTPA-enhanced MR (letter). AIR 1989; 153:884-885.
27.
Ebara
28.
small hepatocellular carcinoma: correlation of MR imaging and tumor histologic studies. Radiology 1986; 159:371-377. Rummeny E, Saini 5, Wittenberg J, et al. MR imaging of liver neoplasms. AJR 1989; 152:493499. Hosoki T, Chatani M, Mon S. Dynamic com-
of
-
29. N.
IR,
Tomogr 1987; 11:651-654. Nokes SR. Baker ME, Spritzer CE, Meyers W, Herfkens RJ. Hepatic adenoma: MR appearance mimicking focal nodular hyperplasia. Comput Assist Tomogr 1988; 12:885-887. Wanless IR, Mawdsley C, Adams R. On the pathogenesis of focal nodular hyperplasia of the liver. 1-lepatology 1985; 5:1194-1200.
sis
Grigsby
dynamic
160:53-58. R, Sironi
specificity, and pathologic correlation. Radiology 1989; 171:323-326. Schiebler ML, Kressel HY, Saul SH, Yeager Axel L, Gefter WB. MR imaging of focal
lar
-
hyperplasia:
MR appearance mimicking focal perplasia. AJR 1987; 149:721-722. Titelbaum OS, Hatabu H, Schiebler
in MR imaging.
1984; 151:12
nodular
Wilbur
RR, et al.
Radiology
of
LE, Francis
23.
Radi-
et al. of ma
focal
ular
MS.
M, HoenningerJ, imaging: effects
GM, Quint
study. Radiology 1986; Rummeny E, Weissleder
Kressel
E,
analysis
MR imaging
tures,
Tomogr
artifact.
GR, Glazer
Central
for elimination
wrapround
pulse sequence performance Radiology 1986; 159:365-370. netic 133.
Assist
i988; 166:231-236. DO, Wittenberg J, Edelman
Detection
11.
19.
26.
Atkinson
pulses:
of motion,
10.
K,
R, Duhmke snapshot
J Comput
RA.
Bree RL, Ensminger WD. MR imaging of hepatic focal nodular hyperplasia: characterization and distinction from primary malignant hepatic tumors. AIR 1987; 148:711-715. Mathieu 0, Bruneton JN, Drouillard J, Pointreau CC, Vasile N. Hepatic adenomas
and
MRI: applications imaging. Magn
Haase
DD, Malt
Mattison
18.
153:1213-1219.
Haase A. Snapshot FLASH to Ti, T2, and chemical-shift Reson Med 1989; 13:77-89.
1990;
17.
HE. Dynamic MR imaging of the Gd-DTPA: initial clinical results. AJR
FLASH MR imaging. 1989; 13:1036-1040. 9.
A, Dupuis
of the
Radiology
16.
T, Mi-
nami M, Yashiro N. Small hepatocellular carcinoma and cavernous hemangioma: differentiation with dynamic FLASH MR imaging with Gd-DTPA. Radiology i989; 171:339-342. Hamm B, Fischer E, Taupitz M. Differentiation of hepatic hemangiomas from metastases by dynamic contrast-enhanced MR imaging. Comput Assist Tomogr i990; 14:205-216. Marchal G, Demaerel P. Decrop E, Van Hecke P, Baert AL. Gadolinium-DOTA enhanced fast imaging of liver tumors at 1.5 T. J Comput Assist Tomogr 1990; 14:217-222. Van Beers B, Demeure R, Pringot J, et al. Dynamic spin-echo imaging with Gd-DTPA: value in the differentiation of hepatic tumors. AJR 1990; 154:515-519. Longmaid liver with
177:515-521. Butch RJ, Stark
MR imaging
contrast.
focal nodular hyperplasia. J Comput Tomogr 1986; 10:874-877. El Rahman M, Li KCP, Ros PR. Hepatic focal nodular hyperplasia: new MR findings. Magn Reson Imaging 1989; 7:687-688.
sign
K, Kokubo
flexible
Assist
Ohtomo K, Itai Y, Yoshikawa K, et al. Hepatic tumors: dynamic MR imaging. Radiology 1987; 163:27-31. Y, Ohtomo
for breath-hold with
RR. Firstwith ultra174:757-762. Atkinson
hepatic
known.
H, Itai
Atkinson DJ, Burstein 0, Edelman pass cardiac perfusion: evaluation fast MR imaging. Radiology 1990; Edelman RR, Wallner B, Singer A, DJ, Saini S. Segmented TurboFLASH: liver
15.
(b) T2-weighted image (7/4/400,
intensity of the scar. (d) Turboa rim enhancement can be the surrounding liver paren-
method
the
lesion has mimicked of a capsule is atypical
diagnosis.
M, Ohto
M, Watanabe
Y, et al.
Diagno-
of
puted
tomography
noma.
AIR
1982;
of hepatocellular
carci-
139:1099-1106.
The dependence nance (NMR)
of nuclear magnetic resoimage contrast on intrinsic and pulse sequence timing parameters. Magn Re-
son
30
scans.
1.
on MR of hepato-
CT
was not visualized
References
parenchyma.
The visualization images
dynamic
images.
Although FNH, the
7.
logic signs of mild chronic hepatitis without cirrhosis were demonstrated in the surrounding
on
role of contrast-enhanced TurboFLASH and postcontrast Ti-weighted images.
the
went surgery. Pathologic examination of the resected right lobe demonstrated a well-differentiated hepatocellular carcinoma with a scar containing edematous fibrosis
the capsule
(Fig Aie).
observed
and
(29),
patient,
How-
the delayed
same This
obvious.
of the lesion
On
images,
scribed T2-weighted
Since the acceptance of our manuscript, we have obtained contrast-enhanced MR images in which the findings simulate those associated with FNH in a case of a well-differentiated hepatocellular carci-
noma.
13.
(27,28). This case illustrates the enhancement of the capsule, as previously de-
for further
#{149}
evaluation.
with a central scar. scar. (c) TurboFLASH
lesion with a low signal The scar is hyperintense; in comparison with
image
identified. chyma.
atitis
shows hypervascular lesion signal intensity of the central
of 10#{176}), obtained iO seconds after intravenous injection, shows hypervascular (7/4/400, flip angle of 10#{176}) obtained 2 minutes after intravenous injection. (e) Delayed postcontrast Ti-weighted image (420/20). The lesion is hyperintense High enhancement of the internal scar and the capsule is seen.
FLASH
e.
d.
after bolus injection, lesion with a high
Imaging
1984;
2:3-16.
July
1991