Downloaded from www.ajronline.org by 117.243.194.68 on 11/05/15 from IP address 117.243.194.68. Copyright ARRS. For personal use only; all rights reserved
267
Gadopentetate DimeglumineEnhanced Chemical-Shift MR Imaging of the Breast
Deborah Rubens1 Saara Totterrnan1 Anna K. Chacko1’2 Karnal Kothari3 Wende Logan-Youn4 Jerzy Szumowski Jack H. Simon1’6 Elizabeth
Zachanah7
Standard Ti-weighted MR images enhanced with gadopentetate dimeglumine show relatively minimal enhancement of breast lesions due to the high background signal from fat in the breast. Strongly enhancing lesions may become isointense relative to the fat signal and become invisible or indistinct after contrast administration. Fat-suppressed chemical-shift imaging (CSI) combined with administration of gadopentetate dimeglumine improves lesion detection and characterization in other areas of the body where a strong lipid signal is present. We evaluated this technique in the breast. Twenty patients with mammographic lesions were studied with standard unenhanced Ti- and T2-weighted images and enhanced Ti-weighted images, as well as with CSI before and after administration of gadopentetate dimeglumine. The series were ranked independently for border and matrix characteristics. The border was assessed for a smooth, irregular, or spiculated margin. The matrix or internal substance was evaluated for visibility and type of enhancement, homogeneous or mnhomogeneous. The enhanced CSI images were superior to all other images in the depiction of border and matrix characteristics. Of 20 patients, a corresponding mass was detected on MR in 14. In two of the 14 patients, the lesion was seen only on the enhanced CSI images. Chemicalshift
Received September 24, revision February 15, 1991.
1990;
accepted
after
This work was supported by Berlex Laboratories, Wayne, NJ. Pulse sequence development was supported in part by a Seed Grant Award from the Radiological Society of North America to J. Szumowski. 1 Department of Radiology, University of Roch-
ester School of Medicine and Dentistry, 601 Elmwood Ave., Rochester, NY 14642. Address reprint requests to D. Rubens. 2 Present address: Department of Radiology, Letterman
Army
Francisco,
CA 941 29-6700.
3
Ridgewood
Rd., Rochester, 4
Ave.,
Breast
Center,
Diagnosticlmaging,
Presidio
of San
1351 Mt. Hope
NY 14620.
address: Department of Health Sciences University,
Radiology, 3181 SW.
Sam Jackson Park Rd., Portland, OR 97201. 6 Present address: MRI, University of Colorado Health Sciences University, 4200 E. Ninth Ave., Denver, CO 80262. 7 492 Portlock Ad, Honolulu, HI 96825. 0361 -803X/91 /1 572-0267 C American Roentgen Ray Society
on enhanced
AJR
157:267-270,
Ti-weighted
images
obscured
border
detail
in several
cases.
August
1991
Recent articles show significant enhancement of carcinomas of the breast after IV injection of gadopentetate dimeglumine [i , 2]. In one series, 20% of lesions were detected only because of the tumor enhancement [i ]. On some enhanced breast images, the abnormal tissue became isointense relative to surrounding fat [i , 2]. This causes a problem because the borders of the lesion become indistinct and small or infiltrating lesions that blend in with the fat may go undetected. Breast lesions are unique in that they may be bordered by glandular tissue, fat, or a combination
3101 W. Ridge
NY 14626.
Clinic of Rochester,
Rochester,
Present The Oregon 5
Medical
artifacts
Enhanced CSI improves visualization of breast lesions as compared with conventional MR imaging with or without enhancement. The enhanced CSI technique produces differential enhancement between glandular tissue and lesions while suppressing the signal from fat This improves the visualization of border and matrix characteristics and depicts lesions that otherwise might be obscured.
of the
contrast
between
Studies
evaluating
two.
An optimal
imaging
technique
the lesion, the fat, and the surrounding lipid
suppression
with
chemical-shift
must
provide
maximal
normal glandular (CSI)
MR
imaging
tissue. have
shown improvement at fat/soft-tissue interfaces in the orbit [3], pelvis [4], and musculoskeletal system [5]. The hybrid method of CSI increases the dynamic range of the gray scale, which accentuates water contrast differences [6]. The combination of CSI with paramagnetic contrast material enhancement [7] should improve lesion
visibility
in the breast,
the lipid and by expanding Materials
and
both by eliminating
the effective
the strong
background
signal
from
gray scale of water.
Methods
MR images of the breast were obtained in 32 women on mammograms and/or in whom palpable abnormalities
in whom lesions had been detected had been found. The MR images
268
RUBENS
of the first 20 patients
with technically
adequate
scans
were
evalu-
ated. Of these 20 patients, five had carcinoma, four had fibroadenomas, four had fibrocystic disease, four had cysts, two had sclerosing adenosls,
and one was
mograms
were
unchanged
(two with carcinoma
Downloaded from www.ajronline.org by 117.243.194.68 on 11/05/15 from IP address 117.243.194.68. Copyright ARRS. For personal use only; all rights reserved
presumed after
Medical
1 8 months.
Three
tumor
after
of the
32 patients
and one with a cyst) were excluded
inadequate CSI MR images. MR imaging was performed Electric
to be without
Systems,
with
a 1 .5-T magnet
Milwaukee,
WI).
The
owing to
(Signa,
patients
mam-
General
were
posi-
ET AL.
AJR:157,
whom no corresponding lesion imaging, five had microcalcifications
no associated evidence
carcinoma.
of benign
The remaining
lesions,
1991
was documented on MR on the mammogram with
mass. One of these patients
4-cm intraductal
August
including
had a i
-
by 3- by
four had histologic
apocrine
metaplasia,
fibrosis, and fibrocystic disease. In the remaining patient in whom no lesion was identified on MR imaging, an asymmetric soft-tissue density was seen on a single mammogram view.
tioned prone, and a single breast was studied with a single-turn solenoid loop-9ap resonator [8]. After an axial image was obtained to localize the lesion, a Ti -weighted spin-echo (SE) series, 600/20
This enhanced
(TR/TE); a proton density-weighted series, 2000/80; and a Ti -weighted
were seen on MR imaging that were not seen on the mammogram. Three of these six lesions were cysts. The other
series, 2000/20; a T2-weighted hybrid (CSI) series tuned for lipid
suppression, 600/20, were performed [6]. All images were obtained in the sagittal plane in the same location of the breast and included the majority of the breast tissue, directed medially or laterally to include the region of the lesion detected on mammograms. The other imaging parameters included 5-mm slices with a 1 .5-mm gap, two excitations, a 256 x 1 28 matrix, and a 1 4- to 1 8-cm field of view. After these unenhanced images were made, gadopentetate dimeglumine (Magnevist, Berlex Labs., Wayne, NJ) was injected IV in 12 mm at a dose of 0.1 mmol/kg. This was followed by immediate (0-
1 5 mm) and delayed
(1 5-25
mm) Ti -weighted
SE and CSI images
with the slice and imaging parameters described above. According to protocol, the Ti-weighted enhanced images were obtained before the enhanced CSI images for both the immediate and delayed images. To enable true evaluation of enhancement after contrast injection, we kept the transmit and receiver gains constant for each standard Ti -weighted SE series. Different constant values were maintained for CSI series. Receiver gain in each unenhanced series was diminished to prevent saturation of the amplifiers from the increased signal intensity
after
contrast
on the CSI series, slices
administration.
16 slices
on the T2-weighted
Twelve
on the Ti-weighted
slices
were
SE series,
available
and 20
SE series.
A total of eight image series were evaluated: Ti -weighted, proton density-weighted (2000/20), T2-weighted (2000/80), and CSl unenhanced series; and immediate Ti , delayed Ti , immediate CSI, and delayed CSI enhanced series. The results of these series were ranked in ascending order; the best score Rankings were given by consensus
were compared
was 8 and 1 was the worst. between two readers. Images
for the demonstration
of the border characteristics
of a lesion (smooth, irregular, or spiculated). The matrix or internal substance of the lesion was evaluated for visibility of heterogeneous enhancement, homogeneous enhancement, or lack of enhancement. Matrix ranking placed heterogeneity above homogeneity if the same lesion
differed
in the
various
series.
If the
pattern
of lesion
in a fashion
similar
to the remainder
of the
breast tissue on MR imaging and was thought to represent asymmetric glandular tissue. In six of the 20 patients, lesions
three were solid tumors with smooth borders and uniformly increased signal intensity on T2-weighted
as well as on all enhanced measured
less
than
fibroadenomas. The ranking
CSI images.
5 mm and were
of the border
showed images
All three
presumed
lesions
to be incidental
characteristics
of the lesions
on
MR images was significant in that the immediately enhanced CSI images had a score of i i 8 out of the total maximum of i 60 points. The sequence with the next highest score was the delayed enhanced CSI with a score of 9i . The other series scored as follows: T2-weighted (2000/80) = 66, Ti -weighted = 62.5, immediately enhanced Ti -weighted = 62, delayed enhanced Ti -weighted = 60.5, unenhanced Ti -weighted CSI = 55, and proton density-weighted (2000-20) = 49. If the four patients with cysts as lesions were excluded, the maximum possible score was 1 28 points. The scores for immediately and delayed enhanced CSI images of solid lesions
dropped to 1 06 and 81 , respectively, but the T2-weighted score dropped to 34. The matrix ranking, including all lesions, resulted in a score of i 26 out
of 1 60 possible
points,
for the immediately
en-
hanced CSI series. The delayed enhanced CSI series scored 98.5, followed by T2-weighted at 77.5, delayed enhanced Ti weighted at 66, immediately enhanced Ti -weighted at 6i, unenhanced Ti -weighted CSI at 60.5, proton densityweighted
at 42.5,
and Ti -weighted
at 33. Again,
were excluded, the immediately and delayed scores dropped to i 05 and 78, respectively, weighted series dropped to 49.
when
cysts
enhanced CSI and the T2-
enhance-
mont was the same, the series with the greatest enhancement scored highest. A tie between two series resulted in an averaging of the two
rankings example,
rather than arbitrarily ranking one above the other. For if two series tied for the best border discrimination, each
would receive a score of 7.5. If a lesion was not visualized on an MR imaging series or a series was not performed, a score of 0 was assigned. After ranking, the results were tabulated by addition of the
scores of each separate
series.
Discussion Our better
results
show
delineation
that
of the
enhanced border
CSI series over the unenhanced
Although all lesions were visible on mammography, the abnormality was detected on MR images in only 14 of the 20 patients. Two of the i4 lesions were seen only with the enhanced CSI series. Both of these lesions were documented by histology
as sclerosing
adenosis.
Of the six patients
in
imaging
provided
of breast
le-
sions than did standard SE sequences either with or without contrast material. In particular, the advantage of the enhanced the improved
Results
CSI
characteristics
Ti -weighted
SE series was
contrast between the lesion and the normal (Fig i .). The expanded gray scale aids in delin-
breast tissue eation of small lesions surrounded either totally or in part by glandular tissue. For lesions adjacent to fat, the elimination of chemical-shift larity
(Fig
artifact
improved
2.). By suppressing
the visibility
of border
irregu-
the signal from fat, enhanced
CSI imaging also prevents an intensely enhancing lesion from becoming isointense relative to the surrounding lipid.
ENHANCED
AJR:157, August 1991
CHEMICAL-SHIFT
MR OF BREAST
269
Fig. i.-47-year-old woman with 5-mm infiltrating tubular carcinoma of the breast A, Craniocaudal mammogram of right breast suggests an 8-mm spiculated lesion (arrows) 3 cm posterior to nipple. B, On lateral mammogram, lesion is difticult to
Downloaded from www.ajronline.org by 117.243.194.68 on 11/05/15 from IP address 117.243.194.68. Copyright ARRS. For personal use only; all rights reserved
distinguish
from dense breast tissue.
C, Ti-weighted SE MR images obtained before (top) and after (bottom) contrast administration show moderate enhancement of lesion (arrows) slmilartothatof adjacentglandulartlssue(g). Note low signal intensity of posterior margin and smooth
border characteristics
anteriorly after contrast ad-
ministration. 0, Ti-weighted unenhanced chemical-shift imaging (CSI) (top) and enhanced CSI (bottom) series (600/20). Note marked enhancement of lesion after contrast administration as compared with adjacent glandular tissue. Border spicules (arrows) are visible on enhanced images because of elimination of chemical-shIft artifact and Improved
contrast
with respect to fat
Matrix characteristics also were shown better on the enhanced CSI images because of the expanded gray scale of water that results from lipid suppression. In particular, peripheral enhancement with a lower-signal center was detected in carcinomas as compared with uniform enhancement in sclerosing adenosis and fibroadenomas. This has not been emphasized previously in the literature. A prospective evaluation with a larger number of patients will be necessary to evaluate the significance of this finding. The next most valuable series to the enhanced CSI series for lesion diagnosis was the T2-weighted SE series. This was the most helpful sequence in evaluating cysts. If cysts are excluded, the value of the T2-weighted series becomes comparable to that of the Ti -weighted SE series. The enhanced Ti -weighted series, either immediate or delayed, was slightly worse at border delineation than the unenhanced Ti -weighted SE series. This was due to pronounced postenhancement chemical-shift
artifacts
at the lesion
margin,
which
interfered
border evaluation. The matrix was evaluated for degree and distribution of enhancement, and therefore was usually better seen on enhanced images. Considering the small number of patients in the series, and the relatively close ranking of the standard Ti - and T2-weighted SE series, the only large difference in scores lies between the two enhanced CSI series and the remaining six series that were evaluated. Previous studies [2, 7] have discussed the importance of signal intensity changes for differentiation of benign from malignant disease of the breast. We found it impossible to make identical slices with the various MR techniques because of involuntary patient motion. Because volume averaging would cause error in intensity measurements, we have not evaluated these series for absolute enhancement values of the lesions. The 5-mm slice thickness was sufficient, particularly in lesions 1 cm or larger, to allow preliminary matrix and border discrimination without significant error from volume averaging. with
RUBENS
270
ET AL.
AJR:157,
August
1991
Considering
the 5-mm slice thickness used, the lesion should missed because of its size. Perhaps the lack of invasiveness contributed to the problem in this carcinoma. In conclusion, enhanced CSI imaging improves MR visualization of breast lesions as compared with conventional SE MR imaging with or without gadopentetate dimeglumine. This is accomplished by expansion of the gray scale, which yields
Downloaded from www.ajronline.org by 117.243.194.68 on 11/05/15 from IP address 117.243.194.68. Copyright ARRS. For personal use only; all rights reserved
not have
been
differential
visible
enhancement
between
the glandular
tissue
of the breast and enhancing lesions. In addition, matrix characteristics of lesions can now be identified. The suppression of the signal from background fat eliminates chemical-shift artifacts and enables visualization oflesion detail in the border, which otherwise may be obscured. Our preliminary results suggest that enhanced CSI imaging should replace conventional Ti -weighted enhanced imaging in the future. Further studies will be required to determine the usefulness of MR imaging
in the characterization
as an adjunct
to mammography
of lesions
and to define
and physical
its role
examination.
REFERENCES WA, Zeitler E. MR imaging of the breast: fast imaging sequences Gd-DTPA. Radiology i989;170:681 -686 2. Heywang SH, Wolf A, Pruss E, Hilbertz T, Eiermann W, Permanetter W. 1 . Kaiser
A
B Fig. 2.-62-year-old
woman
with infiltrating
with and without
intraductal
breast
carci-
noma. A, Ti-weighted SE MR images of a 1.5- by 1.6-cm infiltrating carcinoma before (top) and after (bottom) contrast administration. Marked chemicalshift artifact (arrow) obscures border detail. Matrix appears fairly uniform and little enhancement is appreciated. B, Ti-weighted unenhanced chemical-shift imaging (CSI, top) and enhanced C5l (bottom) series (600/20) Chemical-shift artifact has been eliminated and a small spicule can be seen along anterior margin (arrow). Note heterogeneous matrix, which was not evident on standard 5E series. Signal intensity of lesion has been held constant, while background intensity of enhanced CSI series has decreased, indicating lesion enhancement
Six of 20 lesions seen on mammograms were not detected by MR imaging. The most significant lesion missed on the MR scans was the i - by 3- by 4-cm intraductal carcinoma.
MR imaging of the breast with Gd-DTPA: use and limitations. i989;171 :95-1 03 3. Simon JH, Szumowski J, Totterman 5, et at. Fat suppression resonance imaging of the orbit. AJNR i988;9:961-968
Radiology magnetic
4. Totterman 5, Spataro R, Rubens D, et al. Fat suppression technique in the detection and staging of cervical, bladder and prostate carcinoma (abstr). Radiology i987;165(P):235 5. Totterman 5, Weiss 5, Szumowski J, et al. MR fat suppression technique in the evaluation of normal structures of the knee. J Comput Assist Tomogr i989;13(3):473-479
6. Szumowski
J, Eisen JK, Vinitski 5, Haake PW, Plewes DB. Hybrid methods
of chemical shift imaging. Magn Reson Med i989;9:379-388 7. Simon JH, Szumowski J. chemical shift imaging with paramagnetic contrast material enhancement for improved lesion depiction. Radiology i989;17i :539-543
8. Homak JP, Szumowski imaging
with loop-gap
J, Rubens resonators.
D, Janus J, Bryant Radiology
i986;1
AG. Breast MA
61:832-834
267
Gadopentetate DimeglumineEnhanced Chemical-Shift MR Imaging of the Breast
Deborah Rubens1 Saara Totterrnan1 Anna K. Chacko1’2 Karnal Kothari3 Wende Logan-Youn4 Jerzy Szumowski Jack H. Simon1’6 Elizabeth
Zachanah7
Standard Ti-weighted MR images enhanced with gadopentetate dimeglumine show relatively minimal enhancement of breast lesions due to the high background signal from fat in the breast. Strongly enhancing lesions may become isointense relative to the fat signal and become invisible or indistinct after contrast administration. Fat-suppressed chemical-shift imaging (CSI) combined with administration of gadopentetate dimeglumine improves lesion detection and characterization in other areas of the body where a strong lipid signal is present. We evaluated this technique in the breast. Twenty patients with mammographic lesions were studied with standard unenhanced Ti- and T2-weighted images and enhanced Ti-weighted images, as well as with CSI before and after administration of gadopentetate dimeglumine. The series were ranked independently for border and matrix characteristics. The border was assessed for a smooth, irregular, or spiculated margin. The matrix or internal substance was evaluated for visibility and type of enhancement, homogeneous or mnhomogeneous. The enhanced CSI images were superior to all other images in the depiction of border and matrix characteristics. Of 20 patients, a corresponding mass was detected on MR in 14. In two of the 14 patients, the lesion was seen only on the enhanced CSI images. Chemicalshift
Received September 24, revision February 15, 1991.
1990;
accepted
after
This work was supported by Berlex Laboratories, Wayne, NJ. Pulse sequence development was supported in part by a Seed Grant Award from the Radiological Society of North America to J. Szumowski. 1 Department of Radiology, University of Roch-
ester School of Medicine and Dentistry, 601 Elmwood Ave., Rochester, NY 14642. Address reprint requests to D. Rubens. 2 Present address: Department of Radiology, Letterman
Army
Francisco,
CA 941 29-6700.
3
Ridgewood
Rd., Rochester, 4
Ave.,
Breast
Center,
Diagnosticlmaging,
Presidio
of San
1351 Mt. Hope
NY 14620.
address: Department of Health Sciences University,
Radiology, 3181 SW.
Sam Jackson Park Rd., Portland, OR 97201. 6 Present address: MRI, University of Colorado Health Sciences University, 4200 E. Ninth Ave., Denver, CO 80262. 7 492 Portlock Ad, Honolulu, HI 96825. 0361 -803X/91 /1 572-0267 C American Roentgen Ray Society
on enhanced
AJR
157:267-270,
Ti-weighted
images
obscured
border
detail
in several
cases.
August
1991
Recent articles show significant enhancement of carcinomas of the breast after IV injection of gadopentetate dimeglumine [i , 2]. In one series, 20% of lesions were detected only because of the tumor enhancement [i ]. On some enhanced breast images, the abnormal tissue became isointense relative to surrounding fat [i , 2]. This causes a problem because the borders of the lesion become indistinct and small or infiltrating lesions that blend in with the fat may go undetected. Breast lesions are unique in that they may be bordered by glandular tissue, fat, or a combination
3101 W. Ridge
NY 14626.
Clinic of Rochester,
Rochester,
Present The Oregon 5
Medical
artifacts
Enhanced CSI improves visualization of breast lesions as compared with conventional MR imaging with or without enhancement. The enhanced CSI technique produces differential enhancement between glandular tissue and lesions while suppressing the signal from fat This improves the visualization of border and matrix characteristics and depicts lesions that otherwise might be obscured.
of the
contrast
between
Studies
evaluating
two.
An optimal
imaging
technique
the lesion, the fat, and the surrounding lipid
suppression
with
chemical-shift
must
provide
maximal
normal glandular (CSI)
MR
imaging
tissue. have
shown improvement at fat/soft-tissue interfaces in the orbit [3], pelvis [4], and musculoskeletal system [5]. The hybrid method of CSI increases the dynamic range of the gray scale, which accentuates water contrast differences [6]. The combination of CSI with paramagnetic contrast material enhancement [7] should improve lesion
visibility
in the breast,
the lipid and by expanding Materials
and
both by eliminating
the effective
the strong
background
signal
from
gray scale of water.
Methods
MR images of the breast were obtained in 32 women on mammograms and/or in whom palpable abnormalities
in whom lesions had been detected had been found. The MR images
268
RUBENS
of the first 20 patients
with technically
adequate
scans
were
evalu-
ated. Of these 20 patients, five had carcinoma, four had fibroadenomas, four had fibrocystic disease, four had cysts, two had sclerosing adenosls,
and one was
mograms
were
unchanged
(two with carcinoma
Downloaded from www.ajronline.org by 117.243.194.68 on 11/05/15 from IP address 117.243.194.68. Copyright ARRS. For personal use only; all rights reserved
presumed after
Medical
1 8 months.
Three
tumor
after
of the
32 patients
and one with a cyst) were excluded
inadequate CSI MR images. MR imaging was performed Electric
to be without
Systems,
with
a 1 .5-T magnet
Milwaukee,
WI).
The
owing to
(Signa,
patients
mam-
General
were
posi-
ET AL.
AJR:157,
whom no corresponding lesion imaging, five had microcalcifications
no associated evidence
carcinoma.
of benign
The remaining
lesions,
1991
was documented on MR on the mammogram with
mass. One of these patients
4-cm intraductal
August
including
had a i
-
by 3- by
four had histologic
apocrine
metaplasia,
fibrosis, and fibrocystic disease. In the remaining patient in whom no lesion was identified on MR imaging, an asymmetric soft-tissue density was seen on a single mammogram view.
tioned prone, and a single breast was studied with a single-turn solenoid loop-9ap resonator [8]. After an axial image was obtained to localize the lesion, a Ti -weighted spin-echo (SE) series, 600/20
This enhanced
(TR/TE); a proton density-weighted series, 2000/80; and a Ti -weighted
were seen on MR imaging that were not seen on the mammogram. Three of these six lesions were cysts. The other
series, 2000/20; a T2-weighted hybrid (CSI) series tuned for lipid
suppression, 600/20, were performed [6]. All images were obtained in the sagittal plane in the same location of the breast and included the majority of the breast tissue, directed medially or laterally to include the region of the lesion detected on mammograms. The other imaging parameters included 5-mm slices with a 1 .5-mm gap, two excitations, a 256 x 1 28 matrix, and a 1 4- to 1 8-cm field of view. After these unenhanced images were made, gadopentetate dimeglumine (Magnevist, Berlex Labs., Wayne, NJ) was injected IV in 12 mm at a dose of 0.1 mmol/kg. This was followed by immediate (0-
1 5 mm) and delayed
(1 5-25
mm) Ti -weighted
SE and CSI images
with the slice and imaging parameters described above. According to protocol, the Ti-weighted enhanced images were obtained before the enhanced CSI images for both the immediate and delayed images. To enable true evaluation of enhancement after contrast injection, we kept the transmit and receiver gains constant for each standard Ti -weighted SE series. Different constant values were maintained for CSI series. Receiver gain in each unenhanced series was diminished to prevent saturation of the amplifiers from the increased signal intensity
after
contrast
on the CSI series, slices
administration.
16 slices
on the T2-weighted
Twelve
on the Ti-weighted
slices
were
SE series,
available
and 20
SE series.
A total of eight image series were evaluated: Ti -weighted, proton density-weighted (2000/20), T2-weighted (2000/80), and CSl unenhanced series; and immediate Ti , delayed Ti , immediate CSI, and delayed CSI enhanced series. The results of these series were ranked in ascending order; the best score Rankings were given by consensus
were compared
was 8 and 1 was the worst. between two readers. Images
for the demonstration
of the border characteristics
of a lesion (smooth, irregular, or spiculated). The matrix or internal substance of the lesion was evaluated for visibility of heterogeneous enhancement, homogeneous enhancement, or lack of enhancement. Matrix ranking placed heterogeneity above homogeneity if the same lesion
differed
in the
various
series.
If the
pattern
of lesion
in a fashion
similar
to the remainder
of the
breast tissue on MR imaging and was thought to represent asymmetric glandular tissue. In six of the 20 patients, lesions
three were solid tumors with smooth borders and uniformly increased signal intensity on T2-weighted
as well as on all enhanced measured
less
than
fibroadenomas. The ranking
CSI images.
5 mm and were
of the border
showed images
All three
presumed
lesions
to be incidental
characteristics
of the lesions
on
MR images was significant in that the immediately enhanced CSI images had a score of i i 8 out of the total maximum of i 60 points. The sequence with the next highest score was the delayed enhanced CSI with a score of 9i . The other series scored as follows: T2-weighted (2000/80) = 66, Ti -weighted = 62.5, immediately enhanced Ti -weighted = 62, delayed enhanced Ti -weighted = 60.5, unenhanced Ti -weighted CSI = 55, and proton density-weighted (2000-20) = 49. If the four patients with cysts as lesions were excluded, the maximum possible score was 1 28 points. The scores for immediately and delayed enhanced CSI images of solid lesions
dropped to 1 06 and 81 , respectively, but the T2-weighted score dropped to 34. The matrix ranking, including all lesions, resulted in a score of i 26 out
of 1 60 possible
points,
for the immediately
en-
hanced CSI series. The delayed enhanced CSI series scored 98.5, followed by T2-weighted at 77.5, delayed enhanced Ti weighted at 66, immediately enhanced Ti -weighted at 6i, unenhanced Ti -weighted CSI at 60.5, proton densityweighted
at 42.5,
and Ti -weighted
at 33. Again,
were excluded, the immediately and delayed scores dropped to i 05 and 78, respectively, weighted series dropped to 49.
when
cysts
enhanced CSI and the T2-
enhance-
mont was the same, the series with the greatest enhancement scored highest. A tie between two series resulted in an averaging of the two
rankings example,
rather than arbitrarily ranking one above the other. For if two series tied for the best border discrimination, each
would receive a score of 7.5. If a lesion was not visualized on an MR imaging series or a series was not performed, a score of 0 was assigned. After ranking, the results were tabulated by addition of the
scores of each separate
series.
Discussion Our better
results
show
delineation
that
of the
enhanced border
CSI series over the unenhanced
Although all lesions were visible on mammography, the abnormality was detected on MR images in only 14 of the 20 patients. Two of the i4 lesions were seen only with the enhanced CSI series. Both of these lesions were documented by histology
as sclerosing
adenosis.
Of the six patients
in
imaging
provided
of breast
le-
sions than did standard SE sequences either with or without contrast material. In particular, the advantage of the enhanced the improved
Results
CSI
characteristics
Ti -weighted
SE series was
contrast between the lesion and the normal (Fig i .). The expanded gray scale aids in delin-
breast tissue eation of small lesions surrounded either totally or in part by glandular tissue. For lesions adjacent to fat, the elimination of chemical-shift larity
(Fig
artifact
improved
2.). By suppressing
the visibility
of border
irregu-
the signal from fat, enhanced
CSI imaging also prevents an intensely enhancing lesion from becoming isointense relative to the surrounding lipid.
ENHANCED
AJR:157, August 1991
CHEMICAL-SHIFT
MR OF BREAST
269
Fig. i.-47-year-old woman with 5-mm infiltrating tubular carcinoma of the breast A, Craniocaudal mammogram of right breast suggests an 8-mm spiculated lesion (arrows) 3 cm posterior to nipple. B, On lateral mammogram, lesion is difticult to
Downloaded from www.ajronline.org by 117.243.194.68 on 11/05/15 from IP address 117.243.194.68. Copyright ARRS. For personal use only; all rights reserved
distinguish
from dense breast tissue.
C, Ti-weighted SE MR images obtained before (top) and after (bottom) contrast administration show moderate enhancement of lesion (arrows) slmilartothatof adjacentglandulartlssue(g). Note low signal intensity of posterior margin and smooth
border characteristics
anteriorly after contrast ad-
ministration. 0, Ti-weighted unenhanced chemical-shift imaging (CSI) (top) and enhanced CSI (bottom) series (600/20). Note marked enhancement of lesion after contrast administration as compared with adjacent glandular tissue. Border spicules (arrows) are visible on enhanced images because of elimination of chemical-shIft artifact and Improved
contrast
with respect to fat
Matrix characteristics also were shown better on the enhanced CSI images because of the expanded gray scale of water that results from lipid suppression. In particular, peripheral enhancement with a lower-signal center was detected in carcinomas as compared with uniform enhancement in sclerosing adenosis and fibroadenomas. This has not been emphasized previously in the literature. A prospective evaluation with a larger number of patients will be necessary to evaluate the significance of this finding. The next most valuable series to the enhanced CSI series for lesion diagnosis was the T2-weighted SE series. This was the most helpful sequence in evaluating cysts. If cysts are excluded, the value of the T2-weighted series becomes comparable to that of the Ti -weighted SE series. The enhanced Ti -weighted series, either immediate or delayed, was slightly worse at border delineation than the unenhanced Ti -weighted SE series. This was due to pronounced postenhancement chemical-shift
artifacts
at the lesion
margin,
which
interfered
border evaluation. The matrix was evaluated for degree and distribution of enhancement, and therefore was usually better seen on enhanced images. Considering the small number of patients in the series, and the relatively close ranking of the standard Ti - and T2-weighted SE series, the only large difference in scores lies between the two enhanced CSI series and the remaining six series that were evaluated. Previous studies [2, 7] have discussed the importance of signal intensity changes for differentiation of benign from malignant disease of the breast. We found it impossible to make identical slices with the various MR techniques because of involuntary patient motion. Because volume averaging would cause error in intensity measurements, we have not evaluated these series for absolute enhancement values of the lesions. The 5-mm slice thickness was sufficient, particularly in lesions 1 cm or larger, to allow preliminary matrix and border discrimination without significant error from volume averaging. with
RUBENS
270
ET AL.
AJR:157,
August
1991
Considering
the 5-mm slice thickness used, the lesion should missed because of its size. Perhaps the lack of invasiveness contributed to the problem in this carcinoma. In conclusion, enhanced CSI imaging improves MR visualization of breast lesions as compared with conventional SE MR imaging with or without gadopentetate dimeglumine. This is accomplished by expansion of the gray scale, which yields
Downloaded from www.ajronline.org by 117.243.194.68 on 11/05/15 from IP address 117.243.194.68. Copyright ARRS. For personal use only; all rights reserved
not have
been
differential
visible
enhancement
between
the glandular
tissue
of the breast and enhancing lesions. In addition, matrix characteristics of lesions can now be identified. The suppression of the signal from background fat eliminates chemical-shift artifacts and enables visualization oflesion detail in the border, which otherwise may be obscured. Our preliminary results suggest that enhanced CSI imaging should replace conventional Ti -weighted enhanced imaging in the future. Further studies will be required to determine the usefulness of MR imaging
in the characterization
as an adjunct
to mammography
of lesions
and to define
and physical
its role
examination.
REFERENCES WA, Zeitler E. MR imaging of the breast: fast imaging sequences Gd-DTPA. Radiology i989;170:681 -686 2. Heywang SH, Wolf A, Pruss E, Hilbertz T, Eiermann W, Permanetter W. 1 . Kaiser
A
B Fig. 2.-62-year-old
woman
with infiltrating
with and without
intraductal
breast
carci-
noma. A, Ti-weighted SE MR images of a 1.5- by 1.6-cm infiltrating carcinoma before (top) and after (bottom) contrast administration. Marked chemicalshift artifact (arrow) obscures border detail. Matrix appears fairly uniform and little enhancement is appreciated. B, Ti-weighted unenhanced chemical-shift imaging (CSI, top) and enhanced C5l (bottom) series (600/20) Chemical-shift artifact has been eliminated and a small spicule can be seen along anterior margin (arrow). Note heterogeneous matrix, which was not evident on standard 5E series. Signal intensity of lesion has been held constant, while background intensity of enhanced CSI series has decreased, indicating lesion enhancement
Six of 20 lesions seen on mammograms were not detected by MR imaging. The most significant lesion missed on the MR scans was the i - by 3- by 4-cm intraductal carcinoma.
MR imaging of the breast with Gd-DTPA: use and limitations. i989;171 :95-1 03 3. Simon JH, Szumowski J, Totterman 5, et at. Fat suppression resonance imaging of the orbit. AJNR i988;9:961-968
Radiology magnetic
4. Totterman 5, Spataro R, Rubens D, et al. Fat suppression technique in the detection and staging of cervical, bladder and prostate carcinoma (abstr). Radiology i987;165(P):235 5. Totterman 5, Weiss 5, Szumowski J, et al. MR fat suppression technique in the evaluation of normal structures of the knee. J Comput Assist Tomogr i989;13(3):473-479
6. Szumowski
J, Eisen JK, Vinitski 5, Haake PW, Plewes DB. Hybrid methods
of chemical shift imaging. Magn Reson Med i989;9:379-388 7. Simon JH, Szumowski J. chemical shift imaging with paramagnetic contrast material enhancement for improved lesion depiction. Radiology i989;17i :539-543
8. Homak JP, Szumowski imaging
with loop-gap
J, Rubens resonators.
D, Janus J, Bryant Radiology
i986;1
AG. Breast MA
61:832-834
