Content uploaded by Georg M Kacl
Author content
All content in this area was uploaded by Georg M Kacl on May 18, 2016
Content may be subject to copyright.
Introduction
Various imaging modalities including mammography,
sonography, and most recently, MR imaging have been
explored regarding their ability to detect and character-
ize breast lesions. Mammography, which is readily avail-
able, relatively inexpensive and suited to depict micro-
calcifications, remains the primary imaging modality
for breast evaluation at this time. Despite several tech-
nical advances affecting the actual data acquisition as
well as film processing, several limitations of mammog-
raphy remain well documented; these include difficul-
ties in the assessment of dense glandular tissue, as well
as of regions located close to the chest wall or within
the axilla. The presence of breast implants, and postop-
erative changes, can further impede the performance of
mammography [1±3].
Driven by improvements in surface coil and pulse se-
quence design, as well as the increased use of gadolini-
um-based contrast agents, breast MRI has rapidly
evolved in recent years. Many studies have documented
the improvements in the diagnostic performance of
breast MRI, focusing mainly on lesion detection and
characterization. Reflecting differences in technique,
results have varied widely with sensitivity ranging be-
tween 88 and 100 %, and specificity between 37 and
97% [4±15].
The purpose of this study was to assess the diagnostic
performance of conventional mammography and dy-
namic contrast-enhanced MRI regarding the detection
of breast malignancy in the same patient population
separately and in combination. All lesions had been his-
tologically verified.
Materials and methods
Patients
Fifty-six consecutive patients scheduled for histologic
verification of a palpable breast lesion and/or a suspect
Eur. Radiol. 8, 194±200 (1998) ÓSpringer-Verlag 1998
European
Radiology
Original article
Detection of breast cancer with conventional mammography and
contrast-enhanced MR imaging
G. M. Kacl1, P.-F.Liu1, J. F. Debatin1, E.Garzoli1, R. F. Caduff2, G.P.Krestin1
1Department of Diagnostic Radiology, Zurich University Hospital, RaÈmistrasse 100, CH-8091 Zurich, Switzerland
2Department of Pathology, Zurich University Hospital, RaÈmistrasse 100, CH-8091 Zurich, Switzerland
Received 5 May 1997; Revision received 14 July 1997; Accepted 7 August 1997
Abstract. The aim of this study was to compare the di-
agnostic performance of conventional mammogra-
phy and dynamic contrast-enhanced fast 3D gradi-
ent-echo (GRE) MRI regarding the detection and
characterization of breast lesions relative to histo-
pathologic analysis and to assess the results of a com-
bined evaluation of both methods. fifty consecutive
patients with 63 histopathologically verified breast le-
sions underwent dynamic contrast-enhanced GRE
MRI in addition to routine conventional mammogra-
phy. All lesions were classified by both methods on a
five-point scale as benign or malignant, and the re-
sults were correlated to histopathology. Conventional
mammography and dynamic MRI yielded a sensitivi-
ty and specificity of 82 and 64 %, and 92 and 76 %, re-
spectively. The difference between the results was
statistically not significant (p>0.05) with areas under
the receiver-operating-characteristics curves of 0.807
for mammography and 0.906 for MR imaging. Com-
bination of the results of both methods slightly in-
creased the sensitivity for detection of breast cancer
to 95% but decreased specificity to 52 %. In this se-
lected patient subset, including only patients referred
for excisional biopsy, contrast-enhanced dynamic
MRI proved more sensitive and specific than conven-
tional mammography regarding the detection of ma-
lignancy. While a combination of both methods yields
a slightly improved sensitivity, specificity is vastly re-
duced.
Key words: Breast MRI ± Mammography ± Breast
neoplasm ± Contrast-enhanced MRI
Correspondence to: J. F. Debatin
finding on conventional mammography were asked to
undergo an additional contrast-enhanced MR examina-
tion prior to excisional biopsy. Two patients were ex-
cluded due to claustrophobia and other MR contraindi-
cations. Of the remaining 54 patients, 50 women gave in-
formed consent in agreement with the guidelines set
forth by the institutional review board. Excluding 4 pa-
tients who refused consent, the study population con-
sisted of 50 women.
The age range of the examined patients was 28±
87 years (mean 57 years). Nineteen women (38 %)
were aged 49 years or younger; 31 (62 %) were aged
50 years or older. Thirteen patients presented with bilat-
eral disease, resulting in a total of 63 lesions in 50 pa-
tients. Lesions were verified based on mastectomy in 29
cases and on lumpectomy in 5 patients. Twenty-five le-
sions were confirmed based on excisional biopsy, and 4
lesions were verified based on fine needle biopsy.
Image acquisition
Mammography was performed on a CGR 600T unit
(GE Medical Systems, Milwaukee, Wis.). Standard lat-
eral and craniocaudal projections were obtained in all
patients. Mammograms were performed prior to MRI
in all 50 patients. The average interval between mam-
mography and MR examination was 3.6 weeks (range
0±12 weeks).
The MR images of the breast were acquired on two
different 1.5-T superconducting magnet systems (Signa,
GE Medical Systems, Milwaukee, Wis.; Gyroscan,
ACS-NT, Philips Medical Systems, Best, The Nether-
lands). On both systems a bilateral breast surface coil
was used for signal transmission and reception. Patients
were scanned in the prone position. Following the ac-
quisition of sagittal scout scans, axial fast 3D gradient-
echo (GRE) images were acquired prior to as well as 1,
2, 4, and 7 min following bolus injection of 0.2 mmol/kg
gadoterate-meglumine (Dotarem, Guerbet, Aulnay-
sous-Bois, France) over 10 s followed by 20 ml of saline
flush. Imaging parameters were as follows: TR =
12.5 ms, TE = 4.2 ms, FOV = 280±380 mm (adjusted to
breast size), flip angle = 30, bandwidth 32 kHz,
256 ´192 matrix, with one excitation. The two different
MR systems yielded 28±32 contiguous (no intersection
gap) 3.5- to 4.5-mm transverse sections.
Image subtractions were performed between the
post- and pre-contrast images on a pixel-by-pixel basis
by means of the software subtraction function available
on the post-processing workstation (Sparc 20, Sun Mi-
crosystems, Mountain View, Calif.).
Image analysis
Conventional mammograms were assessed retrospec-
tively by two radiologists (G.M. K. and P. F. L.) with
knowledge of all associated clinical information but
blinded to MRI, sonographic and histopathologic re-
sults in a consensus reading. All visible findings, includ-
ing mass lesions, asymmetric opacities, architectural dis-
tortions, and microcalcifications were considered as ab-
normal. Results were scored based on a five-point confi-
dence scale: definitely benign, probably benign, possibly
malignant, probably malignant, and definitely malig-
nant.
All MR studies were assessed separately by three ra-
diologists (E. G., G.M.K. and P.F.L.) with knowledge of
clinical findings but blinded to the conventional mam-
mographic, sonographic, and histopathologic results.
Consensus among two of the three readers was consid-
ered as positive correlation. In order to avoid recogni-
tion bias interpretation of the conventional and MR
mammograms was spaced by at least 3 weeks. Each
reader was asked to determine the number, border,
shape, and enhancement pattern of all visible enhancing
lesions. For quantitative analysis, signal intensity (SI)
changes were measured in several regions of interest
(ROI): the lesion itself, surrounding breast tissue, fat,
muscle, and background noise. Measurements were per-
formed on all image sets prior to and following contrast
administration. The ROI size was determined by the
size of the enhancing lesion. The increase in SI following
the in travenous injection of 0.2 mmol/kg Gd-DOTA was
calculated as the percentage increase relative to the SI
prior to contrast administration using the following for-
mula: Relative SI (%) = [100(SIx-SIo)/SIo], where SIx
indicated SI at each dynamic phase and SIo indicated
SI on the precontrast image.
Lesions were subsequently classified based on their
enhancement and morphologic characteristics. Early fo-
cal enhancement (signal increase exceeding the initial
signal level by at least 90 % on the images acquired in
the first minute after contrast injection), inhomoge-
neous or rim-like contrast uptake, ill-defined margins,
and an irregular shape were considered signs of malig-
nancy. Diffuse or patchy delayed or poor enhancement,
homogeneous contrast uptake, well-defined borders,
and a regular rounded or oval shape were suggestive of
a benign lesion. Results were scored based on the same
five-point confidence scale as used for evaluation of the
conventional mammograms (definitely benign, pro-
bably benign, possibly malignant, probably malignant,
and definitely malignant).
Statistical analysis
Statistical analysis was performed by means of Student's
t-test and chi-square test. Receiver-operating character-
istics (ROC) curves were generated separately for con-
ventional and MR mammography based on the five-
point confidence scales. The area under the ROC curves
(Fig.1) was calculated. Statistical significance was estab-
lished at a p-value of <0.05.
Results
Histopathologic analysis revealed 38 breast carcinomas
in 32 patients (6 patients had bilateral malignant le-
G.M. Kacl et al.: Conventional mammography and MR imaging in breast cancer 195
sions) and 25 benign lesions (4 patients had bilateral dis-
ease) in 21 patients. In 3 patients malignant and benign
disease was found in the same breast. Of the 38 carcino-
mas, 5 were ductal carcinomas in situ (DCIS), 1 was a
lobular carcinoma in situ (LCIS), and 32 were invasive
carcinomas (IDC). Multifocal tumors were present in 9
breasts. Of the 25 benign lesions, 9 were fibroadenomas,
1 was a papilloma, 1 was an abscess, and 14 were other
benign diseases including proliferative disease (lobular
and intraductal hyperplasia with or without atypia or
adenosis) and fibrocystic changes.
Considering the ratings ªpossibly, probably, and defi-
nitely malignantº as positive and the ratings ªdefinitely
and probably benignº as negative for malignancy, 16 of
25 benign and 31 of 38 malignant lesions were correctly
classified with conventional mammography alone. Us-
ing the combined enhancement and morphologic MRI
characteristics 19 of 25 benign and 35 of 38 malignant le-
sions were correctly classified. Conventional mammog-
raphy yielded a sensitivity, specificity and accuracy of
82, 64, and 75 %, respectively, whereas sensitivity, speci-
ficity, and accuracy of MR imaging were 92, 76, and
86%, respectively. The areas under the ROC curves
were 0.807 for conventional mammography and 0.906
for MR imaging (Fig.1). Statistically, however, results
of both modalities for detecting breast cancer did not
differ significantly (p>0.05). Combining the results of
both methods (considering a correct suspect finding at
either method as a positive finding for malignancy)
yielded a sensitivity and specificity of 95 and 52 %, re-
spectively (Table 1).
Malignant lesions
The appearance of all malignant lesions on conventional
and MR mammography is summarized in Tables 2 and
3, respectively. On conventional mammography true
positive findings were associated with spiculated or ill-
defined masses and/or microcalcifications in 28 cases
(multifocal lesions were considered only once). In three
cancers a marked architectural distorsion suggested the
presence of malignancy (rated as probably malignant
by all readers). The 7 false-negative lesions on mammo-
grams demonstrated architectural distortions (n=2;
one DCIS and one IDC), asymmetric densities (n=2;
one LCIS and one IDC), dense parenchyma (n=2; 2
cases of DCIS), and in 1 case only fatty tissue without
any calcifications or suspect opacities.
Histopathologically, 2 of the 3 false-negative read-
ings by MR imaging were found to be DCIS. One of
these lesions was invisible on MRI and consequently
missed by all readers. Conventional mammograms
showed neither microcalcifications nor any other suspi-
cious findings. The lesion was depicted only microscopi-
cally, but there was no alteration detectable on the mac-
G.M. Kacl et al.: Conventional mammography and MR imaging in breast cancer196
Fig.1. Receiver-operating-characteristics (ROC) curves for mam-
mography and MR imaging with measured areas under the ROC
curves of 0.807and 0.906
Table 1. Diagnostic performance of mammography and contrast-
enhanced MR imaging for detection of breast cancer
Mammo-
graphy
MR imaging MR +
mammography
Sensitivity (%) 82 92 95
Specificity (%) 64 76 52
Accuracy (%) 75 86 78
Table 3. Findings of 63 histopathologically proven lesions on MR-
mammography (multicentricity considered as single lesion)
Appearance Carci-
noma
Fibro-
adenoma
Other
benign
Contrast enhancement
>90% first min 28 6 2
Contrast enhancement
<90% first min 9 3 14
Poorly defined lesion margins 22 2 10
Well-defined lesion margins 15 7 6
Irregular lesion shape 30 3 9
Regular lesion shape 7 6 7
Heterogeneous contrast
enhancement
26 4 4
Rim contrast enhancement 5 0 3
Homogeneous contrast
enhancement
65 9
Not detected 1 0 0
Table 2. Findings of 63 histopathologically proven lesions on con-
ventional mammography
Appearance Carci-
noma
Fibro-
adenoma
Other
benign
Spiculated or ill-defined mass 14 0 1
Well-defined mass 0 5 0
Microcalcifications without mass 4 0 3
Microcalcifications and mass 10 0 2
Architectural distortion 5 2 4
Asymmetric density 2 1 3
Dense parenchyma 2 1 2
No lesion detectable 1 0 1
roscopic specimen in this case. The other DCIS present-
ed only minor contrast uptake within the first minute (SI
increase <50%) and a benign-appearing morphology
on MRI, whereas mammography depicted only dense
tissue without microcalcifications. The size of this tumor
was 6 mm at histopathologic analysis. The third false-
negative lesion demonstrated homogeneous early en-
hancement, a rounded shape with well-defined margins,
and was therefore considered to represent a fibroade-
noma at MRI. This lesion had been correctly classified
as malignant by mammography due to diffuse microcal-
cifications and a small opacity. The remaining 5 false-
negative readings at mammography were correctly diag-
nosed as malignancy on MRI. The extent of tumors lo-
cated close to the chest wall was better demonstrated
with MRI than with conventional mammography in 2
of 4 cases.
Multifocal tumors were correctly diagnosed on MRI
in 8 of 9 breasts. The size of a malignant focus of invasive
ductal carcinoma missed on MR images was 2 mm. The
MR imaging technique was 89 % accurate in identifying
tumor multifocality (Fig.2). Conventional mammogra-
phy missed multicentric lesions in 3 of 9 breasts with an
accuracy of merely 67 %. Malignant-appearing micro-
calcifications without a mass were seen in 2 and multiple
lesions in 4 breasts with histopathologically verified
multifocal malignancies.
Benign lesions
The conventional and MR mammographic findings of
fibroadenomas and other benign diseases are shown in
Tables 2 and 3, respectively. Sixteen of the 25 benign le-
sions were correctly classified with conventional mam-
mography as definitely or probably benign. Among the
9 false-positive results there were 2 fibroadenomas and
an abscess presenting with strong architectural distor-
tions and 6 cases of proliferative disease demonstrating
mass lesions with or without microcalcifications.
Of the 25 histologically verified benign breast lea-
sions, 6 were classified as suspicious for cancer on MR
images. Six fibroadenomas presenting with early en-
hancement within the first minute (108±159 %) and
morphologic characteristics suggestive of malignancy.
Three fibroadenomas characterized by slow and inho-
mogeneous enhancement, with ill-defined, irregular
borders, were also misclassified as possibly malignant
by all three readers. Figure 3 demonstrates the wide
range of dynamic contrast uptake in benign fibroadeno-
mas in comparison with malignant lesions (Fig.3). On
conventional mammography these 6 lesions presented
as well-defined masses (n= 5) or asymmetric density
(n= 1) and were correctly classified as probably benign.
In the other 3 cases both MRI and conventional mam-
mography findings were false positive: 2 were proven
histopathologically to be proliferative disease and
showed strong initial inhomogeneous contrast enhance-
ment (104 and 136 %, respectively, within the first min-
ute) with ill-defined contours and irregular shape on
MR images. At mammography, the lesions appeared as
an ill-defined mass in one case and contained microcal-
cifications in the other. Finally, a well-defined mass
with a strong initial rim-like contrast enhancement
(177 % within the first minute) on MR images was surgi-
cally found to be an abscess. Mammography had reveal-
G.M. Kacl et al.: Conventional mammography and MR imaging in breast cancer 197
a
b
Fig.2. Multicentric invasive ductal
carcinoma. aThe mammograms show
an area of diffuse malignant-appearing
microcalcifications and a spiculated
opacity (arrow).bMR images dem-
onstrate two abnormal focal enhancing
lesions matching with the location of
the microcalcifications (arrows)
ed a marked architectural distortion suggestive of ma-
lignancy.
All other 6 false-positive cases at mammography
were correctly classified as benign with dynamic MRI
based on enhancement and morphologic criteria
(Fig.4). Thus, there was correlation between conven-
tional mammography and MRI in 13 true-negative
cases.
Discussion
Conventional mammography is well established as a
screening technique for breast cancer. The reported sen-
sitivity for the detection of breast cancer varies between
69 and 90 % and is related to age [19]. The diagnostic
performance of the technique is particularly limited in
patients with dense parenchyma, postoperative altera-
tions, breast implants as well as in the presence of prolif-
erative disease. High false-negative rates in these pa-
tient subgroups have initially contributed to the rela-
tively poor performance of several early cancer detec-
tion programs [25]. More recent screening trials from
Sweden indicate beneficial effects of mammography, es-
pecially for patients aged 50±69 years [26]. When clini-
cal findings are present to direct the focus of the study,
physical examination and ultrasonography frequently
allow improved diagnostic management. However, bi-
opsy frequently remains the only means to prove or ex-
clude the presence of malignancy.
The sensitivity of contrast-enhanced MR imaging for
the detection of breast malignancies has been reported
to range between 86 and 100 % [4±15]. With the appro-
priate technique, malignancies are well visualized, even
within dense or mammographically distorted tissue. Re-
ported specificity of this technique has been more vari-
able, ranging from 37 to 97.4 % [4±15]. Benign lesions,
particularly fibroadenomas in premenopausal women
can mimic the enhancement characteristics of carcino-
ma [9, 16, 22, 23]. Other benign lesions can also present
as focal areas of abnormal enhancement on MR imag-
ing, and be similarly difficult to distinguish from malig-
nancies [8, 23, 24]. Considerably higher specificity can
be achieved with the incorporation of qualitative mor-
phologic criteria [27].
In this retrospective study, findings at mammography
and MRI were compared with the final histopathologic
results of 63 breast lesions. The MRI technique was
more sensitive and specific than conventional mam-
mography in the identification of breast carcinoma.
The overall sensitivity of mammography in this study
was 82 %. The high rate of false-negative mammograms
was due to a number of cancers presenting only as archi-
tectural distortions, asymmetric densities or dense pa-
renchyma. Contrast-enhanced MRI correctly identified
cancers in 5 of 7 false-negative mammographic readings.
Conventional mammography, on the other hand, detect-
ed only one of three false-negative MRI findings.
For an improved MRI detection rate, however, mor-
phologic parameters have to be incorporated in the as-
sessment of enhancing lesions: nine invasive carcinomas
in our study showed delayed enhancement, mimicking
benign lesions if diagnosis had been based on quantita-
tive analysis alone. Six of the carcinomas presenting de-
layed enhancement were correctly classified due to their
obvious malignant-appearing morphology.
One of the main problems of reduced sensitivity of
MRI remains the detection and correct classification of
DCIS [13, 16±18]. Sensitivity of MRI for detection of
DCIS is lower than the sensitivity for detecting invasive
carcinoma, and the enhancement patterns associated
with DCIS are often indistinguishable from benign le-
sions. In our study all 3 false-negative lesions at MRI
were DCIS: one DCIS was not seen at all on the MR im-
ages and the other two were misclassified as benign le-
sions based on the outlined diagnostic criteria.
The incidence of multifocality has been reported to
range between 14±47 % [20]. The findings of multifocal-
ity are of particular significance to the surgeon with re-
gard to breast conservation therapy. Patient selection
for breast cancer treatment is based on many factors,
perhaps the most important being determination of
multifocality and the extent of tumor to adjacent tissue.
Heywang-KoÈ brunner and Oellinger [21] reported that
MR imaging was able to show 80 % of all malignant
foci, whereas mammography showed only 20 %. In our
patient population multifocality was depicted histologi-
cally in 9 breasts and correctly diagnosed with MRI in
8. On conventional mammography multifocality was de-
tected in only 6 breasts. Our results indicate that MR
imaging demonstrates multifocal tumors, their location,
and extent to a better advantage than conventional
mammography.
With regard to specificity of MRI, this study confirms
the need to combine quantitative enhancement and
qualitative morphologic parameters into the assessment
of a particular lesion. Six of 9 histopathologically con-
firmed fibroadenomas did indeed demonstrate early en-
hancement indistinguishable from a malignant lesion.
Incorporating morphologic characteristics (regular
shape, well-defined margins and homogeneous contrast
G.M. Kacl et al.: Conventional mammography and MR imaging in breast cancer198
Fig.3. Enhancement profiles for carcinoma, fibroadenoma, and
other benign lesions. Ranges indicate standard deviation
uptake) led to a false classification of only 2 of these 6 fi-
broadenomas. Nevertheless, 3 false-positive diagnoses
on MRI represented proliferative (n= 2) and fibrocystic
(n= 1) disease. These changes, particularly when caus-
ing focal areas of contrast enhancement, cannot be dif-
ferentiated from malignancies. In this selected patient
group, MRI yielded a specificity for detection of malig-
nant lesions in the breast of 76 %. It is likely that the
specificity would be considerably lower in an unselected
patient population. This is a serious drawback of MR
mammography concerning its use as a screening meth-
od. Specificity would decrease even more if the correct
timing of the MR examination (first half of the menstru-
al cycle) is not adhered to.
While combining conventional with MR mammogra-
phy sensitivity would have minimally improved in corre-
lation to the performance of MR mammography alone.
But the overall diagnostic performance would have
been poorer reflecting a significant decrease in specific-
ity. Hence, whereas the availability of conventional
mammograms may be useful in pinpointing the abnor-
mality on the MR mammogram, final interpretation
should be based mainly on the MR mammogram for
best diagnostic performance.
The encouraging diagnostic yield of MR imaging in
the detection and characterization of breast lesions
must be weighed against its considerable costs. Its possi-
ble role as a screening tool for women at increased risk
for breast cancer must be defined in context with other,
less expensive modalities. In the meantime, the high
sensitivity of dynamic contrast-enhanced MR imaging
should be studied in those patients at high risk for mam-
mographically occult breast cancer, and in patients with
already proven cancer when multifocality needs to be
excluded prior to breast conservation therapy. Further-
more, young premenopausal women with a positive
family history as well as patients with breast implants
should be considered candidates for breast MR. To
avoid excessive false-positive detection in these patient
subgroups, it is imperative to base the overall assess-
ment of a breast lesion on all available enhancement
and morphologic criteria, instead of on quantitative pa-
rameters alone. The results of our study suggest that
MRI with its high sensitivity and acceptable specificity
could be used as a single diagnostic method in such a se-
lected patient population. However, the possible role of
MR mammography remains to be proven in series of
unselected patients. Given current cost constraints, it re-
mains unlikely, however, that MRI will ever replace
conventional mammography and ultrasound as the pri-
mary screening modality.
References
1. Page DL, Winfield AC (1986) The dense mammogram. AJR
147: 487±489
2. Hayes H, Vandergrift J, Diner WC (1988) Mammography and
breast implants. Plast Reconstr Surg 82: 1±6
3. Sickles EA, Herzog KA (1980) Intramammary scar tissue: a
mimic of the mammographic appearance of carcinoma.
AJR135: 349±352
4. Kaiser WA, Zeitler E (1989) MR imaging of the breast: fast im-
aging sequences with and without Gd-DTPA. Radiology 170:
681±686
5. Kaiser WA (1994) False-positive results in dynamic MR mam-
mography. Magn Reson Imaging Clin North Am 2: 539±555
G.M. Kacl et al.: Conventional mammography and MR imaging in breast cancer 199
a
b
Fig.4a, b. Proliferative disease.
aMammography demonstrates false
positively a suspect retroareolar hazy
opacity associated with three micro-
calcifications (arrow).bTrue negative
MRI without significant contrast up-
take of the lesion (arrow)
6. Bone
ÂB, Aspelin P, Bronge L, Isberg B, Perbeck L, Veress B
(1996) Sensitivity and specificity of MR mammography with
histopathological correlation in 250 breasts. Acta Radiol 37:
208±213
7. Heywang-KoÈ brunner SH (1994) Contrast-enhanced magnetic
resonance imaging of the breast. Invest Radiol 29: 94±104
8. Harms SE, Flamig DP, Hesley KL et al. (1993) MR imaging of
the breast with rotating delivery of excitation off resonance:
clinical experience with pathologic correlation. Radiology 187:
493±501
9. Kaiser WA, Reiser M (1992) False-positive cases in dynamic
MR mammography Radiology 185: 245
10. Oellinger H, Heins S, Sander B et al. (1993) Gd-DTPA en-
hanced MRI of breast: the most sensitive method for detecting
multicentric carcinomas in female breast? Eur Radiol 3: 223±
226
11. Fischer U, Heyden D, Vosshenrich R, Vieweg I, Grabbe E
(1993) Signalverhalten maligner und benigner LaÈsionen in der
dynamischen 2D-MRT der Mamma. RoFo 158: 287±292
12. Heywang-KoÈ brunner SH (1993) Brustkrebsdiagnostik mit
MR-Ueberblick nach 1250 Patientenuntersuchungen. Electro-
medica 2: 43±52
13. Gilles R, Guinebretie
Áre JM, Lucidarme O et al. (1994) Nonpal-
pable breast tumors: diagnosis with contrast-enhanced subtrac-
tion dynamic MR imaging. Radiology 191: 625±631
14. Lewis-Jones HG, Whitehouse GH, Leinster SJ (1991) The role
of magnetic resonance imaging in the assessment of local recur-
rent breast carcinoma. Clin Radiol 43: 197±204
15. Hulka CA, Smith BL, Sgroi DC et al. (1995) Benign and malig-
nant breast lesions: differentiation with echo-planar MR imag-
ing. Radiology 197: 33±38
16. Orel SG, Schnall MD, LiVolsi VA, Troupin RH (1994) Suspi-
cious breast lesions: MR imaging with radiologic±pathologic
correlation. Radiology 190: 485±493
17. Stomper PC, Herman S, Klippenstein DL et al. (1995) Suspect
beast lesions: findings at dynamic gadolinium-enhanced MR
imaging correlated with mammographic and pathologic fea-
tures. Radiology 197: 387±395
18. Rubens D, Totterman S, Chacko AK et al. (1991) Gadopente-
tate dimeglumine enhanced chemical-shift MR imaging of the
breast. AJR157: 267±270
19. Boetes C, Mus RDM, Holland R et al. (1995) Breast tumors:
comparative accuracy of MR imaging relative to mammogra-
phy and US for demonstrating extent. Radiology 197: 743±747
20. Oellinger H, Heins S, Sander B et al. (1993) Gd-DTPA en-
hanced MRI of breast: the most sensitive method for detecting
multicentric carcinomas in female breast? Eur Radiol 3: 223±226
21. Heywang-Koebrunner SH, Oellinger H (1994) Breast. Magne-
vist monograph. Blackwell, Oxford, pp 79±84
22. Heywang SH, Hahn D, Schmidt H et al. (1986) MR imaging of
the breast using gadolinium-DTPA. J Comput Assist Tomogr
10: 199±204
23. Flickinger FW, Allison JD, Sherry RM, Wright JC (1993) Dif-
ferentiation of benign from malignant breast masses by time-
intensity evaluation of contrast-enhanced MRI. Magn Reson
Imaging 11: 617±620
24. Heywang SH, Wolf A, Pruss E, Hilbertz T, Eiermann W, Per-
manetter W (1989) MR imaging of the breast with Gd-DTPA:
use and limitations. Radiology 171: 95±103
25. Fournier D von, Weber E, Hoeffken W, Bauer M, Kubli F,
Barth V (1980) Growth rate of 147 mammary carcinomas.
Cancer 45: 2198±2207
26. Larsson LG, Nystrom L, Wall S et al. (1996) The Swedish ran-
domized mammography screening trials: analysis of their effect
on the breast cancer related excess mortality. J Med Screen 3:
129±132
27. White Nunez L, Schnall MD, Orel SG et al. (1997) Breast MR
imaging: interpretation model. Radiology 202: 833±841
G.M. Kacl et al.: Conventional mammography and MR imaging in breast cancer200
Notice to authors
European Radiology encourages all authors to submit
manuscripts printed on both sides of the page. This will save
paper and reduce the cost of postage.