An Italian multicenter study showed that Real-time Elastography of the breast is accurate and reproducible and may eas-

ily and quickly integrate conventional ultrasound and other breast imaging.
Elastography was performed in 784 women (mean age, 52.5 years) who had 874 lesions with a definitive diagnosis (614
benign, 260 malignant) proved with core biopsy or fine needle aspiration. US images were classified according the BI-RADS
criteria for US1); Elastography images were assigned an elastographic score (1 to 5) according a new proposal of score that
modifies the Tsukuba score. The new three-layered score, observed in cystic lesions, is described.
Elastography showed a very high specificity in benign lesions, including BI-RADS 3 lesions. With the best cut-off point
between elasticity scores 3 and 4 the negative predictive value was 98% for the whole series, 96.3% for all the BI-RADS 3
lesions, and 100% for those with a size 5 mm.
Elastography scores were insensitive to the thickness and the echogenicity of the breast, and to the depth and the size of
the lesion.
Elastography scores interpretation resulted well reproducible. K indexes of intra-observer (0.93) and inter-observer (0.90)
agreement resulted to be very good.
Basing on their experience the Italian experts have defined new guidelines for standard acquisition and interpretation of
the breast Elastography scores.

Key Words: Breast Ultrasound, Breast Elastography, Elastography, Breast Lesions, BI-RADS

At 8 Italian institutions high-resolution ultrasound (US) imaging experts: C di Maggio (Padua), G M Giuseppetti
and real-time Elastography were performed with the same (Ancona), E Lattanzio (Bari), A Martegani (Como) and G
technology and procedure. Rizzatto (Gorizia). Statistical analysis was performed by
The patients were studied by L Aiani (Como), S Bal- an independent institution (I Floriani, from Mario Negri
dassarre (Ancona), A Bulzacchi (Padua), S Della Sala Institute in Milan).
(Trento), M Locatelli (Gorizia), G Mangialavori (Merano), P The research was approved according to the Italian reg-
Monno (Bari) and G Scaperrotta (Milan) (Fig. 1). ulations.
Their research was supervised by well-known breast

12 MEDIX Suppl. 2007

 adjacent frames is usually small ( 0.5 mm). The echo
intensity inside the cyst is extremely low. The displace-
ment at the center is erroneously estimated as almost 0
and is represented as a green band. The 2 areas near the
cystic wall have a different displacement value, with a
strain that is lower in the front (blue) and higher in the
back (red).
We think that our proposed score (Fig. 3) is more con-
sistent with the normal clinical settings of breast imaging.
A score of 1 indicates a three-layered pattern.
A score of 2 indicates a lesion with even elastic pattern
(diffuse green).
A score of 3 indicates a lesion with mostly even elastic
Fig. 1 : The Italian group at work pattern, but with some areas of no strain (blue areas).
Fig. 4 shows a fibroadenoma with a score 3 pattern.
A score of 4 indicates that most of the lesion has no
strain.
A score of 5 represents a lesion with no strain; there
Elastography was performed in 784 women (mean age, may be also a blue rim surrounding the lesion as defined by
52.5 years) who had 874 lesions with a definitive diagnosis conventional US.
(614 benign, 260 malignant) proved with core biopsy or fine
needle aspiration.
The size was 10 mm in 59% and 5 mm in 13.2% of
the lesions. SCORE 1
US images were classified according the BI-RADS cri-
teria for US 1); Elastography images were assigned an elas-
tographic score (1 to 5) according to the distribution and SCORE 2
degree of strain induced by light compression. Scores were
related both to solid and cystic lesions.
Our classification differs from the Tsukuba Elasto- SCORE 3
graphy Score proposed by Itoh A et al 2). In fact the Japan-
ese score is related only to solid lesions while BI-RADS for
US consider also non solid lesions; moreover in our prelim-
SCORE 4
inary practice we had observed that the cysts showed a
typical three-layered pattern (Fig. 2).
This pattern is due to an artifact. With Elastography
scanning many elasticity images are obtained by compar- SCORE 5

ing two adjacent frames to evaluate the displacement gen-

erated by the probe with continuous compression and
relaxation movements. The displacement of these two Fig. 3 : The proposed Italian Elastography score

Fig. 2 : Cysts always exhibit a typical three-layered pattern Fig. 4 : Fibroadenoma with a score 3 pattern
(score 1)

MEDIX Suppl. 2007 13

 Fig. 5 shows a 7 mm medullary carcinoma with a clear
acoustic distal enhancement in B-mode but with a Elasto- 1
graphy score 5.

0.8

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1

Fig. 5 : Medullary carcinoma with distal acoustic enhance-

ment but a clear score 5 pattern Fig. 6 : ROC curves for all the lesions

Considering the receiver operating curves (ROC) the 0.8

overall diagnostic performance of US was slightly better
than Elastography (area under the curve 0.94 for BI- 0.6
RADS, in pink and 0.90 for Elastography, in blue) (Fig. 6).
ROC curves demonstrated that Elastography works 0.4
better in lesions with a diameter 15 mm, with the best
results obtained with lesions 5 mm (Fig. 7).
0.2
Elastography showed a very high specificity in benign
lesions, including BI-RADS 3 lesions (329 lesions, 37.6%).
0
With the best cut-off point between elasticity scores 3 0 0.2 0.4 0.6 0.8 1
and 4 the true negative predictive value (TNPV) was 98%
for the whole series, 96.3% for all the BI-RADS 3 lesions,
Fig. 7 : ROC curves for lesions with a size 5 mm
and 100% for those with a size 5 mm.
The high specificity of Elastography in this series con-
firms the results obtained by Tardivon A et al 3) and Zhi H (1) Elastography may increase the specificity of US in the
et al 4). evaluation of focal breast lesions. It is not indicated for
Our higher TNPV is strictly correlated to the changes surgical scars, diffuse lesions or lesions larger than the
included in our new proposed score that clearly defines the transducer field of view;
cysts as a separate, well defined group. (2) Elastography interpretation requires a global experi-
Elastography scores were insensitive to the thickness ence in breast imaging with evaluation of all the avail-
and the echogenicity of the breast, and to the depth and able images. There is a learning curve; training must
the size of the lesion. include scanning and interpretation of at least 30 cases
Elastography scores interpretation resulted well repro- under the supervision of an expert;
ducible. K indexes of intra-observer (0.93) and inter- (3) At least 2 Elastography correct acquisitions of 5 sec-
observer (0.90) agreement resulted to be very good. onds must be obtained foe each lesion. The lesion must
be in the centre of the scanning area; the Elastography
area should cover almost all the field of view. With
lesions with mixed texture on B-mode, 2 Elastography
The Italian Elastography experts have defined a group scores must be acquired through perpendicular scan-
of guidelines to suggest the more appropriate approach to ning planes;
Elastography scanning and interpretation. (4) The pressure applied with the probe must be constant
and perpendicular to both the front margin of the lesion

14 MEDIX Suppl. 2007

 and the thoracic plane. Lateral movements must be Elastography score cannot work alone; it is only a new
avoided because they usually produce artefacts; descriptor (as margins, type of growth, echo texture, etc)
(5) Elastography acquisition should be considered correct that must be always integrated with all other US and
when the value of the reference LEDs on the monitor is imaging findings. Fig. 9 shows a 4 mm lesion with a Elasto-
constant and with a value of at least 2 or 3. Color homo- graphy score 5 due to a small cyst with partially calcified
geneity through all the scanning area surrounding the walls, as was clearly evident on the mammogram.
lesion indicates a good technical approach.

Elastography scores are accurate and reproducible. 1) American College of Radiology: Breast imaging report-
Diagnostic scores are acquired in almost all patients in a ing and data system (BI-RADS). Ultrasound. 4th ed.
few minutes and after a short learning curve. American College of Radiology 2003.
They help conventional US in characterizing small 2) Itoh A, et al. Breast disease: clinical application of US
breast lesions. If incorporated in the diagnostic flow chart Elastography for diagnosis. Radiology 2006, 239:341-350.
Elastography scores may avoid the use of biopsy in BI- 3) Tardivon A, et al. Elastosonography of the breast:
RADS 3 for US and may postpone to 1 year the follow-up prospective study of 122 lesions. J Radiol 2007; 88:657-
schedule. 662.
Elastography score may also suggest the most appro- 4) Zhi H, et al. Comparison of ultrasound Elastography,
priate workup for most of the cancers that present with mammography, and sonography in the diagnosis of solid
indeterminate or even benign descriptors. Fig. 8 shows a 9 breast lesions. J Ultrasound Med 2007; 26:807-815.
mm, homogeneous solid lesion, with oval shape and lobu-
lated margins; these descriptors and the absence of alter-
ations in the surrounding tissues clearly suggest a benign
fibroadenoma. In this case the Elastography score 4 is due
to the presence of a lobular carcinoma.

Fig. 8 : Lobular carcinoma showing benign descriptor but

with a score 4 pattern

Fig. 9 : Small calcified cyst on mammography shows a

clear score 4 pattern

MEDIX Suppl. 2007 15