Breast Lesions Quantitative Elastography with Supersonic Shear

Document Sample
Breast Lesions Quantitative Elastography with Supersonic Shear Powered By Docstoc
					         Note: This copy is for your personal, non-commercial use only. To order presentation-ready copies for
         distribution to your colleagues or clients, contact us at

                                                                                                                                                              ORIGINAL RESEARCH
                                                                   Breast Lesions: Quantitative
                                                                   Elastography with Supersonic Shear

                                                                                                                                                           n ULTRASONOGRAPHY
                                                                   Imaging—Preliminary Results1

Alexandra Athanasiou, MD
                                                                            Purpose:    To determine the appearance of breast lesions at quanti-
Anne Tardivon, MD
                                                                                        tative ultrasonographic (US) elastography by using super-
Mickael Tanter, PhD
                                                                                        sonic shear imaging (SSI) and to assess the correlation be-
Brigitte Sigal-Zafrani, MD                                                              tween quantitative values of lesion stiffness and pathologic
Jeremy Bercoff, PhD                                                                     results, which were used as the reference standard.
Thomas Deffieux, PhD
Jean-Luc Gennisson, PhD                                                 Materials and   This study was approved by the French National Com-
Mathias Fink, PhD                                                           Methods:    mittee for the Protection of Patients Participating in Bio-
Sylvia Neuenschwander, MD                                                               medical Research Programs. All patients provided written
                                                                                        informed consent. Conventional US and SSI quantitative
                                                                                        elastography were performed in 46 women (mean age,
                                                                                        57.6 years; age range, 38–71 years) with 48 breast lesions
                                                                                        (28 benign, 20 malignant; mean size, 14.7 mm); patho-
                                                                                        logic results were available in all cases. Quantitative lesion
                                                                                        elasticity was measured in terms of the Young modulus
                                                                                        (in kilopascals). Sensitivity, specificity, and area under the
                                                                                        curve were obtained by using a receiver operating charac-
                                                                                        teristic curve analysis to assess diagnostic performance.

                                                                             Results:   All breast lesions were detected at SSI. Malignant lesions
                                                                                        exhibited a mean elasticity value of 146.6 kPa 6 40.05
                                                                                        (standard deviation), whereas benign ones had an elastic-
                                                                                        ity value of 45.3 kPa 6 41.1 (P , .001). Complicated cysts
                                                                                        were differentiated from solid lesions because they had
                                                                                        elasticity values of 0 kPa (no signal was retrieved from
                                                                                        liquid areas).

                                                                          Conclusion:   SSI provides quantitative elasticity measurements, thus
                                                                                        adding complementary information that potentially could
                                                                                        help in breast lesion characterization with B-mode US.

                                                                                            RSNA, 2010

  From the Departments of Radiology (A.A., A.T., S.N.) and
Tumor Biology (B.S.), Institut Curie, 26 rue d’Ulm, 752048
Paris Cedex 05, France; Laboratoire Ondes et Acoustique,
Centre National de Recherche Scientifique, Unité Mixte de
Recherche 7587, Ecole Supérieure de Physique et Chimie
Industrielle, Paris, France (M.T., T.D., J.L.G., M.F.); and
SuperSonic Imagine, Aix-en-Provence, France (J.B.).
Received March 3, 2009; revision requested April 13; final
revision received August 14; accepted September 4; final
version accepted February 3, 2010. Address correspon-
dence to A.A. (e-mail: ).

    RSNA, 2010

Radiology: Volume 256: Number 1—July 2010      n                                                                                297
ULTRASONOGRAPHY: Elastography of Breast Lesions with Supersonic Shear Imaging                                                              Athanasiou et al

       alpation is a standard medical          movement has certain disadvantages:             speed and, consequently, the mapping of
       practice relying on qualitative es-     the elasticity map obtained is highly de-       local tissue stiffness. Our purpose was to
       timation of tissue Young modulus        pendent on the organ’s compressibility          determine the appearance of breast le-
E = s/´, where s is the external compres-      limits under stress and on the extent           sions at quantitative US elastography by
sion (stress) and ´ is the deformation of      of tissue compression applied. The dis-         using SSI and to assess the correlation
the tissue because of this compression         played information is about local strain        between quantitative values of lesion stiff-
(strain). Generally, benign lesions tend       estimated at a given location in tissues,       ness and pathologic results, which were
to be harder than normal breast tissue         but it depends on surrounding mechani-          used as the reference standard.
but softer than cancers (1). Exceptions        cal properties and it is not quantitative.
can occur, but soft malignant lesions, in-     Despite these caveats, the value of this
cluding medullary, mucinous, papillary,        technique in the evaluation of breast le-         Materials and Methods
and some necrotic infiltrating ductal           sions already has been reported (4–8).          The patent holder for this technique is
carcinomas, are uncommon (2). On the                To overcome this problem, we devel-        Laboratoire Ondes et Acoustique, Ecole
other hand, some benign lesions, such as       oped a quantitative elastography tech-          Supérieure de Physique et Chimie In-
hyalinized fibroadenomas and fat necro-         nique, supersonic shear imaging (SSI),          dustrielle (Paris, France), where M.T.,
sis, can be hard at palpation. At B-mode       that combines two concepts. Instead             T.D., J.L.G., and M.F. are employees.
ultrasonography (US), there is no corre-       of using mechanical external compres-           M.T. is a cofounder of and shareholder
lation between the lesion’s echo pattern       sion, the system itself remotely induces        in, J.B. is a founder of, and M.F. is a
and the Young modulus; hard and soft           mechanical vibration by using acoustic          shareholder in and scientific adviser to
lesions may exhibit similar echogenicity.      radiation force created by a focused ul-        SuperSonic Imagine (Aix-en-Provence,
Elastography has emerged as a prom-            trasound beam. A very fast (5000 frames         France), which provided the equipment
ising technique for improving lesion           per second) US acquisition sequence is          and the technical support for the study
differentiation. It can be used to esti-       used to capture the propagation of shear        (programming of US sequences and
mate breast tissue elasticity by measur-       waves. Both steps are performed by using        data postprocessing). Data were con-
ing tissue strain or tissue displacement       the same conventional US probe (9–12).          trolled by authors (A.A. and A.T.) who
during sonologist-induced mechanical           The displacement induced at the focus           do not have any financial interest.
excitation (3). The principle applied is       (for example, a breast lesion) generates
that strain will be less in harder tissue.     a shear wave that conveys information           Patients
Some currently available US systems            linked to the local viscoelastic properties     This study was approved by the French
include a strain imaging software pack-        of the tissue, thus enabling a quantitative     National Committee for the Protection
age: strain images, usually color coded,       approach to elasticity values. The shear        of Patients Participating in Biomedi-
are displayed in real time as the user         wave speed v is linked with shear modulus       cal Research Programs. All patients
applies light compression on breast tis-       m by the equation m = rv2, where r is the       provided written informed consent.
sue with the transducer. With freehand         local density (constant and equal to 1000       We performed SSI in 62 consecutive
compression, the influence of probe             kg · m23 in soft tissues). In soft tissues,     women with mammographically oc-
                                               local stiffness is described by the Young       cult lesions seen only at US who were
 Advances in Knowledge                         modulus E and can be approximated by            referred to the Curie Institute (Paris,
                                               E ' 3m (13). With imaging in real time,
 n Supersonic shear imaging (SSI)
                                               the propagation of the shear wave en-
   elastography can improve differ-                                                             Published online before print
                                               ables the recovery of local shear wave
   entiation of solid from atypical                                                             10.1148/radiol.10090385
   B-mode cystic breast lesions
                                                 Implications for Patient Care                  Radiology 2010; 256:297–303
   (manifesting as hypoechoic
   lesions) because no signal is                 n The addition of the elasticity               Abbreviations:
   retrieved from liquid areas (atyp-              parameter in the overall Breast              BI-RADS = Breast Imaging Reporting and Data System
                                                                                                SSI = supersonic shear imaging
   ical cyst elasticity value, 0 kPa).             Imaging Reporting and Data
 n No external mechanical compres-                 System classification potentially             Author contributions:
   sion is needed with SSI elastog-                could improve the specificity of              Guarantors of integrity of entire study, A.A., J.L.G.; study
                                                   breast US.                                   concepts/study design or data acquisition or data analysis/
   raphy, which is based on the
                                                                                                interpretation, all authors; manuscript drafting or manu-
   mechanical excitation that inter-             n The detection of cystic lesions              script revision for important intellectual content, all authors;
   rogates breast tissue mechanical                potentially could reduce the                 manuscript final version approval, all authors; literature
   properties by means of remote                   number of fine-needle aspiration              research, A.A., M.T., B.S., T.D., J.L.G., M.F., S.N.; clinical
   palpation induced by the radia-                 biopsies performed for proteina-             studies, A.A., A.T., M.T., B.S., J.B., T.D., J.L.G.; statistical
   tion force of ultrasound beams                  ceous or hemorrhagic cysts that              analysis, M.T., J.L.G.; and manuscript editing, A.A., A.T.,
                                                                                                M.T., B.S., J.B., J.L.G., S.N.
   transmitted by a conventional                   have an atypical B-mode
   US probe.                                       appearance.                                  See Materials and Methods for pertinent disclosures.

298                                                                               n   Radiology: Volume 256: Number 1—July 2010
ULTRASONOGRAPHY: Elastography of Breast Lesions with Supersonic Shear Imaging                                                            Athanasiou et al

France) between September 2006 and                              (,1.9); spatial-peak temporal-average        of a 150-µsec burst at 5 MHz. Succes-
March 2007 for evaluation and percuta-                          intensity, 603 mW/cm2 (,720 mW/              sive focusing depths were separated by
neous procedures. Twelve patients were                          cm2); and thermal index, 0.48 (,6).          5 mm, making the pushing line 30 mm
excluded because of technical hardware                          Spatial-peak temporal-average intensity      long. This supersonic source generated
problems during SSI acquisition; the                            and thermal index were calculated by         plane shear waves that propagated
prototype power supply encountered a                            assuming a 1-Hz SSI frame rate, which        transversally in tissue in a few tens of
stability problem, and no raw US data                           was much lower in practice because two       milliseconds. These shear waves were
were acquired for these patients. Four                          successive acquisitions were launched        imaged by means of a very fast imaging
patients refused to undergo a percu-                            with a pause of 5 to 10 seconds.             sequence; 60 flat insonications (5-MHz
taneous interventional procedure and                                 The radiologist (A.A., A.T.) per-       burst of 1 msec) were performed at a
were excluded because of lack of patho-                         forming the SSI first obtained B-mode         4000-Hz pulse repetition frequency.
logic result. Final analysis was based                          images that were saved on the sys-           Raw radiofrequency data correspond-
on data in 46 patients (mean age, 57.6                          tem hard drive (36 3 44-mm rectan-           ing to back-scattered echoes were
years; age range, 38–71 years) who had                          gular window). For each lesion, two          recorded and stored. The imaging
48 nonpalpable breast lesions (7–16                             SSI sequences were performed; each           sequence provided a US video of the
mm; mean size, 14.7 mm 6 1.3 [stan-                             sequence comprised three successive          tissue during shear wave propagation.
dard deviation]) detected at US. Six le-                        pushing beams. During the first se-           On the basis of this video, displace-
sions were classified as Breast Imaging                          quence, the radiologist located the le-      ments induced by the shear wave in
Reporting and Data System (BI-RADS)                             sion in the center of the screen. The        tissues were calculated by comparing
3; 28 lesions, as BI-RADS 4; and 14                             first pushing beam was centered along         US frames by using classic speckle-
lesions, as BI-RADS 5. Percutaneous                             the central axis of the lesion; the sec-     tracking algorithms. Shear wave prop-
procedures for BI-RADS 3 lesions were                           ond, on the right side of the lesion; and    agation speed then was deduced lo-
performed in patients at high risk.                             the third, on the left side of the lesion.   cally by using a one-dimensional cross-
                                                                Each pushing beam lasted about 3 sec-        correlation algorithm to estimate the
Imaging Workflow                                                 onds, and three pushing beams were           time shift between the two tempo-
Lesions were depicted at B-mode US                              necessary to produce a full elastogra-       ral displacement signals separated by
by using the hospital US system (Aplio                          phy image (12). During the second SSI        1 mm. Finally, the local Young modulus
XG, model SSA-790 A; Toshiba Medi-                              sequence, the radiologist located the        E was deduced.
cal Systems Europe, Zoetermeer, the                             lesion on the left or the right side of          Elasticity maps deduced from each
Netherlands) equipped with a 12-MHz                             the screen to have both the lesion and       of three acquisitions, corresponding to
probe. Patients originally underwent                            surrounding parenchyma on the screen;        three different pushing lines, were then
scanning performed together by A.A. and                         the three pushing beams were used to         merged to compute the final elasticity
A.T. (radiologists specialized in breast                        obtain a full elastography image of both     map that covered the entire width of the
imaging, with 8 and 15 years of experi-                         the lesion and the parenchyma.               US image. The result was displayed on
ence, respectively) for clinical indications.                        The total time of the SSI examina-      a color scale ranging from 0 to 240 kPa;
Mammograms were available in all cases,                         tion was less than 5 minutes. Examina-       the corresponding shear wave speed
but lesions were mammographically                               tion results were transferred to an of-      thus roughly ranged from 0 to 8 m/sec.
occult because of either small size or                          fline computer by using an Ethernet           Spatial resolution was 1 mm2.
dense breast parenchyma. Lesion loca-                           cable limited to a 1 Mb/sec transfer rate
tion, size, echo pattern, and BI-RADS clas-                     and were processed with dedicated elas-      Data Analysis
sification (14) were noted. The BI-RADS                          tography software. Postprocessing time       Lesion dimensions were measured by
score was recorded by means of con-                             was 5 seconds for each quantitative elas-    using calipers on recorded B-mode US
sensus by A.A. and A.T.                                         ticity image. The accuracy of SSI values     images obtained with the hospital sys-
    SSI sequences were performed by                             was compared previously with calibrated      tem and SSI postacquisition images.
using a modified system (ATL HDI 1000;                           phantoms in other studies (68% error         Two quantitative elasticity values were
Philips Medical Systems, Best, the Neth-                        margin) (16). US-guided procedures           extracted, the first corresponding to
erlands) with a linear array (L7-4; Philips                     were then performed during the same          the elasticity value of the lesion and the
Medical Systems). The system was re-                            session by using the hospital US system.     second to the elasticity value of normal
programmed to provide, in addition to                                                                        tissue. Two 3 3 3-mm boxes were po-
its standard capabilities, a special SSI                        Elastography Sequence and Data               sitioned, one box in the lesion area at
acquisition mode by means of button ac-                         Processing                                   the location of maximum stiffness and
tivation. All safety considerations were                        Each SSI sequence corresponded to            one in normal-appearing tissue (glandu-
satisfied according to the U.S. Food and                         the generation of a vibration force by       lar and fatty tissue, which was distin-
Drug Administration (15). The high-                             means of successively focusing ultra-        guished by its lobular form, compress-
est values recorded for our prototype                           sound beams at five different depths.         ibility, and low echogenicity), by A.A.
were as follows: mechanical index, 1.4                          Each pushing focused beam consisted          and A.T. together. Cystic lesions were

Radiology: Volume 256: Number 1—July 2010   n                                                                                   299
ULTRASONOGRAPHY: Elastography of Breast Lesions with Supersonic Shear Imaging                                                                                Athanasiou et al

not used for the estimation of average
values because cysts do not have a stiff-          Final Pathologic Diagnosis in 48 Breast Lesions
ness value (E = 0 kPa). The whole pro-
cedure lasted about 5 minutes.                     Pathologic Diagnosis                                                                                      No. of Lesions

                                                    Malignant lesions (n = 20)
Pathologic Examination                                Infiltrating ductal carcinoma grade I                                                                   3
Twenty-eight lesions were benign at per-              Infiltrating ductal carcinoma grade II                                                                  9*
cutaneous US-guided biopsy (n = 16) or                Infiltrating ductal carcinoma grade III                                                                 5
fine-needle aspiration (n = 12) (Table);               Infiltrating lobular carcinoma grade III                                                                2
no further surgery was performed ex-                  Infiltrating mixed ductal and lobular carcinoma                                                         1
cept in one lesion that corresponded                Benign lesions (n = 28)
to a grade I phyllodes tumor. Seven le-               Fibroadenoma                                                                                           8
sions corresponded to cysts (four had                 Aberrations of normal development and involution                                                       9
a proteinaceous or hemorrhagic com-                   Cyst                                                                                                   7
ponent). Twenty lesions were malignant                Benign grade I phyllodes tumor                                                                         1
at biopsy, and further surgical excision              Fat necrosis                                                                                           2
                                                      Intramammary lymph node                                                                                1
was performed. Final histologic diagno-
sis and tumor grade were defined by us-             * Includes one lesion with extensive intraductal component.
ing the surgical specimen. All diagnoses
were made by a pathologist (B.S., with
18 years of experience in breast patho-        Then, 21 was added to the normalized                                malignant lesions demonstrated mean
logic examination).                            score (2E/Emax 2 1). Any value lower than                           values of 146.6 kPa 6 40.05 (P , .001)
                                               Emax/2 would be given a score of 21, and                            (Figs 3, 4).
Statistical Analysis                           any value higher than the mean would be                                  Mean values of the new variable
The role of elasticity values in discrimi-     given a score of +1. Only values exactly                            b were then compared by using the
nating malignant from benign lesions           equal to Emax/2 would be given a score                              Mann-Whitney test, and they were also
was studied by using a nonparametric           of 0. The value calculated was added to                             different (P , .001). The discriminating
Mann-Whitney test. Receiver operating          the BI-RADS score, thus modifying the                               power for detection of malignancy of
characteristic curves were used to com-        BI-RADS score by 0, 21, or +1.                                      the variable b was significantly higher
pare the performance of estimators cor-            Areas under the curve were com-                                 than that of the BI-RADS score alone,
responding to the established BI-RADS          pared by using a Hanley and McNeil                                  with an area under the receiver operat-
variable and to a modified BI-RADS              test (17). Sensitivity and specificity                               ing characteristic curve of 0.985 6 0.01
variable taking into account the lesion’s      values and their respective 95% confi-                               (95% confidence interval: 0.884, 0.995)
mean Young modulus E, named b. Be-             dence intervals were estimated. P , .05                             compared with the BI-RADS score area
cause the elasticity parameter is not en-      was considered to indicate a statistically                          under the receiver operating character-
visioned as a stand-alone parameter but        significant difference. All data were                                istic curve of 0.917 6 0.046 (95% con-
rather as an added value to US exami-          analyzed by using software (MedCalc                                 fidence interval: 0.764, 0.968; P , .05)
nation, the combination of the BI-RADS         version; MedCalc Software,                                  (Fig 5).
score with the Young modulus variable          Mariakerke, Belgium).                                                    Comparison between receiver oper-
E was estimated. The distribution of E                                                                             ating characteristic curves with a Han-
was rescaled, shifted to the interval (21,                                                                         ley and McNeil test for BI-RADS and
+1), and added linearly to the BI-RADS           Results                                                           b gave a difference of 0.068 6 0.033
score. The rescaling process was based                                                                             (P = .039). The most important effect
on the following equation: b = BI-RADS         Normal Breast Tissue and Breast Cysts                               of the new variable b was for BI-RADS
+ 2E/Emax 2 1, where b stands for the          Quantitative elasticity values in normal                            4 lesions (n = 28). Adding the elastic-
combined parameter BI-RADS + elastic-          breast tissue clearly delineated the dif-                           ity score modified the BI-RADS 4 score
ity, which allows the SSI elasticity value     ferent structures. Fatty tissues displayed                          to BI-RADS 3 for 13 lesions, meaning
to change the BI-RADS score by at most         low values of elasticity (E, approximately                          that we could have avoided perform-
61 unit. To shift elasticity to the inter-     7 kPa), whereas breast parenchyma val-                              ing biopsy in 46% of cases and had a
val [21, +1], we normalized each lesion        ues ranged from 30 to 50 kPa (Fig 1).                               short 6-month follow-up instead. Nine
elasticity value by comparing it to the        All breast cysts had elastography values                            of 28 BI-RADS 4 cases remained at the
maximum elasticity found during SSI im-        of 0 (Fig 2).                                                       same score after the elasticity value
aging in the entire study population, de-                                                                          was added, and six cases were recat-
fined as Emax. The elasticity value (E) was     Solid Lesions                                                       egorized as BI-RADS 5. No cancers
measured for each lesion and was then          Benign lesions had a mean elastic-                                  were missed during this modification
normalized by a factor of 2 (2E/Emax).         ity value of 45.3 kPa 6 41.1, whereas                               of BI-RADS score. The specificity of b

300                                                                                                   n   Radiology: Volume 256: Number 1—July 2010
ULTRASONOGRAPHY: Elastography of Breast Lesions with Supersonic Shear Imaging                                                                           Athanasiou et al

  Figure 1                                                           Figure 2                                            Figure 3

                                                                                                                        Figure 3: Images in 45-year-old woman with
                                                                Figure 2: Images in 48-year-old woman with              fibroadenoma. (a) B-mode US depicts a hypoechoic,
                                                                complex cystic lesion of left breast. (a) B-mode US     homogeneous, lobular mass lesion classified as
                                                                depicts a hypoechoic mass lesion with lobulated         BI-RADS 3. (b) SSI shows a low elasticity score
                                                                margins and slight posterior enhancement, classi-       (,50 kPa).
                                                                fied as BI-RADS 4. (b) SSI shows no elasticity value
Figure 1: Normal breast tissue in 52-year-old                   for this lesion. Fine-needle aspiration was performed
woman. (a) B-mode US does not depict any par-                                                                           for lesion margin and posterior acoustic
                                                                with US guidance, and a yellow liquid was evacu-
ticular lesion. (b) Elasticity map obtained by using                                                                    features, the agreement is moderate (k =
                                                                ated. Final diagnosis was a cyst containing inflam-
SSI. The shear wave speed is coded on a color                                                                           0.40 for both), and for lesion echo pat-
                                                                matory cells and debris.
scale ranging from 0 to 8 m/sec, corresponding                                                                          tern, the agreement is poor (k = 0.29).
to a Young modulus ranging between 0 and 192                                                                            These description criteria often lead to
kPa. The elasticity map exhibits the delineation                biopsy is also recommended for BI-RADS                  false-positive findings and a number of
between soft fatty tissues (E, approximately 7 kPa)             4 lesions.                                              unnecessary biopsies.
and breast parenchyma (E, approximately                                                                                     We have evaluated a quantitative
30–50 kPa).                                                                                                             elastography technique (ie, SSI) that
                                                                     Discussion                                         could complement US examination in
was 0.96, and the sensitivity was 0.95,                         Breast lesion detection and character-                  an efficient and easy-to-use way. The
whereas the specificity and sensitivity                          ization at B-mode US is based mainly on                 results of our preliminary work indicate
of the BI-RADS score were 0.63 and                              specific description criteria defined by                  that SSI was useful in demonstrating
0.96, respectively. No modification was                          the American College of Radiology in the                breast masses and characterizing cys-
noted in the BI-RADS 3 category. Only                           BI-RADS lexicon. For US descriptors,                    tic ones. This potential interesting ap-
one BI-RADS 5 lesion was reduced to                             results of a prior study (18) showed that               plication of SSI elastography relies on
BI-RADS 4 according to the variable                             there is substantial agreement for lesion               its capacity to reject cystic lesions, re-
b, but, in practice, the therapeutic                            orientation, shape, and boundary (k =                   gardless of their typical or complicated
decision remained unchanged because                             0.61, 0.66, and 0.69, respectively), but                B-mode US appearance, because of the

Radiology: Volume 256: Number 1—July 2010   n                                                                                                  301
ULTRASONOGRAPHY: Elastography of Breast Lesions with Supersonic Shear Imaging                                                                                  Athanasiou et al

  Figure 4                                                   Figure 5                                                     The SSI approach also overcomes
                                                                                                                     the intrinsic limitation of conventional
                                                                                                                     elastography, which provides only quali-
                                                                                                                     tative and relative elasticity measure-
                                                                                                                     ments. By providing local measurements
                                                                                                                     of the true Young modulus of tissues
                                                                                                                     across a wide stiffness range (from 1 to
                                                                                                                     240 kPa), SSI could be considered as a
                                                                                                                     quantification tool for breast cancer di-
                                                                                                                     agnosis. Our results demonstrate that
                                                                                                                     quantitative elasticity has clinical per-
                                                                                                                     tinence. The b variable, taking into ac-
                                                           Figure 5: Box-and-whisker plot of Young modulus           count both the BI-RADS categorization
                                                           E estimated by using the SSI modality with respect        and the Young modulus E, can ame-
                                                           to malignancy. According to the Mann-Whitney              liorate the overall diagnostic specific-
                                                           test, E values were significantly different between        ity (0.96 vs 0.63 for BI-RADS alone),
                                                           malignant and benign lesions (P , .001). Boxes =          whereas sensitivity remains high (0.95
                                                           values from lower to upper quartiles, central lines =     vs 0.96 for BI-RADS alone). Larger
                                                           medians; whiskers extend from minimal to maximal          clinical trials are necessary to validate
                                                           values. Dots = outliers.                                  these preliminary results.
                                                                                                                          Quantitative elasticity maps at SSI
                                                           perform a percutaneous procedure such                     seem to allow finer assessment of tis-
                                                           as biopsy or fine-needle aspiration.                       sue mechanical properties than does
                                                               From the user’s perspective, one of                   qualitative elastography by using ex-
                                                           the major differences of SSI compared                     ternal compression. At conventional
                                                           with conventional elastography is that                    elastography, the measured strain at
                                                           the mechanical vibration is induced au-                   one location depends on the surround-
Figure 4: Images in 61-year-old woman already              tomatically by the system by using the                    ing tissues’ mechanical properties. For
treated for breast cancer and referred because of          radiation force of ultrasound beams.                      example, a lesion having a softer cen-
suspicion of recurrence. (a) B-mode US depicts a           The reliability of the imaging tech-                      ter surrounded by a hard periphery
15-mm hypoechoic mass lesion (L) with irregular            nique does not depend on the skills of                    would be impossible to detect because
margins adjacent to the previous lumpectomy scar           the sonologist in correctly vibrating or                  the whole lesion would move in a block
(S). This suspicious lesion was classified as BI-RADS       stressing the tissue. The elasticity infor-               under external stress. The local infor-
4. (b) SSI depicts a hard lesion (E . 200 kPa)             mation is provided by using the same                      mation provided by the SSI technique
highly suggestive of recurrence. The 90-kPa region         workflow and scanning conditions used                      could lead to a much finer analysis of
on the left of this very stiff lesion corresponds to the   for conventional B-mode US.                               the spatial distribution and extension of
scar. SSI depicted the two distinct neighboring zones          An interesting consequence is that                    tissue stiffness.
of different elasticity (very stiff malignant lesion       the user does not need, as in conven-                          Our preliminary study had some
and moderately stiff scar). Biopsy results confirmed        tional elastography, to subjectively or                   limitations. All malignant lesions in-
infiltrating ductal carcinoma.                              arbitrarily select one image in a com-                    cluded in the study corresponded to
                                                           plete image cine loop (7). With the SSI                   infiltrating carcinomas, so we did not
absence of any shear wave propagation                      approach, a single set of successive US                   evaluate elasticity mapping in cases of
in liquid areas. During our preliminary                    sequences provides a relevant elasticity                  ductal carcinoma in situ. Another limi-
study, no solid lesions presented a simi-                  image. Acquisition and postprocessing                     tation concerned the research platform
lar phenomenon of absence of shear                         in our preliminary study on a dedicated                   used for SSI; our modified system was
wave propagation.                                          research platform were limited by the                     limited in terms of acquisition time,
     In solid, nonpalpable masses, quanti-                 slow transfer rate from electronics to                    and B-mode images were substandard
tative elasticity may add potentially valu-                personal computer boards and thus                         because of the 7.4-MHz array. This
able information that could help radiolo-                  did not permit real-time elastography.                    system was not capable of performing
gists better differentiate breast lesions.                 However, no technologic issue prevents                    real-time SSI. Availability of real-time
As already shown in the literature (8),                    the SSI mode from providing elasticity                    SSI quantitative elastography would be
less experienced radiologists, in particu-                 images several times per second in the                    of great interest for evaluating mode ro-
lar, could be helped in their assessment                   near future. In addition to conventional                  bustness, reproducibility, and operator
by elasticity imaging. We believe that the                 B-mode imaging, elasticity frame rates                    independence.
SSI technique may have the potential to                    of three to four images per second are                         A major factor that remains to
improve the decision about whether to                      technologically feasible today.                           be understood is the effect of static

302                                                                                                     n   Radiology: Volume 256: Number 1—July 2010
ULTRASONOGRAPHY: Elastography of Breast Lesions with Supersonic Shear Imaging                                                                              Athanasiou et al

compression on the method. Human                                     3. Garra BS, Cespedes EI, Ophir J, et al.           11. Bercoff J, Chaffai S, Tanter M, et al. In vivo
tissues have hardening properties; that                                 Elastography of breast lesions: initial clini-       breast tumor detection using transient elas-
                                                                        cal results. Radiology 1997;202(1):79–86.            tography. Ultrasound Med Biol 2003;29(10):
is, their elasticity increases if static
compression is applied. A study should                               4. Hiltawsky KM, Krüger M, Starke C, Heuser
be performed to quantify the influence                                   L, Ermert H, Jensen A. Freehand ultra-           12. Tanter M, Bercoff J, Athanasiou A, et al.
                                                                        sound elastography of breast lesions: clinical       Quantitative assessment of breast lesion
of typical static compressions on the di-
                                                                        results. Ultrasound Med Biol 2001;27(11):            viscoelasticity: initial clinical results using
agnostic value of the mode.                                                                                                  supersonic shear imaging. Ultrasound Med
    Our clinical investigation in 48 le-                                                                                     Biol 2008;34(9):1373–1386.
sions showed that quantitative elasticity                            5. Thomas A, Kümmel S, Fritzsche F, et al. Real-
                                                                        time sonoelastography performed in addition      13. Sarvazyan AP. Biophysical bases of elasticity
mapping of breast tissue is feasible in
                                                                        to B-mode ultrasound and mammography:                imaging. In: Acoustical imaging. Vol 21. New
vivo by using the SSI method. Malignant                                                                                      York, NY: Plenum, 1995; 223–240.
                                                                        improved differentiation of breast lesions?
lesions were significantly different from
                                                                        Acad Radiol 2006;13(12):1496–1504.               14. American College of Radiology. Ultrasound.
benign solid ones with regard to lesion
                                                                                                                             In: Breast imaging reporting and data sys-
elasticity quantitative value; because no                            6. Itoh A, Ueno E, Tohno E, et al. Breast dis-
                                                                                                                             tem (BI-RADS). 4th ed. Reston, Va: Ameri-
shear wave propagation occurs in liquid                                 ease: clinical application of US elastogra-
                                                                                                                             can College of Radiology, 2003.
                                                                        phy for diagnosis. Radiology 2006;239(2):
areas, cystic lesions were diagnosed,
                                                                        341–350.                                         15. U.S. Food and Drug Administration. In-
regardless of their B-mode appearance.                                                                                       formation for manufacturers seeking mar-
SSI could be a valuable complementary                                7. Burnside ES, Hall TJ, Sommer AM, et al.
                                                                                                                             keting clearance of diagnostic ultrasound
method for characterizing indetermi-                                    Differentiating benign from malignant solid
                                                                                                                             system and transducers. http://www.fda
                                                                        breast masses with US strain imaging. Radi-
nate breast lesions, thus obviating un-                                                                                      .gov/cdrh/ode/ulstran.pdf. Published 1997.
                                                                        ology 2007;245(2):401–410.
necessary biopsies. Large prospective                                                                                        Accessed October 10, 2009.
trials are necessary to determine the                                8. Scaperrotta G, Ferranti C, Costa C, et al.       16. Bercoff J, Criton A, Bacrie CC, et al. Shear-
role of SSI in a clinical setting.                                      Role of sonoelastography in non-palpable             WaveTM Elastography: a new real time im-
                                                                        breast lesions. Eur Radiol 2008;18(11):              aging mode for assessing quantitatively soft
Acknowledgments: The authors are grateful to                            2381–2389.                                           tissue viscoelasticity. Proceedings of the
Anne Badel, PhD, and Lilliane Ollvier, MD, for
                                                                                                                             2008 IEEE International Ultrasonics Sympo-
their essential contribution to this work.                           9. Bercoff J, Tanter M, Fink M. Supersonic
                                                                                                                             sium, Beijing, China, November 2-5, 2008.
                                                                        shear imaging: a new technique for soft
                                                                        tissue elasticity mapping. IEEE Trans Ultra-     17. Hanley JA, McNeil BJ. A method of com-
                                                                        son Ferroelectr Freq Control 2004;51(4):             paring the areas under receiver operating
References                                                              396–409.                                             characteristic curves derived from the same
 1. Sewell CW. Pathology of benign and malig-                                                                                cases. Radiology 1983;148(3):839–843.
                                                                10. Bercoff J, Tanter M, Muller M, Fink M.
    nant breast disorders. Radiol Clin North Am
                                                                    The role of viscosity in the impulse diffrac-        18. Lazarus E, Mainiero MB, Schepps B, Koel-
                                                                    tion field of elastic waves induced by the                liker SL, Livingston LS. BI-RADS lexicon
 2. Tavassoli FA, Devilee P. Pathology and ge-                      acoustic radiation force. IEEE Trans Ultra-              for US and mammography: interobserver
    netics: tumours of the breast and female                        son Ferroelectr Freq Control 2004;51(11):                variability and positive predictive value.
    genital organs. Lyon, France: IARC, 2003.                       1523–1536.                                               Radiology 2006;239(2):385–391.

Radiology: Volume 256: Number 1—July 2010   n                                                                                                      303

Shared By: