7/25/2019 Us Diagnosis of Breast Cancer
1/14
Copyright 2013 American Scientific PublishersAll rights reservedPrinted in the United States of America
R E S E A R C H A R T I C L E
Journal of Medical Imaging and
Health Informatics
Vol. 3, 114, 2013
Ultrasound Diagnosis of Breast Cancer
Yufeng Zhou
School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore
Ultrasound is a popular imaging modality for its safety and low cost. Its role in the diagnosis of breast cancer
is discussed, and its performance is then compared those of mammography (gold standard) and MRI. Besides
conventional B-mode and color or power Doppler ultrasound images, latest development of acoustic radiation
force impulse (ARFI) and supersonic shear imaging and their application in breast diagnosis are introduced.
Keywords: Breast Cancer, Sonography, Magnetic Resonance Imaging, Doppler Ultrasound, AcousticRadiation Force Impulse Imaging, Supersonic Shear Imaging.
1. SONOGRAPHY IN BREAST
CANCER DIAGNOSIS
Ultrasound (US) is becoming a popular clinical diagnosis modal-ity in the past decades with the major advances in transducer,electrical circuit, digital signal processing, and system control,
and has already been applied to large varieties of diseasesbecause of its uniqueness of low cost, flexibility, non-invasion,
and non-ionization. US has the ability to image and evaluatepatients internal anatomy structure and physiology in real-timewith astounding clarity. Therefore, it makes significant contribu-
tions to healthcare.The breast examination by US started from 1951 with an opti-
mistic opinion that US would replace mammography eventually
in detecting breast cancer.1 However, with more comprehensivestudies, it illustrates that US is only valid for the discrimination
between cysts and solid masses. The performance of US dependson the size, number, location, and properties of the lesions, theoperation skills, and the system specifications (i.e., resolution and
frequency).The diagnosis of cysts is clinically and socially important for
remarkable reduction on the number of breast biopsies and early
therapy, particularly for those non-palpable ones. Aspiration orbiopsy is not required to simple cysts,2 which reduces the health-
care cost, anxiety and discomfort of patients during with surgery.However, for palpable masses, aspiration is advocated due to its
less expense, availability at many centers, and often accompanieswith therapy, but not much desirable for many women due toits discomfort. US is the appropriate method of lesion diagnosiswith accuracy of 96100% if it is not palpable and applicable
for aspiration.25 A simple cyst usually has smooth walls, sharpanterior and posterior borders, no internal echoes, and posteriorenhancement (Fig. 1). Using a gel or fluid offset and aligning
the target to the focal region with better resolution could definethe anterior wall, especially for superficial lesions. In compari-
son, posterior enhancement is mostly inconsistent. The smallest
detectable cysts in the breast depends on the type, location, breast
size, and US system, and are usually 23 mm.4 Scanning the
lesion carefully in two projections and discriminating wall irreg-
ularities are useful in finding intracystic tumors.3 6 Occasion-
ally, inappropriate equipment setting (i.e., time compensated gain
(TCG), locations of focal point(s), or brightness), technical lim-
itation (i.e., low driving frequency of transducer, inherent noise
in beam forming, or poor resolution), or missing subtle changes
by the operator may result in diagnostic error.
However, it is hard to judge whether a visible mass is benign
or malignant in the sonography because of the similar features.
Only circumscribed masses, such as a cyst, require differen-
tial US diagnosis (Fig. 2). A mass with stellate or spiculated
borders as well as a circumscribed mass contains suspicious
micro-calcifications does not require US for further evaluation.
However, calcium deposition inside a cyst is easily mistaken as
micro-calcifications because of the similar granular appearance.7
Both benign and malignant breast masses may be partially or
completely obscured by the normal tissue. When a palpable mass
is not visible in the mammography, US will be applied to deter-
mine whether it is cystic or solid. Because all solid masses may
not be visible in the sonography even in dense breasts, the false-
negative detection of breast cancer is about 25%.8 A palpable
mass that is invisible in both mammography and sonography
strongly needs biopsy histology (Fig. 3).
Occasionally, diffuse mastitis does not have an appropriateresponse to antibiotics, which suggests abscess formation. There-
fore, mammography may not be appropriate for pain and edema
of the breast and will not display a discrete abscess cavity
with inflammation-induced higher density. US is an alternative
approach for diagnosing and guidance of surgical drainage of
abscesses, which have various sonographic characteristics, such
as irregular hypoechoic or anechoic cavities, occasional fluid-
fluid or fluid-debris levels, posterior enhancement, distortion on
surrounding, and overlying skin thickening (Fig. 4).
J. Med. Imaging Health Inf. Vol. 3, No. 2, 2013 2156-7018/2013/3/001/014 doi:10.1166/jmihi.2013.1157 1
7/25/2019 Us Diagnosis of Breast Cancer
2/14
R E S E A R C H A R T I C L E J. Med. Imaging Health Inf. 3, 114, 2013
(a) (b)
(c)
Fig. 1. A palpable lesion in the left upper inner quadrant of breast shown in (a) oblique and (b) craniocaudal mammograms (arrows), and (c) a simple cyst
in the transverse sonogram (arrows) with smooth borders, no internal echoes, posterior enhancement, and lateral shadows from the smoothly curved walls.
US seems a more sensitive imaging modality for positive axil-
lary lymph nodes in the breast cancer that is commonly involved
but rarely evaluated than physical examination and axillary mam-
mography with compression.9 The capability of metastatic nodes
by US and physical examination are 72.7% and 45.4%, respec-
tively, with equal specificities of these two techniques (97.3%)in a study of 60 patients.10 Furthermore, US has success in the
(a) (b)
Fig. 2. A 3-cm palpable fibroadenoma found by biopsy is shown in (a) oblique mammogram (arrows) and (b) the longitudinal sonogram as a well-outlined,
slightly lobulated, and homogeneous solid mass (arrows).
guidance of fine-needle aspiration of cyst, biopsy of solid masses,
preoperative needle, and wire localization.1113
Fibroadenomas are smoothly round, oval, or lobulated lesions
with homogeneous internal echoes and posterior enhancement
while carcinomas are irregular solid masses with internal echoes
and attenuated brightness.1415
Therefore, their US images havecertain overlaps, and accurate determination is possible with
2
7/25/2019 Us Diagnosis of Breast Cancer
3/14
R E S E A R C H A R T I C L EJ. Med. Imaging Health Inf. 3, 114, 2013
(a) (b)
(c)
Fig. 3. False-negative (a) oblique and (b) craniocaudal mammograms with moderate density but no discrete mass and (c) the longitudinal sonogram of a
4-cm palpable mass in the lower inner quadrant of the right breast, which was revealed as ductal carcinoma in biopsy.
conventional sonography.3 1619 Ratios of the length to the antero-
posterior diameter of the cancer and fibroadenomas are signifi-
cantly different, and the oblong configuration in fibroadenomas
is more apparent in superficial position.20
US has been implemented in screening breast cancer, par-ticularly those with dense tissue, in Japan, Europe, and Aus-
tralia. However, its inability of detecting all types of breastcancers and a substantial number of non-palpable cancers who
are visible and invisible at mammography, respectively, pre-
vents US being a valuable screening modality because of its
unacceptably high false-negative rate, (0.347% with mean of
20.7%), which may be even higher for small and clinically occult
cancers.21 Most importantly, US detection of a significant number
of non-palpable carcinomas with good-quality negative mammo-
grams is not always satisfactory. In addition, US has a remark-able false-positive rate in asymptomatic patients because of the
Fig. 4. Longitudinal sonogram of (a) a mastitis and large area of induration that has an irregular abscess cavity with echoes and posterior enhancement, and
(b) a severe mastitis with multiple small (3-9 mm) scattered abscesses (solid arrows), thick skin (open arrows), high echogenicity, structure distortion, and a
sharp interface between normal and abnormal tissue (curved arrow).
shadowing produced by many normal structures22 and similar
sonographic features between fat lobules and solid tumors.23 US
diagnosis is unable to exclude malignancy and identify an under-
lying reason for the asymmetric fibroglandular tissue, but mam-
mography and physical examination.8
In summary, sonography at high quality and high frequency
used in a judicious manner is a valuable tool in the clinics. Sonog-
raphy is applied to the evaluation of circumscribed lesions found
in the mammography, palpable masses invisible in the mam-
mography, and a cyst is in the differential diagnosis. The need
for biopsy on a simple cyst could be eliminated after the US
diagnosis. Otherwise, a biopsy or mammographic follow-up is
required no matter of the sonography results, in which US pro-
vides not much clinically useful information and, subsequently, is
not suggested.4
3
7/25/2019 Us Diagnosis of Breast Cancer
4/14
R E S E A R C H A R T I C L E J. Med. Imaging Health Inf. 3, 114, 2013
2. COMPARISON OF MAMMOGRAPH,
SONOGRAPHY AND MRI
The diagnostic sonography in the breast has been investigated
for at least 30 years.1 Sonography did not detect any proven
cancers that were missed by mammography. Mammography was
found superior in detecting 97% of the 64 pathologically con-firmed cancers, while sonography can only seek 58% of them.
Mammography detected more than 90% in all cancer categories,
including those amenable to cure, but the value for sonography
is only 48% (40% of the non-palpable malignancies and 8% of
the cancers smaller than 1 cm that did not yet spread to axillary
lymph nodes). Tumor size and axillary lymph node status are
the most important prognostic indicators for breast cancer, and
the mammography done far outperformed sonography in detect-
ing the smallest cancers and those that did not yet spread to
axillary nodes. A major factor limiting the ability of sonogra-
phy for non-palpable breast cancers seems to be its inability to
image the micro-calcifications (individual particles 0.20.5 mm).
Mammography-positive sonography-negative cancers usually are
small, non-palpable, and have not yet spread to axillary lymph
nodes, whereas very rare sonography-positive mammography-negative cancers are always detectable by physical examination
and more likely to have metastasized. Therefore, upgrading to
state-of-the-art mammography is preferred to improving the can-
cer detection ability rather than purchasing an US system. 24
The role of sonography in breast diagnosis is an ongoing
investigation.25 US is a widely accepted method for discriminat-
ing cysts from solid masses and guiding interventional proce-
dures. Sensitivities and accuracy of the US in the discrimination
between benign and malignant breast nodules are not high
enough to reply on, and its value in comparison or addition to
mammography is still in debate.26 Thus, US is not recommended
as a screening tool due to the failure in establishing its efficacy. 27
US was performed to detect
(1) circumscribed lesions (possible cysts),
(2) palpable lesions visible in mammography,(3) palpable lesions not visible in mammography, and
(4) non-palpable lesions visible in mammography in a 2-year
prospective study of 4,811 cases.
As a result, 1,103 cases (23%) were reclassified their suspicion
levels of malignancy.25
US can achieve a certain improvement in breast cancer diagno-
sis as an adjunct to mammography. Although it not very dramatic
for the total cohort of patients, such an improvement was consid-
erable for the subgroup of patients, especially among the young
patients with low sensitivity of mammography. Mammographic
classification based on a relatively high threshold for biopsy pro-
vides US opportunity of increasing sensitivity. If further advances
are achieved in US diagnosis, especially for diffusely growing
cancers, a further acceptance of US in the breast cancer diagnosisis expected.25
Young breasts, despite low occurrence, are more sensitive to
radiation so that the limited exposure is desired. US imaging
is usually performed in the initial study. No further evalua-
tion is necessary for a cyst. If the mass is solid or invisible in
sonography, at least one mammogram will be obtained to seek
micro-calcifications. Although mammography allows detection of
almost all palpable masses, adequate positioning may not be pos-
sible for very deep lesions adjacent to the chest wall or in a
slim woman with a mass at the extreme periphery of the breast.
Altogether, US is not a substitute for mammography, nor does a
negative sonogram rule out carcinoma.4
3. DIAGNOSIS OF MICROCALCIFICATION
The identification of micro-calcification (i.e., smaller than0.5 mm) on mammography has been widely studied and is indis-
pensable in the early detection of breast cancer.12 3545% of
discovery of non-palpable breast cancers depends on the pres-
ence of clusters of micro-calcification on mammography.2 Micro-
calcifications are brighter reflectors than the surrounding breast
parenchyma without an acoustic shadow in sonography.6 In astudy of 89 tumors found in 84 patients, micro-calcifications
were visible in 44 breast cancers using high resolution ultrasound
(HRUS, 49%), 40 cancers using X-ray mammography (XRM)
(45%) and 46 breast cancers on histology (53%).28 HRUS has
the sensitivity of 95%, specificity of 87.8%, and accuracy of
91% in the detection of micro-calcification. The correspondingvalues for histology are 80%, 71.4% and 75.3%, respectively.
Therefore, US is a sensitive and reliable diagnosis modality
for micro-calcification in breast cancer presented within a masslesion.28 HRUS detected micro-calcification in 6 cancers that
were negative on XRM, among them 4 were positive on his-
tology. These false positives may be due to the requirement of
sufficient deposition of calcium phosphate for the identificationof micro-calcifications for XRM but not necessary for US. In
addition, XRM is a survey of the entire breast, whilst US is
tomographic and its multi-section analysis may increase detec-
tion accuracy.28
Calcifications that occur within masses are more visible on US,
which is partially because most malignant solid nodules providea great echogenicity. In contrast, sonography for benign calcifi-
cations with many hyperechoic and heterogeneous fibers are less
reliable. Hence, malignant are more visible in the sonography
than benign calcifications. Although the sensitivity of sonogra-
phy for calcifications is lower than that of mammography, sono-graphically visible calcifications within a solid mass have high
possibility of malignant.29
The different types of ductal carcinoma in situ (both comedo
and non-comedo) correlate with mammographic patterns of
micro-calcifications and the latter inconsistent foci of micro-
calcifications. Malignant micro-calcifications within ductal car-
cinoma in situ and microscopically invasive ductal carcinoma,
which do not have associated sonographically demonstrablemasses, are difficult to identify on US.28
4. DIAGNOSIS OF LYMPH NODE
METASTASES
The presence of axillary lymph node metastases in breast cancer
is an important symptom in assessing prognosis and determiningthe treatment plan. Axillary staging is conventionally performed
by axillary lymph node dissection. The use of sonography in
detecting metastases is feasible and would reduce the number offalse-negatives at sentinel node biopsy.30
In the studies including palpable and non-palpable nodes,
if the size (>5 mm) or node visibility in sonography was
used as the criterion for positivity, sensitivity varying between
66.1% (95% confidence interval: 52.677.9%) and 72.7% (49.8
89.3%), with no heterogeneity between them, including both
4
7/25/2019 Us Diagnosis of Breast Cancer
5/14
R E S E A R C H A R T I C L EJ. Med. Imaging Health Inf. 3, 114, 2013
(a) (b)
(c)
Fig. 5. Extensive clustered pleomorphic micro-calcification within a large area of (a) increased soft tissue density, highly suspicious of malignancy in mam-
mogram, (b) multiple bright echogenic spots (small white arrows) within a large markedly hypoechoic mass lesion in sonography, corresponding to the
mammographic findings of micro-calcification, and (c) infiltrative ductal carcinoma with micro-calcifications (arrows) in H&E histology with magnification of 125.
gold standard of axillary lymph node dissection and sentinel
node biopsy. However, the variation of specificity is from 44.1%
(34.354.3%) to 97.9% (88.799.9%), and heterogeneity is found
between the results.30 If the lymph node morphology was used
for positivity, variations of sensitivity and specificity are from
between 54.7% (41.767.2%) and 80.4% (73.986.2%) to 92.3%
(74.999.1%) and 97.1% (9099.6%), respectively. Whist in thestudies, including only non-palpable nodes, if the node size in
sonography (>5 mm) or its visibility was used as a criterion for
positivity, sensitivity varies from 48.8% (39.658%) to 87.1%
(76.194.3%) and specificity from 55.6% (44.766.3%) to 97.3%
(86.199.9%). If the node morphology was used as the crite-
rion for positivity, variations of sensitivity and specificity were
from 26.4% (15.340.3%) and 88.4% (82.193.1%) to 75.9%
(56.489.7%) and 98.1% (90.199.9%), respectively. In the US
guided biopsy, the sensitivity varies between 43.5% (3354.7%)
and 94.9% (88.598.3%) and specificity between 96.9% (91.3
99.4%) and 100% (96.2100%), although sensitivity is reduced
because it is necessary to visualize the node or to fulfill the sono-
graphic criteria for malignancy.
Therefore, sonography is moderately sensitive and specific in
the diagnosis of axillary metastases in breast cancer. However,
it cannot be used as a sole method for decision, whether to per-
form axillary lymph node dissection. When suspicious metastatic
axillary nodes are found, a US guided biopsy can be performed,
which increases the specificity (100% vs. 96.5% use of sonogra-
phy alone) at the cost of certain aggression and extra resources.
Subsequently, about half of the axillae with metastases would be
detected with a high specificity (96.5%) and a good sensitivity
(48.4%), and then those positive patients would undergo axillary
lymph node dissection. The remaining negative would be can-
didates for sentinel node biopsy, which improves the negative
predictive value of the sentinel node biopsy because of the lower
prevalence of metastases and thereby increases the certainty of
sonography-based diagnosis.30
5. SURVEILLANCE OF WOMEN AT HIGH
FAMILIAL RISK FOR BREAST CANCER
Breast cancer susceptibility gene (BRCA) mutation carriers, who
exhibit adverse histopathologic features of biologic aggressive-
ness, has a lifetime risk for up to 6580% to develop breast
cancers rather than sporadic ones.31 A comparative cohort study
was carried out to investigate the effectiveness of mammogra-
phy, US, and MRI in 529 women with increased familial risk.32
Annual conventional mammography was performed with at least
two views (medio-lateral oblique and cranio-caudals) per breast,
and additional or spot compression views where appropriate.
Diagnoses were coded according to the Breast Imaging Report-
ing and Data system (BI-RADS) diagnostic categories. Breast
ultrasound was performed with 7.5- to 13-MHz probes. Stan-dard contrast-enhanced MRI of both entire breasts was performed
on a 1.5 T system after injection of 0.1 mmol/kg gadopentetate
dimeglumine.
43 breast cancers were identified in the total cohort (34 inva-
sive, 9 ductal carcinoma in situ). Overall sensitivity of diagnostic
imaging was 93% (40 of 43); overall node-positive rate was 16%,
and one interval cancer occurred (1 of 43, or 2%). In the anal-
ysis by modality, sensitivity was low for mammography (33%)
and US (40%) or the combination of both (49%). MRI offered
5
7/25/2019 Us Diagnosis of Breast Cancer
6/14
R E S E A R C H A R T I C L E J. Med. Imaging Health Inf. 3, 114, 2013
Fig. 6. (A) Mammogram and (B) sonogram of suggestive of cancer (arrowhead) on a 53-year-old patient with a family history of breast cancer and personal
history of benign breast biopsy on the left breast revealed no clinical findings. (C) MRI showed only scar tissue on the left (arrowhead), but revealed a
suspicious lesion in the right breast (long arrow), which was an invasive ductal cancer, pT1b, G3, N0, M0 by biopsy. Absence of cancer in the left breast was
confirmed by 4-year follow-up.
a much higher sensitivity (91%). The sensitivity of mammog-
raphy in the higher-risk groups was 25%, compared to 100%
for MRI. Specificity of MRI (97.2%) was equivalent to that ofmammography (96.8%).33 Mammography, either alone or com-
bined with sonography, seems insufficient for diagnosis of early
breast cancer in patients who are at increased familial risk with
or without BRCA mutation. If MRI is used for surveillance, a
significantly higher sensitivity, specificity, and positive predictive
value (PPV) could be achieved for diagnosis of intraductal and
invasive familial or hereditary cancer at a more favorable stage.33
Indeed, not even half of all cancers were prospectively diagnosed
with a combination of mammography and sonography, whereas
breast MRI alone diagnosed 91% (39 of 43). However, MRI is
still an investigational technique for surveillance and screening of
asymptomatic women with normal conventional diagnosis. Apart
from cost, the most important reason of breast MRI is low PPV,
low specificity, and allegedly low sensitivity for ductal carcinoma
in situ(DCIS). However, MRI has the highest sensitivity for inva-
sive as well as intraductal cancers, which was not achieved at the
expense of similar specificity as that of mammography.33 Com-
bined with mammography, sonography can compensate some but
not all the shortcomings of mammography with a substantial
number of false-positive diagnoses. In comparison to surveillance
by MRI, mammography was of limited and US of no additional
value. US screening may, however, be useful in the long interval
between the annual surveillances.33
Altogether, surveillance with MRI allows an earlier diagno-
sis of familial breast cancer and has better performance than
mammography or the combination of mammography with high-
frequency sonography.
6. DOPPLER ULTRASOUND
The well-known phenomenon of tumor angiogenesis is asso-
ciated with an increase in malignancy.34 35 These abnormali-
ties included tumor stains, irregular large caliber vessels, and
either prolonged or rapid emptying of vessels presumably due
to blood pooling, leaky vessels, and/or arterio-venous shunts.
Among the established breast diagnosing techniques, mammog-
raphy and sonography have undisputed contributions. However,
no adequate information on the growth pattern and the prognosis
of breast humps are available. Doppler US has been investigated
to differentiate benign lesion from malignant solid breast masses
from their different Doppler characteristics, (i.e., symmetric sig-nals in normal tissue, no signals in cysts, and higher maximum
systolic and end-diastolic pressures in malignant tumors)3638
with high sensitivity and specificity despite considerable overlap
in the benign and malignant types.39 40 Highly sensitive color
Doppler on even minute tumor vessels could map the tumor
blood flow both quantitatively and qualitatively.
Color Doppler in 2 of 39 malignant breast disease patients
with the tumor size of 0.68.0 cm (median 2.0 cm) did not
show any vascularity. In comparison, no blood vessels were
6
7/25/2019 Us Diagnosis of Breast Cancer
7/14
R E S E A R C H A R T I C L EJ. Med. Imaging Health Inf. 3, 114, 2013
found in 10 of 73 benign masses (0.34.7 cm with the median
of 1.4 cm). In patients with puerperal mastitis, abscess, phyl-
loides tumor, and haemangioma, vascularization was extremely
high. Benign and malignant breast lesions have significantly dif-ferent Doppler US features. There is a remarkable overlap of
carcinoma and benign tumor in peak flow velocity.41 The dis-
crepancies between reported studies may be related to the USsystem and the scanning techniques.42 The accuracy for smaller
blood vessels, especially for poorly vascularized masses, could
be improved using a high-frequency and high-resolution sys-
tem. Furthermore, color Doppler may also be able to reduce thenumber of biopsy and histological evaluations for patients with
suspicious mammograms.41
In another prospective study, the color Doppler flow images of
55 proven breast cancers were performed, and 82% of them were
classified on a three-level scale of vascularity (minimal: 14%,
moderate: 29%, marked: 53%), suggesting its clinical potentialuse. 4% of the flow images had no detectable flow. 69% of the
normal breasts had moderate or marked vascularity (minimal:
28%, moderate: 41%, marked: 28%), and 3% were avascular.
Because of poor distinction between normal tissues and cancer,
more sensitive Doppler methods are required for the low vesselflow that is rather specific for malignancy.
Power Doppler sonography has advantages over color Doppler
type with high sensitivity in the detection of vascular flow. Power
Doppler findings were considered positive when at least one
vessel was associated with the solid breast mass (Fig. 7). In
one study comprising 176 breast cancers in 176 patients (2791
years, mean std: 56 15 years), 65% and 10% of the caseswere invasive ductal carcinoma and invasive lobular carcinoma,
(a) (b)
(c) (d)
Fig. 7. Patterns of tumor vascularity in breast cancer revealed by power Doppler sonography. (a) penetrating, irregularly branching prominent vessels,
(b) peripheral and penetrating vessels, (c) few central vessels, and (d) no vessels.
respectively.43 Among them, 73% showed vascularity and 27%
showed no vascularity on power Doppler sonography. The sizes
of the lesions in which power Doppler sonography revealed ves-
sels and no vessels were 780 (21 15) mm and 355 (14
14) mm, respectively (p < 001); however, these two categories
overlapped. Tumor vascularity revealed by power Doppler sonog-
raphy correlated strongly with detection of lymph node involve-ment and lymphatic vascular invasion with sensitivities of 93%
and 90%, but low specificities of 32% and 35%, respectively.More importantly, patients with breast cancer in whom ves-
sels were not revealed by power Doppler sonography were also
unlikely to have lymph node involvement and lymphatic vascu-
lar invasion with negative predictive values of 90% and 87%,
respectively.In conclusion, Doppler US could detect moderately small ves-
sels around and within tumors, even if they were too small to be
displayed on conventional B-mode images. Although only can-
cer patients were involved, the presence of similar vessels in the
normal breast indicates Doppler imaging as a technique with apresumably higher specificity but much lower sensitivity.
7. ACOUSTIC RADIATION FORCEIMPULSE IMAGING
The limitations of palpation and biopsy as well as CT, MRI,
and US imaging require a noninvasive, cost-effective, safe, and
accurate modality for detecting changes in tissue pathology. Sev-
eral groups have developed elasticity-based imaging modality inorder to exploit the relationship between pathology and tissues
mechanical properties.
7
7/25/2019 Us Diagnosis of Breast Cancer
8/14
R E S E A R C H A R T I C L E J. Med. Imaging Health Inf. 3, 114, 2013
Acoustic radiation force is due to the momentum transfer from
the propagation of acoustic waves to the dissipative medium. 44
The absorption is in the direction of wave propagation, whereasthe reflection depends on the angular scattering properties of the
target.45 Under plane-wave assumptions, the generated radiation
force in the tissue is.44 46
F =2I
c(1)
where F is a force per unit volume, is tissue attenuation, I is
the acoustic intensity, and c is the sound speed of tissue.Acoustic radiation force impulse (ARFI) imaging, a novel tran-
sient elastography method, generates radiation force inside the
tissue, detects the consequent localized displacements by corre-lating the ultrasonic echoes, and then estimate the mechanical
properties of target.47 All tissues are inherently viscoelastic and
response differently to mechanical excitation, on the order of ten
micrometers, which can be monitored both spatially and tempo-rally and is inversely proportional to local tissue stiffness. The
tissue volume exposed to radiation force is determined by the
focal characteristics of the transmitting transducer, and the tem-
poral profile of the force is dependent on transmitted pulse shape,usually with the duration less than 1 ms. In this method, a single
diagnostic transducer is used both to apply localized radiation
forces inside the tissue and to track the resulting tissue displace-ments, which guarantees good alignment and ease of real-time
implementation. ARFI imaging has many potential advantages,
such as identifying and characterizing a wide variety of soft tis-sue lesions, atherosclerosis, plaque, and thrombosis in clinics.
There is good correlation between the ARFI image and the
matched B-mode image in the breast at the depth of 525 mm
(Fig. 8(a)). In the ARFI image, the boundary of an infectedlymph node as a palpable lesion appears stiffer than its interior
and the tissue above it (i.e., smaller displacements). The oval
structure in the B-mode image immediately above and to the leftof the lesion (upper arrow) is outlined as a softer region of tissue
than its surroundings in the ARFI image. The transient responseto ARFI excitation depends on tissue structure and mechanical
properties. In Figure 8(b) the region of tissue spanning 1318 mmmoves further, and exhibits a later peak displacement than the
others, which is slightly darker in the matched B-mode image.
Fig. 8. (a) ARFI image of tissue displacement at 0.8 ms (right), and matched B-mode image (left) in an in vivo female breast. The transducer is located at
the top of the images, and the colorbar scale is microns. (b) Displacement through time at different depths in the center of the ARFI image (0 mm laterally).
Peak displacements of ARFI images range from 5 to 13 m
in vivo, and the lesion in the B-mode image exhibits a stiff
outer boundary and a softer interior in the matched ARFI image
(Fig. 9). Core biopsy of this lesion demonstrated an infected
lymph node with a more liquid abscessed component. While
these findings are circumstantial, they do suggest potential cor-
relation between the clinical pathology and the ARFI image.The discrepancy of the brightness in the B-mode image and
detected stiffness indicates no direct relationship between them
as expected.
There is no speckle in the ARFI images. Thus, the tradition-
ally defined speckle SNR of a conventional US imaging system
is in general lower than that of the ARFI imaging system. Com-
parison of the contrast of the ARFI and B-mode images yields
variable results. The applied radiation force and resulting tissue
displacements are predominantly in the direction of wave propa-
gation. With the current imaging geometry, only these axial dis-
placements are tracked, and the anticipated error is 0.14m.48
While useful information might be derived from the lateral dis-
placement, tracking of such small lateral displacements (
7/25/2019 Us Diagnosis of Breast Cancer
9/14
R E S E A R C H A R T I C L EJ. Med. Imaging Health Inf. 3, 114, 2013
Fig. 9. ARFI images at times of (a) 0.4 ms, (b) 1.2 ms, (c) 2.1 ms, and (d) 3.0 ms in the in vivo female breast (the same dataset as shown in Fig. 8).
visco-elastic behavior. Although these findings are preliminary,they present several opportunities for ARFI imaging with a con-
siderable clinical promise.
8. SUPERSONIC SHEAR IMAGING
Supersonic shear imaging (SSI) is another transient elastogra-
phy approach and combines the remote palpation of the ARFI
technique and the ultrafast echographic imaging approach, which
provides a quantitative elasticity map with less dependence on
operator in comparison to static elastography.52 53 The initial
Fig. 10. Generation of a conical shear wave front propagating in the imag-
ing plane of the echographic probe.
clinical investigation illustrates its potential as an adjunct forsonography.
SSI generates a remote radiation force by focused ultrasonic
beams as ARFI. Consequently, a transient shear wave will be
formed by the remote tissue vibration. Several pushing beams
at increasing depths are transmitted to generate a quasi-plane
shear wave front that propagates throughout the region-of-interest
(Fig. 10). After that, successive raw radiofrequency (RF) data
is acquired at an ultrafast frame rate (2,000 frames/s). Contrary
to conventional sonography formed using line-by-line scanning,
ultrafast echoic images are achieved by transmitting a single
quasi-plane ultrasonic wave which has slight diffraction along the
transducer elevation direction and then performing the imaging
process (i.e., beam forming, array signal processing) only in the
receiving mode. Because of the memory limit, only 128 succes-
sive ultrafast echoic images can be stored at a total duration of30 ms.54 55
3 successive SSI sequences were performed to locate the lesion
(Fig. 11). The first SSI sequence was performed using pushing
Fig. 11. Protocol of quantitative elastography for a lesion located in the
center using the SSI method. Three SSI sequences are performed. The first
SSI sequence corresponds to a pushing line along the central line of the
ultrasound image. The second and third SSI sequences correspond to two
successive pushing lines located on the right and left edges of the image.
These three pushing modes enable the final recovery of a quantitative elas-
ticity map over the entire ultrasound region.
9
7/25/2019 Us Diagnosis of Breast Cancer
10/14
R E S E A R C H A R T I C L E J. Med. Imaging Health Inf. 3, 114, 2013
Fig. 12. Comparison between the B-mode ultrasound and the elasticity image obtained in the SSI with colorbar presenting the shear wave speed (09 m/s,
E= 0240 kPa) with the delineation between soft fatty tissues (7 kPa) and breast parenchyma (30 kPa) in normal breast tissue.
beams centered along the central line, which allowed elasticity
imaging on both left and right parts. Then, the second and third
SSI sequences were performed with a left and a right pushing
line, respectively, and allow elasticity imaging in the middle of
the imaging plane.
In the initial clinical trial, a total of 15 lesions were assessed
quantitatively for its elasticity at an image window of 38
44 mm2. The elasticity map displays the local shear wave speed
cs on a color scale ranging from 09 m/s with the corresponding
Youngs Modulus, E= 3c2s , ranging from 0240 kPa. Elastic-
ity maps exhibit good concord with the spatial heterogeneities
and structures in normal breast tissue and provide quantitative
highlighting for benign and malignant lesions. In general, benignsolid and malignant lesions in that study had a mean Youngs
modulus of 4580 kPa and 100 kPa (in some cases >180 kPa),
Fig. 13. Comparison between B-mode ultrasound and quantitative elasticity map in the SSI mode for infiltrating ductal carcinoma grade III. Hypoechoic lesion
with indistinct margins, slightly posterior shadowing classified as BI-RADS category 5.
respectively. It is known, however, that some cancers such as the
mucinous subtype can be rather soft, and some mature fibroade-
nomas can be extremely stiff.
8.1. Normal Breast Tissue
The quantitative elasticity in normal breast tissue clearly delin-
eated the different structures of breast. Youngs modulus ranged
between 3 kPa for fatty tissues to 45 kPa in the parenchyma.
Figure 12 compares the US gray scale and Youngs modulus
image, which presents a fibrous mass (benign lesion correspond-
ing to a fibrocystic disease). The Youngs modulus is easily
recovered in the normal breast tissue areas. There is nice delin-
eation between fatty tissue (
7 kPa) and breast parenchyma(4050 kPa) on both the US grayscale and elasticity images.
These characteristic values were found in all healthy patients.55
10
7/25/2019 Us Diagnosis of Breast Cancer
11/14
R E S E A R C H A R T I C L EJ. Med. Imaging Health Inf. 3, 114, 2013
Fig. 14. Comparison between B-mode ultrasound and quantitative elasticity map in the SSI for small (5-mm diameter) hypoechoic lesion classified as BI-RADS
category 5, which is difficult to detect on B-mode ultrasound. The elasticity map clearly delineates a small and stiff region ( 165 kPa), which was confirmed
by biopsy exam as an infiltrating ductal carcinoma grade III.
8.2. Malignant Lesions
A typical example of a BI-RADS category 5 lesion and its cor-
responding elasticity map is shown in Figure 13. The B-mode
US depicted a 10-mm hypoechoic mass with indistinct margins
and posterior shadowing, which strongly indicates a high suspi-
cion of malignancy and is confirmed in the pathologic diagnosis
as an infiltrating ductal adenocarcinoma (grade III with a high
mitotic activity). SSI clearly exhibits higher stiffness of the lesion
with mean elasticity value of 150175 kPa and better delineation
of the lesion margins than B-mode US images. The lesion size
measured on the elasticity map was confirmed by the pathologic
analysis (810 mm). In addition, the Youngs modulus image
in the SSI enables a satisfactory discrimination between fatty tis-
sues (5 kPa), breast parenchyma (40 kPa), and lesion location
(170 kPa) in the resolution of roughly 1.5 mm.The potential ability of the SSI for the guided percuta-
neous procedures (core biopsies or fine needle aspirations) with
Fig. 15. Comparison between B-mode ultrasound and quantitative elasticity map in the SSI for hypoechoic, homogeneous, lobular-shaped lesion classified as
BI-RADS category 4, which was diagnosed as fibrocystic disease by biopsy and has comparable elasticity with the surrounding healthy tissues ( E:1227 kPa).
clinically satisfactory location precision was illustrated in
Figure 14 for a small, slightly hypoechoic nodular lesion with
indistinct contours measuring 5 mm, which was classified as
BIRADS category 4. This lesion is rather difficult to discrim-
inate on B-mode US because of its low echogenicity contrast
in comparison to the normal parenchyma after local anesthesia,
thus necessitating multiple sampling. In comparison, the elastic-
ity map clearly delineates a small and stiff region (165 kPa)
and properly depicted margins with an average 5 mm diameter.
This lesion size was confirmed after biopsy by the pathologist.
8.3. Benign Solid Lesions
SSI was also able to detect benign solid lesions such as fibroade-
nomas or fibrocystic disease changes. In all cases, these lesions
were detected on the elasticity mapping as rather soft struc-tures with mean Youngs modulus 100 kPa. A hypoechoic,
11
7/25/2019 Us Diagnosis of Breast Cancer
12/14
R E S E A R C H A R T I C L E J. Med. Imaging Health Inf. 3, 114, 2013
(a)
(b)
Fig. 16. Comparison between B-mode image and quantitative elasticity map in the SSI of (a) hypoechoic lesions with lobulated margins, discretely reinforcing
ultrasound beam, classified as BI-RADS category 4, which was diagnosed as a cyst containing inflammatory cells and debris in fine-needle aspiration, and
(b) another patient with a benign cyst nodule.
homogenous, lobular-shaped lesion classified as BI-RADS cat-
egory 4 can be observed in Figure 15. SSI describes struc-
tures with different elasticity in good concord with the structures
depicted in the B-mode image, but comparable elasticity with the
surrounding healthy tissues (E:815 kPa). Biopsy was performed
under US guidance and led to a histologic diagnosis in favor of
fibrocystic disease.
8.4. Benign Cysts
SSI was also useful in the diagnosis of cystic lesions.
Figure 16(a) corresponds to hypoechoic lesions with lobulated
margins and a discretely reinforced US beam, which was clas-
sified as BI-RADS category 4. Fine-needle aspiration was per-formed under US guidance, and a yellow-colored liquid was
evacuated, which led to a histologic examination as a cyst con-
taining inflammatory cells and debris. The SSI elasticity map
provided local Youngs modulus in healthy surrounding tissues
except in the lesion, which is consistent with the fact that shear
waves do not propagate in liquids. All data corresponding to non-
propagating shear waves are intrinsically filtered by the imag-
ing post processing algorithm. Two reasons for this filtering can
be evoked. First, the strong acoustic streaming induced in the
liquid can lead to a de-correlation of the successive US data.
Second, the strong modification of the shear displacement versus
the propagation direction yields to a de-correlation of shear dis-
placement time profiles at neighboring locations, resulting in afalse time-of-flight estimation. This de-correlation can be in both
cases filtered, leading to an absence of Youngs modulus esti-
mation in the liquid. Figure 16(b) corresponds to another more
clearly hypo-echoic cystic lesion which is identified as a liquid
area surrounded by soft tissues in the SSI.
In summary, quantitative mapping of breast tissue elasticity is
feasible in vivo using the SSI approach. Discrimination between
breast fat and parenchyma and identification of malignant lesion,benign solid lesion, cystic lesions are feasible, reliable, and
clearly visible. This novel imaging modality is significantly lessoperator dependent than static elastography as the mechanical
excitation that interrogates breast tissues is induced by the sys-
tem itself. It could have a potential in clinics for breast lesion
diagnosis.
9. CONCLUSION
High-frequency high-quality sonography system has significant
technical improvements with high resolution, great contrast, large
12
7/25/2019 Us Diagnosis of Breast Cancer
13/14
R E S E A R C H A R T I C L EJ. Med. Imaging Health Inf. 3, 114, 2013
dynamic range, less speckle noise, high frame rate, and multi-
ple ultrasound imaging modality (i.e., color Doppler and real-
time three-dimensional scanning) in the past decades. Although
digital signal/image processing techniques aided the automated
tumor/cancer detection and enhanced the outcome, the supe-
riority of state-of-the-art mammography over sonography has
been shown in a variety of clinical studies. However, in thedetection and diagnosis of benign lesions, for example, a dis-
tinction between cystic and solid masses, mammography is not
necessarily the preeminent examination, and sonography is the
useful procedure of choice. Meanwhile, development of novel
ultrasound-based elastography, especially the transient type that
generates remote pushing force to the target, enables the detec-
tion of the mechanical properties of tissue, which has higher
sensitivity, specificity and contrast than the conventional B-mode
ultrasound images. Although the preliminary results are very
promising, its role in breast cancer diagnosis will be carried out
in multiple and randomized clinic centers and then be compared
with mammography for performance evaluation.
References and Notes1. J. J. Wild and D. Neal, Use of high frequency ultrasonic waves for detecting
changes of texture in living tissues. Lancet1, 655 (1951).
2. E. A. Sickles, R. A. Filly, and P. W. Callen, Benign breast lesions: Ultrasound
detection and diagnosis.Radiology151, 467 (1984).
3. S. W. Hilton, G. R. Leopold, K. K. Olson, and S. A. Willson, Real-time
breast sonography: Application in 300 consecutive patients. AJR 147, 479
(1986).
4. V. P. Jackson, The role of US in breast imaging. Radiology 177, 305
(1990).
5. J. Jellins, Combining imaging and vascularity assessment of breast lesion.
Ultrasound in Medicine and Biology14, 121 (1988).
6. F. Kasumi and H. Tanaka, Detection of microcalcifications in breast carci-
noma by ultrasound, Ultrasound Examination of the Breast, edited by J. J.
and T. Kobayashi, John Wiley & Sons, New York (1983).
7. D. R. Pennes and M. Rebner, Layering granular calcifications in macroscopic
breast systs. Breast Dis. 1, 109 (1988).
8. D. B. Kopans, C. A. Swann, G. White et al., Asymmetric breast tissue.Radi-
ology171, 639 (1989).
9. M. Pamilo, M. Soiva, and E. M. Lavast, Real-time ultrasound, axillarymammography, and clinical examination in the detection of axillary lymph
node metastases in breast cancer patients. J. Ultrasound Med. 8, 115
(1989).
10. J. N. Bruneton, E. Caramella, M. Hery, D. Aubanel, J. J. Manzino, and J. L.
Picard, Axillary lymph node metastases in breast cancer: Preoperative detec-
tion with US. Radiology158, 325 (1986).
11. C. J. DOrsi and E. B. Mendelson, Interventional breast ultrasonography.
Semin. Ultrasound CT MR10, 132 (1989).
12. S. A. Feig, The role of ultrasound in a breast imaging center. Semin. Ultra-
sound CT MR10, 90 (1989).
13. D. B. Kopans, J. E. Meyer, K. K. Lindfors, and S. S. Bucchianeri, Breast sonog-
raphy to guide cyst aspiration and wire localization of occult solid lesions.
AJR143, 489 (1984).
14. P. Harper and E. Kelly-Fry, Ultrasound visulaization of the breast in symp-
tomatic patients. Radiology137, 465 (1980).
15. T. Kobayashi, Gray-scale echography for breast cancer.Radiology 122, 207
(1977).
16. C. Cole-Beuglet, R. Z. Soriano, A. B. Kurtz, and B. B. Goldberg, Ultrasound
analysis of 104 primary breast carcinomas classified according to histopatho-
logic type.Radiology147, 191 (1983).
17. S. H. Heywang, P. S. Dunner, E. R. Lipsit, and L. M. Glassman, Advantages
and pitfalls of ultrasound in the diagnosis of breast cancer. JCU 13, 525
(1985).
18. S. H. Heywang, E. R. Lipsit, L. M. Glassman, and M. A. Thomas, Specificity
of ultrasonography in the diagnosis of benign breast masses. J. Ultrasound
Med. 3, 453 (1984).
19. D. B. Kopans, J. E. Meyer, and R. T. Steinbock, Breast cancer: The appear-
ance as delineated by whole breast water-path ultrasound scanning. JCU
10, 313 (1982).
20. B. D. Fornage, J. G. Lorigan, and E. Andry, Fibroadenoma of the breast:
Sonographic appearance.Radiology172, 671 (1989).
21. V. P. Jackson, E. Kelly-Fry, P. W. Rothschild, R. W. HHolden, and S. A. Clark,
Automated breast sonography using a 7.5 MHz PVDF transducer: Preliminary
clinical evaluation. Radiology159, 679 (1986).
22. G. Kossoff, Causes of shadowing in breast sonography. Ultrasound in
Medicine and Biology14, 211 (1988).
23. C. Kimme-Smith, L. W. Bassett, and R. H. Gold, High frequency breast
ultrasound: Hand-held versus automated units-examination for palpable mass
versus screening. J. Ultrasound Med.7, 77 (1988).
24. E. A. Sickles, Sonographic detectability of breast calcifications. Proc. SPIE
419, 51 (1983).
25. H. M. Zonderland, E. G. Coerkamp, J. Hermans, M. J. van de Vijver, and
A. E. van Voorthuisen, Diagnosis of breast cancer: Contribution of US as an
adjunct to mammography.Radiology213, 413 (1999).
26. V. P. Jackson, The current role of ultrasonography in breast imaging. Radiol.
Clin. North Am. 33, 1161 (1995).
27. L. A. Venta, C. M. Dudiak, C. G. Salomon, and M. E. Flisak, Sonographic
evaluation of the breast. RadioGraphics14, 29(1994).
28. W. T. Yang, M. Suen, A. Ahuja, and C. Metreweli, In vivodemonstration
of microcalcification in breast cancer using high resolution ultrasound. The
British Journal of Radiology70, 685 (1997).
29. A. T. Stavros, D. Thickman, C. L. Rapp et al., Solid breast nodules: Use of
sonography to distinguish between benign and malignant lesions. Radiology
196, 123 (1995).
30. S. Alvarez, E. Anorbe, P. Alcorta, F. Lopez, I. Alonso, and J. Cortes, Role
of sonography in the diagnosis of axillary lymph node metastases in breast
cancer: A systematic review.AJR186, 1342 (2006).
31. D. Ford, D. F. Easton, D. T. Bishop et al., Risks of cancer in BRCA1-mutation carriers: Breast cancer linkage consortium. Lancet 343, 692
(1994).
32. K. Metcalfe, H. T. Lynch, P. Ghadirian et al., Contralateral breast can-
cer in BRCA1 and BRCA2 mutation carriers. J. Clin. Oncol. 22, 2328
(2004).
33. C. K. Kuhl, S. Schrading, C. C. Leutner, N. Morakkabati-Spitz, E. Wardelmann,
R. Fimmers, W. Kuhn, and H. Schild, Mammography, brest ultrasound, and
magnetic resonance imaging for surveillance of women at high familial risk
for breast cancer. J. Clin. Oncol. 23, 8469 (2005).
34. J. Folkman, How is blood vessel growth regulated in normal and neoplastic
tissue?Cancer Res. 46, 467 (1986).
35. J. Folkman, K. Watson, and D. Ingber, Induction of angiogenesis during the
transition from hyperplasia to neoplasia. Nature 339, 58 (1989).
36. P. N. Burns, M. Halliwell, P. N. T. Wells, and A. J. Webb, Ultrasonic
Doppler studies of the breast.Ultrasound in Medicine and Biology 8, 127
(1982).
37. P. N. T. Wells, M. Halliwell, R. Skidmore, A. J. Webb, and J. P. Woodcock,
Tumour detection by ultrasonic Doppler blood-flow signals.Ultrasonics15, 231
(1977).38. D. N. White and P. R. Cledgett, Breast carcinoma detection by ultrasonic
Doppler signals.Ultrasound in Medicine and Biology4, 329 (1978).
39. M. Bohm-Velez and E. B. Mendelson, Computed tomography, duplex Doppler
ultrasound, and magnetic resonance imaging in evaluating the breast. Semin.
Ultrasound CT MR10, 171 (1989).
40. V. P. Jackson, Duplex sonography of the breast. Ultrasound in Medicine and
Biology14, 131 (1988).
41. C. Peters-Engl, M. Medl, and S. Leodolter, The use of colour-coded and spec-
tral Doppler ultrasound in the differentiation of benign and malignant breast
lesions.British Journal of Cancer71, 137 (1995).
42. W. Dock, F. Grabenwoger, and V. Metz, Tumor vascularization: Assessment
with Duplex sonography. Radiology181, 241 (1991).
43. T. S. Mehta and S. Raza, Power Doppler sonography of breast cancer: Does
vascularit correlate with node status or lymphatic vascular invasion? AJR
173, 303 (1999).
44. G. Torr, Thge acoustic radiation force.Am. J. Phys. 52, 402 (1989).
45. D. Christensen, Ultrasonic Bioinstrumentation, John Wiley & Sons, New York
(1988).
46. W. Nyborg, Acoustic Streaming, Academic Press, New York(1965).47. K. Nightingale, M. Palmeri, R. Nightingale, and G. Trahey, On the feasibil-
ity of remote palpation using acoustic radiation force. J. Acoust. Soc. Am.
110, 625 (2001).
48. W. Walker and G. Trahey, A fundamental limit on delay estimation using par-
tially correlated speckle signals. IEEE Trans. Ultrason. Ferroelec. Freq. Contr.
42, 301 (1995).
49. K. Nightingale, M. S. Soo, R. Nightingale, and G. Trahey, Acoustic radiation
force impulse imaging: In vivo demonstration of clinical feasibility. Ultrasound
in Medicine and Biology28, 227 (2002).
50. M. Fatemi and J. Greenleaf, Ultrasound-stimulated vibro-acoustic spectrogra-
phy.Science 280, 82 (1998).
13
7/25/2019 Us Diagnosis of Breast Cancer
14/14
R E S E A R C H A R T I C L E J. Med. Imaging Health Inf. 3, 114, 2013
51. W. Walker, F. Fernandez, and L. Negron, A method of imaging viscoelastic
parameters with acoustic radiation force. Phys. Med. Bio.45, 437 (2000).
52. J. Bercorff, M. Tanter, and M. Fink, Sonic boom in soft materials: The elastic
Cerenkov effect. Appl. Phys. Lett. 84, 2202 (2004).
53. J. Bercorff, M. Tanter, and M. Fink, Supersonic shear imaging: A new tech-
nique for soft tissue elasticity mapping. IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control51, 396 (2004).
54. J. Bercoff, S. Chaffai, M. Tanter, L. Sandrin, S. Catheline, M. Fink, J. L.
Gennisson, and M. Meunier, In vivo breast tumors detection using transient
elastography. Ultrasound in Medicine and Biology29, 1387 (2003).
55. M. Tanter, J. Bercoff, A. Athanasiou, T. Deffieux, J.-L. Gennisson, G. Notaldo,
M. Muller, A. Tardivon, and M. Fink, Quantitative assessment of breast lesion
viscoelasticity: Initial clinical results using supersonic shear imaging. Ultra-
sound in Medicine and Biology34, 1373 (2008).
Received: 10 December 2012. Revised/Accepted: 12 January 2013.
14