BREAST RADIOLOGY

Comparison between different imaging techniquesin the evaluation of malignant breast lesions:can 3D ultrasound be useful?

Paola Clauser • Viviana Londero • Giuseppe Como •

Rossano Girometti • Massimo Bazzocchi •

Chiara Zuiani

Received: 6 April 2012 / Accepted: 10 January 2013

� Italian Society of Medical Radiology 2013

Abstract

Purpose This study was done to assess the feasibility of

three-dimensional ultrasonography (3D-US) for volume

calculation of solid breast lesions.

Materials and methods The volumes of 36 malignant

lesions were measured using conventional 2D-US, 3D-US

and magnetic resonance imaging (MRI) and compared with

that obtained with histology (standard of reference). With

2D Ultrasouns, volume was estimated by measuring three

diameters and calculating volume with the mathematical

formula for spheres. With 3D-US, stored images were

retrieved and boundaries of masses were manually out-

lined; volume calculation was performed with VOCAL

software. For MRI, volume measurements were obtained

with special software for 3D reconstructions, after each

lesion had been manually outlined. Histology measured the

three main diameters and the volume was estimated using

the mathematical formula for spheres. Interclass correlation

coefficient (ICC) and Bland–Altman plots were used to

assess agreement between the volumes measured.

Results ICC indicated that a good level of concordance

was identified between 3D-US and histology (0.79).

According to the Bland–Altman analysis, limits of agree-

ment of mean differences of the volumes measured with

the three imaging modalities were comparable with his-

tology: -271.5 cm3 for 3D-US; -2.371.3 cm3 for 2D-

US and -2.271.6 cm3 for MRI.

Conclusions 3D-US is a reliable method for the volu-

metric assessment of breast lesions. 3D-US is able to

provide valuable information for the preoperative evalua-

tion of lesions.

Keywords 3-Dimensional ultrasonography �Breast cancer � Breast ultrasound � Breast MRI

Introduction

Ultrasound (2D-US) in breast imaging proved to be an

accurate, available, noninvasive and inexpensive one

which is frequently used in the study of breast disease. It

can be combined with mammography especially in women

with dense breasts, in the differential diagnosis of palpable

lesions or in presence of equivocal mammographic findings

[1, 2]. However, it is an operator-dependent examination

and its quality is strongly related to the doctor’s expertise

and knowledge, with limited possibilities for standardisa-

tion [1].

When a space-occupying lesion is identified on ultra-

sound, its volume can be estimated using the three

orthogonal diameters [antero-posterior (AP), latero-lateral

(LL) and cranio-caudal (CC)] and applying the formula for

calculating the volume of a sphere; however, this may be

only an approximation when the lesion has irregular

morphology.

In the last few years, 3D ultrasound (3D-US) has been

increasingly used in clinical practice, also in breast imag-

ing [3, 4]. A 3D analysis starts with the acquisition of the

static volume of the lesion and the surrounding tissues. The

data acquired represent a scan of the area of interest with

the target lesion and surrounding tissues. Images can be

archived and then visualised and reviewed at a later date.

The possibility of saving a series of images of the lesion,

rather than a single image, makes the examination easier to

P. Clauser (&) � V. Londero � G. Como � R. Girometti �M. Bazzocchi � C. Zuiani

Institute of Diagnostic Radiology, University of Udine,

P.le Santa Maria della Misericordia 15, 33100 Udine, Italy

e-mail: clauser.p@hotmail.it

123

Radiol med

DOI 10.1007/s11547-013-0338-z

review and less operator-dependent. The lesion can be

subsequently measured by tracing its borders manually,

automatically or semi-automatically and, on the basis of

these measurements, volume can be calculated using spe-

cific software programmes (i.e. VOCAL, visual organ

computer-aided analysis) [5]. Currently, the bulkiness of

3D probes with square or circular arrays makes their use

difficult in everyday clinical practice: 2D-US is still irre-

placeable for the identification of suspicious lesions.

Magnetic resonance imaging (MRI) of the breast is a

very good technique for the detection and staging of dis-

ease prior to surgery in women with breast cancer [6]. It

allows for accurate volume measurement of each lesion

both in the preoperative setting and in the follow-up of

patients undergoing neoadjuvant chemotherapy; some

studies have, however, underlined that breast MRI can

underestimate the real volume of lesions, especially small

lesions, and overestimate the volume of larger lesions [7,

8].

A precise evaluation of lesion volume could be espe-

cially useful in several clinical settings: for example, in the

preoperative evaluation of malignant lesions or in the fol-

low-up of benign or probably benign nodules (ACR BI-

RADS 2 and 3 [9]).

The aim of our study was to evaluate the reliability of

3D-US in the measurement of solid malignant breast

lesions, compared to the same evaluation with 2D-US and

breast MRI, using the measurements obtained at histopa-

thological assessment of the surgical specimen as the

standard of reference.

Materials and methods

Patients

In the period between July and September 2010, we con-

secutively evaluated all patients with a suspicious finding

at breast ultrasound (ACR BI-RADS 4 and 5 [9]) who

underwent 14-gauge core-needle biopsy with a histological

diagnosis of malignancy. All women gave their informed

consent and the study obtained the approval of the local

Ethics Committee.

We evaluated 36 lesions in 34 patients aged between 39

and 89 years (mean age 60.0 years). All lesions included in

the study appeared at MRI as an area of mass-like

enhancement. MRI was performed after biopsy in 27

lesions (75.0 %) and before biopsy in nine (25.0 %). In

these nine women, the biopsy was performed because of

suspicious findings on MRI.

Images obtained with a 3D probe were collected for 26

lesions (72.0 %) before biopsy and for 10 lesions (28.0 %)

when the patient came for positioning of the needle-wire,

about 1 month after biopsy. Histology revealed 28 invasive

ductal carcinomas (IDC) (9 grade I, 15 grade II and 4 grade

III, one of which with a widespread intraductal compo-

nent), one high-grade ductal carcinoma in situ (DCIS), two

intermediate grade DCIS with a lobular component, two

grade-II invasive lobular carcinomas (ILC), two mucinous

carcinomas (one grade I and the other grade II), and one

papillary carcinoma. These findings were confirmed by

analysis of the surgical specimen (23 quadrantectomies and

13 mastectomies).

Imaging evaluation

Ultrasound

All patients underwent 2D-US, performed by a resident

following specific training in breast imaging and by a

radiologist experienced in breast imaging, with the aim of

locating and characterising lesions, and measuring the three

diameters (antero-posterior, latero-lateral and cranio-cau-

dal). The examination was performed with a Logiq E9

ultrasound (GE Healthcare, Milwaukee, USA) with an ML

6–15 MHz matrix–array transducer. Then, before the

biopsy or while positioning the wire, the resident acquired

the volumetric images on the same ultrasound unit using a

6–16 MHz 3D–4D RSP square–array probe, which auto-

matically scans the area of interest. This probe is able to

obtain 3D images on an area with a maximum diameter of

4.0 cm; lesions larger than 4.0 cm cannot be correctly

assessed.

Breast magnetic resonance imaging

Magnetic resonance imaging was performed on a 1.5-T

system (Magnetom Avanto, Siemens Medical System, Er-

langen, Germany) using a dedicated bilateral multichannel

coil. Patients were studied in prone position.

Axial T1-weighted images were obtained with a 3D

FLASH sequence with the following parameters: TR/TE,

9/4.7 ms; flip angle, 25�; matrix, 512 9 512; field of view,

340 9 340 mm; slice thickness, 2 mm; acquisition time,

80 s. Gadobenate dimeglumine 0.5 M (Multihance, Brac-

co, Milan, Italy) was used as a contrast medium. This was

given at a dose of 0.1 ml/kg of body weight injected with

an automatic pump at a rate of 2 ml/s, followed by 20 ml

of saline solution. A series of dynamic images were

acquired before contrast medium injection and five times

after administration. At the end of the examination, images

underwent post-processing with subtraction of pre-contrast

images from post-contrast images, multiplanar recon-

struction (MPR) and maximum intensity projection (MIP).

Dynamic curves of enhancement were obtained by posi-

tioning a region of interest (ROI) on the suspicious area.

Radiol med

123

T2-weighted STIR images were acquired with the follow-

ing parameters: TR/TE, 5930/73 ms; TI, 150 ms; flip

angle, 150�; matrix, 384 9 230; slice thickness, 3 mm;

field of view, 320x320; distance factor, 0.6; no. of aver-

ages, 1; oversampling, 7; acquisition time, 239 s.

Volume measurement with 3D-US

Volumes were calculated at the end of the session by the

same resident who performed the 3D–4D acquisition.

Specific software was used (VOCAL, virtual organ com-

puter-aided analysis, 4D View; GE Healthcare, Kretz-

technik, Zipf, Austria). Of the three planes available for

image display [axial (A), sagittal, (B), and coronal (C)], we

selected plane A to trace the lesion margins. The image is

turned 360� using variable rotation intervals, which can be

chosen by the operator. Intervals of 30� were used and the

borders were traced for each of the six images visualised.

At the end of the process, the software automatically cal-

culates the volume and provides a surface rendering

reconstruction of the area (Fig. 1).

Volume measurement with magnetic resonance

imaging

Volumes were calculated using dedicated software for 3D

reconstruction (Vitrea 2-Vital Images, Plymouth, Minne-

sota, USA). The same trainee radiologist who measured the

volumes with 3D-US traced the lesion borders in each axial

plane, using the first acquisition of subtracted post-contrast

images.

Volume evaluation based on the three diameters

On the images acquired with breast MRI, the cranio-cau-

dal, latero-lateral and anteroposterior diameters of each

lesion were measured, and the mean diameter was calcu-

lated. The three diameters defined by the pathologist on

assessment of the surgical specimen were used. Volume

was calculated using the mathematical formula for sphere

volume (V = 4/3pr3, where radius (r) is intended as half of

the mean diameter).

Statistical analysis

All evaluations were done with MedCalc software v.

9.1.0.1 and a Microsoft Excel 2003 spreadsheet (Microsoft

Corporation, Redmond, WA).

To preliminarily evaluate if the imaging and histological

data were comparable, the Student t test was applied on the

largest diameter measured by 2D-US, MRI and surgical

specimens. The same test was then used to identify a sig-

nificant difference in volumes evaluated with imaging and

at histopathologic examination.

Concordance was evaluated between the volumes cal-

culated by: 3D-US and 2D-US; 3D-US and MRI and

between those three imaging modalities and histology

using the interclass correlation coefficient (ICC). ICC was

also calculated after dividing the lesions into two groups:

\1 cm3 or C1 cm3 at histological examination. Finally,

the Bland–Altman analysis was used to assess the accuracy

of the techniques in estimating volume compared to his-

tology [10].

Fig. 1 Image obtained with the volumetric 3D-4D probe of a lesion

considered suspicious for malignancy. Biopsy identified an invasive

ductal carcinoma grade 3 (a). The figure is visualised in three planes;

the axial and longitudinal planes can also be obtained with a 2D

probe, while the third plane is a reconstruction in the coronal plane.

After the operator traces the borders of the lesion, dedicated software

(VOCAL, virtual organ computer-aided diagnosis) calculates the

volume and represents it with a surface rendering modality (b)

Radiol med

123

Results

The maximum diameters of the lesions considered are

described in Table 1, in terms of range, mean and standard

deviation. After each volume had been measured with 2D-

and 3D-ultrasound, MRI and histology, range, mean and

standard deviation were calculated (Table 2). The Student

t test found no significant differences between the maxi-

mum diameter measured by MRI and that identified on the

surgical specimen (p [ 0.05), whereas it did show a sta-

tistically significant difference between histology and 2D-

US (p \ 0.05). The mean volume measured by the three

imaging modalities was lower than that determined on the

surgical specimens, but the Student t test found no signif-

icant differences (p [ 0.05) (an example is shown in

Fig. 2).

We then evaluated concordance between the volume

measured with 3D-US and the other imaging modalities

using ICC. Values showed a good concordance between

3D-US and 2D-US (0.85) and between 3D-US and MRI

(0.82). Concordance between imaging and histology was

also good, with 0.79 for 3D-US, 0.82 for 2D-US and 0.79

for MRI, respectively.

Lesions with a volume \1 cm3 were slightly overesti-

mated by 3D-US (16/20 cases), while 12 of 16 lesions with a

volume C1 cm3 were underestimated. Breast MRI, on the

other hand, underestimated the dimensions defined by his-

tology in 19/36 cases (52.7 %), and overestimated 13/36

cases (36.1 %), without differences related to lesion volume.

In the other four cases (11.2 %), volumes measured with

breast MRI and determined on the basis of histology were

the same in four patients. Concordance between 3D-US and

histology was better for lesions C1 cm3 (ICC = 0.84).

Bland–Altman analysis showed mean differences and

95 % limits of agreement of: -0.22 cm3 (0.72 to

-1.17 cm3) between 3D-US and 2D-US (Fig. 3); 0.02 cm3

(1.26 to -1.21 cm3) between 3D-US and MRI (Fig. 4).

Comparison with histology showed mean differences and

95 % limits of agreement of: -0.5 cm3 (1.3 to -2.3 cm3)

for 2D-US (Fig. 5); -0.3 cm3 (1.5 to -2.0 cm3) for 3D-US

(Fig. 6) and -0.3 cm3 (1.6 to -2.2 cm3) for MRI (Fig. 7).

Volumetric scan of the area of interest of the breast

required 3–6 s, depending on lesion size. Image processing

performed by a single operator to calculate volumes took

2–4 min, depending on dimensions and morphological

features of the area.

Discussion

2D-US is an easy and useful approach, which is irre-

placeable in recognising and characterising breast lesions.

Once identified, the lesion can be studied with a 3D probe

and the area of interest can be acquired and stored with a

volumetric acquisition that includes the lesion and a small

portion of surrounding tissues. This can be achieved with a

liner-array probe that scans the area of interest, with or

without systems to localise (tracked freehand systems or

untracked freehand systems), obtaining 2D-images from

which volumetric images are reconstructed [4].

As an alternative, one can use a probe with a square or

circular 2D-matrix (two-dimensional arrays) that is kept

still while it automatically scans the area of interest to give

3D real-time information [4]. In our study, we used a

square two-dimensional array. The images acquired can be

presented with surface rendering (showing the acquired

volume with 3D appearance), multiplanar reformatting

(showing the three perpendicular planes, with the possi-

bility to move within the slices acquired) or volume ren-

dering [4]. Once recorded, the images can be re-evaluated

and re-processed at a later date (even by more than one

operator). This is not possible with the static 2D images

obtained with 2D ultrasound, so that 3D-US proves to be

less operator dependent. Moreover, 3D images allow the

analysis of lesion morphology and dimension, with the

identification of characteristic patterns [11–14]. Several

studies comparing the sensitivity, specificity, predictive

values and accuracy of 3D-US with those 2D imaging in

the differential diagnosis of breast masses showed that 3D-

US does not improve diagnostic accuracy [14]. Cho [14]

found a high level of concordance between different

operators in identifying a lesion as benign or malignant

using 3D probes, and concluded that this technique could

improve diagnostic confidence and reproducibility. More

Table 1 Range, mean and standard deviation of the greatest diameter

of the malignant breast lesions examined

Range (cm) Mean (cm) Standard

deviation (cm)

2D-US 0.4–3.4 1.4 0.6

MRI 0.6–3.4 1.5 0.6

Histology 0.5–3.5 1.5 0.7

Table 2 Volumes obtained with 2D ultrasound, 3D ultrasound and

breast magnetic resonance imaging compared to the measurements

obtained on histological examination

Range (cm3) Mean (cm3) Standard

deviation (cm3)

2D-US 0.1–10.2 1.0 1.8

3D-US 0.1–8.7 1.2 1.6

MRI 0.1–8.9 1.2 1.6

Histology 0.1–9.5 1.5 1.9

The table shows the range, mean and standard deviation of the

measurements

Radiol med

123

than one study underlined the usefulness of the coronal

plane (C), which is perpendicular to the two planes usually

visualised on 2D ultrasound. This plane is able to give

more information on the involvement of surrounding tis-

sues and better depicts spiculations, if present [11–13],

even though no study to date has established whether this

information actually improves diagnostic accuracy (Fig. 8).

Several studies have emphasised the limitations of vol-

ume calculations based on the three diameters both for

organs and phantoms: in fact, this method tends to under-

estimate the real volume [15, 16]. Similarly, in our study

we found that the maximum diameter measured with 2D-

US was significantly smaller than that measured by the

pathologist. The mathematical estimation of volume has

also lower intra- and inter-operator reproducibility.

Many authors have shown that reproducibility in volume

measurement is better when using a volumetric probe, with

an inter-operator concordance of almost 100 %, for

example, in the study of thyroid or focal liver lesions [17,

19]. The use of ultrasound with 2D probes has also been

proposed for measuring renal volume in healthy subjects

and in patients with impaired renal function: the accuracy,

Fig. 2 a US showed a hypoechoic lesion with undefined margins and

a suspicious appearance. Diagnosis after biopsy was an invasive

ductal carcinoma G2. The three diameters (antero-posterior, latero-

lateral and cranio-caudal) were initially measured on the images

obtained with 2D-US. b An image of the lesion was acquired with the

volumetric 3D–4D probe before biopsy. Once the perimeter of the

interested area has been manually drawn, the VOCAL software

automatically calculates the volume. c Breast MR acquired in the

early phase after contrast injection shows an area of inhomogeneous,

intense and early enhancement, with undefined margins, suspicious

for malignancy. d Volume is calculated on breast MR with dedicated

software (Vitrea 2-Vital Images, Plymouth, MN, USA), after drawing

the area of the lesions in all the slices. Volume measured with 2D-US

was smaller than that calculated by 3D-US and MR (2.2, 3.1 and

3.3 cm3, respectively). Histology of the lesion after surgery confirmed

the dimensions identified by 3D-US and breast MR imaging

Radiol med

123

reproducibility and correlation with other techniques such

as computed tomography were interesting [15, 16].

The limits of measuring the three diameters are partic-

ularly evident for masses or organs with irregular margins

and morphology [17, 18], which is a typical aspect of

malignant breast lesions [11].

Acquiring 3D images requires only a few seconds

(3–6 s), and overall the examination is extended by no

more than 1 min, necessary to activate the probe and

identify the lesion. This operation entails a short learning

curve: the operator has to become familiar with a slightly

different image representation from that of the 2D probe.

The volume measurement can be performed at the end of

the session after the acquisition, by a different operator and

by more than one operator, as can the morphological

evaluation of the lesion and surrounding tissues. Volume

measurement requires a maximum of 4 min, especially in

lesions with a greater diameter and undefined margins.

The VOCAL software used in our study has already

been reported to have good accuracy in the volumetric

measurement of irregular lesions, especially when margins

are traced manually and not automatically [5]. Its reliability

improves with the number of planes used to describe the

margins of the area of interest, but at the same time

increases the time needed for the analysis. On the basis of a

study by Pang et al. [20] as well as our own experience, we

decided to use a 30� angle, which appears to be a good

compromise, as it is simple, with no significant differences

with respect to measurements obtained at 6�, 9� or 15�, but

at the same time less time-consuming.

Fig. 3 Graph showing the results of Bland–Altman analysis applied

to the comparison between the volumes measured with 2D and 3D

ultrasound

Fig. 4 Graph showing the results of Bland–Altman analysis applied

to the comparison of the volumes measured with 3D ultrasound and

breast MR imaging

Fig. 5 Graph showing the results of Bland–Altman analysis applied

to the comparison of the volumes measured with 2D ultrasound and

histology

Fig. 6 Graph showing the results of Bland–Altman analysis applied

in the comparison of the volumes measured with 3D ultrasound and

histology

Radiol med

123

It has already been underlined that MRI can be superior

in describing lesion characteristics and extension: for this

reason, we decided to compare also the volumes obtained

with this technique [7, 8, 21], even though our standard of

reference has been histology of the specimen.

In the literature, we have found only one article [22]

comparing the volumes of benign and malignant breast

lesions measured with 2D-US, 3D-US and mammography,

and the authors reported a good correlation.

We are not aware of studies that evaluated the corre-

spondence of 2D-US, 3D-US, MRI and histology for

malignant breast lesions. The study of volume of breast

lesions with MRI has already shown a good reproducibility

[21, 23]. MRI does, however, tend to underestimate small

lesions and overestimate nodules with a greater diameter

[21]. In our experience, MRI slightly underestimated vol-

umes compared to histology, and that was not related to the

dimensions determined on the basis of the histology mea-

surements. MRI underestimated volume in 19 cases of 36

(52.7) and overestimated it in 13 of 36 lesions (36.1 %).

Interclass correlation coefficient showed a good con-

cordance between volumes measured with different imag-

ing modalities, not only for 3D and 2D ultrasound but also

for 3D ultrasound and MRI. Also, the concordance with

histology was satisfactory both for ultrasound (2D and 3D)

and for MRI. In particular, 3D-US had an ICC with

Fig. 7 Graph showing the results of Bland–Altman analysis applied

to the comparison of the volumes measured with breast MR Imaging

and histology

Fig. 8 a US showed a hypoechoic lesion. The histological diagnosis

was invasive ductal carcinoma grade 2. b The acquisition and

reconstruction in three planes allowed for a better identification of the

irregular borders and the distortion of the perilesional tissue. c, d In

this case the volume calculated with the three methods was

comparable to that obtained by histology after surgery (2D-

US = 1.2 cm3; 3D-US = 1.05 cm3; breast MR = 0.87 cm3;

histology = 0.9 cm3)

Radiol med

123

histology comparable to that of MRI, which is considered

the standard of reference among imaging modalities for a

correct evaluation of lesion extension. This correlation has

not always been underlined in literature: in a study by

Londero et al. [7], 2D-ultrasound underestimated tumour

diameter during the follow-up of patients treated with

neoadjuvant chemotherapy, while MRI showed a better

performance with a slight tendency to overestimation.

However, the particular situation and hence the modifica-

tions related to chemotherapy could probably explain the

limited ability of US in identifying post-therapy changes,

especially compared to MRI, which is able to visualise

contrast-enhancement only where there is vascularisation.

The 95 % limits of agreement calculated with Bland–Alt-

man analysis between imaging and histology overlapped

for all modalities, proving that 3D-US is at least compa-

rable to other imaging techniques in evaluating the volume

of malignant breast lesions.

In the literature, other possible applications of 3D-US

can be found. For example, its use in interventional pro-

cedures: 3D images allow a better evaluation of needle

position, thus improving the accuracy of tissue sampling

and reducing the number of cores necessary [24, 25].

In addition, the possibility of using CAD (computer-

aided diagnosis) on volumetric images acquired with 3D-

US has already been evaluated, with the aim of improving

sensitivity and specificity in the identification of lesions,

especially those classified as BI-RADS 3 [26].

Conclusions

On the basis of our initial experience, 3D ultrasound

appears to be a reliable and easy-to-use method which is at

least as accurate as 2D ultrasound and MRI in measuring

volumes of malignant breast lesions. 3D-US is fast and

practical to use and allows memorisation of volumetric

images that can be processed and re-evaluated, even at a

later date and by more than one operator.

Conflict of interest Paola Clauser, Viviana Londero, Giuseppe

Como, Rossano Girometti, Massimo Bazzocchi, Chiara Zuiani

declare no conflict of interest.

References

1. Zonderland HM, Coerkamp EG, Hermans J et al (1999) Diag-

nosis of breast cancer: contribution of US as an adjunct to

mammography. Radiology 213:413–422

2. Kolb TM, Lichy J, Newhouse JH (1998) Occult cancer in women

with dense breasts: detection with screening US-diagnostic yield

and tumor characteristics. Radiology 207:191–199

3. Weismann CF, Datz L (2007) Diagnostic algorithm: how to make

use of new 2D, 3D and 4D ultrasound technologies in breast

imaging. Eur J Radiol 64:250–257

4. Downey DB, Fenster A, Williams JC (2000) Clinical utility of

three-dimensional US. Radiographics 20:559–571

5. de Sa Barreto EQ, Figuinha Milani HJ, Araujo Junior E et al

(2010) Reliability and validity of in vitro volume calculation by

3-dimensional ultrasonography using the multiplanar, virtual

organ computer-aided analysis (VOCAL), and extended imaging

VOCAL methods. J Ultrasound Med 29:767–774

6. Esserman L, Hylton N, Yassa L et al (1999) Utility of magnetic

resonance imaging in the management of breast cancer: evidence

for improved preoperative staging. J Clin Oncol 17:110–119

7. Londero V, Bazzocchi M, Del Frate C et al (2004) Locally

advanced breast cancer: comparison of mammography, sonog-

raphy and MR imaging in evaluation of residual disease in

women receiving neoadjuvant chemotherapy. Eur Radiol 14:

1371–1379

8. Lorenzon M, Zuiani C, Londero V et al (2009) Assessment of

breast cancer response to neoadjuvant chemotherapy: is volu-

metric MRI a reliable tool? Eur J Radiol 71:82–88

9. D’Orsi CJ, Mendelson EB, Ikeda DM et al (2003) Breast imaging

reporting and data system: ACR BI-RADS—Breast Imaging

Atlas. American College of Radiology, Reston

10. Bland JM, Altman DG (1986) Statistical methods for assessing

agreement between two methods of clinical measurement. Lancet

1:307–310

11. Rotten D, Levaillant JM, Zerat L (1999) Analysis of normal

breast tissue and of solid breast masses using three-dimensional

ultrasound mammography. Ultrasound Obstet Gynecol 14:

114–124

12. Watermann DO, Foldi A, Beck Hanjalic et al (2005) Three-

dimensional ultrasound for the assessment of breast lesions.

Ultrasound Obstet Gynecol 25:592–598

13. Meyberg-Solomayer GC, Kraemer B, Bergmann A et al (2004)

Does 3-D sonography bring any advantage to non-invasive breast

diagnostics? Ultrasound Med Biol 30:583–589

14. Cho N, Moon WK, Cha JH et al (2006) Differentiating benign

from malignant solid breast masses: comparison of two-dimen-

sional and three-dimensional US. Radiology 240:26–32

15. Kim HC, Yang DM, Lee SH et al (2008) Usefulness of renal

volume measurements obtained by a 3-dimensional sonographic

transducer with matrix electronic arrays. J Ultrasound Med

27:1673–1681

16. Kim HC, Yang DM, Jin W et al (2010) Relation between total

renal volume and renal function: usefulness of 3D sonographic

measurements with matrix array transducer. AJR Am J Roent-

genol 194:W186–W192

17. Ng E, Chen T, Lam R et al (2004) Three dimensional ultrasound

measurement of thyroid volume in asymptomatic male Chinese.

Ultrasound Med Biol 30:1427–1433

18. Ying M, Sin M, Pang S (2005) Sonographic measurement of

thyroid gland volume: a comparison of 2D and 3D ultrasound.

Radiography 11:242–248

19. Lang H, Wolf GK, Prokop M et al (1999) 3-Dimensional

sonography for volume determination of liver tumors—report of

initial experiences. Der Chirurg 70(3):246–250

20. Pang BSF, Kot BCW, Ying M (2006) Three-dimensional ultra-

sound volumetric measurements: is the largest number of image

planes necessary for outlining the region-of-interest? Ultrasound

Med Biol 32:1193–1202

21. Rominger MB, Fournell D, Nadar BT et al (2009) Accuracy of

MRI volume measurements of breast lesions: comparison

between automated, semiautomated and manual assessment. Eur

Radiol 19:1097–1107

22. Kotsianos-Hermle D, Hiltawsky KM, Wirth S et al (2009) Ana-

lysis of 107 breast lesions with automated 3D ultrasound and

comparison with mammography and manual ultrasound. Eur J

Radiol 71:109–115

Radiol med

123

23. Levrini G, Sghedoni R, Mori C et al (2011) Size assessment of

breast lesions by means of a computer-aided detection (CAD)

system for magnetic resonance mammography. Radiol Med

116:1039–1049

24. Wiesmann CF, Forstner R, Prokop E et al (2000) Three-dimen-

sional targeting: a new three-dimensional ultrasound technique to

evaluate needle position during breast biopsy. Ultrasound Obstet

Gynecol 16:359–364

25. Delle Chiaie L, Terinde R (2004) Three-dimensional ultrasound-

validated large-core needle biopsy: is it a reliable method for the

histological assessment of breast lesions? Ultrasound Obstet

Gynecol 23:393–397

26. Sahiner B, Chan HP, Roubidoux MA et al (2007) Malignant and

benign breast masses on 3D US volumetric images: effect of

computer-aided diagnosis on radiologist accuracy. Radiology

242:716–724

Radiol med

123