The Automated Breast Volume Scanner (ABVS): initial experiences in lesion detection compared with conventional handheld B-mode ultrasound: a pilot study of 50 cases.

Abstract

The idea of an automated whole breast ultrasound was developed three decades ago. We present our initial experiences with the latest technical advance in this technique, the automated breast volume scanner (ABVS) ACUSON S2000(™). Volume data sets were collected from 50 patients and a database containing 23 women with no detectable lesions in conventional ultrasound (BI-RADS(®)-US 1), 13 women with clearly benign lesions (BI-RADS(®)-US 2), and 14 women with known breast cancer (BI-RADS(®)-US 5) was created. An independent examiner evaluated the ABVS data on a separate workstation without any prior knowledge of the patients' histories. The diagnostic accuracy for the experimental ABVS was 66.0% (95% confidence interval [CI]: 52.9-79.1). The independent examiner detected all breast cancers in the volume data resulting in a calculated sensitivity of 100% in the described setting (95% CI: 73.2%-100%). After the ABVS examination, there were a high number of requests for second-look ultrasounds in 47% (95% CI: 30.9-63.5) of the healthy women (with either a clearly benign lesion or no breast lesions at all in conventional handheld ultrasound). Therefore, the specificity remained at 52.8% (95% CI: 35.7-69.2). When comparing the concordance of the ABVS with the gold standard (conventional handheld ultrasound), Cohen's Kappa value as an estimation of the inter-rater reliability was κ = 0.37, indicating fair agreement. In conclusion, the ABVS must still be regarded as an experimental technique for breast ultrasound, which definitely needs to undergo further evaluation studies.

Full text

Page 1

© 2011 Wojcinski et al, publisher and licensee Dove Medical Press Ltd. This is an Open Access article
which permits unrestricted noncommercial use, provided the original work is properly cited.
International Journal of Women’s Health 2011:3 337–346
International Journal of Women’s HealthDovepress
submit your manuscript | www.dovepress.com
Dovepress
337
O r I g I n A L r e s e A r c H
open access to scientific and medical research
Open Access Full Text Article
The Automated Breast Volume scanner (ABVs):
initial experiences in lesion detection compared
with conventional handheld B-mode ultrasound:
a pilot study of 50 cases
http://dx.doi.org/10.2147/IJWH.S23918
sebastian Wojcinski1
Andre Farrokh1
Ursula Hille2
Jakub Wiskirchen3
samuel gyapong1
Amr A soliman1,4
Friedrich Degenhardt1
Peter Hillemanns2
1Department of OB/gYn, Franziskus
Hospital, Bielefeld, germany; 2Department
of OB/gYn, Hannover Medical school,
Hannover, germany; 3Department of
radiology, Franziskus Hospital, Bielefeld,
germany; 4Department of OB/gYn, Faculty
of Medicine, University of Alexandria,
Alexandria, egypt
correspondence: sebastian Wojcinski
Franziskus Hospital, Department
of OB/gYn, Kisker str 27, 33615
Bielefeld, germany
Tel +49 521 589 1501
Fax +49 521 589 1504
email s@wojcinski.de
Abstract: The idea of an automated whole breast ultrasound was developed three decades ago.
We present our initial experiences with the latest technical advance in this technique, the automated
breast volume scanner (ABVS) ACUSON S2000TM. Volume data sets were collected from 50
patients and a database containing 23 women with no detectable lesions in conventional ultrasound
(BI-RADS®-US 1), 13 women with clearly benign lesions (BI-RADS®-US 2), and 14 women with
known breast cancer (BI-RADS®-US 5) was created. An independent examiner evaluated the ABVS
data on a separate workstation without any prior knowledge of the patients’ histories. The diagnostic
accuracy for the experimental ABVS was 66.0% (95% confidence interval [CI]: 52.9–79.1). The
independent examiner detected all breast cancers in the volume data resulting in a calculated sensi-
tivity of 100% in the described setting (95% CI: 73.2%–100%). After the ABVS examination, there
were a high number of requests for second-look ultrasounds in 47% (95% CI: 30.9–63.5) of the
healthy women (with either a clearly benign lesion or no breast lesions at all in conventional handheld
ultrasound). Therefore, the specificity remained at 52.8% (95% CI: 35.7–69.2). When comparing
the concordance of the ABVS with the gold standard (conventional handheld ultrasound), Cohen’s
Kappa value as an estimation of the inter-rater reliability was κ = 0.37, indicating fair agreement.
In conclusion, the ABVS must still be regarded as an experimental technique for breast ultrasound,
which definitely needs to undergo further evaluation studies.
Keywords: breast cancer, automated breast ultrasound, automated breast volume scanner,
ABVS
Introduction
Breast ultrasound is a commonly accepted and reliable diagnostic method for women
with clinically or radiologically suspicious breast lesions.1 Although it is not estab-
lished as a routine part of screening programs, bilateral whole breast ultrasound has
even demonstrated diagnostic advantages in screening asymptomatic women.2–6 Breast
ultrasound has been performed for more than 50 years.7,8 Conventional handheld
ultrasound (HHUS) is the gold standard for performing the examination. Obviously,
sonography is an examiner-dependent method and the examiner has to be present at
the time of image acquisition.
The concept of automated breast ultrasound dates back to the 1970s when the first
applicable systems were reported by Maturo et al.9 In the current report we present our
initial experiences with the latest technical advance in automated breast ultrasound,
the Automated Breast Volume Scanner (ACUSON S2000TM ABVS; Siemens Medical
Video abstract
Point your SmartPhone at the code above. If you have a
QR code reader the video abstract will appear. Or use:
http://dvpr.es/wojcinski
Number of times this article has been viewed
This article was published in the following Dove Press journal:
International Journal of Women’s Health
10 October 2011

Page 2

International Journal of Women’s Health 2011:3
submit your manuscript | www.dovepress.com
Dovepress
Dovepress
338
Wojcinski et al
Solutions, Inc, Mountain View, CA). This system can be
operated by an assistant medical technician or radiographer.
The ABVS acquires a whole series of consecutive B-mode
pictures and reconstructs 3D data sets of the entire breast
volume. These data can be sent to a separate workstation to
be independently analyzed by a specialist.
We evaluated whether or not breast lesions, previously
detected by means of conventional ultrasound, could also be
detected and correctly classified by an independent examiner
who used only ABVS data. To the best of our knowledge, this
is one of the first descriptive studies dealing with the clinical
application of the ABVS system.
Materials and methods
general design and creation
of the patient database
Our study was conducted at the Breast Cancer Center of
Franziskus Hospital in Bielefeld, Germany, between March
2010 and May 2010. The responsible ethical committee did
not require additional approval for this noninterventional
study design. All examinations were performed using the
Siemens ACUSON S2000TM ultrasound system (Siemens
Medical Solutions, Inc, Mountain View, CA).
The study cohort was recruited from patients who
attended the outpatient service of our breast cancer center.
The patients were referred to our clinic due to specific diag-
nostic queries such as palpable breast lesions, breast pain,
suspicious mammograms, and intensified screening in high-
risk populations. All patients received conventional HHUS as
the standard diagnostic method and subsequent examinations
whenever necessary. For each patient, the B-mode ultrasound
pictures were categorized according to the Breast Imaging
Reporting and Data System criteria of the American College
of Radiology (ACR BI-RADS®-US).10
Patients with a final categorization of BI-RADS®-US 1,
2, and 5 in the conventional ultrasound examination were
regarded as being suitable for our study. We excluded patients
with BI-RADS®-US 0, 3 and 4 lesions as the focus of our
study was on the bare detection of clearly benign and clearly
malignant lesions and not on the evaluation of questionable
lesions.
Patients with a history of breast surgery, bra cup sizes
greater than D, inflammatory conditions of the breast, and
skin disorders were also excluded.
Patients who met the inclusion criteria and who agreed
to be enrolled in our study signed an informed consent
form and subsequently received the additional ABVS
examination.
Thus, we created a database containing 50 patients exhib-
iting BI-RADS®-US 1, 2, and 5 findings in conventional
examinations. In our cohort the age ranged from 32 to 72
years (median 52 years). According to the BI-RADS®-US
categorization, 46% (n = 23) of our cases were assigned as
BI-RADS®-US 1, 26% (n = 13) had BI-RADS®-US 2 lesions,
and 28% (n = 14) of the cases had BI-RADS®-US 5 lesions.
All cases of BI-RADS®-US 5 were histologically confirmed
after the study setting was completed.
The mean tumor size for the malignant lesions was
23.4 mm (range 13 to 55).
After anonymization, the ABVS database was then
available for evaluation by an independent second examiner.
We compared the performance of the ABVS (experimental
method) to the HHUS (gold standard).
Technical background of the ABVs
The ACUSON S2000TM ABVS is an ultrasound system that
automatically surveys and acquires full-field volumes of the
breast (Figure 1).
For automated scanning, we used the integrated Siemens
14L5BV linear transducer (14 MHz, 15.4 cm) with 768
piezoelectric elements. These images have a depth of up to
6 cm. In the process of automated scanning, the transducer
covers a distance of 16.8 cm, acquiring 318 high-resolution
slices for post-processing. Therefore, the maximum volume
is 1552.3 cm³ and the slice thickness is about 0.5 mm.
In order to optimize the ABVS results there is a wide range
of known imaging modes including tissue harmonic imaging
(THI) and Advanced SieClearTM spatial compounding and
Dynamic TCETM (tissue contrast enhancement) technology,
Figure 1 Installation of the AcUsOn s2000TM ABVs ultrasound system. On the
left-hand side is the AcUsOn s2000TM machine, on the right-hand side is the
14L5BV volume transducer attached to a mechanical arm.

Page 3

International Journal of Women’s Health 2011:3
submit your manuscript | www.dovepress.com
Dovepress
Dovepress
339
The automated breast volume scanner
as well as new processing algorithms for nipple shadow and
reverberation artifacts that are automatically applied when
using the ABVS. The Reverberation Removal Algorithm
processes the 3D data and determines whether tissue contact
is present and where it is not. The data corresponding to the
area with no tissue contact are removed. This is intended to
suppress the reverberation artifacts from the noncontact areas.
The adaptive Nipple Shadow Reduction Tool analyzes data on
a case-by-case basis and is thought to enhance the structures
in the retro-areolar area and to improve the visualization of
this important region. Finally, a Gain Collection Algorithm
analyzes the 3D data and adjusts for the brightness variation
artifacts caused by transducer channel-to-channel variations.
After acquisition, the data are automatically sent from the
ultrasound system to a breast ultrasound review workstation,
which provides comprehensive analysis and manipulation of
the 3D data: the workstation presents images through multi-
planar reconstruction (MPR, Figure 2). All volume data sets
can also be viewed in multiple orientations, including the
standard transverse, sagittal, coronal, radial, and anti-radial
planes, as well as any other user-defined plane (Figure 3).
Secondary images are reconstructed from the acquisition
volume in any plane in real-time (Figure 4). A standardized
scanning technique, which was also used in our study, has been
described elsewhere.11
Standard HHUS examination (definition
of the gold standard)
Clinical history taking and clinical examination, and the
HHUS, were performed by the author SW, a DEGUM
(Deutsche Gesellschaft für Ultraschall in der Medizin,
German society for ultrasound in medicine) level II certified
senior consultant in gynecology with 7 years’ experience in
breast ultrasound.12 This examiner also knew the results of
the other imaging modalities when available (mammography,
magnetic resonance imaging [MRI]).
For HHUS we used the integrated Siemens 18L6 HD
linear transducer (5.5–18 MHz, 5.6 cm). The patient was
in the supine position with the ipsilateral hand raised
above the head. All examinations were performed with
the ultrasound probe oriented perpendicular to the chest
wall. During each examination, all necessary B-mode
pictures were first obtained according to the diagnostic
standards. All images were digitally recorded. The exam-
iner categorized the breast lesions detected according to
the ACR BI-RADS®-US classification system. These results
were defined as the gold standard (reference standard) for
comparison with the experimental method. All standard
examinations were completed before the patients were
enrolled in the study. Eligible patients were subsequently
examined using the ABVS.
Figure 2 Multi-planar reconstruction of the volume data displayed on the automated
breast volume scanner workplace.
Figure 3 coronal view of the volume data at a depth of 10.4 mm from the skin line.
The yellow spot indicates the position of the nipple.
Figure 4 This view provides the coronal (left), transverse (upper right) and sagittal
(lower right) planes. The body marker indicates that this volume was acquired at
the apex of the left breast. The yellow spot marks the position of the nipple. A
point of interest can be chosen and marked by two orthogonal lines. Then, the
corresponding cross-sections are calculated in real-time and shown in the other
planes. The images can be optimized by adjusting the magnification, brightness, and
contrast.

Page 4

International Journal of Women’s Health 2011:3
submit your manuscript | www.dovepress.com
Dovepress
Dovepress
340
Wojcinski et al
Acquisition of the ABVs data
The experimental ABVS examination was also performed
by the author, SW. For the ABVS examination, the patient
was placed in the same position as for the HHUS. A special
ultrasound lotion that provides optimal imaging results in
combination with the ABVS was applied to the skin instead
of the usual ultrasound gel (Polysonic Ultrasound Lotion,
Parker Laboratories, Inc, Fairfield, NJ).
According to other studies, we used the bra cup size to
describe the size of the breast.13 Bra cup size is determined
by measuring the horizontal chest circumference in the infra-
mammary fold, adding 5 inches (12.7 cm) and subtracting
the horizontal chest circumference at the fullest part of the
breasts. A difference of 1 inch (2.54 cm) is equivalent to an
A cup, 2 inches a B cup, and so on.14 We did not perform
this measurement in the current study, but instead asked the
patients for their bra cup size.
Depending on the bra cup size, the examiner chose the
number of scans to be taken from each side. Usually, breasts
with A-cup and B-cup sizes can be fully displayed by per-
forming two volume scans (medial and lateral, Figure 5).
In breasts with C-cup or D-cup sizes it is often necessary
to choose additional views (usually a separate view of the
apex and the axillary process of the breast, Figure 5).
As standard, there is a replaceable membrane fixed to
the transducer to assure sufficient contact with the skin of
the entire area. In order to acquire the ultrasound image
volume, the 14L5BV transducer was positioned on the
breast with slight pressure. The mechanical arm that is
part of the scanner column was locked in the chosen posi-
tion. A preset scanning program was chosen according to
the bra cup size of the breast. In addition, all previously
known ultrasound features were available in order to opti-
mize picture quality. The automated scan took between
55 and 65 seconds. There was no need for breath holding
during this time. After a particular scan was completed,
the coronal plane of the volume acquired was displayed at
the height of skin level. This enabled the examiner to set
a fixed marker on the nipple, allowing orientation in the
subsequent analysis. Finally, the entire set of volume scans
was sent to the workstation.
Independent interpretation of the
experimental ABVs data (by examiner 2)
The independent interpretation of the ABVS data sets was
performed by the co-author AF, a DEGUM level I certified
senior resident in gynecology with 4 years’ experience in
breast ultrasound.
The second examiner exclusively analyzed the 3D data
sets without prior knowledge of the patients’ histories, clini-
cal findings, or results of the other imaging modalities. The
second examiner had no information about the proportion
of BI-RADS®-US 1, 2, and 5 cases in the database, but
he did know that BI-RADS®-US 3 and 4 cases had been
excluded.
This examiner was able to use a variety of tools for image
manipulation, including the standard views (axial, sagittal,
coronal, radial, anti-radial), user-defined views, rotation
around x, y, z axes, free rotation around any point of interest,
a magnifier and interactive zoom, marking and annotation of
areas of interest, and the generation of snapshots.
The second examiner used the following standard proce-
dure to analyze the ABVS data sets: first, the whole volume
was analyzed in the coronal plane moving slowly from the
skin to the chest wall. Suspicious lesions were marked with
the system’s default tool. In the next step, all lesions were
evaluated by generally re-examining them in the sagittal
and axial planes (and optionally in any other plane), using
adequate magnification, brightness, and contrast. Finally,
the examiner moved through the whole volume in the sagit-
tal and axial planes in order to potentially detect additional
lesions that were not seen in the coronal plane. Lastly, the
examiner assigned the lesions a category according to the
ACR BI-RADS®-US system.
Despite the fact that the second examiner knew that there
were no BI-RADS®-US 0, 3, or 4 cases in the database, he
was allowed to categorize lesions as BI-RADS®-US 0, 3, or
4 whenever he requested a second-look ultrasound in order
to scrutinize suspicious lesions.
When a second-look ultrasound was requested for a lesion
that eventually turned out to be benign, the result of the AVBS
examination was defined as “nonconcordant” and “false-
positive”. On the other hand, when a second-look ultrasound
was requested for a lesion that turned out to be malignant,
the result was classified as “true-positive” because the cancer
could then be correctly detected in the subsequent conventional
ultrasound.
Figure 5 Predefined positions of the scanner, which are used to cover the entire
volume of an individual breast.

Page 5

International Journal of Women’s Health 2011:3
submit your manuscript | www.dovepress.com
Dovepress
Dovepress
341
The automated breast volume scanner
Ultrasound quality assurance
Due to the national regulating authority statutes, all breast
ultrasound systems have to fulfill basic technical requirements
and undergo regular quality control measures.15 The technical
requirements include a linear transducer covering a least
38 mm, an ultrasound frequency $7 MHz, an adjustable
digital focus, a time gain control signal generator, a frame rate
$15 fps, an image depth $6 cm, the detection of a returning
echo $ 60 dB below the transmitted signal, and at least 8-bit
processing of the images.
In order to guarantee long-lasting performance of the
transducer and conformance of the whole device with the
standards, a compliance test is mandatory when setting up a
new ultrasound system and a consistency inspection is needed
every 4 years. All of the abovementioned standards applied
to the equipment used in our study.
statistical analysis
The software package SPSS Statistics (v 17.0; SPSS, Inc,
Chicago, IL) was used for the statistical analysis. The results
obtained from the experimental method (ABVS) were com-
pared with the results from the gold standard (HHUS). The
data were analyzed using descriptive statistics. The diagnostic
sensitivity and specificity, as well as the accuracy of the ABVS,
were calculated based on the Bayesian theorem using Fisher’s
exact test.
The Chi-square test was used to assess the correlation
between the expected and the observed rate of second-look
ultrasounds.
Calculation of the concordance with the BI-RADS®
classification was based on Cohen’s Kappa test.16 Although
both examiners (for both HHUS and ABVS) had differ-
ent degrees of information about the patients, this test can
provide a satisfactory estimation of the true inter-rater
reliability. We used the magnitude guidelines published by
Landis and Koch, who characterized the values of κ , 0 as
indicating no agreement and κ 0–0.20 slight, κ 0.21–0.40
fair, κ 0.41–0.60 moderate, κ 0.61–0.80 substantial, and κ
0.81–1 as almost perfect agreement.17
Statistical significance was assumed at P , 0.05 for all
tests.
Results
Accuracy, sensitivity, and specificity
of the ABVs
The sensitivity of the ABVS in detecting malignant lesions
ranged from 73.2 to 100% (95% CI), and all of the 14
cancer patients were detected by the second examiner.
The diagnostic accuracy of the experimental ABVS was
66.0% (95% CI: 52.9–79.1). The specificity seemed to
be quite low at 52.8% (95% CI: 35.7–69.2) as there was
a high number of requests for second-look ultrasounds
after the ABVS and these cases are, as a matter of fact,
false-positive if there is no cancer. These results must be
carefully interpreted, bearing in mind that the experimental
ABVS was performed on a subset of patients with results
defined by the HHUS.
rate of second-look ultrasounds
We expected 14 (28%) requests for second-look ultrasound
after the ABVS examination as there were 14 cases of
BI-RADS®-US 5 lesions. We did not expect requests for
the other 36 cases (BI-RADS®-US 1 or 2 lesions). Nev-
ertheless, the observed rate of second-look ultrasounds
was significantly high, totaling 62% (95% CI: 48.6–75.5,
P , 10−8).
Regarding the subgroups, there was a request for a
second-look ultrasound in 61.5% (95% CI: 35.1–88.0) of
the women with benign lesions and even in 39.1% (95%
CI: 19.2–59.1) of the women with no breast lesions at all.
Overall, examiner 2 requested second-look ultrasounds
for all 14 cases of breast cancer but also in 17 of the 36
controls (47% [95% CI: 30.9–63.5]). These data are shown
in Table 1.
Table 1 Agreement between examiner 1 (gold standard) and examiner 2 (experimental ABVs data interpretation) focusing on the
correct clinical decision of whether the patient should undergo a control ultrasound due to a suspicious finding or whether the patient
should be defined as healthy as there is no suspicious lesion
Classification by examiner 1using all information and HHUS
(gold standard)
Total
Disease - (BI-RADS 1, 2)Disease + (BI-RADS 5)
Classification by examiner 2 analyzing the ABVS data (experimental method)
ABVs − (BI-rADs 1, 2)
ABVs + (BI-rADs 0, 3, 4, 5)
Total
19
17
36
0
14
14
19
31
50
Abbreviations: ABVs, automated breast volume scanner; HHUs, handheld ultrasound.