Suggestions for topics suitable for these Point/Counterpoint debates should be addressed to Colin G. Orton, Professor
Emeritus, Wayne State University, Detroit: firstname.lastname@example.org. Persons participating in Point/Counterpoint discussions are
selected for their knowledge and communicative skill. Their positions for or against a proposition may or may not
reflect their personal opinions or the positions of their employers.
Cone beam x-ray CT will be superior to digital x-ray tomosynthesis in
imaging the breast and delineating cancer
Andrew Karellas, Ph.D.
Radiology Department, University of Massachusetts Medical School, Worcester, Massachusetts 01655
(Tel: 508-856-1238, E-mail: Andrew.Karellas@umassmed.edu)
Joseph Y. Lo, Ph.D.
Radiology Department, Duke University Medical Center, Durham, North Carolina 27705
(Tel: 919-684-7763, E-mail: email@example.com)
Colin G. Orton, Ph.D., Moderator
?Received 30 October 2007; accepted for publication 30 October 2007; published 9 January 2008?
Recent advances in cone beam CT and digital x-ray tomo-
synthesis suggest that three-dimensional ?3D? systems may
soon replace conventional planar mammography as the mo-
dality of choice for imaging the breast and delineating can-
cer. Both of these new technologies exhibit clear advantages
over planar mammography but which one of these two is
most likely to dominate is debatable. This is the topic of this
Arguing for the Proposition is
Andrew Karellas, Ph.D. Dr.
Karellas received his Ph.D. in
Medical Physics from UCLA
in 1984 and is currently Pro-
fessor of Radiology in the Uni-
MA. He is a Diplomate of the
ABR in Diagnostic Radiologic
Physics and a Fellow of the
AAPM. His interests include
digital mammography, tomo-
synthesis, and tomographic and 3D imaging of the breast. He
is a member of the Medical Physics Board of Editors and
serves as a Deputy Editor, and has been the Chairman of the
AAPM Diagnostic X-ray Imaging Committee and TG 15 on
Digital Mammography for Stereotactic Localization. He is a
Past President of both the New England and Southeast Chap-
ters of the AAPM.
Arguing against the Proposi-
tion is Joseph Lo, Ph.D. Dr. Lo
received his Ph.D. in Biomedi-
cal Engineering from Duke
University, Durham, NC. Be-
tween 1993 and 1995 he was a
postdoctoral research associate
in the Department of Radiol-
ogy, Duke University Medical
Center. He is currently Assis-
tant Professor of Radiology
and Biomedical Engineering,
and serves on the faculty of the
medical physics graduate program at Duke. His interests
cover many aspects of breast cancer research including
breast tomosynthesis and CT, bioinformatics, computer-aided
diagnosis, and digital image processing.
FOR THE PROPOSITION: Andrew Karellas, Ph.D.
The general concepts of digital breast tomosynthesis
?DBT? and dedicated breast computed tomography ?DBCT?
have been known for many years, but they could not be
practically implemented without advanced flat panel detec-
tors of the type that are now used for digital radiography and
mammography.1–6Recent advances in flat panel detector
technology have provided a strong impetus for the develop-
ment of improved and computationally efficient image re-
construction algorithms for DBT and DBCT.7In parallel, re-
search and development efforts in digital mammography
have been directed toward improvements in the physical as-
pects of planar imaging of the breast. However, imaging in
planar mammography is limited by the inability to visualize
tissues in a tomographic or three-dimensional mode. There-
409409 Med. Phys. 35 „2…, February 2008 0094-2405/2008/35„2…/409/3/$23.00© 2008 Am. Assoc. Phys. Med.
fore a suspicious abnormality can be obscured by interfering
breast tissue because the three-dimensional anatomy is rep-
resented in a two-dimensional image. DBT and DBCT hold
considerable promise in overcoming the limitations of mam-
mography, particularly in dense breasts, but DBT may be
viewed as a limited tomographic extension of digital mam-
mography rather than a true tomographic and 3D imaging
modality. Breast tomosynthesis can be performed in a num-
ber of ways by varying the projection geometry, detector
characteristics, exposure technique, reconstruction algorithm,
and mode of image display. Developers of the technology
may claim unique advantages of a particular tomosynthesis
approach based on the implementation of various improve-
ments. For example, we are likely to see improvements in
radiation dose efficiency, speed of acquisition, image recon-
struction speed, and reconstruction artifacts. Despite such ad-
vances, DBT is fundamentally limited by its constraints in
the projection geometry. In DBT the tomographic slice is not
well defined, which can cause loss of resolution in the axial
direction that can affect visualization of subtle features such
as amorphous microcalcifications.
Dedicated computed tomography can image the entire
breast in a more complete tomographic approach and with
essentially isotropic resolution. This technology is in its in-
fancy and several improvements have yet to be made that
relate to parameters like voxel size, cone beam reconstruc-
tion, x-ray scatter suppression, radiation dose, and breast
coverage. Dedicated breast CT can generate true tomogra-
phic and 3D images of the breast hitherto unavailable by any
other x-ray imaging technique of the breast, and it does not
require physical compression of the breast. It is likely to be
of particular value for imaging dense breasts and breasts with
implants. Given the choice between limited tomography with
breast compression offered by tomosynthesis and full tomog-
raphy with 3D imaging of the breast without compression,
dedicated breast CT offers a more powerful alternative to
tomosynthesis. Although I am strongly in favor of continued
research on DBT, we should make an even greater commit-
ment in DBCT because of its true tomographic and 3D ca-
AGAINST THE PROPOSITION: Joseph Lo, Ph.D.
Digital tomosynthesis will replace mammography, and
soon, while breast CT will not. This strong claim is justified
because tomosynthesis ?often abbreviated as “tomo”? has all
the advantages of mammography, while providing 3D im-
ages to address mammography’s main problem of overlap-
Breast tomo is based upon modifications to existing full-
field digital mammography ?FFDM? systems. The result is
high resolution in the x-y plane parallel to the compression
paddle, with lower but acceptable resolution ?e.g., 1 mm? in
the z, or depth, direction. In comparison, breast CT resolu-
tion within each slice is likely to be several times worse,
possibly affecting the ability to detect and characterize cal-
cification morphology. Even for masses where resolution is
likely not the limiting factor, one study showed no significant
difference in performance between breast tomo and CT.8Al-
though research continues in order to optimize tomo
acquisition9,10and reconstruction,11–14clinical trials with fed-
eral and industrial funding are already in progress involving
multiple sites/vendors and well over 3000 subjects to date.
Tomo is technically just limited-angle cone beam CT, but
angular range is not the only important difference between
the modalities. Tomo compresses the breast in a standing
position just like mammography, while breast CT uses no
compression and thus requires prone positioning. This dis-
tinction is actually a very big deal for many clinically rel-
evant reasons. First and foremost, tomo provides far better
posterior tissue coverage than CT. Because the patient lies on
a table with finite thickness and there is no compression to
pull the breast into position, the chest wall and axilla cannot
be effectively imaged. Moreover, just getting some patients
into the prone position will slow down the workflow or just
be impossible, such as for women who may be arthritic, mor-
bidly obese, or otherwise infirm. For these women, posterior
tissue coverage will be even more challenging.
Breast tomo also has several practical advantages. Mini-
mal re-training is necessary for technologists ?as positioning
and operation are virtually the same as for FFDM? or radi-
ologists ?because tomo images already resemble mammo-
grams?. In contrast, breast CT is a whole new modality and
may require substantial re-training. Consider as well infra-
structure and cost. While a tomo system can do double duty
as a conventional FFDM system, a breast CT unit cannot,
which means hospitals would need to buy both mammogra-
phy and CT units. The unit cost for breast CT is also likely to
be much greater than that of a much simpler FFDM/tomo
system. Finally, the larger footprint of the breast CT table
may not fit into existing mammography rooms, which would
require even more costly renovations and downtime.
In summary, although breast CT will play some role in
future breast imaging, it is not practical for primary screen-
ing. Breast tomo will have comparable performance and
much wider clinical acceptance than breast CT. Ultimately,
breast tomo will likely replace mammography, at least for
screening and, perhaps, for diagnostic examinations as well.
Rebuttal: Andrew Karellas, Ph.D.
Breast tomosynthesis is not likely to provide the final so-
lution to circumventing the limitations of planar mammogra-
phy. Tomosynthesis systems are not likely to evolve as
simple upgrades of existing digital mammography systems.
Major redesign with regard to the mechanics of the motion
of the x-ray tube and detector must be made and the beam
quality ?kVp, target, filtration? is likely to be different from
that in planar mammography in order to maintain low radia-
tion dose. This would also require modifications of the de-
tector in the form of larger pixel size, pixel binning and
thicker x-ray detector that would result in lower spatial reso-
lution than in digital mammography. The adaptation of breast
tomosynthesis systems to a dual function for digital mam-
mography and tomosynthesis is attainable but such systems
410Karellas and Lo: Point/Counterpoint410
Medical Physics, Vol. 35, No. 2, February 2008
will not deliver true tomographic and 3D information. Breast
tomosynthesis relies on limited projections ?typically about
10–25? and the reconstructed images are inherently prone to
artifacts that may render some features difficult to interpret.
Its spatial resolution in the depth ?z? direction presents a
particular concern in depicting the geometry and morphology
of clustered microcalcifications due to its non-isotropic reso-
lution and propensity to artifacts.
By comparison, dedicated breast CT delivers isotropic
spatial resolution for true tomographic and 3D depiction of
anatomic detail, and it is less prone to image reconstruction
artifacts in the absence of highly attenuating tissues and large
cone beam angle. DBCT is critically dependent on advances
in detector and electronics technology for attaining good spa-
tial resolution, fast data acquisition and lower radiation dose.
Slip ring technology can be implemented for fast acquisition
in order to avoid any motion effects. In view of recent ad-
vances in detectors and electronics and gantry design and
rotational mechanics, further improvements in DBCT are
very realistic. Simultaneous imaging of the axial and medial
aspects of the breast presents a significant challenge with
DBCT but, with innovative gantry design, rotational me-
chanics and patient positioning techniques, this challenge
can be met. Finally, unlike tomosynthesis, DBCT does not
require physical compression of the breast and this represents
a radical departure and a great improvement over existing
Rebuttal: Joseph Lo, Ph.D.
I agree with Dr. Karellas that the limited angle acquisition
of tomosynthesis is a limitation. I disagree, however, on its
impact. Tomo consistently provides compelling images that
have radiologists clamoring for the technology. In a recent
clinical study, tomosynthesis outperformed mammography in
sensitivity and specificity.15Radiologists do not mind the
artifacts or depth resolution. They are concerned, however,
with workflow implications of interpreting dozens of images
per breast for tomo and possibly 200 or more for breast CT.
Breast CT will probably play an important role as a diag-
nostic adjunct to mammography or tomosynthesis. I agree
that CT will image implants far better. As for 100% dense
breasts, tomo can often image them quite well already, but
CT may do even better. CT may also facilitate future quan-
titative applications such as contrast enhanced imaging.
Microcalcifications are a controversial issue for both mo-
dalities. Tomosynthesis has higher in-plane resolution but
may suffer from artifacts. Clinical results to date are mixed,
and this is an active area of research. Others have suggested
hybrid scan sequences which avoid the whole problem by
acquiring a conventional FFDM in mid-tomosynthesis
scan.16In comparison, breast CT’s lower in-plane resolution
may make calcification detection and characterization quite
At this moment, neither of these modalities has received
approval in the US or elsewhere. This may change quickly,
of course, but clearly both are nascent technologies and there
is much potential for improvement. On this point, my col-
league and I certainly agree.
In conclusion, while breast CT yields “true” 3D images, it
has important practical limitations including prone position-
ing, poor posterior coverage, and likely higher cost. Tomo-
synthesis delivers practical and effective 3D images without
such limitations. From the patient’s point of view, we may
use slightly less compression to achieve dose reduction, po-
sitioning, and immobilization while avoiding pain. Tomosyn-
thesis is therefore likely to become the new standard for
screening and perhaps diagnostic breast imaging.
1C. H. Chang, J. L. Sibala, and J. H. Gallagher et al., “Computed tomog-
raphy of the breast: A preliminary report,” Radiology 124, 827–829
2V. Raptopoulos, J. K. Baum, M. Hochman, and A. Karellas et al., “High
resolution CT mammography of surgical biopsy specimens,” J. Comput.
Assist. Tomogr. 20, 179–184 ?1996?.
3L. T. Niklason, B. T. Christian, and L. E. Niklason et al., “Digital tomo-
synthesis in breast imaging,” Radiology 205, 399–406 ?1997?.
4S. Suryanarayanan, A. Karellas, and S. Vedantham et al., “Comparison of
tomosynthesis methods used with digital mammography,” Acad. Radiol.
7, 1085–1097 ?2000?.
5J. M. Boone, T. R. Nelson, K. K. Lindfors, and J. A. Seibert, “Dedicated
breast CT: Radiation dose and image quality evaluation,” Radiology 221,
6B. Chen and R. Ning, “Cone-beam volume CT breast imaging: Feasibility
study,” Med. Phys. 29, 755–770 ?2002?.
7T. Wu, A. Stewart, and M. Stanton et al., “Tomographic mammography
using a limited number of low-dose cone-beam projection images,” Med.
Phys. 30, 365–380 ?2003?.
8X. Gong, S. J. Glick, B. Liu, A. A. Vedula, and S. Thacker, “A computer
simulation study comparing lesion detection accuracy with digital mam-
mography, breast tomosynthesis, and cone-beam CT breast imaging,”
Med. Phys. 33, 1041–1052 ?2006?.
9I. Sechopoulos, S. Suryanarayanan, S. Vedantham, C. D’Orsi, and A.
Karellas, “Computation of the glandular radiation dose in digital tomo-
synthesis of the breast,” Med. Phys. 34, 221–232 ?2007?.
10Y. Chen, J. Y. Lo, N. T. Ranger, E. Samei, and J. T. Dobbins, III, “Meth-
odology of NEQ ?f? analysis for optimization and comparison of digital
breast tomosynthesis acquisition techniques and reconstruction algo-
rithms,” in Medical Imaging 2007: Physics of Medical Imaging, San Di-
ego, CA, Proc. SPIE 6510, 651011 ?2007?.
11T. Wu, R. H. Moore, E. A. Rafferty, and D. B. Kopans, “A comparison of
reconstruction algorithms for breast tomosynthesis,” Med. Phys. 31,
12Y. Zhang, H. P. Chan, B. Sahiner, J. Wei, M. M. Goodsitt, L. M. Hadji-
iski, J. Ge, and C. Zhou, “A comparative study of limited-angle cone-
beam reconstruction methods for breast tomosynthesis,” Med. Phys. 33,
13J. T. Rakowski and M. J. Dennis, “A comparison of reconstruction algo-
rithms for C-arm mammography tomosynthesis,” Med. Phys. 33, 3018–
14T. Wu, R. H. Moore, and D. B. Kopans, “Voting strategy for artifact
reduction in digital breast tomosynthesis,” Med. Phys. 33, 2461–2471
15S. P. Poplack, T. D. Tosteson, C. A. Kogel, and H. M. Nagy, “Digital
breast tomosynthesis: Initial experience in 98 women with abnormal digi-
tal screening mammography,” AJR, Am. J. Roentgenol. 189, 616–623
16R. M. Nishikawa, I. Reiser, P. Seifi, and C. J. Vyborny, “A new approach
to digital breast tomosynthesis for breast cancer screening,” in Medical
Imaging 2007: Physics of Medical Imaging, San Diego, CA, Proc. SPIE
6510, 65103C ?2007?.
411Karellas and Lo: Point/Counterpoint411
Medical Physics, Vol. 35, No. 2, February 2008