Cancer Imaging (2008) 8, 206?215
Optical imaging of the breast
S.M.W.Y. van de Ven, S.G. Elias, M.A.A.J. van den Bosch, P. Luijten and W.P.Th.M. Mali
Department of Radiology, University Medical Center Utrecht, 3508 GA Utrecht, The Netherlands
Corresponding address: S.M.W.Y. van de Ven, MD, Department of Radiology, University Medical Center Utrecht,
PO Box 85500, 3508 GA Utrecht, The Netherlands.
Date accepted for publication 17 September 2008
This review provides a summary of the current state of optical breast imaging and describes its potential future
clinical applications in breast cancer imaging. Optical breast imaging is a novel imaging technique that uses near-
infrared light to assess the optical properties of breast tissue. In optical breast imaging, two techniques can be
distinguished, i.e. optical imaging without contrast agent, which only makes use of intrinsic tissue contrast, and
optical imaging with a contrast agent, which uses exogenous fluorescent probes. In this review the basic concepts
of optical breast imaging are described, clinical studies on optical imaging without contrast agent are summarized, an
outline of preclinical animal studies on optical breast imaging with contrast agents is provided, and, finally, potential
applications of optical breast imaging in clinical practice are addressed. Based on the present literature, diagnostic
performance of optical breast imaging without contrast agent is expected to be insufficient for clinical application.
Development of contrast agents that target specific molecular changes associated with breast cancer formation is the
opportunity for clinical success of optical breast imaging.
Keywords: Optical imaging; Breast cancer; Fluorescence; Absorption; Molecular imaging.
Breast cancer is a major global health problem. As of
2007, an estimated 1.3 million new cases of invasive
465,000 women are expected to die from this disease
worldwide. X-Ray mammography is used in screening
programs and reduces mortality significantly due to ear-
lier detection of breast cancer[2,3]. For younger women,
the benefit from screening with X-ray mammography is
markedly less than for women over the age of 50 years.
This is probably caused by the lower incidence of breast
cancer at a younger age, the more rapidly growing
tumours, and the higher radiographic breast density in
young women. Sensitivity of X-ray mammography
for breast cancer detection in women with fatty breasts
is approximately 88%, but this sensitivity is strongly
reduced in women with dense breasts, i.e. 62%. This
is an important problem, especially since these women
have an increased risk of breast cancer.
Optical breast imaging is a novel imaging technique
that uses near-infrared (NIR) light to assess the optical
properties of tissue, and is expected to play an important
role in breast cancer detection. It dates back to 1929
when Cutler investigated the shadows of light transmitted
through the breast with a normal lamp (transillumina-
tion). Although large malignant lesions with high vas-
cularization could be detected, the method did not
achieve sufficient sensitivity and specificity to be used
in clinical practice at the time. During the last decade,
progress in source and detector technology, light propa-
gation modelling, and potential fluorescent contrast
agents, has resulted in a renewed interest in optical
imaging. Optical breast imaging uses near-infrared
(NIR) light in the wavelength range of 600?1000nm to
assess the optical properties of tissue. Functional infor-
mation on tissue components, i.e. absorption character-
istics of oxy- and deoxyhaemoglobin, water, and lipid, can
be obtained by combining images acquired at various
wavelengths. When using only intrinsic breast tissue
This paper is available online at http://www.cancerimaging.org. In the event of a change in the URL address, please use the DOI
provided to locate the paper.
? 2008 International Cancer Imaging Society
contrast in optical breast imaging, this is referred to as
optical breast imaging without contrast agent. The other
modality, i.e. optical breast imaging with a contrast agent,
uses exogenous fluorescent probes that target molecules
specific for breast cancer. The use of fluorescent probes
has great potential in early breast cancer detection, since
in vivo imaging of molecular changes associated with
Additional advantages of optical breast imaging are that
it uses no ionizing radiation and it is relatively inexpen-
sive, which can realize repeated use (also in young
women) and easy access to the technique. The aim of
this review is to provide a summary of the current state of
optical breast imaging and to describe its potential future
clinical applications in breast cancer imaging.
The basic concepts of optical
In general, optical imaging devices transmit light through
the breast, where it is both absorbed and scattered by the
tissue components present. NIR in the wavelength range
of 600?1000nm is used to allow for sufficient tissue
penetration. After passing through the breast, the remain-
ing light is registered by detectors and advanced com-
puter algorithms are used to reconstruct the images
(Figs. 1 and 2)[9?11]. Determining tissue properties and
their spatial distribution is complex due to the irregular
and long pathways over which light travels through the
Different optical breast imaging systems have been
investigated. In transillumination, sources and detectors
are positioned at opposite sides of the breast. This gen-
erates two-dimensional projection views, comparable to
system used for clinical research (Philips Healthcare,
Best, The Netherlands).
Prototype of the diffuse optical tomography
breast imaging lay-out (A) with source and detector fibres
covering the entire breast surface. In optical breast
imaging without contrast agent (B) higher absorption by
tumour components (predominantly haemoglobin) results
in decreased light intensity registered by the detectors.
In optical breast imaging with contrast agent (C) a fluores-
cent probe is administered that ideally accumulates at the
tumour site. After excitation, light is emitted at a higher
wavelength by this agent and the excitation wavelength
is filtered to only detect the fluorescent signal.
Concepts of optical breast imaging. Optical
Optical imaging of the breast207
compression[13?17]. In tomography, sources and detectors
are placed over the entire breast surface[18,19]. This
enables the acquisition of three-dimensional optical
breast images. Another approach is the use of handheld
devices that are placed manually at the position of inter-
est, comparable to imaging with ultrasound probes[20?22].
Although companies and academic institutions have
putvast effort into
imaging systems, only three of them are commercially
available at this moment. The ComfortScan?system,
developed by DOBI Medical, is a transillumination
system that requires breast compression to generate
two-dimensional optical images (http://www.dobimedi-
Research Technologies Inc. (ART), is a system that
uses slight breast compression, but is able to generate
tomographic images of a chosen region of interest of
the breast. This is the only commercial system that
uses more than one laser, namely four, to be able to
transmit light of different wavelengths through the
system CTLM?, developed by Imaging Diagnostic
Systems Inc. (IDSI), is a tomographic system that
requires no breast compression to generate volumetric
optical images of the breast (http://www.imds.com/
All optical imaging systems in general, use three differ-
ent illumination methods: time domain, frequency
domain, and continuous wave. The time domain tech-
nique uses short (50?400 ps) light pulses to assess the
temporal distribution of photons[15,16,24]. In this way,
distinction between scattering and absorption can be
made. This technology collects the most information on
the optical properties of tissue and therefore has
better contrast and spatial resolution compared to the
other methods. However, time domain equipment is
more expensive and acquisition times are longer.
Frequency domain devices modulate the amplitude of
the light that is continuously transmitted at high frequen-
cies (50?500MHz). By measuring phase shifts
of photons and their amplitude decay (compared to a
mammography,and usually requiresbreast
reference signal), information on the optical properties
of tissue is acquired and scattering and absorption can be
distinguished. Frequency domain devices could generate
the same information as time domain systems if a large
range of frequencies is used. Continuous wave sys-
tems emit light at constant intensity or modulated at
low frequencies (0.1?100kHz). It is a straightforward
technique, which basically measures the attenuation of
light transmitted between two points on the breast sur-
face. Because of its simplicity, continuous wave equip-
ment is cheap and image acquisition fast. However, it is
very difficult to discriminate scattering from absorption
with this technique and data analysis requires complex
Optical breast imaging without
Optical breast imaging uses NIR light to assess the opti-
cal properties of breast tissue. Light absorption at these
wavelengths is minimal, allowing for sufficient tissue
penetration (up to 15cm). The main components of
the breast all have specific absorption characteristics as
a function of the wavelength. By combining images
acquired at various wavelengths (spectroscopy), concen-
trations of oxy- and deoxyhaemoglobin, water and lipid
can be determined. Fig. 3 demonstrates an example of a
benign cyst imaged with both magnetic resonance ima-
ging (MRI) and optical imaging; spectroscopic analysis
of the optical data confirmed the high water and low
blood concentration in this lesion.
In a malignant tumour, haemoglobin concentration is
directly related to angiogenesis, the key factor required
for tumour growth and metastases. In addition, the
proportions of oxy- and deoxyhaemoglobin change in
such a tumour due to its metabolism. By measuring
the concentrations of the breast components, discrimina-
tion of benign and malignant tumours may be possible
with diffuse optical imaging (Fig. 2A and B).
Clinical studies thus far performed using optical
breast imaging without contrast agent are shown
in Table 1[13?24,31?34]. Case reports are not presented
in this overview. Most studies report the number of
the optical data set. The cyst shows a high signal intensity on the MRI and the enhanced-water map (high water content),
and a low signal intensity on the enhanced-blood map (low blood content).
T2-weighted MRI with fat-suppression compared to the enhanced-water map and the enhanced-blood map of
208S.M.W.Y. van de Ven et al.
Clinical studies on optical breast imaging without contrast agent
In situ carci-
Rinneberg et al.
670, 785, 843,
Yes. X-Ray and
Weak contrast, tumor only detectable provided
exact location of inhomogeneity is known
Floery et al.
Increased absorption; an area clearly more
luminous than the surrounding parenchyma
Taroni et al.
637, 656, 683,
785, 913, 975
Yates et al.
Zhu et al.
Handheld with US
Maximum hemoglobin concentration
G€ otz et al.
690, 750, 790,
Clearly visible contrast
Tomandl et al.
Based on non-specified criteria developed by an
experienced radiologist who compared optical
images with X-ray and US findings
Franceschini et al.
Yes. X-Ray and
Visible optical inhomogeneity corrected for edge
effects (so-called dimensionless N value)
Chance et al.
760, 805, 850
Relatively high hemoglobin content and low
Gu et al.
785, 808, 830
Yes. X-Ray and
Lower absorption and/or scattering coefficients
than surrounding parenchyma
Hsiang et al.
658, 682, 785,810, 830, 850
760, 780, 830,
Malignant lesions show higher haemoglobin content and lower oxygenation than surrounding
Durduran et al.
Blood flow increases to 230% in malignant lesions and to 153% in benign lesions
Cerussi et al.
Malignant lesions show increase in deoxyhaemoglobin, oxyhaemoglobin, and water (450%), and
decrease in lipid (?20%) compared to normal tissue; tumour spectra appeared age-dependent
Cerussi et al.
Yes. X-Ray, US,
Responders to chemotherapy showed significant decrease in deoxyhaemoglobin (27%) and
relative water content (20%) compared to non-responders; oxyhaemoglobin decreased in both
groups, but significantly more in responders (33%) compared to non-responders (18%)
et al. (2002)
combined with MRI
Malignant lesions show higher haemoglobin content and lower oxygenation than surrounding
US, ultrasound; MRI, magnetic resonance imaging.
aWavelengths in bold were used in all measurements, others only in part of the measurements.
bWhen there were less than 5% in situ carcinomas, this group was combined with invasive carcinomas (italic), and also when there was no information on invasiveness available.
Optical imaging of the breast209
lesions detected on the optical images (detection rates),
irrespective of their classification (benign/malignant).
Sensitivity and specificity have not been determined
yet. Detection rates for carcinomas range from 0.50 to
1.00 in these studies. Studies performed with handheld
devices report high detection rates (0.95?1.00)[20?22].
Detection rates for the transillumination approach
range from 0.58 to 0.94[13?17]. With tomography, carci-
nomas were detected in 74%and 50%. Detection
rates of benignlesions
0.94[14?21,31]. Benign cysts were detected with tomogra-
phy in 83%. Malignant lesions were detected by their
higher optical attenuation compared to the surrounding
tissue, mainly related to increased light absorption by
their higher haemoglobin
benign lesions were more difficult to detect, but some-
times showed increased attenuation, although to a lesser
extent than malignant lesions[14?21,31]. As opposed to the
other lesions, benign cysts showed lower optical attenua-
tion, associated with lower light absorption or scattering
by their high water content[16,24,31]. Some groups found
lower oxygenation for carcinomas compared to the sur-
rounding tissue[21,23,24,33]. In addition, Cerussi et al.
described increased water content and decreased lipid
content in malignant lesions, and age-dependency of
the tumour spectra. This group also investigated the
response to chemotherapy in breast cancer patients and
reported significant decrease in deoxyhaemoglobin (27%)
and relative water content (20%) in responders compared
to non-responders; oxyhaemoglobin decreased in both
groups, but significantly more in responders (33%) com-
pared to non-responders (18%).
Optical breast imaging with
In optical breast imaging with contrast agent, fluorescent
probes are used that emit photons at predefined wave-
lengths after excitation by laser light. These photons are
detected while the light of the excitation wavelength is
filtered (Fig. 2C).
Fluorescent probes that target molecules specific for
breast cancer are currently being developed and validated
in preclinical animal studies. An overview of these studies
is provided in Table 2[35?43]. All animal studies were
performed with breast cancer mouse models with NIR
continuous wave optical imaging devices. In most studies,
transillumination was used, but two research groups
applied a tomographic approach[35,41]. A variety of opti-
cal probes for specific breast cancer cell targeting has
been designed. The group of Bremer and Mahmood
et al. developed so-called ?smart? optical probes to
target proteases[35?37]. These probes are non-fluorescent
in their native state, but convert to a highly fluorescent
active state when their backbone is cleaved by cathepsins.
In four animals with human breast cancer xenografts,
tumours showed a strong fluorescence signal in vivo
after injection of the cathepsin-sensing probe. Signal-to-
noise ratio (SNR) after 48h was 21 in tumours with
mean diameters 52mm.
tumour was 51mm in diameter. This technique
using smart optical probes also showed good results in
transgenic mice that spontaneously developed tumours.
With transillumination, all 24 tumours in 10 animals
could be clearly delineated after injection of the cathe-
psin-sensing probe. Tumour fluorescence in vivo was sig-
nificantly higher compared to background fluorescence
measured in the adjacent skin (380?23 AU vs. 179?8
AU; p50.01). Tomography was performed in four ani-
mals; co-registration with MRI revealed a strong fluores-
cence signal within the tumour tissue and virtually no
background fluorescence in the corresponding slices.
Differences in tumour aggressiveness could be depicted
by this technique when comparing eight well-differen-
tiated, with eight highly invasive metastatic human
breast cancer models. The highly aggressive cancers,
which expressed higher levels of proteases, revealed sig-
nificantly higher tumour fluorescence compared to well-
differentiated tumours (861?88 AU vs. 566?36 AU;
p50.01). Tumours in non-injected animals were not vis-
ible due to identical autofluorescence in tumour and adja-
Three research groups focused on targeting the human
epidermal growth factor-2 (HER2) receptor with probes
antibody trastuzumab, Herceptin, coupled to an NIR
dye[39,41,42]. Hilger et al. compared such probes in
three animals with HER2-overexpressing tumours and
three animals with normal HER2-expression. Distinctly
higher relative fluorescence signals were found in
the tumours with HER2-overexpression compared to
the tumours with normal HER2-expression (e.g. 16h
after injection: 2.2?0.1 vs. 1.3?0.2). Sampath and
colleagues designed a dual-labelled probe consisting of
trastuzumab as targeting component, an111In complex
as radiotracer, and an NIR dye as optical signal genera-
tor. Fluorescence signal intensities obtained after injec-
tion with this HER2-specific probe in three mice bearing
HER2-overexpressing tumours, were significantly higher
(tumour-to-muscle ratio (TMR) 2.25?0.2) compared to
fluorescence signal intensities after injection of two non-
specific probes (TMR 1.35?0.1 and 1.44?0.18;
p?0.001), each administered in five mice. TMR in
five mice pre-injected with trastuzumab before receiving
the HER2-specific probe was significantly lower than
in the mice not pre-injected (p¼0.0048). Single photon
emission computed tomography (SPECT) fused with
computed tomography (CT) showed similar patterns
in probe uptake. Montet et al. co injected two optical
marker (Angiosense-750) and Herceptin coupled to an
NIR dye, at the same time in an HER2-overexpressing
breast cancer mouse model. This model showed
significant tumoural uptake of both the vascular
210S.M.W.Y. van de Ven et al.