Biomedical Photoacoustic Imaging

Interface Focus (2011) 1, 602–631
doi:10.1098/rsfs.2011.0028
Published online 22 June 2011
REVIEW
Biomedical photoacoustic imaging

Paul Beard*
Department of Medical Physics and Bioengineering, University College London,
Gower Street, London WC1E 6BT, UK
Photoacoustic (PA) imaging, also called optoacoustic imaging, is a new biomedical imaging
modality based on the use of laser-generated ultrasound that has emerged over the last
decade. It is a hybrid modality, combining the high-contrast and spectroscopic-based speci-
ficity of optical imaging with the high spatial resolution of ultrasound imaging. In essence, a
PA image can be regarded as an ultrasound image in which the contrast depends not on the
mechanical and elastic properties of the tissue, but its optical properties, specifically optical
absorption. As a consequence, it offers greater specificity than conventional ultrasound ima-
ging with the ability to detect haemoglobin, lipids, water and other light-absorbing
chomophores, but with greater penetration depth than purely optical imaging modalities
that rely on ballistic photons. As well as visualizing anatomical structures such as the micro-
vasculature, it can also provide functional information in the form of blood oxygenation,
blood flow and temperature. All of this can be achieved over a wide range of length scales
from micrometres to centimetres with scalable spatial resolution. These attributes lend PA
imaging to a wide variety of applications in clinical medicine, preclinical research and basic
biology for studying cancer, cardiovascular disease, abnormalities of the microcirculation
and other conditions. With the emergence of a variety of truly compelling in vivo images
obtained by a number of groups around the world in the last 2 – 3 years, the technique has
come of age and the promise of PA imaging is now beginning to be realized. Recent
highlights include the demonstration of whole-body small-animal imaging, the first
demonstrations of molecular imaging, the introduction of new microscopy modes and
the first steps towards clinical breast imaging being taken as well as a myriad of in vivo
preclinical imaging studies. In this article, the underlying physical principles of the technique,
its practical implementation, and a range of clinical and preclinical applications are reviewed.
Keywords: photoacoustic; ultrasound; imaging; medical; biomedical
1. OVERVIEW propagating in a gas generated by laser-induced surface

heating are detected with a microphone [2]. This is in con-
Research into the underlying physics of photoacoustic
trast to the direct detection of laser-induced ultrasound
(PA) techniques has a relatively long, if sporadic, history
waves which biomedical PA imaging exploits. Although
dating back to 1880 when Alexander Graham Bell first
the latter direct detection approach was explored for char-
discovered the PA effect following his observation of the
acterizing solids as a potential non-destructive testing
generation of sound owing to the absorption of modulated
tool, it was not until the mid-1990s that it began to be
sunlight [1]. Thereafter, relatively little active scientific
investigated for biomedical imaging and the first images
research or technological development took place until
began to appear thereafter [3–9]. This early work, under-
the development of the laser in the 1960s which provided
taken by a handful of researchers, progressed steadily, if
the high peak power, spectral purity and directionality
not with any notable degree of rapidity, until the early
that many PA sensing applications require. A raft of
to mid-2000s when the first truly compelling in vivo
industrial and scientific sensing applications then began
images began to be obtained. From this point onwards,
to emerge in the 1970s and 1980s. However, these
the field has witnessed major growth in terms of the
applications generally exploited the indirect gas-phase
development of instrumentation, image reconstruction
cell type of PA detection, in which acoustic waves
algorithms, functional and molecular imaging capabilities
*pbeard@medphys.ucl.ac.uk and the in vivo application of the technique in clinical
Electronic supplementary material is available at http://dx.doi.org/ medicine and basic biological research. It is this latter
10.1098/rsfs.2011.0028 or via http://rsfs.royalsocietypublishing.org. period that this review is focused on.
One contribution of 15 to a Theme Issue ‘Recent advances in In PA or thermoacoustic imaging, ultrasound waves
biomedical ultrasonic imaging techniques’. are excited by irradiating tissue with modulated
Received 23 March 2011

Accepted 13 May 2011 602 This journal is q 2011 The Royal Society
Review. Biomedical photoacoustic imaging P. Beard 603
electromagnetic radiation, usually pulsed on a nanose- pressure distribution produced by the deposition of
cond timescale although other modulation techniques the optical energy. With some assumptions, this
can be used [10 – 12]. In PA imaging, optical wave- image can be taken to be proportional to the absorbed
lengths in the visible and near-infrared (NIR) part of optical energy distribution, which depends on the opti-
the spectrum between 550 and 900 nm are most com- cal absorption and scattering properties of the tissue. In
monly used. The NIR spectral range 600 – 900 nm fact, it is the former that dominates and thus PA image
offers the greatest penetration depth extending to sev- contrast is often said to be ‘absorption-based’. As a con-
eral centimetres. Thermoacoustic imaging employs sequence, PA imaging can provide greater tissue
significantly longer wavelengths, beyond the optical differentiation and specificity than US because differ-
spectrum and into the microwave band (300 MHz – ences in optical absorption between different tissue
3 GHz), and can provide even greater penetration types can be much larger than those in acoustic impe-
depths. In the case of optical excitation, absorption dance. A case in point is the strong preferential
by specific tissue chromophores such as haemoglobin, optical absorption of haemoglobin. This makes PA
melanin, water or lipids followed by rapid conversion imaging particularly well suited to imaging the micro-
to heat produces a small temperature rise (approx. vasculature, which can be difficult to visualize with
less than 0.1 K)—well below that required to cause pulse-echo US owing to the weak echogenicity of micro-
physical damage or physiological changes to tissue. vessels. In addition, the spectral dependence of optical
This leads to an initial pressure increase, which sub- absorption enables image contrast provided by specific
sequently relaxes resulting in the emission of tissue chromophores to be selectively enhanced by
broadband (approx. tens of megahertz) low-amplitude tuning the laser excitation wavelength to their peak
(less than 10 kPa) acoustic waves. The latter propagate absorption. Thus, for example, the presence of lipid
to the surface where they are detected either by a single deposits in atheromatous plaques can be revealed by
mechanically scanned ultrasound receiver or an array of choosing an excitation wavelength coincident with the
receivers in order to acquire a sequence of A-lines. By lipid absorption peak at 1210 nm. The spectroscopic
measuring the time of arrival of the acoustic waves nature of the PA effect can be further exploited to
and knowing the speed of sound, an image can be recon- quantify the concentrations of specific chromophores
structed in much the same way that a conventional via their spectral signatures to provide physiological
pulse-echo ultrasound image is formed. Depending on parameters. An important example of this is the spec-
the specific imaging mode (see §4), the recorded A- troscopic measurement of blood oxygen saturation
lines can either be used directly to form an image or (sO2). By acquiring images at multiple wavelengths
alternatively used in conjunction with a reconstruction and applying a spectroscopic analysis, the known spec-
algorithm based on backprojection or phased array tral differences in oxyhaemoglobin (HbO2) and
beamforming principles. An important difference deoxyhaemoglobin (HHb) can be used to quantify
between pulse-echo ultrasound (US) and PA image for- their concentrations and thus obtain a measurement
mation, however, is that with the former, localization of blood sO2.
can be achieved by focusing the transmit beam as well However, although the large variation in the optical
as the receive beam. In PA imaging, for depths absorption properties of tissue provides PA imaging
beyond approximately 1 mm, the overwhelming optical with its high contrast, it comes at a cost, namely
scattering exhibited by most soft tissues precludes penetration depth. For example, whilst the strong pre-
equivalently focusing the ‘transmit signal’—the exci- ferential absorption of haemoglobin is advantageous
tation light—for the purpose of localization. In PA in terms of contrast, it is also a major contributing
imaging, for depths greater than approximately 1 mm, factor to the strong optical attenuation exhibited by
localization can be achieved in reception only. A further most tissues. As a consequence, although penetration
difference between PA and US imaging lies in the mag- depths of several centimetres can be achieved, PA ima-
nitude of the acoustic pressures involved. Diagnostic ging is unlikely to ever match the penetration depth
clinical US scanners can produce focal peak pressures limit of ultrasound which can be 10 cm or more in
in excess of 1 MPa, whereas PA pressure amplitudes soft tissues. On the other hand, the penetration
are several orders of magnitude lower, typically less depth available to PA imaging significantly exceeds
than 10 kPa. Thus, nonlinear acoustic propagation that of purely optical imaging techniques such as mul-
is not encountered in PA imaging, so there is no PA tiphoton or confocal microscopy that rely on
equivalent of US tissue harmonic imaging. The low unscattered or so-called ballistic photons. As well as
PA pressure amplitudes also means that the potential being able to image structure and function via optical
hazards due to US exposure are not of concern—those absorption and its wavelength dependence, other capa-
relating to laser exposure dominate safety-related bilities of PA imaging include blood flow measurement
considerations. by exploiting the acoustic Doppler effect in a manner
Although PA and US image formation and the fac- analogous to conventional Doppler US and temperature
tors that affect spatial fidelity and resolution are sensing via the temperature dependence of the PA gen-
essentially the same, the sources of image contrast are eration process.
fundamentally different. An US image provides a The combination of the strong spectral discrimination
representation of the acoustic impedance mismatch arising from optical tissue interactions and the high
between different tissues. US image contrast therefore, spatial resolution associated with ultrasound propa-
depends on the mechanical and elastic properties of gation lends PA imaging to a broad range of potential
tissue. A PA image, however, represents the initial applications in clinical medicine, preclinical research
Interface Focus (2011)

604 Review. Biomedical photoacoustic imaging P. Beard
and biology. The ability to map the structure, oxygen- vibrational and collisional relaxation. This produces
ation status and flow characteristics of the vasculature an initial pressure increase and the subsequent emission
makes PA imaging well suited to the assessment of of acoustic waves which propagate to the surface where
tumours and other pathologies characterized by they are detected. In this way, the signal generation
abnormalities in the morphology and function of the mechanism can be regarded as one in which the
vasculature. Imaging breast and skin cancers are optically induced initial pressure distribution p0 is
potential clinical applications that exploit this capa- encoded onto a propagating acoustic wave which,
bility. Others include the assessment of the skin upon detection by an ultrasound receiver located on
microvasculature for studying superficial soft tissue the surface, is converted to a time-resolved electrical
damage, such as burns or abnormalities of the microcir- signal. Since the PA image is then formed from a set
culation in patients with lower limb venous disease and of such PA signals detected at different spatial points,
diabetes. By exploiting the contrast provided by lipid it follows that the PA image is a representation of p0.
absorption, there is the potential to identify vulnerable Given this, what physical properties of the tissue does
plaques prone to rupture in the coronary arteries using p0 then depend upon? As might be expected, p0 is
an intravascular PA imaging probe in a manner analo- related to the heating produced by the deposited laser
gous to intravascular ultrasound (IVUS). Other energy. If impulsive heating is assumed, and in practice,
clinical applications lie in opthalmology, high intensity this requires that the acoustic propagation time is small
focused ultrasound (HIFU) and photothermal treat- compared with the length scale of the heated volume,
ment monitoring. PA imaging has a role to play in then by simple thermodynamic considerations it can
studying mouse models, which are widely used as a pre- be shown that p0 at a point r is proportional to the
clinical research platform for studying human disease absorbed optical energy H(r) [20]
processes. As well as imaging the anatomy and physi-
ology of small-animal models, PA imaging has strong p0 ðrÞ ¼ GH ðrÞ; ð2:1Þ
potential as a preclinical molecular imaging modality
where G is known as the Grüneisen coefficient, a dimen-
through the use of optically absorbing targeted contrast sionless thermodynamic constant that provides a
agents or reporter genes that express absorbing measure of the conversion efficiency of heat energy to
proteins. pressure and is given by G ¼ bc 2/Cp where b is the
Several previous review articles on PA imaging, some volume thermal expansivity, c the sound speed and
focusing on specific aspects of the technique such as its Cp the specific heat capacity at constant pressure. The
role in neuroimaging [13], molecular imaging [14] and
absorbed optical energy distribution H(r) is given by
miocrovascular imaging [15], as well as more general
the product of the local absorption coefficient ma(r)
reviews [16– 19] appear in the literature. The aim of and the optical fluence f(r; ma, ms, g) where ma and
this paper is to provide an overview of the underlying ms are the absorption and scattering coefficients over
physics of PA imaging, its practical implementation, the illuminated tissue volume and g is the anistropy
the functional and molecular imaging capability it can factor. Writing p0 explicitly, we obtain
provide and the biomedical applications it lends itself
to. It is also intended to compare it with conventional p0 ðrÞ ¼ Gma ðrÞfðr; ma ; ms ; gÞ: ð2:2Þ
ultrasound given the similarities between the two mod-
alities. Thus in §2, consideration is given to the physical As equation (2.2) illustrates, p0 depends upon a
properties that underlie PA image contrast and the variety of mechanical, thermodynamic and optical par-
specific tissue constituents that contribute to it. Section ameters. However, in PA imaging, the mechanical and
3 discusses the factors that define penetration depth thermodynamic properties are usually considered to
and spatial resolution. The two principal PA imaging vary sufficiently weakly between different tissue types
modes and their variants, PA tomography (PAT) and that they can be regarded as being spatially invariant.
microscopy, are described in §4. The means by which There are inevitably some limits to this assumption
one can go beyond purely structural imaging and pro- and there is a growing recognition that image contrast
vide functional, physiological and molecular provided by certain tissues may in part originate from
information using PA spectroscopy, flowmetry and the heterogeneities in the Grüneisen coefficient [21].
thermometry are outlined in §5. Finally, a range of cur- However, this notwithstanding, image contrast can be
rent and potential clinical and preclinical imaging assumed to be dominated by the optical absorption
applications are described in §6. and scattering properties of the tissue. In fact, it tran-
spires that optical absorption tends to dominate and
for this reason PA images are often described as being
‘absorption based’. However, it is important to recognize
2. PHOTOACOUSTIC IMAGE CONTRAST
that this does not mean that image contrast is directly
In order to identify the origins of PA image contrast, it proportional to the absorption coefficient ma. As
is instructive to consider the PA signal generation pro- equation (2.2) shows, p0 is proportional to the product
cess. As described above, pulsed laser light is incident of ma and the fluence f which is itself dependent on
on the tissue surface. Depending on the wavelength, ma, so p0 is a nonlinear function of ma. This distinction
the light penetrates to some depth. In doing so, it is is of critical importance when considering the spectro-
multiply scattered and absorbed, the latter by specific scopic capability of the technique as described in §5.1.
light absorbing molecules known as chromophores. The dominance of optical absorption as the primary
The absorbed laser energy is converted into heat by source of PA image contrast lends PA imaging to the

10 000 by water rather than haemoglobin and the strong

absorption coefficient ma (cm–1) lipid absorption peak at 1210 nm becomes predomi-
1000
nant. This peak can be exploited to image localized
100 lipid deposits such as those found in atherosclerotic
10 plaques.
Chromophores such as haemoglobin and melanin
1 absorb much more strongly than other tissue chromo-
0.1 phores and thus provide an obvious source of primary
contrast. However, the more weakly absorbing chromo-
0.01
phores such as water and lipids which may not be
0.001 obviously visible on a PA image can still be detected
by exploiting their characteristic spectral signatures.
0.0001
400 600 800 1000 1200 1400 1600 1800 This involves the use of spectroscopic inversion or
wavelength (nm) unmixing techniques applied to images acquired at
different wavelengths—in much the same way that
Figure 1. Absorption coefficient spectra of endogenous tissue blood oxygen measurements can be made by exploiting
chromophores. Oxyhaemoglobin (HbO2), red line: (http:// the known spectral differences between HHb and HbO2.
omlc.ogi.edu/spectra/hemoglobin/summary.html; 150 gl21), Thus, for example, although the absorption coefficient
deoxyhaemoglobin (HHb), blue line: (http://omlc.ogi.edu/ of water at physiological concentrations in tissue is sig-
spectra/hemoglobin/summary.html; 150 gl21), water, black
nificantly less than that of haemoglobin over much of
line [22] (80% by volume in tissue), lipid(a), brown line [23]
(20% by volume in tissue), lipid(b), pink line [24], melanin, the visible and NIR wavelength range, it can still be
black dashed line (http://omlc.ogi.edu/spectra/melanin/ detected by acquiring images at a range of wavelengths
mua.html; ma corresponds to that in skin). Collagen (green around its absorption peak at 975 nm [25,26] and
line) and elastin (yellow line) spectra from [24]. applying a spectroscopic inversion. The weaker lipid
absorption peaks at 920, 970 and 1040 nm can be
similarly exploited to provide lipid-based contrast.
visualization of anatomical features that contain an As well as endogenous contrast, a variety of targeted
abundance of chromophores such as haemoglobin, contrast agents that absorb at visible and NIR wave-
lipids and water. Of these, haemoglobin is the most lengths can provide additional sources of spectrally
important for wavelengths below 1000 nm. As figure 1 selective PA contrast and be used to image disease-
shows, between 650 and 900 nm the absorption coeffi- specific receptors. These include organic dyes such as
cients of both the oxygenated and deoxygenated indocyanine green (ICG) and methylene blue which
states of haemoglobin at physiologically realistic con- can be used clinically in some circumstances or
centrations are at least an order of magnitude higher nanostructures such as metallic nanorods or shells
than the other major chromophores, such as water, which are currently limited to preclinical use. When
lipids and elastin that are present in connective tissues, appropriately targeted, these agents can bind to a
blood vessels and other organ constituents. At shorter disease-specific cellular or intracellular receptor with the
wavelengths extending into the visible part of the spec- spectral signature of the dye/nanostructure providing
trum, haemoglobin absorption is even higher and can the source of contrast in the PA image. Nanoparticles
exceed that of other chromophores by more than two of various geometries, in particular, are of major interest
orders of magnitude. It is the very strong preferential as their plasmon resonance absorption cross sections can
absorption of haemoglobin that enables the vasculature be orders of magnitude higher than dye molecules. In
to be visualized with such high contrast in PA images. addition, by adjusting their geometric parameters, their
Furthermore, the differences in the absorption spectra peak absorption wavelength can be tuned to NIR
of HbO2 and HHb shown in figure 1 can be exploited wavelengths where tissue penetration is greatest.
to measure blood oxygenation by acquiring images at
multiple wavelengths and applying a spectroscopic
analysis as described in §5.1. In this way, the absorp- 3. PENETRATION DEPTH AND SPATIAL
tion-based contrast of PA imaging allows functional RESOLUTION
as well as structural images of the vasculature to be
3.1. Penetration depth
obtained.
Although melanin has a higher absorption coefficient Penetration depth is limited ultimately by optical and
than blood, it tends to be highly localized in regions, acoustic attenuations. In general, for most soft tissues,
such as the skin or the retina rather than being a although acoustic attenuation is significant, it is optical
major constituent of most tissues. It does not therefore attenuation that dominates. Optical attenuation
tend to dominate PA image contrast in the way that depends on both the absorption and scattering coeffi-
haemoglobin does. Nevertheless, it forms an important cients and, as figure 1 would suggest, is strongly
source of contrast for visualizing melanin-rich struc- wavelength-dependent.
tures such as certain pigmented lesions in the skin In optically scattering media such as tissues, optical
and the retinal-pigmented epithelium (RPE). Absorp- penetration depth is best characterized by the effective
tion by lipids is significantly lower than that of attenuation coefficient meff derived from diffusion
haemoglobin over the visible and NIR range up until theory where meff ¼ (3ma(ma þ ms0 ))1/2 and ma and ms0
around 1100 nm when blood absorption is dominated are the absorption and reduced scattering coefficients,

respectively. In homogeneous scattering media, once at penetration depths, sub-millimetre spatial resolution is
a depth beyond several transport mean free paths possible, decreasing to sub-100 mm for millimetre pen-
(approx. 1 mm), the light becomes diffuse and the irra- etration depths and sub-10 mm spatial resolution for
diance decays exponentially with depth with meff the depths of a few hundred micrometres. Although acous-
exponential constant. 1/meff is therefore the depth at tic attenuation defines the ultimate spatial resolution
which the irradiance has decreased by 1/e and termed limit, other factors such as detector bandwidth, element
the penetration depth. At an excitation wavelength of size and the area over which the PA signals are
700 nm which lies in the spectral region (600– 900 nm) recorded—the detection aperture—can be limiting fac-
where tissue is at its most transparent and using physio- tors in practice. This is particularly so when imaging
logically realistic values of ms (1.6 mm21) and superficial features that lie within a few millimetres of
concentrations of HbO2 and HHb (1% by volume and the surface. The bandwidth of the PA signal can then
assuming 95% sO2), water (74% by volume) and extend to several tens of megahertz presenting signifi-
lipids (25% by volume), gives a value of meff cant challenges in terms of meeting the detection
0.13 mm21. The 1/e optical penetration depth is thus bandwidth and spatial sampling requirements.
approximately 8 mm. This means that, once beyond There is an exception to the acoustically defined
the first millimetre in tissue, light is attenuated by spatial resolution limit discussed above. In the optical
approximately a factor of 4 for each additional resolution microscopy mode of PA imaging (OR-
centimetre of penetration depth. Assuming a value of PAM) described further in §4.2.2, a focused excitation
0.75 dB cm21 MHz21 for plane wave acoustic attenu- laser beam is employed. For very small penetration
ation and a frequency of 10 MHz, the total depths (less than 1 mm), before the focused beam has
attenuation owing to both optical and acoustic attenu- been significantly distorted by optical scattering, the
ation is thus at least one order of magnitude per lateral spatial resolution is defined by the laser beam
centimetre. This represents one of the major challenges diameter at the focus. Under these circumstances, the
in PA imaging as to penetrate several centimetres in lateral resolution is limited by optical diffraction
tissue incurs a signal attenuation of several orders of which depends on the optical wavelength and the
magnitude thus requiring the detection of extremely numerical aperture (NA) of the focusing lens and can
weak ultrasound signals. Despite this, through careful be as small as few micrometres. Vertical resolution,
choice of wavelength, optimization of the light delivery, however, remains limited by acoustic attenuation.
transducer design parameters and signal processing, it
has been demonstrated that penetration depths of sev-
eral centimetres are attainable. A penetration depth of 4. PHOTOACOUSTIC IMAGING
4 cm has been achieved in vivo in the human breast CONFIGURATIONS
[27] using an excitation wavelength of 800 nm. Other Photoacoustic imaging can be divided into several cat-
studies using tissue phantoms and ex vivo tissues have egories: PAT, PA microscopy and its variants. These
suggested depths of 5 – 6 cm may be achievable with categories are, to some extent, all variations on a
the use of contrast agents [28,29]. With regard to the theme and more a consequence of the different imaging
optimum wavelength range, it has been suggested that instruments that have emerged in the last few years,
longer wavelengths, such as 1064 nm, at which blood than fundamental methodological differences.
absorption is low, might provide a penetration depth
advantage if contrast agents that absorb at this
4.1. Photoacoustic tomography
wavelength are used [30].
PAT can perhaps be regarded as the traditional mode of
PA imaging as envisaged by early practitioners. It is
3.2. Spatial resolution
also the most general and least restrictive PA imaging
In common with pulse-echo US imaging, spatial resol- approach with the fewest limitations on imaging
ution depends ultimately on the frequency content of performance imposed by its practical implementation.
the acoustic wave arriving at the detector. In PA ima- In PAT, full field illumination, in which a large diam-
ging, nanosecond excitation laser pulses are most eter pulsed laser beam irradiates the tissue surface, is
often used and can result in extremely broadband employed. At NIR wavelengths where tissue is relatively
acoustic waves with a frequency content extending to transparent, the light penetrates deeply and is also
several tens or even hundreds of megahertz, depending strongly scattered, resulting in a relatively large tissue
on the length scale of the optical absorbers. Under volume becoming ‘bathed’ in diffuse light. Absorption
these conditions, the bandwidth of the PA signal and of the incident radiation by tissue chromophores leads
thus the spatial resolution, is not usually limited by to impulsive heating of the irradiated tissue volume fol-
the generation process itself. Instead, it is the bandli- lowed by the rapid generation of broadband ultrasonic
miting of the propagating PA wave owing to the waves. These propagate to the tissue surface where
frequency-dependent acoustic attenuation exhibited by they are detected by a mechanically scanned ultrasound
soft tissues that limits the maximum frequency content receiver or array of receivers. The time-varying detected
of the PA wave and thus defines the ultimate practically ultrasound signals can then, with knowledge of the
achievable spatial resolution limit. Under these circum- speed of sound, be spatially resolved and back-projected
stances, spatial resolution scales with depth. Acoustic to reconstruct a three-dimensional image. Figure 2
attenuation strongly depends upon tissue type but an shows three commonly used detection geometries:
approximate rule of thumb is that for centimetre spherical, cylindrical and planar. Clearly, the

(a) (b) (c)
Figure 2. PAT detection geometries. (a) Spherical, (b) cylindrical and (c) planar.
cylindrical or spherical detection geometries requires

access to all points around the target and are therefore
limited to applications such as imaging the breast or
small animals such as mice. Planar detection geometries
are more versatile providing access to a greater range
of anatomical targets, especially those superficially
located.
A variety of methods for reconstructing the PAT
image from the detected signals have been developed. r
t
Conceptually, the amplitude at each point t in the =c ultrasound transducer
R t
time record of the PA waveform recorded at a point r array
can be regarded as representing the sum of all points
in the initial pressure distribution po that lie on a
spherical surface centred on r and with a radius equal
time
to the product of the sound speed and t—that is, to
say the PA source distribution is regarded as being com- Figure 3. Backprojection PAT image reconstruction for a
posed of an ensemble of elemental acoustic point planar detection geometry. PA waveforms are recorded by
sources each emitting spherical waves. The image recon- each array element at r, spatially resolved using the sound
struction process can then be thought of as one in which speed c and backprojected over spherical surfaces of radius
each of the detected PA waveforms are spatially R ¼ ct into the image volume. Since the backprojected quan-
resolved using the sound speed, back-projected over tity is the velocity potential, the output of each detector is
spherical surfaces centred on r and summed over the depicted as a time-integrated pressure waveform for illustra-
image volume—figure 3 illustrates this for a planar tive purposes—in practice, the detectors record a pressure
waveform and the time integration is performed
detection geometry. This type of simple ad hoc back-
computationally.
projection is equivalent to delay-and-sum receive
focusing or beamforming employed in phased array
US imaging. Although it provides a simple and intui- applications be achieved. In the more practically
tively amenable description of PAT image formation useful planar case, the implementation involves Fourier
and was used in early implementations [5,31], it is transforming the time-dependent pressure data
non-optimal in terms of accuracy and computational measured over the surface, mapping the temporal fre-
expense. quency to the axial spatial frequency and inverse
More advanced methods that provide a more accu- Fourier transforming to obtain p0 [42]. It is computa-
rate reconstruction and or greater computational tionally advantageous because much of the processing
efficiency have been developed in recent years, many is accomplished via the fast Fourier transform and a
of which are reviewed in Kuchment & Kunyansky simple k-space interpolation via the dispersion relation.
[32,33]. These methods can be divided into several cat- This and its ease of computational implementation
egories depending on the type of algorithm employed. have led to its widespread practical use [29,44,45].
Filtered backprojection-type algorithms involve filter- Time-reversal methods involve computationally re-
ing before or after a backprojection step [3] and can emitting the measured PA waveforms at each detection
provide an exact reconstruction for spherical [34,35] point in temporally reversed order by running a numeri-
cylindrical [36] and planar geometries [36]. Although cal acoustic propagation model backwards [46– 49].
computationally intensive, they have found practical They are perhaps the least restrictive of all algorithms
application for spherical detection geometries used in [48], relying on the fewest assumptions and can be
PAT breast [27] and small-animal imaging [37]. A used for any detection geometry, detector distribution
number of series summation-based methods, such as and can account for known acoustic heterogeneities.
those based on the temporal and spatial spectral In addition, they can be used to mitigate for signal-to-
decomposition of the detected PA waveforms and a sub- noise ratio (SNR) and resolution degrading effects of
sequent mapping to spatial frequency components in p0 acoustic absorption [49]. Although memory require-
have been described [38,39]. They can provide an exact ments are modest, for practical use, a fast numerical
reconstruction for spherical [40], cylindrical [41], planar acoustic propagation model is required since it is necess-
[42,43] and some other geometries such as a cube [39]. ary to compute the entire wavefield for each temporal
Only in the case of planar or cuboidal geometries can backpropagation step. This has perhaps limited its
sufficiently fast computational times for practical practical application, although a time-reversal scheme

using an efficient pseudo-spectral k-space propagation to the detector rather than the projection of spherical
model has been evaluated using experimental data surfaces that a point-like detector output represents.
[49,50]. Model-based inversion techniques employ a An image can then be reconstructed using the inverse
numerical forward model to simulate the detected PA radon transform analogous to that used in X-ray
signals from an initial estimate of p0 or a related quan- computed tomography (CT).
tity [51,52]. An improved updated estimate can then
be obtained by iteratively adjusting p0 at each spatial
point until the difference between the predicted and 4.1.1. Photoacoustic tomography imaging systems.
measured PA signals is minimized. By using matrix Spherical scanners: A variety of three-dimensional
inversion methods and pre-computing for a specific geo- scanning instruments that employ a spherical detection
metry as described in Rosenthal et al. [53], these geometry have been demonstrated. As noted above, for
methods can be fast, albeit at the cost of flexibility applications such as small-animal or breast imaging
and the significant computational expense of the initial that allow the region of interest (ROI) to be enclosed
pre-computing step. by the detection surface, this geometry offers the high-
Inevitably, there are practical limitations. An exact est practically achievable image fidelity on account of
reconstruction usually requires the assumption of an the large solid angular detection aperture that can be
infinite number of wideband omni-directional point- attained.
like detectors distributed over a solid angular detection Figure 4 shows a spherical scanner design used for
aperture of 4p sr for a spherical1 or cylindrical detection PA small-animal [37] and breast imaging [27]. The
geometry and 2p sr for a planar geometry, the latter instrument comprises a hemispherical detector bowl
implying detection over an infinite plane—these aper- with an aperture in the bottom to permit delivery of
ture conditions mean that the entire acoustic the excitation laser light. One hundred and twenty
wavefront is recorded, so there is a complete measured eight unfocused 5 MHz 3 mm-diameter piezoelectric
dataset. Adequately broadband piezoelectric ultra- elements are distributed in a spiral pattern over the sur-
sound receivers that can provide the necessary face. The bowl is mounted on a shaft to allow it to be
megahertz bandwidth are, with some limitations, avail- incrementally rotated with successive excitation pulses
able. Achieving acoustically small detector element so that sufficiently fine spatial sampling can be
sizes in order to provide a near-omnidirectional achieved. A second, smaller-diameter, optically and
response and a spatial sampling interval that fulfils acoustically transparent bowl into which the breast is
the spatial Nyquist criterion (, l/2) at megahertz fre- suspended is inserted inside the detector bowl. Both
quencies is more challenging but possible, depending are filled with water to provide acoustic coupling.
on the upper frequency limit. The detection aperture This arrangement allows the detector bowl to be
requirements can, however, present a more fundamental rotated independently without disturbing the breast.
limitation. For spherical geometries, it is possible, in The excitation light, sourced from a tunable optical
principle, to completely enclose the source region to parametric oscillator (OPO) laser system emitting at
fulfil the requirement of a 4p solid angular aperture. 800 nm and with a PRF of 10 Hz, is directed up through
However, for cylindrical and planar geometries, the the aperture in the bottom of the bowl. A near-isotropic
aperture is always truncated in practice. For planar spatial resolution of approximately 250 mm over a 6.4
detection geometries, in particular, measurements are 6.4 5 cm field of view (FOV) was reported. The image
always restricted to a finite region of the infinite plane acquisition time obtained over this FOV using 240
that the reconstruction algorithm assumes. As a conse- angular steps per complete revolution was 24 s.
quence, only part of the wavefront is recorded resulting Figure 4 shows a maximum intensity projection
in image artefacts and reduced spatial resolution—the (MIP) images of the vasculature in the left breast of a
so-called limited view or partial scan problem [54– 56]. patient volunteer with sub-millimetres vessels visible
Artefacts, image distortion and blurring can also arise to a depth of 4 cm. The corresponding animated
from sound speed heterogeneities and acoustic attenu- three-dimensional MIP movies available online (see
ation which are not accounted for in most electronic supplementary material, movies S1 and S2)
reconstruction methods. Several methods aimed at com- perhaps best illustrate the full extent of the detailed
pensating for sound speed perturbations [48,57] and vascular anatomy revealed by these images. The in
acoustic attenuation [49] have been demonstrated with vivo penetration depth achieved in this study provides
the aim of improving image quality in acoustically a compelling demonstration of the deep tissue imaging
heterogeneous tissues. capability of PAT, especially as the optical fluence
Although most reconstruction methods rely on the used was more than one order of magnitude lower
assumption of detection by point-like omnidirectional than the safe maximum permitted exposure (MPE)
receivers, there are techniques that employ mechani- for skin.
cally scanned large-area directional planar detectors A spherical detection geometry can also be achieved
[58,59] or line detectors [60]. In the former case, the by arranging the detectors over an arc and then axially
time-resolved output of the detector represents a set rotating the target located within the interior space to
of time-retarded projections of p0, over planes parallel synthesize detection over a spherical surface as
described in Kruger et al. [61]. In this study, ex vivo
1
There are some exceptions to this for a spherical geometry. For
images of a mouse were obtained. Another small-
example, an exact reconstruction of an object that lies within a animal scanner that uses a similar approach has been
hemispherical detection surface (ie 2p sr) can be obtained. described in Brecht et al. [62] and was used to acquire

(a)
light Beam
clear aperture
laser
spectrometer
(c)
(b)
64 mm ¥ 50 mm
64 mm ¥ 64 mm
Figure 4. PAT breast scanner with hemispherical detection geometry [27]. (a) Schematic of system. (b) MIP of the left breast of
patient volunteer (lateral projection) over 64 50 mm2 FOV. Arrows at top indicate the direction of the incident excitation light.
Hollow arrow marks the position of a vessel at a depth of 40 mm. Hollow box represents 1 1 cm2 (c) orthogonal (anterior –pos-
terior) projection 64 64 mm2 FOV. Excitation laser wavelength: 800 nm. Three-dimensional animated MIP movies of both
projections can be viewed online at the electronic supplementary material, movies S1 and S2.
in vivo whole-body images (figure 5). In this scheme, features of different length scales and geometries, the
the arc array comprised 64 square (2 2 mm) piezo- raw RF-detected signals were processed with a family
composite transducers of centre frequency 3.1 MHz of wavelet and other filters prior to image reconstruc-
distributed over a two-dimensional angular aperture of tion, the latter being achieved using a spherical
1528. When the object is rotated through 3608, this backprojection algorithm. The spatial resolution was
translates to a solid angular detection aperture of reported to be 0.5 mm and the acquisition time was
10.6 sr and thus close to the ideal 4p sr aperture 8 min based on 150 steps per complete revolution of
required for an exact reconstruction. The animal is the animal and averaging over 32 laser pulses.
immersed in water, with a diving bell arrangement to Figure 5 shows a volume-rendered image obtained by
allow for the delivery of anaesthetic and respiratory the system demonstrating that internal organs, such
gases and placed at the centre of the array as depicted as the spleen, liver and kidney can be visualised. The
in figure 5. Fibres placed orthogonal to the plane of benefits of incorporating prior structural information
the array and directed at the mouse provide 755 nm into the formation of the image, in this case via the
excitation pulses emitted by an Alexandrite laser with wavelet filtering referred to above, in order to selectively
an incident surface fluence of 1 mJ cm22. To preferen- enhance organs, blood vessels and other anatomical
tially emphasize the PA waves emitted by anatomical structures are apparent.

(b) (c)
(a) spleen
partial lobe
inferior right
gas bubbles of liver
vena cava kidney
escaping
diving bell left
kidney
mouse
right
kidney
left
kidney spinal
region
fibre Bundle ovarian

vessel
ovarian
pre stressed vessel inferior
fibreglass vena cava
rods inferior
black tape probe vena cava
5 mm 10 mm
Figure 5. PAT whole body small animal scanner based on a spherical detection geometry [62]. (a) Experimental arrangement
showing 64 element arc array and fibre delivery bundle. (b) Three-dimensional image of a nude mouse illuminated at 755 nm.
Both kidneys are visualized as well as the spleen and a partial lobe of the liver. (c) Image showing spinal region and left and
right kidneys. Three-dimensional-animated movies of the two images can be viewed online at electronic supplementary material,
movies S3 –S5.
(a) (b) 1.5 (c)

computer MCA
step motor trigger
2
mirror Nd:YAG laser
concave lens 1.0
ground glass
(cm)
oscilloscope
3
x y 0.5
transducer L
aerophore
z
RH LH
0 L
0 0.5 1.0 1.5
water (cm)
amplifier jack 4
min optical absorption A max
Figure 6. PAT cylindrical scanner for small animal brain imaging [64] (a) experimental arrangement. (b) Image showing super-
ficial cortical vasculature and a surgically induced lesion. MCA: middle cerebral artery, RH: right cerebral hemisphere, LH: left
cerebral hemisphere. (c) Photograph of cerebral surface following resection of the skull after PAT image acquisition. Laser
wavelength: 532 nm.
Cylindrical scanners: Although a true two- common with the spherical detection geometry, inevita-
dimensional cylindrical detection geometry has rarely bly limited to applications such as small-animal [64–73]
been implemented for three-dimensional imaging in or breast [74,75] imaging.
practice ([63] being an exception), its one-dimensional Figure 6 shows perhaps the simplest one-dimensional
equivalent, recording over a circle or an arc to obtain a cylindrical scanner for two-dimensional cross-sectional
two-dimensional cross-sectional image, has been widely imaging [64]. It comprises a single 3.5 MHz PZT trans-
implemented. Its popularity stems from its ease of ducer, focused in the elevation direction to attenuate
implementation and the ability to acquire a high-fidelity out-of-plane signals, that is mechanically scanned
image with few artefacts—a consequence of the fact that around the target, in this case the mouse head. The
detection over a full 3608 angular aperture can readily be excitation laser light (at 532 nm) is delivered along
achieved for targets that can be enclosed. If high frame the axis of rotation in order to transversely irradiate
rates are not required, it can also be inexpensively the surface. Although the attenuation of light at
implemented as a laboratory system using a single 532 nm in tissues is relatively high, as figure 6 shows,
mechanically scanned receiver and a stepper motor- the superficial cortical vasculature can still be visualized
driven rotational stage. Although widely used, it is, in with high contrast. The spatial resolution of the system

was estimated at approximately 200 mm. Following this (a)

early demonstration, a variety of similar single-element
mechanically scanned laboratory systems have been
used variously to image epileptic events [65], tumour
growth [67] and cerebrovascular changes [66] in mice
and peripheral joints [76]. To overcome the long (b)
image acquisition times of these systems (minutes to ultrasound
hours), several array-based cylindrical scanners that US probe
array probe
provide real-time two-dimensional image frame rates
have been developed and used to study cerebral haemo- excitation excitation
light light
dynamics [70– 72], cardiovascular dynamics [69] and
organ perfusion [68] in mice. One of these systems
employs a 512-element array arranged over a full 3608 fibre-optic
aperture [70]. With 8 : 1 multiplexing, a single acqui- bundle
sition of all elements could be achieved in 1 s. This fibre bundles
permitted dynamic imaging of the wash-in of a systemi-
cally introduced contrast agent as it perfused through
the superficial cortical vasculature in a mouse. A similar
system but employing 64 elements over the arc of a 1808
aperture has been used for whole-body small-animal Figure 7. Use of a conventional ultrasound imaging probe for
imaging [68,69]. Three-dimensional images were PA imaging. (a) Two-dimensional array probe. (b) Implemen-
obtained by axially translating the target in the z-direc- tation using a linear array probe for dual mode PAT-US
tion and concatenating the two-dimensional cross- imaging described in Kim et al. [29].
sectional slices acquired at each axial step. This
system was used to acquire structural images of the function of the vasculature while the US image provides
abdominal, thoracic and heart regions. Its high image information on the surrounding tissue morphology
frame rate (10 Hz) enabled dynamic events such as based on its elasto-mechanical properties. In this way,
motion of the heart chambers and ICG-mediated the different but complementary contrast provided by
kidney perfusion to be visualized in real time. Another each modality can be used in combination to provide
approach employed a fixed 128-element linear ultra- additional diagnostic information.
sound array and a rotating target holder to Although more versatile, particularly for clinical use,
implement a two-dimensional cylindrical detection image quality provided by a planar detection geometry
scheme [77]. This system was used to obtain ex vivo rarely matches that image quality provided by spherical
PA images of the mouse upper body and both PA and cylindrical geometries because of the limited detec-
and US images of phantoms. tion aperture. As well as introducing artefacts and
Planar scanners: Although spherical and cylindrical distortion, this also reduces lateral spatial resolution—
detection geometries can provide the large angular indeed, it is the limited view, not frequency-dependent
aperture required for an accurate image reconstruction, acoustic attenuation, that tends to limit lateral resol-
their applicability is constrained by the need for access ution. Vertical resolution, on the other hand, is
to all sides of the target. They are not suitable for ima- relatively independent of the detection aperture and is
ging highly superficial features, such as the skin limited by acoustic attenuation. This results in a dis-
microvasculature, or if strongly echogenic structures parity between the lateral and vertical resolution and
such as bone or lung are situated along the acoustic gives rise to an anisotropic spatial point spread func-
propagation path. These circumstances call for the tion, itself a source of image distortion. Delivering the
more versatile planar detection geometry in which the excitation laser light can also be problematic if an
detection is performed over a finite plane using a two- array of receivers rather than a single mechanically
dimensional ultrasound array or its one-dimensional scanned receiver is employed. The usual solution is to
equivalent, detection over a line using a linear array. vertically offset the array from the tissue surface, fill
PA imaging instruments that employ this geometry the intervening space with an optically transparent
begin to resemble conventional diagnostic clinical US acoustic couplant and deliver the laser light obliquely
scanners, in some cases comprising a hand-held array to the tissue surface beneath the array as illustrated
probe that is acoustically coupled to the skin and in figure 7a. This is readily achievable if a linear array
moved around while viewing the images in real time. is used for two-dimensional imaging as the excitation
Indeed, a variety of PA imaging instruments use exist- laser beams can be delivered orthogonal to the length
ing commercially available diagnostic scanners, axis of the array and so only have to ‘clear’ the width
suitably modified so that the RF acquisition can be trig- of the transducer elements. However, the requirement
gered by the excitation laser in order to detect PA for a spacer does impose a limitation on the dimensions
waves as well as US echoes [44,77– 80]. Co-registered of a two-dimensional array that can be used—the larger
PA and US images can then be reconstructed either the area, the greater the required spacer thickness which
using the hardware beamformer of the scanner or in a in turn reduces the effective detection aperture and thus
post-processing step using a reconstruction algorithm. image quality.
Thus, the absorption-based contrast provided by the Figure 7b shows a specific implementation in which a
PA image can be exploited to reveal the structure and commercially available diagnostic linear array probe

(a) part a consequence of the limited detection aperture

that linear array-based systems provide. It is also
because the transducers used in clinical US scanners
S tend to operate in the sub-10 MHz range and are res-
S
onant to some degree. They are, therefore, often
insufficiently broadbanded for the detection of PA
waves emitted by superficial structures, the frequency
content of which can extend from the low megahertz
to several tens of megahertz. To address this a custom
(b) designed high-frequency 48-element linear array fabri-
cated from 2 – 2 piezocomposite elements with a centre
frequency of 30 MHz and a 70 per cent fractional band-
S
width has been developed for PA imaging [82,83]. As
S
with other linear arrays used for PA imaging, it pro-
vides a real-time two-dimensional B-scan frame rate,
in this case 50 fps. By rapid mechanical scanning, a
volumetric frame rate of 1 fps for 166 B-scans was
reported and the system used to image the subcutaneous
vasculature in the human and rat.
Figure 8. (left) In vivo PAT images and (right) corresponding
US images of a vein at the interior part of the medial lower leg A different approach, based on an optical method of
obtained using a linear ultrasound array [44]. The image area ultrasound detection, has been explored with a view to
is 2.62 cm (ticks every centimetre) (a) Cross-sectional image overcoming the limitations of piezoelectric-based detec-
from the interior part of the leg. In the PAT image (left), the tion for planar geometries. It employs a transparent
skin (S) is visible as a black line. Black and white arrows Fabry – Perot (FP) polymer film etalon that comprises
identify the location of common blood vessels observed in a polymer film spacer sandwiched between a pair of mir-
both PAT and US images (b) Images acquired at the same rors [84,85]. Acoustically induced changes in the optical
location but in the orthogonal plane. Laser wavelength: thickness of the spacer modulate the reflectivity of the
760 nm. etalon which can be detected by measuring the changes
in the reflected power of an incident laser beam. By
raster scanning a focused laser beam across the surface
and a pair of fibre bundles are integrated to form a of the sensor, an incident PA wavefront can therefore
hand-held dual-mode US-PA imaging head [29]. This be spatially mapped in 2D (figure 9a). There are several
system provides co-registered two-dimensional PA and advantages of this concept. First, the mirrors of the
US images at real-time image frame rate (10 fps) and etalon can be designed to be transparent to the exci-
is intended for a relatively deep tissue imaging appli- tation laser wavelength. The sensor head can therefore
cation: sentinel lymph node detection in the breast. be placed directly on the surface of the skin, and the
Other similar schemes have been demonstrated for laser pulses transmitted through it into the underlying
visualizing superficial vascular anatomy, including one tissue. It thus avoids the detection aperture limitations
that employs a 64-element, 7.5 MHz linear array that imposed by an acoustic spacer required for piezoelectric
also provides real-time laser PRF-limited frame rates detection methods. It also provides an inherently broad-
(7.5 fps) [44]. An example of a PA and US image pro- band response from DC to several tens of megahertz
vided by this system is provided in figure 8 showing and very fine spatial sampling of the incident acoustic
the blood vessels at a depth of around 10 mm in the field. The effective acoustic element size is defined, to
leg. Fronheiser et al. [79] describe the use of a dual a first approximation, by the diffraction limited dimen-
PA – US system that uses a 128-element linear array sions of the focused interrogation laser beam. The
probe from a commercial ultrasound scanner to image notional element size and interelement spacing can
the vasculature in the arm in real time with a view to therefore be on a scale of tens of micrometres. Perhaps,
identifying vessels prior to haemodialysis. A 1.75 D most importantly, the small element size is achieved
phased array comprising 1280 elements has also been with significantly higher detection sensitivity that can
used to provide three-dimensional co-registered PA be provided by similarly broadbanded piezoelectric
and US images [80]. The Visualsonics small-animal receivers of the same element dimensions [85]. The com-
ultrasound scanner has also been adapted to provide a bination of a transparent sensor head and wideband
dual-mode PA – US imaging capability in an instrument acoustic performance attributes makes this type of
that is now commercially available [81]. sensor particularly well suited to imaging superficial
The use of existing, commercially available ultra- features located within a few millimetres of the tissue
sound scanners is a convenient and relatively surface. In these circumstances, the short acoustic
inexpensive means of implementing PA imaging that propagation distances involved mean that the PA
exploits the advances in piezoelectric array technology, signal is only weakly bandlimited by acoustic attenu-
hardware beamformers and RF acquisition electronics ation and can therefore be extremely broadbanded
that have taken place in recent years in diagnostic ultra- with a frequency content extending to several tens of
sound imaging. However, PA image quality obtained megahertz. The requirement for a broadband omnidir-
using these systems tends to be somewhat limited, par- ectional point detector for an accurate image
ticularly for superficial imaging applications. This is in reconstruction then becomes particularly challenging

(a) (b)
2 mm
a b c d e f
z
y
x
fabry–perot sensor head y
scanning interrogation beam x
excitation laser 0
pulses
scanning system
z (mm)
2
x–y scanner a b c e
4 d f
6
0 2 4 6 8 10 12 14 16 18
x (mm)
(c)
1.0 mm
1.5 mm
x z
1.5 mm
2.0 mm
6 mm
20 mm
20 mm
2.0 mm
20 mm 2.5 mm
20 mm A
2.5 mm
4.0 mm
Figure 9. (a) Schematic of FP-based photoacoustic scanner used to acquire a three-dimensional image of the vasculature in the
palm [45]. (b) Lateral MIP image (top) and vertical (x– z) slice image (bottom) taken along horizontal yellow dotted line on MIP.
Grey arrows indicate the contour of the skin surface. (c) Left: photograph of the imaged region, middle: volume rendered image.
An animated representation of this image can be viewed online at electronic supplementary material, movie S6, and right: lateral
slices at different depths. The arrow ‘A’ indicates the deepest visible vessel, which is located 4 mm beneath the surface of the skin.
Laser excitation wavelength: 670 nm.
using conventional piezoelectric detection as it image dataset obtained by scanning the human palm
demands element dimensions of a few tens of micro- [45]. The sensor used for this demonstration had a
metres. Fabricating piezoelectric elements of such peak noise-equivalent-pressure (NEP) of 0.2 kPa (over
small size with adequate detection sensitivity is a measurement bandwidth of 20 MHz) and a 23 dB
problematic since their sensitivity scales with active detection bandwidth from 100 kHz to 22 MHz. The
area. By contrast, the FP sensor sensitivity is, to a interrogation laser beam spot size was 64 mm, spatial
first approximation, independent of element size. sampling interval was 250 mm and an FOV was
A schematic of the system is shown in figure 9a along 20 mm. No signal averaging was used. The excitation
with several representations of the three-dimensional wavelength was 670 nm and the incident fluence

incident on the skin surface was 10 mJ cm22, and thus generated at the surface. This can otherwise obscure
below the MPE of 20 mJ cm22 for skin at this wave- later arriving signals owing to ringing if a resonant
length [86]. Figure 9b shows a lateral MIP and a transducer such as one fabricated from PZT is used.
single vertical slice through the centre of the lateral To acquire a 3D image, both the transducer and the
MIP as indicated. This vertical slice shows the contour excitation beam are mechanically scanned together
of the skin surface as well as several underlying sub- over a planar surface, generating and detecting PA
dermal blood vessels. Figure 9c shows a volume-ren- waves at each step of the scan. The resulting two-
dered representation of the reconstructed image and a dimensional sequence of detected acoustic signals or
series of lateral slices at different depths. These A-lines, each of which represents a depth profile of
images show the subcutaneous vasculature to a depth absorbed energy, is then rectified, envelope detected,
of approximately 4 mm—the deepest lying vessel is spatially resolved and mapped to a greyscale to form
indicated by the arrow ‘A’ on both the volume-rendered a three-dimensional image directly. Unlike PAT, an
image and the deepest lateral slice. The time taken to explicit reconstruction algorithm is not employed
acquire the three-dimensional image data shown in although this is more a difference in implementation
figure 9 was approximately 10 min, limited by the than physical principle—if PAT image reconstruction
sequential nature of the detection and the low PRF of is regarded as in silico point-by-point receive beam
the excitation laser (10 Hz). There is, however, scope forming, then it is apparent that the focused receiver
to increase acquisition speed, potentially obtaining itself performs this function for each x – y scan position.
real-time three-dimensional image acquisition rates, Figure 10 shows an image of the skin vasculature in the
through the use of higher repetition rate laser systems forearm region obtained using AR-PAM illustrating the
or parallelizing the sensor read-out scheme using full- dermal and sub-dermal vasculature. The system was
field illumination and a photodetector array as also used to image a benign-pigmented skin lesion to
described in Lamont & Beard [87]. As well as imaging demonstrate the potential for identifying melanomas
the skin vasculature, the system has been used for ima- via their morphology and composition. The transducer
ging tumour vasculature [45,88], the mouse brain [89] frequency was 50 MHz, the fractional bandwidth 70 per
and embryo [50,90]. cent, the aperture 5.8 mm, depth of focus 0.3 mm and
focal length 6.7 mm giving a lateral resolution at the
4.2. Photoacoustic microscopy focus of 45 mm and a vertical resolution of 15 mm. For
an 8 8 mm FOV and a step size of 20 mm, the
PA microscopy refers to techniques in which a PA acquisition time was approximately 5 min. Similar
image is obtained by mechanically scanning either a high-resolution AR-PAM systems have been used in
focused ultrasound detector or a focused laser beam. preclinical studies of the mouse brain [96] and skin
The image is then formed directly from the set of microvasculature [101]. The length scale that
acquired A-lines, without the aid of a reconstruction AR-PAM can address is scalable. By reducing the
algorithm as in PAT. If a focused ultrasound detector transducer bandwidth and increasing the focal length,
is used, it is termed acoustic resolution PA microscopy it has been used to visualize organs at depths of several
(AR-PAM) since axial and lateral spatial resolution is centimetres in the mouse, albeit with a reduced spatial
defined by the physics of ultrasound propagation and resolution of a few hundred micrometres [95,96].
detection. If a focused laser beam is used, it is termed As figures 4 – 6,9 and 10 show, both AR-PAM and
OR-PAM since the spatial resolution in at least one PAT can provide compelling images over a range of
plane (usually, the lateral) is defined by the spatial spatial scales. However, there are significant differences
characteristics of a focused laser beam propagating in between the two techniques in terms of performance,
tissue. Despite the name, PA microscopy, unlike its cost and complexity. An important difference relates
optical equivalent, does not necessarily imply the obser- to the factors that affect spatial resolution over the
vation of anatomy on a small length scale—AR-PAM, FOV. For a planar detection geometry, the lateral
for example, can be used to image to depths of several spatial resolution of PAT at a specific point in the
centimetres. illuminated volume is ultimately defined by frequency-
dependent acoustic attenuation and the solid angle
4.2.1. ‘Acoustic resolution’ photoacoustic microscopy. subtended by the detection aperture to that point.
The term AR-PAM is usually used to describe Thus at all points in the three-dimensional FOV, the
implementations that employ a single mechanically lateral resolution can be said to be acoustic diffraction-
translated or rotated focused transducer to map the limited which represents the fundamental resolution
PA signals [91– 100]. Figure 10 shows one approach limit. However, in AR-PAM, this is only so at the
that has been extensively used. It comprises a focused depth corresponding to the location of the transducer
receiver around which the excitation light is delivered, focus. Elsewhere, the lateral resolution degrades rapidly.
usually using a conical lens that produces a hollow To obtain true acoustic diffraction-limited resolution
cone of weakly focused light within which the transdu- over the entire three-dimensional FOV necessitates
cer is placed. Optical focusing is not essential as it does axial as well as lateral mechanical scanning with a conse-
not serve to localize the signal and thus does not influ- quent increase in acquisition time. The problem can be
ence the spatial resolution. Full-field illumination could ameliorated to some extent using synthetic aperture
equally be used as in PAT but confining the excitation focusing [102]. Another possibility that has been
beam by weakly focusing it reduces the laser energy explored is the use of an axicon reciever [103]. Based
requirements. It also reduces the large PA signal on the theory of non-diffracting acoustic Bessel beams,

(a) y optical fiber dye Nd:YLF

x laser laser
conical lens
z
trigger
translation motor
stages control
ultrasonic
sync
optical transducer
condenser
PC DAQ
water
tank amplifier
acoustic
focus volunteer’s
weak optical arm/hand
focus
max
(b) (c) 5
optical absorbtion
4
2
1
min
1 mm
(d)
epiderm.-derm.juntion
1 2 4
3 5
dermis subpapillary plexus
1 mm
Figure 10. Acoustic-resolution photoacoustic microscopy (AR-PAM) system used for imaging the skin vasculature [94]. (a) Sche-
matic of system, (b) region of forearm scanned, (c) lateral x–y MIP image (FOV ¼ 8 8 mm), (d) vertical x –z slice image taken
along vertical line in (c). Laser excitation wavelength: 584 nm.
this can, in principle, provide a non-divergent receive inexpensively implemented as a laboratory-based

focus that could potentially eliminate the need for research tool using a single mechanically scanned recei-
depth scanning. ver. However, there is little scope to parallelize the
A further distinction between PAT and AR-PAM detection using an array to overcome the limitations
lies in the complexity and cost of implementation. on acquisition speed imposed by mechanical scanning.
The laser power requirements for AR-PAM are more PAT detection, on the other hand, can readily, if
modest than those of PAT. In PAT, the entire expensively, be parallellized using an array of receivers
three-dimensional FOV must be irradiated whereas in and, in principle, provide real-time three-dimensional
AR-PAM only the region that coincides with the receive image frame acquisition. For high-resolution superficial
beam profile of the transducer is required to be illumi- imaging to depths of few millimetres, AR-PAM has the
nated for each scan position. By confining the internal advantage that, unlike PAT, it does not require acous-
light distribution to this region by weak focusing, a tically small receivers on a tens of micrometre scale.
laser pulse energy that is an order of magnitude lower As discussed in §4.1.1, these are difficult to fabricate
than that required for PAT can be used. This allows a with adequate detection sensitivity using piezoelectric
greater variety of laser sources to be used, particularly receivers although the optical etalon detectors
those that offer higher PRFs and a tunable output. described in the previous section may provide a solution
AR-PAM can also be straightforwardly and to this. In summary, PAT offers the ultimate imaging

(a)
CCD
mirror
BS
PD
objective
US
corl
RAP
SO
RhP
y
x
z membrane petri dish
(b) (c)
1 capillary bed
50 mm
relative HbT
500 mm RBC
0
Figure 11. Optical-resolution photoacoustic microscopy (OR-PAM) scanner used for in vivo imaging of the mouse ear [118]. (a)
Schematic of system BS, beam splitter; PD, photodiode; CorL, correction lens; RAP, right-angle prism; SO, silicone oil; RhP,
rhomboid prism; US, ultrasonic transducer (50 MHz). The CCD is used to view the imaging region. The components that lie
within the dotted rectangle form the scan head, which is mechanically translated in order to acquire an image. (b) In vivo
image of the microvasculature in the mouse ear. (c) Expanded region shows capillary network and red blood cells (RBC)
within a capillary. Excitation wavelength: 570 nm.
performance in terms of spatial resolution and Figure 11 shows a schematic of one implementation.
acquisition speed, but in its most sophisticated form is A high numerical aperture (NA) optical lens is used to
considerably more complex and expensive than focus the excitation laser beam onto the tissue surface
AR-PAM. and an optically transparent acoustic reflector directs
the PA wave to an ultrasound transducer. By mechani-
cally scanning the focused excitation beam and the
4.2.2. ‘Optical resolution’ photoacoustic microscopy. A transducer together and recording the detected
second mode of PA microscopy is the so-called ‘opti- A-lines at each point, a three-dimensional image can
cal-resolution’ mode (OR-PAM) in which optical be formed as in AR-PAM. However, unlike AR-PAM,
rather than acoustic confinement is exploited for local- lateral resolution for depths less than approximately
ization [99,104 – 118] purposes. It is in many ways 1 mm (beyond this optical scattering defocuses the
more akin to optical microscopy than acoustic imaging beam and degrades resolution) is defined by the diffrac-
in that lateral resolution is defined by the dimensions of tion-limited dimensions of the focused laser beam.
a tightly focused diffraction-limited laser beam which is Compared with AR-PAM, much higher lateral resol-
used to generate the PA waves. Also in common with ution of the order of a few microns can therefore be
optical microscopy, it is a strictly superficial imaging achieved over this depth range. To achieve a compar-
technique with a maximum penetration depth of able acoustic diffraction-limited resolution with
approximately 1 mm in most tissues due to optical scat- AR-PAM would require a broadband acoustic fre-
tering. However, an important distinguishing feature is quency content extending to several hundred MHz.
that unlike any of the current variants of optical Although it is possible to generate PA waves with
microscopy, it provides optical absorption-based such a broad bandwidth, acoustic attenuation at such
image contrast. high frequencies is much higher than optical

attenuation and would limit penetration depth to of OR-PAM are inevitably limited owing to the short
approximately 100 mm [105]. By contrast, in OR-PAM, penetration depth it provides.
the maximum penetration depth of approximately
1 mm is a consequence not of optical or acoustic attenu-
4.3. Endoscopic devices
ation but the spreading and distortion of the excitation
laser beam owing to tissue optical scattering which There are several potential clinical applications in which
prevents a tight focus from being maintained beyond the target tissue can only be accessed by introducing a
this depth. miniature endoscopic probe percutaneously or through a
The high lateral resolution available to OR-PAM natural orifice. Among these are the clinical assessment
enables en face images of individual capillaries [105] of coronary artery disease, prostate cancer and gastrointes-
and even individual red blood cells (RBCs) to be visual- tinal pathologies. A number of prototype PA endoscopic
ized in the mouse ear as shown in figure 11 [118]. or intravascular devices, conceptually similar to conven-
Although these images illustrate the very high optically tional US probes, have been developed [122–130] for
defined lateral resolution of OR-PAM, vertical resol- these applications. Intravascular PA imaging employs a
ution is about one of magnitude lower (typically sideways looking rotating probe in a manner analogous to
approx. 10 mm) as it depends on acoustic propagation IVUS. Figure 12 shows one such configuration [127]. It
and detection and is thus limited by acoustic attenu- comprises a 1 mm diameter 30 MHz PZT transducer
ation or the detector bandwidth as in AR-PAM or located at the end of a catheter in close proximity to the
PAT. The disparity in lateral and vertical resolution tip of an angle polished optical fibre that delivers the PA
can be ameliorated in part using nonlinear optical exci- excitation laser pulses. The total outer diameter of the
tation methods which provide optically defined axial as assembly is 1.25 mm and thus of comparable dimensions
well as lateral spatial localization. However, these tech- to those required for intravascular imaging in human cor-
niques require three-dimensional scanning to provide a onary arteries. As well as providing cross-sectional PA
volumetric image [112]. Indeed, even in conventional images by rotating the catheter and acquiring PA A-lines
OR-PAM, a degree of axial scanning is required for at each angular step, the transducer was operated in
three-dimensional imaging as the highest lateral resol- pulse-echo mode enabling a co-registered IVUS image to
ution occurs only at the depth of the optical focus. In be acquired simultaneously. Figure 12 shows PA and US
addition to the configuration shown in figure 11, several images obtained ex vivo in human coronary arteries. In
other implementations have been reported. Xie et al. order to identify lipid-rich regions, the PA images were
[107] describe an alternative scanning method that obtained at the peak lipid absorption wavelength of
was also used to image blood vessels in the mouse ear 1210 nm and, for comparison, at 1230 nm where absorp-
[119] and subsequently the retina [121]. This approach tion is low [131,132]. A variety of other PA intravascular
avoids the frame rate limitations of mechanically probe designs based on different light delivery and
scanned methods by optically scanning a focused PA ultrasound detection mechanisms have also been demon-
excitation beam across the tissue surface using an x – y strated [123,125,128]. As yet, none has been evaluated in
galvanometer scanner and detecting the PA signals vivo. This is in part due to the technical challenges involved
with a single stationary planar ultrasound receiver in integrating the delivery optical fibre with a conventional
offset from the scan area. Although higher frame rates piezoelectric detector while achieving the necessary level of
are attainable with this method, they are achieved at miniaturization for intravascular use. The use of the optical
the cost of a reduced FOV, limited by the directional ultrasound detector referred to in §4.1.1 may be able to
response of the receiver. A further development in address this. A 250 mm diameter sideways looking PA ima-
OR-PAM involves the use of adaptive optics. To date ging probe based on this type of sensor, the smallest PA
this has been used to correct for aberrations in the probe developed to date, has been described [128]. Other
scan lens and light delivery optics [114] but could also miniature probe designs have been proposed or evaluated
potentially be employed to compensate for optical for other applications. One of these employs a commercial
wavefront distortion in tissues to extend penetration 128-element forward-looking endocavity ultrasound
depth. The optical detection system described in probe to detect the PA waves and has been proposed for
§4.1.1 has also been used to perform both PAT prostate imaging [124] applications. Yang et al. [122]
and OR-PAM enabling the different spatial scales describe a sideviewing dual-mode PA and US imaging
of each modality to be addressed with the same probe. This uses a rotating 458 optical and acoustic
instrument [113]. mirror located at the tip of the device to deliver the exci-
Although in vivo OR-PAM is very much a superficial tation light and receive the PA signal thereby avoiding
imaging technique, limited mainly to imaging the mouse the need to rotate the catheter shaft. The diameter of the
ear and brain (with the skin removed), it provides a probe is relatively large at 4 mm although sufficiently
useful adjunct to the current armoury of PA imaging small for gastrointestinal use.
techniques. A key attribute is that PA Doppler flowme-
try (§5.2) and the spectroscopic measurement of blood
oxygenation (§5.1) are significantly less challenging to 5. PHOTOACOUSTIC SENSING
implement using OR-PAM than PAT and AR-PAM. TECHNIQUES
These functional capabilities make OR-PAM a powerful
5.1. Photoacoustic spectroscopy
research tool for basic preclinical investigations of oxygen
supply and delivery at capillary level. With the exception A major advantage of PA imaging is that image con-
of opthalmic applications [119,121] clinical applications trast can be selectively enhanced for specific tissue

(a) (b)
catheter
Ca
laser
trigger
* Pf
delay
Lu
AWG sig exp lim
out
sig
DAQ in bpf amp bpf
Pf
(c) (d)
1 mm
Figure 12. Dual mode intravascular PA and US imaging probe used to image ex vivo human coronary arteries [127]. Left panel:
(upper) schematic of experimental set-up. (lower) photograph of distal end of catheter. Right panel: (a) Histological section show-
ing a lipid-rich plaque (asterisk) and a region of calcification (Ca). Lu, lumen; Pf, peri-adventitial fat. (b) IVUS image, (c)
intravascular PA image obtained using an excitation wavelength of 1210 nm (high lipid absorption) and (d ) intravascular PA
image obtained using an excitation wavelength of 1230 nm (low lipid absorption).
components by tuning the excitation laser wavelength can therefore provide information about disease pro-
to the absorption features of their constituent chromo- cesses at a cellular or molecular level.
phores. For example, as discussed in the previous As the above indicates, PA spectroscopy is an impor-
section, by exciting at a wavelength that corresponds tant adjunt to PA imaging techniques in that it
to the 1210 nm characteristic lipid absorption peak, provides it with a functional and molecular imaging
the presence of lipid-rich atheromatous plaques can be capability. However, achieving absolute or even relative
identified [131,132]. Although useful, it is possible to quantification of chromophore concentrations using PA
go much further than this type of simple contrast spectroscopy represents a significant challenge [21,134].
enhancement. By acquiring images at multiple wave- Although as discussed in §2, PA contrast is often said to
lengths and undertaking a spectroscopic analysis, the be ‘absorption-based’, this does not mean that it is
concentration of specific chromophores can be quanti- directly proportional to the absorption coefficient—a
fied in a manner analogous to conventional optical PA image is a representation of the absorbed optical
transmission spectroscopy [133] assuming that the energy distribution not the absorption coefficient distri-
spectral characteristics of the chromophores and scat- bution as sometimes assumed. The absorbed optical
terers present are known. For example, the absorption energy at a point is the product of the local absorption
spectrum of blood at visible and NIR wavelengths is coefficient and the fluence, the latter being dependent
strongly dependent upon its oxygen saturation (sO2), on the distribution of absorption and scattering coeffi-
a consequence of the significant spectral differences cients over the entire illuminated tissue volume. The
between oxyhaemoglobin (HbO2) and deoxyhaemglo- spectra of all the different chromophores and scatterers
bin (HHb) as shown in figure 1. With knowledge of in this volume will therefore, to some extent, be encoded
these spectral differences, it is then possible to quantify onto the absorbed energy spectrum at a specific point.
the concentrations of HbO2 and HHb and estimate sO2, An obvious example occurs when two spectrally distinct
an important physiological parameter intimately absorbers are located such that one lies directly beneath
related to a broad range of pathophysiological processes the other. Since the light will have passed through the
such as angiogenesis and tissue inflammatory processes. top absorber to reach the one beneath it, the absorbed
As well as imaging endogenous chomophore concen- energy spectrum of the latter will be composed of the
trations, spectroscopic techniques provide a means of spectral characteristics of both. Indeed given that in tis-
detecting and quantifying the accumulation of targeted sues, a photon can be scattered from any point to
contrast agents used in PA molecular imaging. Here, a another, this type of spectroscopic cross-talk can
spectrally distinct absorber such as a dye molecule or occur even if two absorbers are located side by side at
nanoparticle is tagged to targeted compound that the same depth or indeed, if there is significant back-
binds to a disease-specific receptor such as a cell-surface scattering, from a deeply lying absorber to another
protein or enzyme. Using spectroscopic methods to situated directly above it. As a consequence, the pro-
quantify the local accumulation of the contrast agent blem of recovering chromophore concentrations differs

from simple conventional optical transmission spec- concentrations until the difference between the
troscopy [133]. In the latter, the signal is assumed to measured images and those predicted by the model is
be directly proportional to the absorption coefficient minimized. Because the model can be formulated as a
which allows the problem to be formulated as a set of function of an arbitrary spatial distribution of any
simultaneous linear equations (one for each wavelength) number of chromophores, this approach is applicable,
which can be efficiently inverted to recover the chromo- in principle, to any tissue geometry or type, albeit at
phore concentrations. Since PA image contrast is not some computational expense especially if implemented
proportional to the absorption coefficient however, in three dimensions. This method has been demon-
attempting to apply this type of simple linear inversion strated experimentally using simulated data [137] to
pixel-by-pixel to a set of multi-wavelength PA images obtain two-dimensional maps of chromophore concen-
is unlikely to be successful [25]. There are some trations. It has also been used (in a reduced form to
exceptions. For example, if the target is optically homo- minimize computation time), for the quantitative
geneous [135] or if the chromophore of interest is measurement of HbO2 and HHb and blood sO2 from
surrounded by chromophores that are either of uniform single-point PA signals generated in tissue mimicking
spectral characteristics or relatively weakly absorbing or phantoms [25] and subsequently for the recovery of
both. However, these conditions rarely exist in biologi- two-dimensional images of chromophore concentrations
cal tissues in vivo. An exception in which a simple [138]. An alternative strategy is to attempt to use a
linear spectroscopic inversion can be valid occurs light transport model to recover absorption coefficient
when imaging highly superficial features within a few images at individual wavelengths and then apply a
hundred micrometres of the surface as in OR-PAM linear inversion, pixel by pixel, to recover chromophore
(§4.2.2). Under these conditions, the wavelength depen- concentrations. However, because of the absorption-
dence of the external light distribution can largely be scattering non-uniqueness associated with absorbed
ignored. The assumption that image contrast is directly energy images [137], it is not generally possible to
proportional to the absorption coefficient can then obtain absorption coefficient images at a single wave-
reasonably be made and the quantitative measurement length without incorporating additional information
of chromophore concentrations using a linear inversion such as the scattering distribution, which is not usually
becomes possible. known, or fluence measurements obtained from another
It is now widely accepted that addressing the cor- imaging modality (e.g. optical tomography [139,140]),
rupting influence of the wavelength dependence of the which incurs additional experimental complexity. The
fluence distribution is key to achieving accurate quanti- method described above overcomes this by fitting the
tative spectroscopic measurements. One approach is to forward model to all of the measured multi-wavelength
modify the above-mentioned linear spectroscopic PA images simultaneously—in essence the additional
decomposition by incorporating empirical correction information provided by the known spectral character-
or calibration factors that account for the wavelength istics of the constituent chromophores and scatterers
dependence of the external light distribution. For ameliorates the non-uniqueness. Xiao et al. [141]
example, it has been suggested that the spectral charac- describe a different model-based inversion scheme and
teristics of the tissue overlaying the region of interest suggest that both chromophore concentrations and
could be accounted for by measuring the light trans- sound speed can be recovered.
mission as a function of wavelength through a sample Other methods that seek to compensate for the cor-
of excised tissue of the same type [136]. Another rupting effect of the wavelength dependence of the
method of obtaining the same type of information fluence distribution are based on the use of additional
involved measuring the PA spectrum of a black plastic information. One of these relies on the introduction of
absorber (of presumably uniform or known spectral a contrast agent of known absorption spectrum [142].
characteristics) embedded beneath the skin of a live Another employs diffuse reflectance spectroscopy [143]
mouse at the depth of interest [96,101]. The drawback to estimate tissue optical properties which are then
with such empirical correction factors, apart from the used in a light propagation model to correct for fluence
invasive and clinically inapplicable methods required changes. The use of diffuse optical tomography to inde-
to obtain them, is that they are highly dependent pendently estimate the fluence as mentioned above has
upon the tissue structure, composition and physiology, also been explored [139,140].
particularly perfusion and oxygenation status. Given
that these vary significantly between different tissue
5.2. Photoacoustic Doppler flowmetry
types and over time, this inevitably limits their general
applicability. A fundamentally different approach, and A further functional capability is the measurement of
one that has the potential to avoid these limitations, blood flow velocity using Doppler flowmetry techniques.
is to use a nonlinear model-based inversion scheme. In This would be useful in its own right, for example, to
this, the wavelength-dependent light distribution is study flow in tumour vessels where the tortuous
accounted for, not by using empirical correction factors, nature of the microvasculature can lead to chaotic
but through an explicit mathematical representation of and variable blood flow which can inhibit therapeutic
light transport within a forward model. The model pro- response. However, if both blood flow and sO2 can be
vides multi-wavelength images of PA images or signals measured simultaneously (the latter as described in
as a function of the spatial distribution of chromophore §5.1), there is also the prospect of being able to estimate
concentrations. The latter are then estimated by invert- oxygen delivery and thus provide a measure of oxygen
ing the model, achieved by varying the chromophore consumption—an important physiological parameter

related to tissue metabolism that is difficult to measure been estimated to be +1 mm s21 [151]. To date, depth-
non-invasively using other methods without employing resolved PA Doppler flowmetry using the acoustically
contrast agents. PA flow measurements can be made in defined PA modes (AR-PAM or PAT) has not been
a manner analogous to conventional Doppler ultra- demonstrated in blood, only in tissue-mimicking phan-
sound—that is to say by recovering the Doppler toms [147–150]. The challenge in making measurements
frequency, phase or time shift encoded onto PA waves in blood relates to its optical heterogeneity. If the spatial
emitted by moving red blood cells (RBCs). Unlike Dop- separation between individual RBCs or clumps of RBCs
pler ultrasound, however, the detected acoustic signal is is small compared with the minimum detectable acoustic
emitted by the blood cells as opposed to being weakly wavelength, blood will approximate to a spatially homo-
reflected from them. This offers significant SNR advan- geneous absorber and compromise the ability of time
tages especially when measuring flow in microvessels as correlation and possibly other time-domain methods to
these exhibit low echogenicity. Furthermore, Doppler make velocity measurements. Whether this will prove
ultrasound measurements of the relatively low flow to be fundamentally limiting remains to be seen. The suc-
speeds (less than 50 mm s21) in microvessels can be cor- cess of pulsed wave Doppler ultrasound [152] and cross
rupted by the much larger backscattered signal from correlation pulse-echo ultrasound measurements of
the surrounding tissue which can move at comparable blood flow [151] suggests that at least some optimism is
speeds owing to respiratory or cardiac motion. In PA warranted.
Doppler flowmetry, this is less problematic owing to In the optically defined mode, OR-PAM, however,
the much stronger optical absorption of blood com- the spatial heterogeneity issue does not present a funda-
pared with that of the vessel wall and surrounding mental limitation. This is because the excitation spot
tissue. For these reasons, PA Doppler is envisaged pri- size (approx. a few micrometres) is comparable to that
marily for the measurement of the low flow velocities of a single RBC allowing the spatial heterogeneity in
in the microvasculature, not fast moving blood in the the optical absorption of blood to be directly resolved.
major arteries which is currently well served by As a consequence, measurements of blood flow have
conventional pulsed wave Doppler ultrasound. successfully been demonstrated in vivo. One method
Although the principles of PA Doppler flowmetry were is based on recovering the acoustic bandwidth broaden-
outlined over a decade ago [144], it is only relatively ing arising from the frequency difference between the
recently that it has been demonstrated experimentally. extremities of a wavefront emitted by an absorber
PA measurements of velocity in a tissue-mimicking phan- moving orthogonal to the axis of a focused transducer
tom have been obtained by recovering Doppler frequency [153]. The frequency shift is estimated by computing
shifts using continuous wave (CW) excitation [145,146]. the autocorrelation of sequentially acquired A-lines
However, in common with CW Doppler ultrasound, produced by successive laser pulses. Being a pulsed exci-
this approach cannot readily provide depth-resolved tation scheme, spatially resolved flow measurements
measurements. The use of pulsed excitation overcomes can be made. This approach has been demonstrated
this limitation. Time-resolved spectral analysis of tone- in vivo for measuring velocities up to approximately
burst excited PA waves [147] has been successful in 7.4 mm s21 with a resolution of 0.1 mm s21 and thus
quantifying Doppler frequency shifts for velocities within the velocity range of blood flow in microvessels
between 3.5 and 200 mm s21 in phantoms [148]. There [154]. It has also been used to obtain two-dimensional
is inevitably a velocity-spatial resolution compromise maps of flow in the vasculature of the mouse ear and
with this method—a long-duration toneburst provides combined with spectroscopic measurements of sO2 to
high spectral and therefore velocity resolution but at estimate changes in the metabolic rate of oxygen
the cost of spatial resolution. In Sheinfeld et al. [148], a [155], a measure of tissue oxygen consumption. The
toneburst duration of 3 ms was used. This provided a challenge ahead in PA Doppler flowmetry is to translate
velocity resolution of 1.5 mm s21 but a relatively low the success of OR-PAM flowmetry to the acoustically
depth resolution of 4.5 mm. Although a source emitting defined modalities AR-PAM and PAT in order to
in the telecoms C-band (1530 nm–1565 nm) was used make flow measurements beyond the optically defined
to demonstrate the concept in a phantom [148], the lim- 1 mm penetration depth limit of OR-PAM.
ited availability of quasi-CW-modulated laser sources
with sufficient output power at wavelengths suitable for
5.3. Photoacoustic thermometry
in vivo use (i.e. ,900 nm) may present a practical limit-
ation. An alternative Doppler flowmetry method based The temperature dependence of the Grüneisen coeffi-
upon pulsed excitation and time correlation processing cient, a measure of the thermo-mechanical efficiency
has also been explored [149,150]. This approach, inspired of the PA generation process, provides a means of
by time correlation ultrasound flowmetry [151,152], relies non-invasively obtaining maps of temperature distri-
up on measuring the change in the time of arrival of suc- butions in tissue. This has application in
cessive photoacoustic pulses emitted by a moving cluster thermotherapy in which heat is used to ablate tissue
of RBCs using the cross-correlation function. By using pathologies such as tumours. In photothermal cancer
short (nanosecond) laser pulses, it offers high spatial res- therapy, for example, a targeted contrast agent such
olution and can employ the type of readily available as highly absorbing conjugated gold nanoparticles
pulsed Q-switched lasers used routinely in PA imaging. that selective bind to tumour-specific cells is
Velocity range and resolution are scalable with excitation systemically introduced. The tumour is irradiated
pulse separation: for a pulse separation of 5 ms and an with high-intensity ‘therapeutic’ CW laser light at a
upper limit of 180 mm s21, the velocity resolution has wavelength at which the contrast agent is strongly

absorbing. Only those tumour cells to which the highly NIR optical [27,167] or microwave excitation [7,165]
absorbing contrast agent has an affinity to will be suggest this is feasible. A variety of instrument con-
heated and therefore destroyed leaving the surrounding figurations have been developed including the
normal tissue unaffected. PA imaging could be used in hemispherical PAT detection geometry depicted in
two ways to plan and monitor the treatment. Firstly, a figure 4 for high-speed whole breast imaging [27] or
PA image is directly proportional to absorbed optical cylindrical geometries for two-dimensional cross-sec-
energy. Thus if an image is acquired prior to therapy, tional imaging [74,75,165] of the breast. A planar
it will provide an estimation of those parts of the geometry has also been used in the Twente Photoacous-
tumour that will be heated when the ‘therapeutic’ tic Mammoscope [159– 161]. In this scheme, the breast
CW laser light is delivered to the tumour. Secondly, is lightly compressed between a glass plate and a two-
by acquiring PA images during the therapeutic heating dimensional 590-element PVDF array. The system has
phase, the temperature rise and thus the thermal ‘dose’ been evaluated in a small-scale pilot clinical study
could be monitored in real time. In this way, it would be [160]. Regions of increased absorption were observed
possible to determine which regions of the tumour had in the PA images acquired and attributed to tumour
been heated to a temperature beyond the threshold for vascularization. A hand-held commercial US probe
cell death and adjust the treatment parameters accord- comprising a linear array and a fibre bundle to deliver
ingly. The ability to map temperature could similarly the excitation light that can be moved around the
be applied to the monitoring of HIFU therapy for treat- breast providing two-dimensional PA and US images
ment parameter optimization, as described in §6.1.5. in real time has been developed. It has been evaluated
Several studies have been undertaken to explore the using an animal model for the sentinel lymph node
temperature dependence of time-resolved PA signals detection application referred to above [29,166].
measured at a single point in phantoms and ex vivo Although the clinical utility of PA breast imaging has
tissue samples [156 – 158]. In addition, a study that yet to be established with no major clinical studies
demonstrated the potential to obtain two-dimensional undertaken to date, meeting the basic requirements in
maps of optically induced temperature rises in gold terms of range, resolution and the patient – instrument
nanoparticles embedded in ex vivo tissue samples illus- interface for at least some applications appears feasible.
trates the potential for non-invasive monitoring of
photothermal therapy [78]. The temperature resolution
6.1.2. Skin. Imaging the skin is a relatively undemanding
reported in these studies is of the order of 0.158C.
application as the penetration depth required is modest
and well within that achievable by PA imaging as illus-
trated in figures 8–10. Furthermore, a variety of PAT
6. BIOMEDICAL APPLICATIONS
and AR-PAM instruments in one form or another have
6.1. Clinical already been developed that have the required perform-
ance for visualizing the vasculature and other features
6.1.1. Breast imaging. Imaging breast cancer is an
in the skin [44,83,85,95]. The potential to image the
important potential clinical application of PAT. It is
structure and function of the dermal and sub-dermal vas-
based on the hypothesis that the increased optical
culature as well as melanin content suggest PA imaging
absorption arising from the higher vascularity of a
is applicable to the clinical assessment of skin pathol-
tumour will be revealed as a region of elevated contrast
ogies: for example, to provide a more accurate
on a PA image that can then be used to aid tumour
diagnosis and staging of tumours such as malignant mel-
detection and diagnosis [6,27,159 – 161]. In addition,
anomas and help plan their surgical excision. Other
the ability of PA imaging to map blood oxygenation,
potential applications include the assessment of burn
which may provide an indication of therapeutic
depths, wound healing, plastic surgery procedures and
response, can be exploited to help plan and monitor
superficial soft tissue damage such as pressure sore
treatment. The identification of sentinel lymph nodes
onset and ulceration. Some of these applications have
for needle biopsy guidance via the accumulation of
explored in mouse models initially, although none as
an injected optically absorbing contrast agent is a
yet clinically. For example, preclinical studies have
further application that has been investigated
shown that skin burn depths [168,169] and melanomas
[29,33,162,163]. Pulsed microwaves can also be used
[101,170] can be identified. Most human studies have
[7,164] instead of, or in addition to [74,165], optical
demonstrated the ability of PA imaging to visualize the
excitation. Here, increased RF absorption owing to
skin microvasculature rather than characterize specific
the higher conductivity provided by elevated levels of
pathologies. Favazza et al. [94], however, describe a
ionic water content, a consequence of the increased hae-
study in which it is shown that the structure of a
moglobin concentration and other proteins associated
benign-pigmented skin lesion can be imaged—the impli-
with cancer, offers a potential discriminatory mechan-
cation being that PA imaging could be used to provide
ism that is complementary to that provided by optical
tumour volume, depth and thickness as an aid to the
excitation. The use of microwave radiation also offers
clinical assessment of skin tumours.
the prospect of achieving greater penetration depth.
Combining PA imaging with pulse-echo ultrasound
[29,166] may provide a further source of contrast via 6.1.3. Cardiovascular. As described previously, the
differences in acoustic impedance. A key challenge in strong absorption peak at 1210 nm of lipids offers the
breast cancer imaging is the large penetration depths prospect of identifying lipid-rich atheromatous plaques
required (approx. 6 cm) although several studies using which can exhibit an increased propensity to rupture.

970 nm
2 (b)
(a) (b) (c)
(a)
mm
3 3 mm
4 2 mm
5
1200 nm atheromatous plaque

2 (c)
mm
1 cm 3
4
5
2 4 6 8 10 12 14 16 18 20
mm
Figure 14. Combined ultrasound and PA imaging of a coron-
Figure 13. Photoacoustic imaging of lipid-rich atheromatous ary artery stent embedded in a tissue phantom [173]. (a)
plaques [132]. (a) Photograph of human aorta sample with Ultrasound image, (b) photoacoustic image, (c) fused ultra-
raised plaque. The horizontal dotted line represents the sound and photoacoustic image.
photoacoustic scan line. (b) Two-dimensional photoacoustic
image obtained at 970 nm (low lipid absorption), (c) photoa-
coustic image obtained at 1200 nm (high lipid absorption)
showing the high contrast owing to increased lipid content preclinical animal studies have been undertaken. A dual-
within plaque. mode OR-PAM-OCT instrument has been developed
[120] and used to obtain in vivo images of the retinal vas-
culature, optic disc and underlying RPE (figure 15(a–c)).
Such vulnerable plaques can then form an occlusive As well as imaging retinal blood vessels in there is further
thrombus which may lead to heart attack or stroke. scope to visualize the structure and sO2 of the iris micro-
The coronary artery wall could potentially be imaged vasculature and related abnormalities as shown in figure
using a miniature intravascular sideways looking probe 15(d) [121]. Silverman et al. [176] describe an OR-PAM
in a manner analogous to conventional IVUS as scheme combined with a high-frequency pulse-echo ultra-
described in §4.3. However, the spectroscopic capability sound imaging capability. Another study obtained PA
of PA imaging offers the prospect of providing more and US images of the eye using a dual-mode AR-PAM-
specific and quantitative information about plaque com- US scanner revealing a greater variety of ocular anatomy
position than ultrasound alone can provide. Several ex (the cornea, lens iris as well as the retina, RPE and chor-
vivo studies [127,131,132,171] have demonstrated the oid) albeit with lower resolution [177].
potential to reveal the location of lipid-rich plaques using
multi-wavelength imaging as illustrated in figure 13.
6.1.5. HIFU monitoring. Key to the success of HIFU
While lipids present the most obvious source of differ-
cancer therapy are (i) accurate identification of the
ential contrast, other diagnostically relevant plaque
tumour region to be treated, (ii) ensuring that the
constituents such as calcium deposits, macrophage con-
HIFU beam is delivered with sufficient spatial precision
tent and fibrous material may also have sufficiently
and selectivity to the ROI, (iii) monitoring the tissue
distinctive absorption spectra to allow their content to
status during the treatment via its temperature time
be identified and quantified. To detect macrophages
course or through the measurement of temperature-
(which may be implicated in the formation of the lipid
dependent physical properties of the tissue and (iv)
pool in vulnerable plaques), a novel scheme based on
post-treatment assessment of the extent of tissue abla-
the plasmon resonance coupling-induced wavelength
tion. PA imaging offers the opportunity to address all
shift in the absorption spectrum of gold nanoparticles
four requirements. First, the increased vascularity of a
has been described [172]. In this approach, gold nanopar-
tumour provides intrinsic PA image contrast that can
ticles are taken up by macrophages. The close proximity
be exploited to locate the tumour. Once located, accu-
of the nanoparticles within the cell results in the wave-
rate targeting of the HIFU beam is required so that it
length shift owing to plasmon resonance coupling.
selectively destroys the tumour with minimal collateral
Nanoparticles outside the cell are not in sufficiently
damage to the surrounding tissue. However, spatial het-
close proximity to each other to produce the shift and
erogeneities in sound speed can distort the HIFU
can therefore be differentiated from those internalized
wavefront in such a way that it is not brought to a
in the cell. A further application is imaging stent place-
tight focus, thus compromising targeting accuracy.
ment during surgical insertion and thereafter in follow
There is the potential to mitigate this by using the
up. Figure 14 shows a combined PA and US volumetric
detected PA signals in conjunction with time-reversal
image of a stent inserted in a phantom [173]. Non-inva-
methods [178 – 180]. This involves transmitting the
sive cardiovascular applications that have been
measured PA waveforms at each detection point in
proposed include the detection of deep vein thombosis
temporally reversed order from an array of HIFU trans-
[174] in the venous system and the characterization of
ducers, ideally the same ones that detected the PA
carotid artery plaques [175].
signals. The wavefront distortion produced by the
sound speed variations will be unravelled by the time-
6.1.4. Opthalmology. The absorption-based contrast of reversal process so that the transmitted HIFU field
PA imaging lends it to mapping the structure and oxygen- converges on the tumour. Since the PA waves originate
ation status of the retinal vasculature which is of interest in from the tissue region that is required to be ablated, a
studying conditions such as age-related macular degener- high degree of targeting accuracy should be achievable.
ation and diabetic retinopathy. Additionally, it offers the Preliminary phantom studies have been undertaken
prospect of imaging melanin content in the RPE which demonstrating the feasibility of this approach [178,180].
is relevant to the study of the ageing process. Several There are two approaches which could be used either

(a) (d)
blood vessel
(c)
blood vessel RIA
HA
(b)
RCB
MIC
CP 200 µm
RPE RPE boundary optical absorption SO2
min max 0.6 1.0
Figure 15. In vivo photoacoustic ocular imaging: (a) PA vertical B-scan image acquired along the horizontal red line in (c) show-
ing retinal blood vessel and the underlying retinal pigment epithelium (RPE) in the rat eye [120]. (b) OCT image acquired
simultaneously showing corresponding features [120], (c) lateral MIP showing the retinal vasculature over the optic disc. HA sig-
nifies shadowing owing to hyaloid artery remnant [120]. (d ) Lateral MIP image of iris vasculature in mouse eye [121]. Dual
wavelength spectroscopy was used to map oxygen saturation over the rectangular region indicated by the yellow dotted outline.
(d ) CP, ciliary process; MIC, major iris circle; RCB, recurrent choroidal branch and RIA, radial iris artery.
individually or together in a complementary fashion to rats. These are widely used as a research platform for
non-invasively monitor treatment progression. The first studying human disease processes and to develop new
exploits the temperature dependence of the Gruneisen therapies. PA imaging is particularly well suited to this
coefficient [78,156–158] to non-invasively image the application. The small size of the animal means that
spatial–temporal evolution of the heating and cooling the penetration depth requirements are not excessively
phases of the treatment [181]. The second approach is challenging enabling high SNR to be achieved. The
based on tracking changes in the optical and thermo- modest spatial scale also allows the PA signals to be
mechanical properties of the tissue during and after recorded around the whole body of the animal, thus pro-
treatment to assess tissue ablation. Several studies viding the necessarily large angular detection aperture for
using ex vivo tissue samples have shown that the optical a high-quality image reconstruction. These factors have
properties of tissues are temperature-dependent, particu- enabled high-resolution anatomical images to be obtained
larly beyond the temperature at which coagulation with high contrast as illustrated in figures 5 and 6.
occurs when optical scattering increases significantly A key application is the characterization of small-
[182–184]. These studies suggest that HIFU exposure animal models of brain injury and disease processes,
should result in a local and permanent increase in PA particularly those that require studying vascular anat-
image contrast once beyond this temperature. However, omy and function such as stroke, epilepsy and
a recent study in which an HIFU lesion was induced in traumatic brain injury. A number of studies have been
the kidney of a live mouse appeared to show reduced undertaken using all three PA imaging modes. PAT
PA image contrast [185] at 1064 nm. A possible expla- has been used to image the vascular anatomy of the
nation is that although scattering may have increased, mouse brain [64,66,70,89], the haemodynamic response
the ablative process induced a greater reduction in the due to whisker stimulation [64] and epileptic events [65].
absorption by water or blood, the two dominant chromo- AR-PAM has been used to study temporal variations in
phores at 1064 nm. The blanching effect of HIFU that is cortical blood oxygenation by altering the inspired
visually apparent on highly vascularized tissues would oxygen fraction to induce hypoxia and hyperoxia
seem to be consistent with this. However, another in [188]. Haemodynamic changes due to electrical stimu-
vivo mouse study [186], reported enhanced PA image lation have also been investigated [110]. OR-PAM can
contrast as a result of HIFU exposure although the use be used for small-animal brain imaging, but unlike
of gold nanorods to provide contrast in the ROI may PAT and AR-PAM, requires removal of the scalp. It
be a complicating factor precluding direct comparisons. has been used to map sO2 at capillary level in the
It has also been shown that it is possible to image an mouse brain [117]. Other OR-PAM studies have focused
HIFU lesion in ex vivo tissues using microwave excitation, on demonstrating that spatial –temporal haemodyamic
the source of contrast being thermally induced changes in events such as vasomotion and vasodilation [109] can
tissue conductivity due to water evaporation [187]. be monitored in the mouse. Maps of blood sO2 and
flow over the vasculature have also been acquired with
a view to estimating MRO2 [155].
Mouse models are also used extensively to study the
7. PRE-CLINICAL
pathophysiology of a variety of tumours to aid the
PA imaging has a role to play as a non-invasive investiga- preclinical development of new cancer therapies.
tive tool for characterizing small animals such as mice or Characterizing the supplying tumour vasculature is key

to this endeavour. PA imaging offers the prospect of pupa [196]. New reporters that absorb at longer wave-
being able to achieve this through its ability to non-inva- lengths in the NIR are required in order to achieve the
sively map the structure, oxygenation and flow status of penetration depths required for imaging in small animals
tumour vasculature, all of which have a major impact on such as mice.
tumour development and therapeutic response. Several
preliminary studies have been undertaken with the aim
of imaging abnormal tumour vessels [189] or the evol-
8. SUMMARY
ution of tumour vasculature over time owing to
angiogenesis [67,190]. Several other studies have shown In terms of obtaining high-quality three-dimensional
that subcutaneous melanomas can be visualized, either anatomical images over a range of spatial scales, the
by exploiting intrinsic contrast [170] or through the use feasibility of PA imaging has surely been demonstrated.
of a targeted contrast agent [191]. Major progress in the development of instrumentation,
Other preclinical studies that have been undertaken image reconstruction methods and functional and mol-
include imaging the vascular anatomy of the mouse ecular PA imaging has been made in the last 5 years
embryo [50,90] studying cardiovascular dynamics [69] and the translation to clinical and other applications
and kidney perfusion [68]. has begun. However, there are a number challenges to
As well as providing structural and functional infor- be addressed if the full potential of the technique is to
mation; there is the prospect of using PA imaging to be realized.
help elucidate disease processes at a cellular or molecu- The challenges ahead in terms of instrument devel-
lar level through the use of exogenous contrast agents opment lie at both a component and system level. At
and reporter genes—a detailed review of PA molecular a component level, there is a need for highly sensitive
imaging is provided in Kim et al. [14]. Briefly, there are suitably broadband receivers for PA detection, a need
two principal approaches. The first involves systemi- that may be met by advanced piezoelectric transducers,
cally introducing a targeted contrast agent that based on single-crystal materials or new piezocompo-
selectively accumulates at a disease-specific receptor sites designed specifically to meet the detection
such as a cell-surface protein or enzyme. The absorbing sensitivity and bandwidth requirements of PA imaging
component of the contrast agent can be an organic dye rather than being non-optimally adapted from other
such as ICG or a fluorophore [192] (the absorption band ultrasound applications. New approaches based on opti-
of the fluorophore providing the source of contrast). cal ultrasound sensing such as the FP sensor described
The advantage of using fluorophores is that they have in §4.1.1 or other methods [197] may also play a role in
been extensively developed for fluorescence molecular this respect. The lack of suitable laser systems is widely
imaging and can readily be conjugated to target a recognized as a major bottleneck in the clinical trans-
wide range of specific cellular and intracellular recep- lation of the technique. For PAT and AR-PAM,
tors. Their disadvantages are a relatively small commercially available Q-switched Nd : YAG pumped
absorption cross section and a propensity to bleach or OPO, Ti Sapphire or dye laser systems have proved to
otherwise degrade when irradiated with the high peak be extremely effective for demonstrating feasibility in
powers characteristic of PA excitation laser pulses. Met- a laboratory environment. However, their large size,
allic nanoparticles such as nanoshells, nanorods and lack of practical utility (e.g. the need for re-alignment)
other structures provide a promising alternative. Their and requirement for external cooling systems as well as
plasmon resonance absorption cross sections can be the need to be operated by technically experienced per-
orders of magnitude higher than dye molecules and, sonnel inhibits their practical use, particularly within a
by adjusting their geometrical parameters, their peak clinical environment. Furthermore, those systems that
absorption wavelength can be tuned to NIR wave- are able to provide the milliJoule pulse energies required
lengths. Furthermore, in some cases, nanoparticulate for PAT tend to have low PRFs (less than 50 Hz).
structures (e.g. single-walled carbon nanotubes) can As well as limiting image frame rates this inhibits the
exhibit higher photostability than organic dyes. Specific implementation of time domain signal processing tech-
targeting using contrast agents has now been demon- niques such as pulsed Doppler flowmetry that require
strated in vivo in tumours using carbon nanotubes high PRFs. Slow tuning speeds or insufficient tuning
[100,193], gold nanorods [194] and cages [191] to pro- range also inhibits the practical application of spectro-
vide selective contrast. The second method involves scopic methods and thus the functional and molecular
incorporating engineered reporter genes that express imaging capability of the technique. There are some
optically absorbing proteins in mice. This offers the signs that progress is being made in this direction.
prospect of studying the role of genetic processes in Compact, portable, high PRF fibre laser systems are
disease pathways. LacZ gene expression has been beginning to find application [99,198] although, with
detected by PA imaging via the change in the optical one exception [199], only in OR-PAM as the pulse ener-
absorption of X-gal when catalysed by the LacZ-encoded gies of most currently available systems are insufficient
enzyme b-galactosidase [195]. However, achieving this for use in AR-PAM and PAT. Laser diode-based systems
required the local injection of the X-gal substrate into have been also demonstrated for superficial imaging
the ROI. PA imaging of cells expressing flourescent pro- applications [200,201]. Although advances in high-
teins (the absorption band of the protein providing the power pulsed laser diode technology and novel signal
PA image contrast) has been demonstrated in a non- processing schemes are likely to increase their practical
interventional manner but only in the relatively trans- utility, the fundamental peak power limit imposed by
parent Zebrafish or small-scale Drosophila (fruitfly) the facet damage threshold is likely to preclude their

use for deep-tissue PAT imaging using pulsed excitation. here is to develop genetically expressed proteins that
Alternative excitation schemes such as those based on fre- absorb at longer wavelengths in order to enable greater
quency domain methods may provide an opportunity to penetration depth to be achieved, are stable under the
overcome this limitation [10–12]. high peak power of PA excitation laser pulses and can
At a systems level, the challenge in PAT is to achieve be expressed by mammalian cells.
real-time image frame rates without excessively compro- In terms of biomedical application, preclinical PA
mising SNR. For PAT, there is a fundamental limit. imaging is already becoming established as a useful
Irrespective of the availability of suitable lasers, average tool and beginning to be applied as opposed to merely
power safety-related considerations mean that to being demonstrated. Translation to clinical application
achieve real time frame rates and the highest possible is less well advanced, perhaps inevitably with relatively
SNR, a PRF in the low tens of hertz and parallel detec- few clinical studies having been undertaken, in part due
tion, signal conditioning and acquisition on all channels to the more stringent requirements in terms of
is ideally required. Operating at a higher PRF and mul- acquisition speed, portability and patient – instrument
tiplexing down to reduce the number of truly parallel interface considerations. However, with the emergence
channels can be achieved as in conventional US. However, of dedicated clinical systems, especially those based on
it results in reduced SNR as the pulse energy has to be existing diagnostic ultrasound scanners, the rate of pro-
reduced to limit average power. For three-dimensional gression towards clinical application in oncology,
imaging, which can require two-dimensional array dermatology, cardiovascular medicine and other
channel counts of the order of 104 [3], the cost of a fully specialities is set to increase in the near future.
parallel system would be prohibitive. The challenge,
therefore, is to find the optimal compromise between The author acknowledges helpful discussions with and
contributions from members of the UCL Photoacoustic
PRF and thus SNR, multiplexing ratios and frame
Imaging Group, among them Ben Cox, Jan Laufer, Thomas
rates. These considerations apply principally to PAT. Allen, Edward Zhang, Jo Brunker and Bradley Treeby. The
Different considerations in respect of PRF and acquisition author was supported by an EPSRC Leadership Fellowship.
speed apply to the PA microscopy modes which are not
readily parallelizable.
Spectroscopic techniques are key to realizing the
functional and molecular imaging capability of PA ima- REFERENCES
ging. Although much progress has been made in the 1 Bell, A. G. 1880 On the production and reproduction of
development of inversion algorithms, addressing the sound by light. Am. J. Sci. 20, 305–324.
corrupting influence of the wavelength dependence of 2 Tam, A. 1986 Applications of photoacoustic sensing tech-
the light fluence remains a major challenge. Model- niques. Rev. Mod. Phys. 58, 381– 431. (doi:10.1103/
based inversion schemes that fully account for the RevModPhys.58.381)
physics involved in the PA signal generation and 3 Kruger, R., Liu, P. & Appledorn, C. 1995 Photoacoustic
detection process have promise but in their most general ultrasound (PAUS)—reconstruction tomography. Med.
Phys. 22, 1605–1609. (doi:10.1118/1.597429)
form are still too computationally expensive for many
4 Esenaliev, R. O., Karabutov, A. A., Tittel, F. K.,
practical applications. Limiting the spatial domain,
Fornage, B. D., Thomsen, S. L., Stelling, C. & Oraevsky,
constraining the solution with additional information, A. A. 1997 Laser optoacoustic imaging for breast cancer
prior or otherwise, and developing ad hoc, albeit diagnostics: limit of detection and comparison with
rigorously validated, approximations for specific appli- x-ray and ultrasound imaging. Proc. SPIE 2979,
cations may yield progress with these methods. 71 –82. (doi:10.1117/12.280213)
Accurate quantitation of chromophore concentrations 5 Hoelen, C. G., de Mul, F. F., Pongers, R. & Dekker, A.
is the ultimate goal of quantitative PA spectroscopy 1998 Three-dimensional photoacoustic imaging of blood
and essential for the measurement of physiological par- vessels in tissue. Opt. Lett. 23, 648–650. (doi:10.1364/
ameters such as blood sO2. However, spectroscopic OL.23.000648)
methods are important even if it is required only to 6 Oraevsky, A. A., Andreev, V. A., Karabutor, A. A.,
Declan Fleming, K. R., Gatalica, Z., Singh, H. & Esena-
detect rather than quantify the presence of specific
lier, R. O. 1999 Laser optoacoustic imaging of the breast:
chromophores. This is especially so in molecular ima- detection of cancer angiogenesis. Proc. SPIE 3597,
ging given the relatively low concentrations that 352–363. (doi:10.1117/12.356829)
targeted molecular imaging contrast agents or geneti- 7 Kruger, R. A., Miller, K. D., Reynolds, H. E., Kiser, W. L.,
cally expressed reporters are likely to accumulate in. Reinecke, D. R. & Kruger, G. A. 2000 Breast cancer in
Accurate spectroscopic inversion or unmixing techniques vivo: contrast enhancement with thermoacoustic CT at
will be essential for the sensitive detection of molecular 434 MHz-feasibility study. Radiology 216, 279–283.
imaging agents in the face of the overwhelming PA 8 Ku, G. & Wang, L. 2000 Scanning thermoacoustic tom-
contribution from haemoglobin. ography in biological tissue. Med. Phys. 27, 1195–1202.
In PA molecular imaging, specific targeting using (doi:10.1118/1.598984)
systemically introduced contrast agents has been 9 Beard, P. C. 2002 Photoacoustic imaging of blood vessel
equivalent phantoms. Proc. SPIE 4618, 54–62. (doi:10.
demonstrated and much effort is now being devoted 1117/12.469848)
to optimizing the labelling component of the agent 10 Fan, Y., Mandelis, A., Spirou, G. & Alex Vitkin, I. 2004
using a variety of nanoparticulate structures. The use Development of a laser photothermoacoustic frequency-
of engineered genetic reporters has much promise swept system for subsurface imaging: Theory and exper-
having been demonstrated using existing fluorescent iment. J. Acoust. Soc. Am. 116, 3523–3533. (doi:10.
reporters to provide PA contrast. The challenge ahead 1121/1.1819393)

11 Maslov, K. & Wang, L. V. 2008 Photoacoustic imaging of using a clinical ultrasound array system. Biomed. Opt.
biological tissue with intensity-modulated continuous- Exp. 1, 335–340.
wave laser. J. Biomed. Opt. 13, 024006. (doi:10.1117/1. 30 Homan, K., Kim, S., Chen, Y.-S., Wang, B., Mallidi, S. &
2904965) Emelianov, S. 2010 Prospects of molecular photoacoustic
12 Lashkari, B. & Mandelis, A. 2010 Photoacoustic radar imaging at 1064 nm wavelength. Opt. Lett. 35, 2663–
imaging signal-to-noise ratio, contrast, and resolution 2665. (doi:10.1364/OL.35.002663)
enhancement using nonlinear chirp modulation. Opt. 31 Beard, P. C. 2002 Photoacoustic imaging of blood vessel
Lett. 35, 1623–1625. (doi:10.1364/OL.35.001623) equivalent phantoms. Proc. SPIE 4618, 54 –62. (doi:10.
13 Hu, S. & Wang, L. V. 2010 Neurovascular photoacoustic 1117/12.469848)
tomography. Front. Neuroenergetics 2, 10. (doi:10.3389/ 32 Kuchment, P. & Kunyansky, L. 2008 Mathematics of
fnene.2010.00010) thermoacoustic tomography. Eur. J. Appl. Math. 19,
14 Kim, C., Favazza, C. & Wang, L. V. 2010 In vivo photo- 191–224. (doi:10.1017/S0956792508007353)
acoustic tomography of chemicals: high-resolution 33 Kuchment, P. & Kunyansky, L. 2011 Mathematics of
functional and molecular optical imaging at new depths. photoacoustic and thermoacoustic tomography. In Springer
Chem.l Rev. 110, 2756–2782. (doi:10.1021/cr900266s) Handb. Math. Methods Imag. (ed. O. Scherzer), pp.
15 Hu, S. & Wang, L. V. 2010 Photoacoustic imaging and 819–865. New York: Springer.
characterization of the microvasculature. J. Biomed. 34 Finch, D., Patch, S. K. & Rakesh, S. J. 2004 Determining
Opt. 15, 011101. (doi:10.1117/1.3281673) a function from its mean values over a family of spheres.
16 Wang, L. V. 2009 Multiscale photoacoustic microscopy SIAM J. Math. Anal. 35, 1213–1240. (doi:10.1137/
and computed tomography. Nat. Photonics 3, 503–509. S0036141002417814)
(doi:10.1038/nphoton.2009.157) 35 Kunyansky, L. A. 2007 Explicit inversion formulae for
17 Li, C. & Wang, L. V. 2009 Photoacoustic tomography the spherical mean Radon transform. Inverse Probl. 23,
and sensing in biomedicine. Phys. Med. Biol. 54, R59– 373–383. (doi:10.1088/0266-5611/23/1/021)
R97. (doi:10.1088/0031-9155/54/19/R01) 36 Xu, M. & Wang, L. 2005 Universal back-projection algor-
18 Xu, M. & Wang, L. V. 2006 Photoacoustic imaging in ithm for photoacoustic computed tomography. Phys.
biomedicine. Rev. Sci. Instrum. 77, 041101. (doi:10. Rev. E 71, 1–7.
1063/1.2195024) 37 Lam, R. B., Kruger, R. A., Reinecke, D. R., DelRio, S. P.,
19 Mallidi, S., Luke, G. P. & Stanislav, E. 2011 Photoacous- Thornton, M. M., Picot, P. A. & Morgan, T. G. 2010
tic imaging in cancer detection, diagnosis, and treatment Dynamic optical angiography of mouse anatomy using
guidance. Trends Biotechnol. 29, 213 –221. (doi:10.1016/ radial projections. Proc. SPIE 7564, 756405. (doi:10.
j.tibtech.2011.01.006) 1117/12.841024).
20 Cox, B. T. & Beard, P. C. 2005 Fast calculation of 38 Norton, S. J. & Linzer, M. 1981 Ultrasonic reflectivity
pulsed photoacoustic fields using k-space methods. J imaging in three dimensions: exact inverse scattering sol-
Acoust. Soc. Am. 117, 3616–3636. (doi:10.1121/1. utions for plane, cylindrical, and spherical apertures.
1920227) IEEE Trans. Bio-med. Eng. 28, 202– 220. (doi:10.1109/
21 Cox, B. T., Laufer, J. G. & Beard, P. C. 2009 The chal- TBME.1981.324791)
lenges for quantitative photoacoustic imaging. Proc. 39 Kunyansky, L. A. 2007 A series solution and a fast algor-
SPIE 7177, 717713. (doi:10.1117/12.806788) ithm for the inversion of the spherical mean Radon
22 Hale, G. M. & Querry, M. R. 1973 Optical-constants of transform. Inverse Probl. 23, S11– S20. (doi:10.1088/
water in 200 nm to 200 mum wavelength region. Appl. 0266–5611/23/6/S02)
Opt. 12, 555 –563. (doi:10.1364/AO.12.000555) 40 Anastasio, M. A., Zhang, J., Modgil, D. & Rivière,
23 van Veen, R. L. P., Sterenborg, H. J. C. M., Pifferi, A., P. J. L. 2007 Application of inverse source concepts to
Torricelli, A., Chikoidze, E. & Cubeddu, R. 2010 Deter- photoacoustic tomography. Inverse Probl. 23, S21 –S35.
mination of visible near-IR absorption coefficients of (doi:10.1088/0266-5611/23/6/S03)
mammalian fat using time- and spatially resolved diffuse 41 Xu, Y., Xu, M. & Wang, L. V. 2002 Exact frequency-
reflectance and transmission spectroscopy. J. Biomed. domain reconstruction for thermoacoustic tomography–
Opt. 10, 054004. (doi:10.1117/1.2085149) II: Cylindrical geometry. IEEE Trans. Med. Imag. 21,
24 Tsai, C. L., Chen, J. C. & Wang, W. J. 2001 Near-infra- 829–833. (doi:10.1109/TMI.2002.801171)
red absorption property of biological soft tissue 42 Köstli, K. P., Frenz, M., Bebie, H. & Weber, H. P. 2001
constituents. J. Med. Biol. Eng. 21, 7– 14. Temporal backward projection of optoacoustic pressure
25 Laufer, J., Delpy, D., Elwell, C. & Beard, P. 2007 Quan- transients using fourier transform methods. Phys.
titative spatially resolved measurement of tissue Med. Biol. 46, 1863–1872. (doi:10.1088/0031-9155/46/
chromophore concentrations photoacoustic spectroscopy: 7/309)
application to the measurement of blood oxygenation 43 Xu, Y., Feng, D. & Wang, L. V. 2002 Exact frequency-
and haemoglobin concentration. Phys. Med. Biol. 52, domain reconstruction for thermoacoustic tomography.
141 –168. (doi:10.1088/0031-9155/52/1/010) I: planar geometry. IEEE Trans. Med. Imag. 21,
26 Xu, Z., Li, C. & Wang, L. V. 2010 Photoacoustic tom- 823–828. (doi:10.1109/TMI.2002.801172)
ography of water in phantoms and tissue. J. Biomed. 44 Niederhauser, J. J., Jaeger, M., Lemor, R., Weber, P. &
Opt. 15, 036019. (doi:10.1117/1.3443793) Frenz, M. 2005 Combined ultrasound and optoacoustic
27 Kruger, R. A., Lam, R. B., Reinecke, D. R., Del Rio, S. P. & system for real-time high-contrast vascular imaging in
Doyle, R. P. 2010 Photoacoustic angiography of the breast. vivo. IEEE Trans. Med. Imag. 24, 436– 440. (doi:10.
Med. Phys. 37, 6096. (doi:10.1118/1.3497677) 1109/TMI.2004.843199)
28 Ku, G. & Wang, L. V. 2005 Deeply penetrating photoa- 45 Zhang, E. Z., Laufer, J. G., Pedley, R. B. & Beard, P. C.
coustic tomography in biological tissues enhanced with 2009 In vivo high-resolution 3D photoacoustic imaging of
an optical contrast agent. Opt. Lett. 30, 507–509. superficial vascular anatomy. Phys. Med. Biol. 54, 1035–
(doi:10.1364/OL.30.000507) 1046. (doi:10.1088/0031-9155/54/4/014)
29 Kim, C., Erpelding, T., Jankovic, L. & Pashley, M. D. 46 Xu, Y. & Wang, L. V. 2004 Time reversal and its appli-
2010 Deeply penetrating in vivo photoacoustic imaging cation to tomography with diffracting sources. Phys.

Rev. Lett. 92, 033902. (doi:10.1103/PhysRevLett.92. 62 Brecht, H.-P., Su, R., Fronheiser, M., Ermilov, S. A., Con-
033902) justeau, A. & Oraevsky, A. 2009 Whole-body three-
47 Burgholzer, P., Matt, G. J., Haltmeier, M. & Paltauf, G. dimensional optoacoustic tomography system for small ani-
2007 Exact and approximative imaging methods for mals. J. Biomed. Opt. 14, 064007. (doi:10.1117/1.3259361)
photoacoustic tomography using an arbitrary detection 63 Wang, X., Pang, Y., Ku, G., Stoica, G. & Wang, L. V.
surface. Phys. Rev. E 75, 046706. (doi:10.1103/Phys- 2003 Three-dimensional laser-induced photoacoustic tom-
RevE.75.046706) ography of mouse brain with the skin and skull intact. Opt.
48 Hristova, Y., Kuchment, P. & Nguyen, L. 2008 Recon- Lett. 28, 1739–1741. (doi:10.1364/OL.28.001739)
struction and time reversal in thermoacoustic 64 Wang, X., Pang, Y., Ku, G., Xie, X., Stoica, G. & Wang,
tomography in acoustically homogeneous and inhomo- L. V. 2003 Noninvasive laser-induced photoacoustic tom-
geneous media. Inverse Probl. 24, 055006. (doi:10.1088/ ography for structural and functional in vivo imaging of
0266-5611/24/5/055006) the brain. Nat. Biotechnol. 21, 803–806. (doi:10.1038/
49 Treeby, B. E., Zhang, E. Z. & Cox, B. T. 2010 Photoa- nbt839)
coustic tomography in absorbing acoustic media using 65 Zhang, Q., Liu, Z., Carney, P. R., Yuan, Z., Chen, H.,
time reversal. Inverse Probl. 26, 115003. (doi:10.1088/ Roper, S. N. & Jiang, H. 2008 Non-invasive imaging of
0266-5611/26/11/115003) epileptic seizures in vivo using photoacoustic tomogra-
50 Treeby, B. E., Laufer, J. G., Zhang, E. Z., Norris, F. C., phy. Phys. Med. Biol. 53, 1921–1931. (doi:10.1088/
Lythgoe, M. F., Beard, P. C. & Cox, B. T. 2011 Acoustic 0031-9155/53/7/008)
attenuation compensation in photoacoustic tomography: 66 Yang, S., Xing, D., Zhou, Q., Xiang, L. & Lao, Y. 2007
application to high-resolution 3D imaging of vascular Functional imaging of cerebrovascular activities in
networks in mice. Proc. SPIE 7899, Y178992–Y978992. small animals using high-resolution photoacoustic tom-
51 Paltauf, G., Viator, J. A., Prahl, S. A. & Jacques, S. L. ography. Med. Phys. 34, 3294 –3301. (doi:10.1118/1.
2002 Iterative reconstruction algorithm for optoacoustic 2757088)
imaging. J. Acoust. Soc. Am. 112, 1536. (doi:10.1121/ 67 Lao, Y., Xing, D., Yang, S. & Xiang, L. 2008 Noninvasive
1.1501898) photoacoustic imaging of the developing vasculature
52 Yuan, Z. & Jiang, H. 2007 Three-dimensional finite- during early tumor growth. Phys. Med. Biol. 53, 4203–
element-based photoacoustic tomography: reconstruction 4212. (doi:10.1088/0031-9155/53/15/013)
algorithm and simulations. Med. Phys. 34, 538. (doi:10. 68 Buehler, A., Herzog, E., Razansky, D. & Ntziachristos, V.
1118/1.2409234) 2010 Video rate optoacoustic tomography of mouse
53 Rosenthal, A., Razansky, D. & Ntziachristos, V. 2010 kidney perfusion. Opt. Lett. 35, 2475–2477. (doi:10.
Fast semi-analytical model-based acoustic inversion for 1364/OL.35.002475)
quantitative optoacoustic tomography. IEEE Trans. 69 Taruttis, A., Herzog, E., Razansky, D. & Ntziachristos,
Med. Imag. 29, 1275–1285. (doi:10.1109/TMI.2010. V. 2010 Real-time imaging of cardiovascular dynamics
2044584) and circulating gold nanorods with multispectral optoa-
54 Patch, S. K. 2004 Thermoacoustic tomography—consist- coustic tomography. Opt. Express 18, 19592 –19602.
ency conditions and the partial scan problem. Phys. Med. (doi:10.1364/OE.18.019592)
Biol. 49, 2305–2315. (doi:10.1088/0031-9155/49/11/013) 70 Li, C., Aguirre, A., Gamelin, J., Maurudis, A., Zhu, Q. &
55 Xu, Y., Wang, L. V., Ambartsoumian, G. & Kuchment, Wang, L. V. 2010 Real-time photoacoustic tomography of
P. 2004 Reconstructions in limited-view thermoacoustic cortical hemodynamics in small animals. J. Biomed. Opt.
tomography. Med. Phys. 31, 724– 733. (doi:10.1118/1. 15, 010509. (doi:10.1117/1.3302807)
1644531) 71 Gamelin, J., Maurudis, A., Aguirre, A., Huang, F., Guo,
56 Cox, B. T., Arridge, S. R. & Beard, P. C. 2007 Photoa- P., Wang, L. V. & Zhu, Q. 2009 A real-time photoacous-
coustic tomography with a limited-aperture planar tic tomography system for small animals. Opt. Express
sensor and a reverberant cavity. Inverse Probl. 23, 17, 10489–10498. (doi:10.1364/OE.17.010489)
S95–S112. (doi:10.1088/0266-5611/23/6/S08) 72 Gamelin, J., Aguirre, A., Maurudis, A., Huang, F., Castillo,
57 Modgil, D., Anastasio, A. M., Wang, K. & LaRiviere, P. D., Wang, L. V. & Zhu, Q. 2010 Curved array photoacous-
J. 2009 Image reconstruction in photoacoustic tomogra- tic tomographic system for small animal imaging.
phy with variable speed of sound using a higher order J. Biomed. Opt. 13, 024007. (doi:10.1117/1.2907157)
geometrical acoustics approximation. Proc. SPIE 7177, 73 Song, K. H., Stoica, G. & Wang, L. V. 2006 In vivo three-
A71771 –A717718. dimensional photoacoustic tomography of a whole mouse
58 Burgholzer, P., Hofer, C., Paltauf, G., Haltmeier, M. & head. Opt. Lett. 31, 2453–2455. (doi:10.1364/OL.31.
Scherzer, O. 2005 Thermoacoustic tomography with inte- 002453)
grating area and line detectors. IEEE Trans. Ultrason. 74 Pramanik, M., Ku, G., Li, C. & Wang, L. V. 2008 Design
Ferroelectr. Freq. Cont. 52, 1577–1583. (doi:10.1109/ and evaluation of a novel breast cancer detection system
TUFFC.2005.1516030) combining both thermoacoustic (TA) and photoacoustic
59 Haltmeier, M., Scherzer, O., Burgholzer, P. & Paltauf, G. (PA) tomography. Med. Phys. 35, 2218–2223. (doi:10.
2004 Thermoacoustic computed tomography with large 1118/1.2911157)
planar receivers. Inverse Probl. 20, 1663–1673. (doi:10. 75 Ermilov, S. A., Khamapirad, T., Conjusteau, A., Leo-
1088/0266-5611/20/5/021) nard, M. H., Lacewell, R., Mehta, K., Miller, T. &
60 Paltauf, G., Nuster, R., Haltmeier, M. & Burgholzer, P. Oraevsky, A. A. 2009 Laser optoacoustic imaging
2007 Experimental evaluation of reconstruction algor- system for detection of breast cancer. J. Biomed. Opt.
ithms for limited view photoacoustic tomography with 14, 024007. (doi:10.1117/1.3086616)
line detectors. Inverse Probl. 23, S81–S94. (doi:10. 76 Wang, X., Chamberland, D. L. & Jamadar, A. D. 2007
1088/0266-5611/23/6/S07) Noninvasive photoacoustic tomography of human periph-
61 Kruger, R. A., Kiser, W. L., Reinecke, D. R., Kruger, G. eral joints toward diagnosis of inflammatory arthritis.
A. & Miller, K. D. 2003 Thermoacoustic Molecular Ima- Opt. Lett. 32, 3002– 3004. (doi:10.1364/OL.32.003002)
ging of Small Animals. Mol. Imag. 2, 113 –123. (doi:10. 77 Kruger, R. A., Kiser, W. L., Reinecke, D. R. & Kruger, G.
1162/153535003322331993) 2003 Thermoacoustic computed tomography using a

conventional linear transducer array. Med. Phys. 30, 94 Favazza, C. P., Jassim, O., Cornelius, L. A. & Wang,
856 –860. (doi:10.1118/1.1565340) L. V. 2011 In vivo photoacoustic microscopy of human
78 Shah, J., Park, S., Aglyamov, S., Larson, T., Ma, L., cutaneous microvasculature and a nevus. J. Biomed.
Sokolov, K., Johnston, K., Milner, T. & Emelianov, Opt. 16, 016015. (doi:10.1117/1.3528661)
S. Y. 2010 Photoacoustic imaging and temperature 95 Favazza, C. P., Cornelius, L. A. & Wang, L. V. 2011 In
measurement for photothermal cancer therapy. vivo functional photoacoustic microscopy of cutaneous
J. Biomed. Opt. 13, 034024. (doi:10.1117/1.2940362) microvasculature in human skin. J. Biomed. Opt. 16,
79 Fronheiser, M. P., Ermilov, S. A., Brecht, H.-P., Conjus- 026004. (doi:10.1117/1.3536522)
teau, A., Su, R., Mehta, K. & Oraevsky, A. A. 2011 Real- 96 Stein, E. W., Maslov, K. & Wang, L. V. 2009 Noninva-
time optoacoustic monitoring and three-dimensional sive, in vivo imaging of the mouse brain using
mapping of a human arm vasculature. J. Biomed. Opt. photoacoustic microscopy. J. Appl. Phys. 105, 102027.
15, 021305. (doi:10.1117/1.3370336) (doi:10.1063/1.3116134)
80 Aguirre, A., Guo, P., Gamelin, J., Yan, S., Sanders, 97 Song, K. H. & Wang, L. V. 2008 Noninvasive photoa-
M. M., Brewer, M. & Zhu, Q. 2011 Coregistered three- coustic imaging of the thoracic cavity and the kidney in
dimensional ultrasound and photoacoustic imaging small and large animals. Med. Phys. 35, 4524–4529.
system for ovarian tissue characterization. J. Biomed. (doi:10.1118/1.2977534)
Opt. 14, 054014. (doi:10.1117/1.3233916) 98 Wang, X., Chamberland, D. L. & Xi, G. 2008 Noninva-
81 Ephrat, P., Needles, A., Bilan, C., Trujillo, A., Theodoro- sive reflection mode photoacoustic imaging through
poulos, C., Hirson, D., Foster, S. & Visualsonics. 2010 infant skull toward imaging of neonatal brains.
Photoacoustic imaging of Murine tumors using the Vevow J. Neurosci. Methods 168, 412–421. (doi:10.1016/j.jneu-
2100 micro-ultrasound system (Visualsonics White Paper). meth.2007.11.007)
82 Song, L., Maslov, K., Bitton, R., Shung, K. & Wang, 99 Shi, W., Kerr, S., Utkin, I., Ranasinghesagara, J., Pan,
L. V. 2008 Fast 3–D dark-field reflection-mode photoa- L., Godwal, Y., Zemp, R. J. & Fedosejevs, R. 2010
coustic microscopy in vivo with a 30 –MHz ultrasound Optical resolution photoacoustic microscopy using novel
linear array. J. Biomed. Opt. 13, 1–10. high-repetition-rate passively Q-switched microchip and
83 Song, L., Maslov, K., Shung, K. K. & Wang, L. V. 2010 fiber lasers. J. Biomed. Opt. 15, 056017. (doi:10.1117/1.
Ultrasound-array-based real-time photoacoustic 3502661)
microscopy of human pulsatile dynamics in vivo. 100 De la Zerda, A. et al. 2008 Carbon nanotubes as photoa-
J. Biomed. Opt. 15, 021303. (doi:10.1117/1.3333545) coustic molecular imaging agents in living mice. Nat.
84 Beard, P. C., Perennes, F. & Mills, T. N. 1999 Transduc- Nanotechnol. 3, 557–562. (doi:10.1038/nnano.2008.231)
tion mechanisms of the Fabry– Perot polymer film 101 Zhang, H. F., Maslov, K. & Wang, L. V. 2006 Functional
sensing concept for wideband ultrasound detection. photoacoustic microscopy for high-resolution and nonin-
IEEE Trans. Ultrason. Ferroelect. Freq. Cont. 46, vasive in vivo imaging. Nat. Biotechnol. 24, 848–851.
1575–1582. (doi:10.1109/58.808883) (doi:10.1038/nbt1220)
85 Zhang, E., Laufer, J. & Beard, P. 2008 Backward-mode 102 Li, M.-L., Zhang, H. E., Maslov, K., Stoica, G. & Wang,
multiwavelength photoacoustic scanner using a planar L. V. 2006 Improved in vivo photoacoustic microscopy
Fabry –Perot polymer film ultrasound sensor for high-res- based on a virtual-detector concept. Opt. Lett. 31,
olution three-dimensional imaging of biological tissues. 474–476. (doi:10.1364/OL.31.000474)
Appl. Opt. 47, 561 –577. (doi:10.1364/AO.47.000561) 103 Passler, K., Nuster, R., Gratt, S., Burgholzer, P., Berer,
86 1994. British Standard: Safety of laser products BS EN T. & Paltauf, G. 2010 Scanning acoustic-photoacoustic
60825– 1. microscopy using axicon transducers. Biomed. Opt.
87 Lamont, M. & Beard, P. C. 2006 2D imaging of ultrasound Express 1, 318–323. (doi:10.1364/BOE.1.000318)
fields using a CCD array to detect the output of a Fabry– 104 Savateeva, E. V., Karabutov, A. A., Bell, B., Johnigan,
Perot polymer film sensor. Electron. Lett. 42, 187–189. R., Motamedi, M. & Oraevsky, A. A. 2000 Noninvasive
88 Laufer, J., Johnson, P., Zhang, E., Treeby, B., Cox, B., detection and staging of oral cancer in vivo with confocal
Pedley, B. & Beard, P. 2011 In vivo longitudinal photoa- optoacoustic tomography. Proc. SPIE 3916, 55–66.
coustic imaging of subcutaneous tumours in mice. SPIE (doi:10.1117/12.386341)
Proc. 7899, 789915. (doi:10.1117/12.876782) 105 Maslov, K., Zhang, H. F., Hu, S. & Wang, L. V. 2008
89 Laufer, J., Zhang, E., Raivich, G. & Beard, P. 2009 Optical-resolution photoacoustic microscopy for in vivo
Three-dimensional noninvasive imaging of the vascula- imaging of single capillaries. Opt. Lett. 33, 929–931.
ture in the mouse brain using a high resolution (doi:10.1364/OL.33.000929)
photoacoustic scanner. Appl. Opt. 48, D299–D306. 106 Hu, S., Maslov, K. & Wang, L. V. 2009 In vivo functional
(doi:10.1364/AO.48.00D299) chronic imaging of a small animal model using optical-
90 Laufer, J. G., Cleary, J. O., Zhang, E. Z., Lythgoe, M. F. resolution photoacoustic microscopy. Med. Phys. 36,
& Beard, P. C. 2010 Photoacoustic imaging of vascular 2320–2323. (doi:10.1118/1.3137572)
networks in transgenic mice. Proc. SPIE 7564, 75641A. 107 Xie, Z., Jiao, S., Zhang, H. F. & Puliafito, C. A.
(doi:10.1117/12.842204) 2009 Laser-scanning optical-resolution photoacoustic
91 Maslov, K., Stoica, G. & Wang, L. V. 2005 In vivo dark- microscopy. Opt. Lett. 34, 1771–1773. (doi:10.1364/
field reflection-mode photoacoustic microscopy. Opt. OL.34.001771)
Lett. 30, 625– 627. (doi:10.1364/OL.30.000625) 108 Hu, S., Yan, P., Maslov, K., Lee, J.-M. & Wang, L. V.
92 Zhang, H. F., Maslov, K., Stoica, G. & Wang, L. V. 2006 2009 Intravital imaging of amyloid plaques in a trans-
Functional photoacoustic microscopy for high-resolution genic mouse model using optical-resolution
and noninvasive in vivo imaging. Nat. Biotechnol. 24, photoacoustic microscopy. Opt. Lett. 34, 3899–3901.
848 –851. (doi:10.1038/nbt1220) (doi:10.1364/OL.34.003899)
93 Zhang, H. F., Maslov, K. & Wang, L. V. 2007 In vivo ima- 109 Hu, S., Maslov, K. & Wang, L. H. V. 2009 Noninvasive
ging of subcutaneous structures using functional label-free imaging of microhemodynamics by optical-res-
photoacoustic microscopy. Nat. Protocols 2, 797–804. olution photoacoustic microscopy. Opt. Express 17,
(doi:10.1038/nprot.2007.108) 7688–7693. (doi:10.1364/OE.17.007688)

110 Liao, L.-D. et al. 2010 Imaging brain hemodynamic Intravascular photoacoustic imaging of human coronary
changes during rat forepaw electrical stimulation using atherosclerosis. Opt. Lett. 36, 597–599. (doi:10.1364/
functional photoacoustic microscopy. NeuroImage 52, OL.36.000597)
562 –570. (doi:10.1016/j.neuroimage.2010.03.065) 128 Zhang, E. Z. & Beard, P. C. 2011 A miniature all-optical
111 Shelton, R. L. & Applegate, B. E. 2010 Off-axis photoa- photoacoustic imaging probe. Proc. SPIE 7899, 78991F.
coustic microscopy. IEEE Trans. Bio-Med. Eng. 57, (doi:10.1117/12.874883)
1835– 1838. (doi:10.1109/TBME.2010.2043103) 129 Xi, L., Sun, J., Zhu, Y., Wu, L., Xie, H. & Jiang, H. 2010
112 Shelton, R. L. & Applegate, B. E. 2010 Ultrahigh resol- Photoacoustic imaging based on MEMS mirror scanning.
ution photoacoustic microscopy via transient Biomed. Opt. Express 1, 1278–1283. (doi:10.1364/BOE.
absorption. Biomed. Opt. 1, 676 –686. (doi:10.1364/ 1.001278)
BOE.1.000676) 130 Yang, J.-M., Favazza, C., Chen, R., Maslov, K., Cai, X.,
113 Zhang, E. Z., Laufer, J., Povaay, B., Alex, A., Hofer, B., Zhou, Q., Shung, K. K. & Wang, L. V. 2011 Volumetric
Drexler, W. & Beard, P. 2010 Multimodal simultaneous photoacoustic endoscopy of upper gastrointestinal tract:
photoacoustic tomography, optical resolution ultrasonic transducer technology development. Proc.
microscopy, and OCT system. Proc. SPIE 7564, SPIE 7899, 78990D1–78990D6. (doi:10.1117/12.
75640U. (doi:10.1117/12.841615) 875377)
114 Jiang, M., Zhang, X., Puliafito, C. A., Zhang, H. F. & 131 Allen, T. J. & Beard, P. C. 2009 Photoacoustic character-
Jiao, S. 2010 Adaptive optics photoacoustic microscopy. isation of vascular tissue at NIR wavelengths. Proc. SPIE
Opt. Express 18, 21770 –21776. (doi:10.1364/OE.18. 7177, 71770A. (doi:10.1117/12.808777)
021770) 132 Allen, T. J., Hall, A., Dhillon, A., Owen, J. S. & Beard,
115 Rao, B., Li, L., Maslov, K. & Wang, L. 2010 Hybrid-scan- P. C. 2010 Photoacoustic imaging of lipid rich plaques
ning optical-resolution photoacoustic microscopy for in in human aorta. Proc. SPIE 7564, 75640C. (doi:10.
vivo vasculature imaging. Opt. Lett. 35, 1521–1523. 1117/12.842205)
(doi:10.1364/OL.35.001521) 133 Wray, S., Cope, M., Delpy, D. T., Wyatt, J. S. & Rey-
116 Ku, G., Maslov, K., Li, L. & Wang, L. V. 2010 Photoacous- nolds, E. O. R. 1988 Characterisation of the near-
tic microscopy with 2–microm transverse resolution. infrared absorption spectra of cytochrome-AA3 and hae-
J. Biomed. Opt. 15, 021302. (doi:10.1117/1.3339912) moglobin for the non-invasive monitoring of cerebral
117 Hu, S., Maslov, K., Tsytsarev, V. & Wang, L. V. 2011 oxygenation. Biochim. Biophys. Acta 933, 184–192.
Functional transcranial brain imaging by optical-resol- (doi:10.1016/0005-2728(88)90069-2)
ution photoacoustic microscopy. J. Biomed. Opt. 14, 134 Wang, L. V. 2009 Photoacoustic imaging and spec-
040503. (doi:10.1117/1.3194136) troscopy, ch. 11.Boca Raton, FL: CRC Press.
118 Hu, S., Maslov, K. & Wang, L. V. 2011 Second- 135 Laufer, J., Elwell, C., Delpy, D. & Beard, P. 2005 In vitro
generation optical-resolution photoacoustic microscopy measurements of absolute blood oxygen saturation using
with improved sensitivity and speed. Opt. Lett. 36, pulsed near-infrared photoacoustic spectroscopy: accu-
1134– 1136. (doi:10.1364/OL.36.001134) racy and resolution. Phys. Med. Biol. 50, 4409–4428.
119 Jiao, S., Xie, Z., Zhang, H. F. & Puliafito, C. A. 2009 Simul- (doi:10.1088/0031-9155/50/18/011)
taneous multimodal imaging with integrated photoacoustic 136 Wang, X., Xie, X., Ku, G., Wang, L. V. & Stoica, G. 2006
microscopy and optical coherence tomography. Opt. Lett. Noninvasive imaging of hemoglobin concentration and
34, 2961–2963. (doi:10.1364/OL.34.002961) oxygenation in the rat brain using high-resolution photo-
120 Jiao, S., Jiang, M., Hu, J., Fawzi, A., Zhou, Q., Shung, acoustic tomography. J. Biomed. Opt. 11, 024015.
K. K., Puliafito, C. A. & Zhang, H. F. 2010 Photoacoustic (doi:10.1117/1.2192804)
ophthalmoscopy for in vivo retinal imaging. Opt. Express 137 Cox, B. T., Arridge, S. R. & Beard, P. C. 2009 Estimating
18, 3967–3972. (doi:10.1364/OE.18.003967) chromophore distributions from multiwavelength
121 Hu, S., Rao, B., Maslov, K. & Wang, L. V. 2010 Label- photoacoustic images. J. Opt. Soc. Am. A Opt.
free photoacoustic ophthalmic angiography. Opt. Lett. Image Sci. Vis. 26, 443– 455. (doi:10.1364/JOSAA.26.
35, 1–3. (doi:10.1364/OL.35.000001) 000443)
122 Yang, J.-M., Maslov, K., Yang, H.-C., Zhou, Q., Shung, 138 Laufer, J., Cox, B., Zhang, E. & Beard, P. 2010 Quanti-
K. K. & Wang, L. V. 2009 Photoacoustic endoscopy. tative determination of chromophore concentrations from
Opt. Lett. 34, 1591–1593. (doi:10.1364/OL.34.001591) 2D photoacoustic images using a nonlinear model-based
123 Karpiouk, A. B., Wang, B. & Emelianov, S. Y. 2010 inversion scheme. Appl. Opt. 49, 1219–1233. (doi:10.
Development of a catheter for combined intravascular 1364/AO.49.001219)
ultrasound and photoacoustic imaging. Rev. Sci. 139 Yin, L., Wang, Q., Zhang, Q. & Jiang, H. 2007
Instrum. 81, 01490. (doi:10.1063/1.3274197) Tomographic imaging of absolute optical absorption
124 a Yaseen, M., Ermilov, S., Brecht, H.-P., Su, R., Conjus- coefficient in turbid media using combined photoa-
teau, A., Fronheiser, M., Bell, B., Motamedi, M. & a coustic and diffusing light measurements. Opt. Lett.
Oraevsky, A. 2010 Optoacoustic imaging of the prostate: 32, 2556– 2558. (doi:10.1364/OL.32.002556)
development toward image-guided biopsy. J. Biomed. 140 Bauer, A. Q., Nothdurft, R. E., Erpelding, T. N., Wang,
Opt. 15, 021310. (doi:10.1117/1.3333548) L. V. & Culver, P. 2011 Quantitative high-resolution
125 Hsieh, B.-Y., Chen, S.-L., Ling, T., Guo, L. J. & Li, P.-C. photoacoustic spectroscopy by combining photoacoustic
2010 Integrated intravascular ultrasound and photoa- imaging with diffuse optical tomography. Proc. SPIE
coustic imaging scan head. Opt. Lett. 35, 2892–2894. 7899, 789930. (doi:10.1117/12.875549)
(doi:10.1364/OL.35.002892) 141 Xiao, J., Yuan, Z., He, J. & Jiang, H. 2010 Quantitative
126 Yuan, Y., Yang, S. & Xing, D. 2010 Preclinical photoa- multispectral photoacoustic tomography and wavelength
coustic imaging endoscope based on acousto-optic optimization. J. X-ray Sci. Technol. 18, 415–427.
coaxial system using ring transducer array. Opt. Lett. 142 Rajian, J. R., Carson, P. L. & Wang, X. 2009 Quantitat-
35, 2266–2268. (doi:10.1364/OL.35.002266) ive photoacoustic measurement of tissue optical
127 Jansen, K., van der Steen, A. F. W., van Beusekom, absorption spectrum aided by an optical contrast agent.
H. M. M., Oosterhuis, J. W. & van Soest, G. 2011 Opt. Exp. 17, 4879–4889. (doi:10.1364/OE.17.004879)

143 Ranasinghesagara, J. C. & Zemp, R. J. 2010 Combined 161 Piras, D., Steenbergen, W., van Leeuwen, T. G. & Man-
photoacoustic and oblique-incidence diffuse reflectance ohar, S. G. 2010 Photoacoustic imaging of the breast
system for quantitative photoacoustic imaging in turbid using the twente photoacoustic mammoscope: present
media. J. Biomed. Opt. 15, 046016. (doi:10.1117/1.3470336) status and future perspectives. IEEE J. Select. Topics
144 Beard, P. C. 2001. Flow velocity measurements. UK Quantum Electron. 16, 730–739. (doi:10.1109/JSTQE.
Patent Application. WO 03/0393. 2009.2034870)
145 Fang, H., Maslov, K. & Wang, L. V. 2007 Photoacoustic 162 Song, K. H., Stein, E. W., Margenthaler, J. A. & Wang,
doppler effect from flowing small light-absorbing par- L. V. 2008 Noninvasive photoacoustic identification of
ticles. Phys. Rev. Lett. 99, 184501. (doi:10.1103/ sentinel lymph nodes containing methylene blue in vivo
physRevlett.99.184501) in a rat model. J. Biomed. Opt. 13, 054033. (doi:10.
146 Fang, H., Maslov, K. & Wang, L. V. 2007 Photoacoustic 1117/1.2976427)
Doppler flow measurement in optically scattering media. 163 Kim, C., Song, K. H., Gao, F., Wang, L. V. & Wang,
Appl. Phys. Lett. 91, 264103. (doi:10.1063/1.2825569) L. V. 2010 Sentinel lymph nodes and lymphatic
147 Sheinfeld, A., Gilead, S. & Eyal, A. 2010 Photoacoustic vessels: noninvasive dual-modality in vivo mapping
Doppler measurement of flow using tone burst excitation. by using indocyanine green in rats—volumetric spec-
Opt. Exp. 18, 4212–4221. (doi:10.1364/OE.18.004212) troscopic. Radiology 255. 442– 450. (doi:10.1148/
148 Sheinfeld, A., Gilead, S. & Eyal, A. 2010 Simultaneous radiol.10090281)
spatial and spectral mapping of flow using photoacoustic 164 Nie, L. & Xing, D. 2008 Microwave-induced thermoa-
Doppler measurement. J. Biomed. Opt. 15, 066010. coustic scanning CT for high-contrast and noninvasive
(doi:10.1117/1.3509113) breast cancer imaging. Med. Phys. 35, 4026 –4032.
149 Brunker, J. & Beard, P. 2010 Pulsed photoacoustic Dop- (doi:10.1118/1.2966345)
pler flowmetry using a cross correlation method. Proc. 165 Ku, G., Fornage, B. D., Jin, X., Xu, M., Hunt, K. K. &
SPIE 7564, 756426. (doi:10.1117/12.841760) Wang, L. V. 2005 Thermoacoustic and photoacoustic
150 Brunker, J. & Beard, P. 2011 Pulsed photoacoustic Dop- tomography of thick biological tissues toward breast ima-
pler flow measurements in blood-mimicking phantoms. ging. Technol. Cancer Res. Treat. 4, 559– 566.
Proc. SPIE 7899, 78991K. (doi:10.1117/12.874469) 166 Erpelding, T., Kim, C., Pramanik, M. & Jankovic, L.
151 a Hein, I. & O’Brien, W. R. 1993 Current time-domain 2010 Sentinel lymph nodes in the rat: noninvasive
methods for assessing tissue motion by analysis from photoacoustic and us imaging with a clinical US
reflected ultrasound echoes-a review. IEEE Trans. Ultra- system. Radiology 256, 102–110. (doi:10.1148/radiol.
son. Ferroelect. Freq. Control 40, 84 –102. (doi:10.1109/ 10091772)
58.212556) 167 Khokhlova, T. D., Pelivanov, I. M., Kozhushko, V. V.,
152 Evans, D. & McDicken, W. 2000 Doppler Ultrasound: Zharinov, A. N., Solomatin, V. S. & Karabutov, A.
physics, instrumentation and signal processing 2nd ed. 2007 Optoacoustic imaging of absorbing objects in a
Chichester, UK: Wiley. turbid medium: ultimate sensitivity and application to
153 Yao, J. J. & Wang, L. H. V. 2010 Transverse flow imaging breast cancer diagnostics. Appl. Opt. 46, 262–272.
based on photoacoustic Doppler bandwidth broadening. (doi:10.1364/AO.46.000262)
J. Biomed. Opt. 15, 021304.(doi:10.1117/1.3339953) 168 Zhang, H. F., Maslov, K. m, Soica, G. & Wang, L. V.
154 Yao, J., Maslov, K., Shi, Y., Taber, L. & Wang, L. 2010 2006 Imaging accute thermal burns by photoacoustic
In vivo photoacoustic imaging of transverse blood flow by microscopy. J. Biomed. Opt. 11, 054031–054033.
using Doppler broadening of bandwidth. Opt. Lett. 35, (doi:10.1117/1.2355667)
1419–1421. (doi:10.1364/OL.35.001419) 169 Yamazaki, M., Sato, S., Ashida, H., Saito, D., Okada, Y.
155 Yao, J., Maslov, K. I. & Wang, L. V. 2011 Noninvasive & Obara, M. 2005 Measurement of burn depths in
quantification of metabolic rate of oxygen (MRO2) by rats using multiwavelength photoacoustic depth
photoacoustic microscopy. Proc. SPIE 7899, 78990N. profiling. J. Biomed. Opt. 10, 064011. (doi:10.1117/1.
(doi:10.1117/12.874777) 2137287)
156 Larina, I. V., Larin, K. V. & Esenaliev, R. O. 2005 Real- 170 Oh, J.-T., Li, M.-L., Zhang, H. F., Maslov, K., Stoica, G.
time optoacoustic monitoring of temperature in tissues. & Wang, L. V. 2010 Three-dimensional imaging of skin
J. Phys. D Appl. Phys. 38, 2633–2639. (doi:10.1088/ melanoma in vivo by dual-wavelength photoacoustic
0022-3727/38/15/015) microscopy. J. Biomed. Opt. 11, 34032. (doi:10.1117/1.
157 Schule, G., Huttmann, G., Framme, C., Roider, J. & 2210907)
Brinkmann, R. 2004 Noninvasive optoacoustic tempera- 171 Wang, B., Su, J. L., Amirian, J., Litovsky, S. H., Smal-
ture determination at the fundus of the eye during laser ling, R. & Emelianov, S. 2010 Detection of lipid in
irradiation. J. Biomed. Opt. 9, 173 –179. (doi:10.1117/1. atherosclerotic vessels using ultrasound-guided spectro-
1627338) scopic intravascular photoacoustic imaging. Opt.
158 Pramanik, M. & Wang, L. V. 2010 Thermoacoustic and Express 18, 4889–4897. (doi:10.1364/OE.18.004889)
photoacoustic sensing of temperature. J. Biomed. Opt. 172 Wang, B., Yantsen, E., Larson, T., Karpiouk, A. B.,
14, 054024. (doi:10.1117/1.3247155) Sethuraman, S., Su, J. L., Sokolov, K. & Emelianov,
159 Manohar, S., Kharine, A., van Hespen, J. C. G., Steen- S. Y. 2009 Plasmonic intravascular photoacoustic imaging
bergen, W. & van Leeuwen, T. G. 2005 The Twente for detection of macrophages in atherosclerotic plaques.
Photoacoustic Mammoscope: system overview and per- Nano Lett. 9, 2212–2217. (doi:10.1021/nl801852e)
formance. Phys. Med. Biol. 50, 2543–2557. (doi:10. 173 Su, J. L.-S., Wang, B. & Emelianov, S. Y. 2009 Photoa-
1088/0031-9155/50/11/007) coustic imaging of coronary artery stents. Opt. Express
160 Manohar, S., Vaartjes, S. E., van Hespen, J. C. G., 17, 19894 –19901. (doi:10.1364/OE.17.019894)
Klaase, J. M., van den Engh, F. M., Steenbergen, W. & 174 Karpiouk, A. B., Aglyamov, S. R., Mallidi, S., Shah, J.,
van Leeuwen, T. G. 2007 Initial results of in vivo non- Scott, W. G., Rubin, J. M. & Emelianov, S. Y. 2010 Com-
invasive cancer imaging in the human breast using bined ultrasound and photoacoustic imaging to detect and
near-infrared photoacoustics. Opt. Express 15, 12277 – stage deep vein thrombosis: phantom and ex vivostudies.
12285. (doi:10.1364/OE.15.012277) J. Biomed. Opt. 13, 054061. (doi:10.1117/1.2992175)

175 Iulia, M. G., Jimmy, S., Doug, Y., James, A., Richard, S. microscopy. J. Biomed. Opt. 14, 020502. (doi:10.1117/
& Stanislav, E. 2011 Methodical study on plaque charac- 1.3095799)
terization using integrated vascular ultrasound, strain 189 Ku, G., Wang, X., Xie, X., Stoica, G. & Wang, L. V. 2005
and spectroscopic photoacoustic imaging. SPIE Proc. Imaging of tumor angiogenesis in rat brains in vivo by
7899, 789902. photoacoustic tomography. Appl. Opt. 44, 770–775.
176 Silverman, R. H., Kong, F., Chen, Y. C., Lloyd, H. O., (doi:10.1364/AO.44.000770)
Kim, H. H., Cannata, J. M., Shung, K. K. & Coleman, 190 Siphanto, R. I., Thumma, K. K., Kolkman, R. G. M., van
D. J. 2010 High-resolution photoacoustic imaging of Leeuwen, T. G., de Mul, F. F. M., van Neck, J. W., van
ocular tissues. Ultrasound Med. Biol. 36, 733–742. Adrichem, L. N . A. & Steenbergen, W. 2005 Serial non-
(doi:10.1016/j.ultrasmedbio.2010.02.006) invasive photoacoustic imaging of neovascularization in
177 de la Zerda, A., Paulus, Y. M., Teed, R., Bodapati, S., tumor angiogenesis. Opt. Express 13, 89 –95. (doi:10.
Dollberg, Y., Khuri-Yakub, B. T., Blumenkranz, M. S., 1364/OPEX.13.000089)
Moshfeghi, D. M. & Gambhir, S. S. 2010 Photoacoustic 191 Kim, C. et al. 2010 In vivo molecular photoacoustic
ocular imaging. Opt. Lett. 35, 270 –272. (doi:10.1364/ tomography of melanomas targeted by bioconjugated
OL.35.000270) gold nanocages. ACS Nano 4, 4559– 4564. (doi:10.1021/
178 Bossy, E., Daoudi, K., Boccara, A. C., Tanter, M., nn100736c)
Aubry, J. F., Montaldo, G. & Fink, M. 2006 Time rever- 192 Razansky, D., Vinegoni, C. & Ntziachristos, V. 2007
sal of photoacoustic waves. Appl. Phys. Lett. 89, 184108. Multispectral photoacoustic imaging of fluorochromes in
(doi:10.1063/1.2382732) small animals. Opt. Lett. 32, 2891– 2893. (doi:10.1364/
179 Funke, A. R., Aubry, J.-F., Fink, M., Boccara, A.-C. & OL.32.002891)
Bossy, E. 2009 Photoacoustic guidance of high intensity 193 de la Zerda, A., Liu, Z., Bodapati, S., Teed, R., Vaithilin-
focused ultrasound with selective optical contrasts and gam, S., Khuri-Yakub, B. T., Chen, X., Dai, H. &
time-reversal. Appl. Phys. Lett. 94, 054102. (doi:10. Gambhir, S. S. 2010 Ultrahigh sensitivity carbon nano-
1063/1.3077018) tube agents for photoacoustic molecular imaging in
180 Funke, A. R. 2010 On the feasibility of photoacoustic gui- living mice. Nano Lett. 10, 2168–2172. (doi:10.1021/
dance of high-intensity focused ultrasound.PhD thesis, nl100890d)
ESPCI ParisTech, Université Pierre et Marie Curie, 194 Li, P.-C., Wang, C.-R. C., Shieh, D.-B., Wei, C.-W., Liao,
Paris, France. C.-K., Poe, C., Jhan, S., Ding, A.-A. & Wu, Y.-N. 2008
181 Chitnis, P. V., Mamou, J., Mclaughlan, J., Murray, T. & In vivo photoacoustic molecular imaging with
Roy, R. A. 2009 Photoacoustic thermometry for thera- simultaneous multiple selective targeting using anti-
peutic hyperthermia. IEEE Ultrason. Symp. 1757–1760. body-conjugated gold nanorods. Opt. Express 16,
182 Essenpreis, M. 1992 Thermally induced changes in opti- 18605 –18615. (doi:10.1364/OE.16.018605)
cal properties of biological tissues. PhD thesis, 195 Li, L., Zemp, R. J., Lungu, G., Stoica, G. & Wang, L. V.
University College London, London, UK. 2007 Photoacoustic imaging of lacZ gene expression in
183 Larin, K. V., Larina, I. V. & Esenaliev, R. O. 2005 vivo. J. Biomed. Opt. 12, 020504. (doi:10.1117/1.
Monitoring of tissue coagulation during thermotherapy 2717531)
using optoacoustic technique. J. Phys. D Appl. Phys. 38, 196 Razansky, D., Distel, M., Vinegoni, C., Ma, R. & Perri-
2645–2653. (doi:10.1088/0022-3727/38/15/017) mon, N. 2009 Multispectral opto-acoustic tomography
184 Pelivanov, I., Sapozhnikov, O. & Khokhlova, A. A. 2006 of deep-seated fluorescent proteins in vivo. Nat. Photon.
Opto-acoustic diagnostics of the thermal action of high- 3, 412–417. (doi:10.1038/nphoton.2009.98)
intensity focused ultrasound on biological tissues: the 197 Maxwell, A., Huang, S.-W., Ling, T., Kim, J.-S., Ashke-
possibility of its applications and model experiments. nazi, S. & Guo, L. J. 2008 Polymer microring
Quant. Electron. 36, 1097–1102. (doi:10.1070/ resonators for high-frequency ultrasound detection and
QE2006v036n12ABEH013261) imaging. IEEE J. Select. Topics in Quantum Electron.
185 Chitnis, P. V., Brecht, H.-P., Su, R. & a Oraevsky, A. 14, 191–197.
2010 Feasibility of optoacoustic visualization of high- 198 Wang, Y., Maslov, K., Zhang, Y., Hu, S., Yang, L., Xia,
intensity focused ultrasound-induced thermal lesions in Y., Liu, J. & Wang, L. V. 2011 Fiber-laser-based
live tissue. J. Biomed. Opt. 15, 021313. (doi:10.1117/1. photoacoustic microscopy and melanoma cell detection.
3339977) J. Biomed. Opt. 16, 011014. (doi:10.1117/1.3525643)
186 Cui, H. & Yang, X. 2010 In vivo imaging and treatment 199 Allen, T. J., Shaiful, A., Zhang, E. Z., Laufer, J. G.,
of solid tumor using integrated photoacoustic imaging Richardson, D. J. & Beard, P. C. 2011 Use of a pulsed
and high intensity focused ultrasound system. Med. fibre laser as an excitation source for photoacoustic tom-
Phys. 37, 4777–4781. (doi:10.1118/1.3480963) ography. Proc SPIE 7899, 78991V. (doi:10.1117/12.
187 Jin, X., Xu, Y., Wang, L. V., Fang, Y. R., Zanelli, C. I. & 875291)
Howard, S. M. 2005 Imaging of high-intensity focused 200 Allen, T. J. & Beard, P. C. 2006 Pulsed NIR laser diode
ultrasound-induced lesions in soft biological tissue using excitation system for biomedical photoacoustic imaging.
thermoacoustic tomography. Med. Phys. 32, 5 –11. Opt. Lett. 31, 3462– 3464.
(doi:10.1118/1.1829403) 201 Kolkman, R. G. M., Steenbergen, W. & Leeuwen, T. 2006
188 Stein, E. W., Maslov, K. & Wang, L. V. 2010 Non- In vivo photoacoustic imaging of blood vessels with a
invasive, in vivo imaging of blood-oxygenation pulsed laser diode. Lasers Med. Sci. 21, 134–139.
dynamics within the mouse brain using photoacoustic (doi:10.1007/s10103-006-0384-z)

Biomedical Photoacoustic Imaging

Uploaded by

Copyright:

Available Formats

Biomedical Photoacoustic Imaging

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Biomedical Photoacoustic Imaging

Uploaded by

Copyright:

Available Formats

Interface Focus (2011) 1, 602–631

Biomedical photoacoustic imaging

1. OVERVIEW propagating in a gas generated by laser-induced surface

Received 23 March 2011

Interface Focus (2011)

Interface Focus (2011)

10 000 by water rather than haemoglobin and the strong

Interface Focus (2011)

Interface Focus (2011)

(a) (b) (c)

cylindrical or spherical detection geometries requires

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

fibre Bundle ovarian

(a) (b) 1.5 (c)

Interface Focus (2011)

was estimated at approximately 200 mm. Following this (a)

Interface Focus (2011)

(a) part a consequence of the limited detection aperture

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

(a) y optical fiber dye Nd:YLF

this can, in principle, provide a non-divergent receive inexpensively implemented as a laboratory-based

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

1200 nm atheromatous plaque

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

Interface Focus (2011)

You might also like