IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 14, NO.

1, JANUARY/FEBRUARY 2008 171

Tutorial on Photoacoustic Microscopy

and Computed Tomography
Lihong V. Wang

(Invited Paper)

Abstract—The field of photoacoustic tomography has experi- tion) and other thermal and mechanical properties—propagates
enced considerable growth in the past few years. Although several as an ultrasonic wave, which is referred to as a photoacoustic
commercially available pure optical imaging modalities, includ- wave. The photoacoustic wave is detected by an ultrasonic trans-
ing confocal microscopy, two-photon microscopy, and optical co-
herence tomography, have been highly successful, none of these ducer, producing an electric signal. The electric signal is then
technologies can provide penetration beyond ∼1 mm into scat- amplified, digitized, and transferred to a computer. To form an
tering biological tissues, because they are based on ballistic and image, a single-element ultrasonic transducer is scanned around
quasi-ballistic photons. Heretofore, there has been a void in high- the tissue; alternatively, an ultrasonic array can be used to ac-
resolution optical imaging beyond this penetration limit. Photoa- quire data in parallel.
coustic tomography, which combines high ultrasonic resolution and
strong optical contrast in a single modality, has broken through this PAI has two major forms of implementation. One is based on
limitation and filled this void. In this paper, the fundamentals of a scanning-focused ultrasonic transducer. Dark-field confocal
photoacoustics are first introduced. Then, scanning photoacoustic photoacoustic microscopy belongs to this category. The other is
microscopy and reconstruction-based photoacoustic tomography based on an array of unfocused ultrasonic transducers in com-
(or photoacoustic computed tomography) are covered. bination with a reconstruction algorithm. A circular detection
Index Terms—Microwave-induced acoustic imaging, optical configuration will be illustrated.
imaging, optoacoustic imaging, photoacoustic imaging (PAI), The image contrast of PAI is based on optical absorption in
photoacoustic microscopy (PAM), photoacoustic tomography, the photoacoustic excitation phase. Selective optical absorption
thermoacoustic tomography.
is associated with molecules such as oxygenated and deoxy-
genated hemoglobin and melanin. Concentrations of multiple
I. INTRODUCTION chromophores whose spectra of absorption coefficient are differ-
ent can be quantified simultaneously by varying the wavelength
HOTOACOUSTIC IMAGING (PAI) refers to imaging that
P is based on the photoacoustic effect. Although the photoa-
coustic effect as a physical phenomenon was first reported by
of the irradiating laser. Such quantification of oxygenated and
deoxygenated hemoglobin, for example, can provide functional
imaging of the total concentration and oxygen saturation of
Alexander Graham Bell in 1880 [1], PAI as an imaging tech-
hemoglobin.
nology was developed only after the advent of ultrasonic trans-
The spatial resolution of PAI is derived from ultrasonic detec-
ducers, computers, and lasers [2]–[31]. In this paper, we will
tion in the photoacoustic emission phase. Due to strong optical
focus on the fundamentals of photoacoustics, which is adapted
scattering, pure optical imaging in biological tissue has either
from a chapter in [32], then illustrate PAI with two implemen-
shallow imaging depth or low spatial resolution. Although light
tations. A comprehensive review on biomedical photoacoustics
scattering broadens and attenuates light reaching an embedded
and imaging is available elsewhere [33].
object, the resulting ultrasonic signal provides better resolution
The motivation for PAI is to combine optical-absorption con-
than the optical signal in the optical quasi-diffusive or diffusive
trast with ultrasonic spatial resolution for deep imaging in the
regime, because ultrasonic scattering is two to three orders of
optical quasi-diffusive or diffusive regime, which is >1 mm in
magnitude weaker than optical scattering. The image resolution,
most biological tissues. In PAI, the tissue is irradiated usually
as well as the maximum imaging depth, is scaleable with the
by a short-pulsed laser beam to achieve a thermal and acoustic
ultrasonic frequency within the reach of diffuse photons. Specifi-
impulse response. Locally absorbed light is converted into heat,
cally, as the ultrasonic center frequency and bandwidth increase,
which is further converted to a pressure rise via thermoelastic
the spatial resolution improves at the expense of imaging depth,
expansion. The initial pressure rise—determined by the local
because ultrasonic attenuation increases with frequency. In ad-
optical energy deposition (specific or volumetric optical absorp-
dition, PAI provides images devoid of speckle artifacts, which
are conspicuous in images acquired with either ultrasonography
Manuscript received September 10, 2007; revised November 12, 2007. This or optical coherence tomography.
work was supported by the National Institutes of Health under Grant R01
EB000712 and Grant R01 NS46214.
The author is with the Optical Imaging Laboratory, Department of Biomedical
Engineering, Washington University, St. Louis, MO 63130-4899 USA (e-mail:
lhwang@[Link]). II. INITIAL PHOTOACOUSTIC PRESSURE
Color versions of one or more of the figures in this paper are available online
at [Link] Upon short laser pulse excitation, local fractional volume
Digital Object Identifier 10.1109/JSTQE.2007.913398 expansion dV /V of the heated tissue at position r can be

1077-260X/$25.00 © 2008 IEEE

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.
172 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 14, NO. 1, JANUARY/FEBRUARY 2008

expressed as III. PHOTOACOUSTIC EQUATION

dV The photoacoustic wave generation and propagation in an in-
= −κp(r) + βT (r). (1) viscid medium is described by the general photoacoustic equa-
V
tion
Here, κ denotes the isothermal compressibility (∼5 ×

10−10 Pa−1 for water or soft tissue), β denotes the thermal coef- 1 ∂2 β ∂ 2 T (r, t)
∇ − 2 2 p(r, t) = − 2
2
(9)
ficient of volume expansion (∼4 × 10−4 K−1 for muscle), and vs ∂t κvs ∂t2
p and T denote the changes in pressure (in Pascal) and temper- where p(r, t) denotes the acoustic pressure at location r and
ature (in Kelvin), respectively. The isothermal compressibility time t, and T denotes the temperature rise. The left-hand side
κ can be expressed as of this equation describes the wave propagation, whereas the
CP right-hand side represents the source term.
κ= . (2) For a short laser pulse that satisfies the condition shown in
ρvs2 CV
(3), the following thermal equation holds:
Here, ρ denotes the mass density (∼1000 kg/m3 for water and ∂T (r, t)
soft tissue), vs denotes the speed of sound (∼1480 m/s in water), ρCV = H(r, t). (10)
∂t
and CP and CV [∼4000 J/(kg·K) for muscle] denote the specific
Here, H is the heating function defined as the thermal en-
heat capacities at constant pressure and volume, respectively. It
ergy converted per unit volume and per unit time; it is related
is important to distinguish between CP and CV for gasses but
to the optical absorption coefficient µa and fluence rate Φ by
not for tissue.
H = µa Φ. As a result, we obtain the following photoacoustic
Here, we assume that the laser pulse duration tL is less than
equation:
the acoustic confinement time—which is less than the thermal

confinement time 1 ∂2 β ∂H (r, t)
∇ − 2 2 p (r, t) = −
2
. (11)
dc d2 vs ∂t CP ∂t
tL < < c (3)
vs 4αth Because the source term is related to the first time derivative
of H, time-invariant heating does not produce a pressure wave;
where dc is the characteristic length of heat heterogeneity (the
only time-variant heating does.
dimension of the optically absorbing target of interest or the
decay constant of the optical energy deposition, whichever is
IV. FORWARD SOLUTION
smaller), and αth is the thermal diffusivity (∼0.1 mm2 /s for
tissue). The general photoacoustic equation shown in (9) can be
For such a short laser pulse, the fractional volume expansion solved by the Green’s function approach [35]. The Green’s
is negligible and the local pressure rise p0 immediately after the function is defined here as the response to a spatial and tem-
laser excitation can be derived from (1) [34] as poral impulse source term


1 ∂2
p0 (r) =
βT (r)
. (4) ∇ − 2 2 G(r, t; r , t ) = −δ(r − r )δ(t − t ) (12)
2

κ vs ∂t
It can be shown that each millikelvin temperature rise yields where r and t denote the source location and time, respectively.
an 8 mbar (or 800 Pa) pressure rise. If we assume that all ab- In an infinite space, where no boundary exists, the Green’s
sorbed optical energy is converted into heat and nonthermal function is given by
relaxation such as fluorescence is negligible, the temperature δ(t − t − |r − r |/vs )
rise generated by the short laser pulse is G(r, t; r , t ) = (13)
4π|r − r |
Ae which represents an impulse diverging spherical wave. The fol-
T = (5)
ρCV lowing reciprocity relation holds:
where Ae is the specific or volumetric optical absorption (in G(r, t; r , t ) = G(r , −t ; r, −t). (14)
joules per centimeter cubed, optical energy deposition density).
To see this relationship more clearly, one observes G(r, t;
From (4) and (5), we have
r , 0) = G(r , 0; r, −t) by setting t = 0.
β Applying the Green’s function approach to (9), we get
p0 = Ae . (6)
κρCV  t+ 
 β ∂ 2 T (r , t )
p(r, t) = dt dr G(r, t; r , t ) 2 (15)
We define the Grueneisen parameter (dimensionless) as −∞ κvs ∂t2
β which represents the pressure in response to an arbitrary source.
Γ= . (7) Substituting (13) into (15) leads to
κρCV
 
Hence, (6) becomes β  1 ∂ 2 T (r , t ) 
p(r, t) = dr  . (16)
4πκvs2 |r − r | ∂t2 t =t−
|r −r  |
vs
p0 = ΓAe . (8)

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.
WANG: TUTORIAL ON PHOTOACOUSTIC MICROSCOPY AND COMPUTED TOMOGRAPHY 173

In thermal confinement, substituting (10) into (16), we get Similarly, for a spherical object of radius Rs , we have
  
β 1 ∂H(r , t )  r − vs t
p(r, t) = dr  (17) p(r, t) = p0 U (Rs − vs t − r) +
4πCP |r − r | ∂t t =t−
|r −r  | 2r
vs

or × U (r − |Rs − vs t|)U (Rs + vs t − r) . (25)


β ∂  1  |r − r |
p(r, t) = dr H r ,t − . (18)
4πCP ∂t |r − r | vs The radial coordinate r originates at the center of the sphere.
If we write the initial pressure as
If the heating function
 can be decomposed as H(r , t ) =
Ae (r )Ht (t ), where Ht (t )dt = 1, (18) can be further sim-
  p0 (r) = p0 U (r)U (−r + Rs ), for 0 ≤ r < Rs (26)
plified to
we have


β ∂ Ae (r ) |r − r | r + vs t r − vs t
p(r, t) = dr H t t − . (19) p(r, t) = p0 (r + vs t) + p0 (−r + vs t)
4πCP ∂t |r − r | vs 2r 2r
If the condition in (3) holds, we have approximately Ht (t ) = r − vs t
+ p0 (r − vs t). (27)
δ(t ). In this case, (19) becomes 2r
 
 The first term on the right-hand side represents a converging
∂ β 1   |r − r |
p(r, t) = dr Ae (r )δ t − spherical wave; the second term represents a diverging spher-
∂t 4πCP vs t vs ical wave that originates from the initially converging wave
(20)
propagating through the center, and the third term represents a
where the quantity within the square brackets is the step-heating
diverging spherical wave.
response of an arbitrary absorbing object and its time differen-
When an ultrasonic detector is placed outside the sphere at
tiation yields the delta-heating response. Using (7) and (8), we
various radial locations, the detected pressures are shown in
rewrite (20) as
Fig. 1. Upon delta heating, an initial pressure p0 —constant
 

1 ∂ 1   |r − r | across the entire heated sphere—is generated. This initial pres-
p(r, t) = dr p 0 (r )δ t − . (21) sure is divided into two equal parts, each initiating a spherical
4πvs2 ∂t vs t vs
wave. One travels outward as a diverging spherical compression
This equation can be used to compute photoacoustic pressure wave, yielding the first-arriving positive pressure in the plot.
generated by an arbitrarily heterogeneous optically absorbing The other travels inward as a converging spherical compression
object. wave. When reaching the center of the heated spherical object,
We now consider an optically thin slab of thickness d, heated the converging spherical wave becomes a diverging spherical
up homogeneously by a delta excitation pulse. The laser beam, rarefaction wave, yielding the late-arriving negative pressure.
infinitely broad and uniform, is incident normally on the slab.
The attenuation inside the slab is negligible. An initial pressure V. DARK-FIELD CONFOCAL PHOTOACOUSTIC MICROSCOPY
p0 is first built up within the slab, and then, propagated outward
in both the positive and the negative z-directions, where the z- In this section, we introduce the dark-field confocal photoa-
axis is perpendicular to the slab with the origin on the middle coustic microscopy (PAM) (Fig. 2) based on a scanning-focused
plane of the slab. The initial pressure distribution can be written ultrasonic transducer [36], [37]. In conventional dark-field trans-
as mission optical microscopy, an opaque disc is placed between


 the light source and the condenser lens, so that ballistic light is
d d
p0 (z) = p0 U z + U −z + (22) rejected; as a result, only nonballistic light—which is scattered
2 2 by the sample—is detected. In dark-field PAM, the excitation
where U is the Heaviside step function defined as laser beam has a donut-shaped cross section; therefore, the pho-
 toacoustic signal from the tissue surface in the field of view
1 for z ≥ 0. is minimized. The PAM can image optical-absorption contrast
U (z) = (23)
0 for z < 0. beyond the existing depth limit for high-resolution optical imag-
Solving (21) leads to the pressure distribution at any later ing; it can achieve a ratio of maximum imaging depth-to-depth
time resolution (uninterpolated pixel count) greater than 100, which
is defined here as high relative spatial resolution.
1 1
p(z, t) = p0 (z − vs t) + p0 (z + vs t). (24) Acoustic coupling requires contact measurement of the sam-
2 2 ple. The components within the dashed-box in the photograph
The first term on the right-hand side represents a right prop- (Fig. 2) are translated in a water bath. A window at the bottom
agating (along the +z-axis) plane wave and the second term of the water container is sealed with an optically and ultrasoni-
represents a left propagating (along the −z-axis) plane wave. cally transparent disposable polyethylene membrane (thickness:
On delta heating, pressure p0 is generated within the slab. Then, 44 µm). After commercial ultrasound gel is applied to the region
p0 is split into two plane waves, each having a magnitude of of interest on the sample for acoustic coupling, the sample is
p0 /2 but propagating in opposite directions. placed between the water container and the sample supporter

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.
174 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 14, NO. 1, JANUARY/FEBRUARY 2008

short laser pulses are able to generate photoacoustic waves of a

>100 MHz bandwidth (approximately the reciprocal of the laser
pulse width) at the target. Laser light is delivered through an op-
tical fiber to the PAM scanner. Because the optical fiber homog-
enizes the light beam, no additional measure needs to be taken
to avoid hot spots in the laser beam. For hemoglobin imaging
using PAM, the laser wavelength is typically set in the Q-band
of hemoglobin (∼560 nm) to achieve good SNR. The energy
of each laser pulse is recorded by a photodiode for shot-to-shot
calibration. The laser beam from the fiber passes through a con-
ical lens to form a ring-shaped illumination pattern. It is then
weakly focused into the tissue with the focal region coaxially
overlapping the ultrasonic focus inside the tissue. The optical
illumination on the skin surface is donut-shaped with a dark cen-
ter. In an optically clear medium, the optical focus is measured
to be 2 mm in diameter, which is much wider than the ultra-
sonic focus. The time-resolved photoacoustic wave is recorded
at each location of the ultrasonic transducer for 2 µs, and sub-
sequently, converted into a 1-D depth-resolved image (A-line)
based on the sound velocity in soft tissue (1.54 mm/µs). Then,
raster scanning of the dual optical ultrasonic foci in the hori-
zontal (x–y) plane with a step size of 50 µm produces a 3-D
image. No signal averaging is required in data acquisition for
this particular system.
Spatial resolution is determined by the ultrasonic parameters.
The lateral resolution of PAM is determined by the focal di-
ameter of the ultrasonic transducer at the center frequency [38],
whereas the axial resolution is inversely related to the bandwidth
of the ultrasonic transducer. The lateral resolution is usually es-
timated by the peak-to-zero width of the diffraction pattern on
the focal plane, and is given by 0.61 (λ0 /NA), where λ0 is the
center acoustic wavelength and NA is the numerical aperture
of the ultrasonic transducer. Therefore, achieving high spatial
Fig. 1. Bipolar (positive followed by negative) pressure profiles from a homo- resolution requires the use of an ultrasonic transducer with a
geneously heated sphere versus time observed at various radial positions [(a) large NA, a high center frequency, and a wide bandwidth. In
2R S , (b) 4R S , and (c) 6R S ] from the center of the sphere. Note the amplitude
and duration of each bipolar pulse. the PAM reported here, the ultrasonic detector (V214-BB-RM,
Panametrics, WA; element diameter: 6 mm) has a 50 MHz cen-
ter frequency and a 35 MHz nominal bandwidth. A spherically
focusing lens provides an NA of 0.44, a focal length of 6.7 mm,
and a focal zone of 0.3 mm. This PAM system is found to
have an axial resolution of ∼15 µm and a lateral resolution of
∼45 µm. An imaging depth of more than 3 mm in biological
tissue has been reached in animals. Therefore, this PAM sys-
tem has a depth-to-resolution ratio of 200 and is considered
to possess high relative spatial resolution. In a soft tissue, the
ultrasonic penetration limit is inversely proportional to the ul-
trasonic frequency because the acoustic attenuation coefficient
is proportional to the ultrasonic frequency. As a result, both the
spatial resolution and the maximum imaging depth of PAM are
Fig. 2. Photoacoustic microscopy (PAM) system. Reproduced with permis- scaleable with ultrasonic parameters within the reach of excita-
sion from [37]. tion photons.
The major optical absorbers in a biological tissue in the vis-
for imaging. Note that the sample here can also refer to live ible spectral region include deoxyhemoglobin (HbR), oxyhe-
subjects. moglobin (HbO2 ), and melanin. At the 584 nm optical wave-
Image formation starts from the laser. For the generation length, PAM can image the morphological relationship between
of photoacoustic waves, 6.5 ns laser pulses from a tunable a melanoma tumor and the surrounding blood vessels in the x–y
dye laser that is pumped by an Nd:YAG laser are used. Such plane, because both melanin and hemoglobin have comparably

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.
WANG: TUTORIAL ON PHOTOACOUSTIC MICROSCOPY AND COMPUTED TOMOGRAPHY 175

for the melanoma is 67 and 64 dB at the 584 and 764 nm wave-

lengths, respectively. In addition to the functional imaging of
total hemoglobin concentration, PAM is capable of providing
functional imaging of oxygen saturation of hemoglobin using
multiwavelength measurements.
PAM is safe for human subjects since it delivers a transient
fluence of only ∼6 mJ/cm2 at the optical focus (pulse energy:
0.2 mJ; focal diameter: ∼2 mm), which is well within the
American National Standard Institute (ANSI) safety standard
(20 mJ/cm2 ) in the visible spectrum (400–700 nm) [39]. If a
point on the skin is exposed to laser light for more than 10 s,
ANSI stipulates that the mean irradiance should not exceed
200 mW/cm2 , which results in a maximum permissible repe-
tition rate of ∼30 Hz here. Since PAM works in raster scan-
ning mode, the exposure duration of any given point on the
skin surface is typically less than 10 s. In this case, ANSI
requires that the maximum permissible exposure is limited
1/4
Fig. 3. Images of a subcutaneously inoculated B16-melanoma in a nude mouse by 1100te in megajoules per centimeter cube, where te de-
acquired in vivo by PAM at 584 and 764 nm, respectively. (a) Photograph of a notes the √ exposure duration in seconds. This limit translates
melanoma. It shows no tumor morphology or vessel structure. (b) Composite of into E × 4 PRF ≤ 2.75 × 102 πd5/4 σ 3/4 , where E denotes the
the two maximum amplitude projections (MAP) images projected along the z-
axis. Six orders of vessel branching (No. 1–6) can be observed. (c) 3-D rendering pulse energy in megajoules, PRF denotes the pulse repetition
of the melanoma acquired at the 764 nm wavelength. Two MAP images at this frequency in hertz, d denotes the diameter of the illumination
wavelength projected along the x- and y-axes are shown on the sidewalls. The area at the ultrasonic focus in centimeter, and σ denotes the
composite image in panel b is redrawn at the bottom. The top of the tumor
is 0.32 mm below the skin surface and the thickness of the tumor is 0.3 mm. scanning step size in centimeter [40]. From the parameters of
(d) Close-up B-scan image of the melanoma parallel with the z–x plane at the the current PAM, PRF can be as high as 13 kHz. The subcuta-
location marked with a dashed line in panel b. (e) HE-stained section at the same neous microvasculature of a human palm has been imaged by
marked location. M, melanoma. Reproduced with permission from [37].
PAM in vivo (results not shown). The largest vessel in the image
is approximately 350 µm in diameter, whereas the smallest is
strong absorption at this wavelength. At the 584 nm isosbestic within one pixel (50 µm).
optical wavelength, where the two forms of hemoglobin have PAM is fundamentally a high-speed technology. The acqui-
an identical molar extinction coefficient, the contrast of PAM is sition time of PAM is currently limited by the 10 Hz laser pulse
sensitive to the total hemoglobin concentration but insensitive repetition frequency. While scanning for a single B-scan image
to the hemoglobin oxygenation. However, visible light at this takes only 10 s, a 2-D scan over an 8 mm × 8 mm area takes
wavelength is unsuitable for measuring the tumor thickness be- more than 18 min for single-wavelength imaging. The data ac-
cause it cannot easily penetrate through the melanin-rich tumor. quisition can be shortened considerably, however. Each A-line
To overcome this limitation, we form another image using near- is acquired within only 2 µs, which equals the maximum imag-
infrared light at the 764 nm wavelength, where light can infiltrate ing depth (3 mm) divided by the speed of sound (∼1.5 mm/µs).
the tumor because of both the decreased optical absorption of The time lapse between two consecutive laser pulses should be
melanin and the minimal absorption of blood. The combination long enough for the photoacoustic wave induced by the first one
of the two images from these two spectral regions reveals the to subside. If the time lapse is set to 20 µs, for example, the
3-D morphology of both the melanoma and the surrounding laser pulse repetition rate can be as high as 50 kHz, which is
vasculature, where some parallel arterioles and venules are ev- greater than the aforementioned 13 kHz limited by the ANSI
ident (Fig. 3). In the PAM images, the blood vessels from the safety standard. The author’s laboratory is working on acceler-
584-nm image are shown in red and the melanoma from the ating data acquisition using a much higher laser pulse repetition
764 nm image is shown in brown. In Fig. 3(b), microvessels frequency (2 kHz).
with diameters of less than a single pixel (50 µm), presumably
VI. PHOTOACOUSTIC COMPUTED TOMOGRAPHY
resulting from angiogenesis, are observed clearly surrounding
the tumor. At the 584 nm wavelength, the average ratio of the In this section, we consider the reconstruction-based pho-
vessel to the background in photoacoustic signal amplitude is toacoustic computed tomography (PAT) that is formed by
13 ± 0.89, and the average ratio of the melanoma tumor to scanning an unfocused ultrasonic transducer—ideally a point
the surrounding blood vessels is 0.92 ± 0.02. However, at the transducer—followed by image reconstruction. The initial pho-
764 nm wavelength, the average ratio of the melanoma tu- toacoustic pressure excited by laser pulse δ(t) equals p0 (r) [see
mor to the blood vessels is 29 ± 3, and the average ratio of (8)]. The acoustic pressure p(r0 , t) at position r0 and time t,
the melanoma tumor to the background is as high as 68 ± 5. initiated by source p0 (r), is measured around the tissue by an
Such high optical contrast and specificity are due to the mini- unfocused ultrasonic transducer.
mal background absorption, and the strong absorption of both Spherical, cylindrical, and planar detection configurations are
hemoglobin and melanin. The average contrast-to-noise ratio considered (Fig. 4), where the detection surface S0 encloses the

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.
176 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 14, NO. 1, JANUARY/FEBRUARY 2008

Fig. 4. Reconstruction-based photoacoustic computed tomography. (a) Spher-

ical or cylindrical detection configuration. During measurement, an ultrasonic
point detector at position r 0 on surface S 0 receives photoacoustic signals emit-
ted from source p 0 (r). During image reconstruction, a quantity related to the
measurement at position r 0 projects backward via a spherical surface (not
shown) centered at r 0 . (b) Planar detection configuration.

Fig. 6. In vivo PAT images of a rat brain acquired. (a) Before three sequential
administrations of nanoshells. (b) ∼20 min after three sequential administrations
of nanoshells. (c) Differential image (c = b − a). Field of view: 20 × 20 mm.
Ae, optical absorption; ∆A e , differential optical absorption. (d) Differential
signal versus time. Adapted with permission from [43].

the skin surface to conform to the ANSI standards [39]. Upon

laser excitation, photoacoustic waves are produced in the head.
An unfocused ultrasonic transducer (XMS-310, Panametrics)
with a central frequency of 10.4 MHz, a bandwidth of 100% at
∼6 dB and an active element of 2 mm in diameter is submerged
in water to detect the photoacoustic signals. A rotational de-
tection system that is driven by a computer-controlled stepper
motor scans the transducer around the brain with a radius of
Fig. 5. Diagram of a circular-scanning PAT system for small-animal imaging. 3 cm and a step size of 1.5◦ . The data acquisition time to obtain
Reproduced with permission from [43]. a single 2-D image slice is ∼24 min at this pulse repetition rate.
source p0 (r). For brevity, we simply provide the final back- The photoacoustic signals detected by the transducer are ampli-
projection formula [41] without showing the derivation fied by an amplifier, and then, digitized by an oscilloscope (TDS
   540B, Tektronix). Finally, a computer acquires the signals for
1 2t∂p(r0 , t) cos θ0 image reconstruction.
p0 (r) = 2p(r0 , t) − dS0
Ω0 S 0 ∂t |r − r0 |2 Since a circle—rather than a full spherical surface—is usu-
(28) ally scanned, the reconstruction algorithm given by (28) is only
where Ω0 = 2π for the planar geometry, Ω0 = 4π for the spher- approximately applicable. Nevertheless, good images are still at-
ical or cylindrical geometry, and θ0 denotes the angle between tainable. The in-plane spatial resolution of this imaging system,
the detection surface normal and the vector pointing to the re- limited mainly by the bandwidth of the detected photoacous-
construction point r, as illustrated in Fig. 4(a). tic signals, is ∼60 µm. The out-of-plane resolution, limited by
A representative PAT system [42], [43] is shown in Fig. 5. An the aperture effect of the transducer, is ∼2 mm. Alternatively,
Nd:YAG laser (Brilliant B, Bigsky), operating at a wavelength ultrasonic transducer cylindrically focused in the direction per-
of 532 nm with a pulsewidth of 6.5 ns and a pulse repetition pendicular to the imaging plane can be used to improve the
rate of 10 Hz, pumps a tunable dye laser (ND6000, Continuum). out-of-plane resolution. In addition, a cylindrical scan can be
A small animal (rat or mouse) to be imaged is immobilized on taken to form a 3-D image with better out-of-plane resolution.
an animal holder. The head of the animal protrudes into a wa- We demonstrate the imaging of nanoshells (silica core diam-
ter tank through an opening at the bottom of the tank, where eter: 125 ± 5 nm; gold shell thickness: 10–12 nm), which serve
the opening is sealed using a clear polyethylene membrane. as a contrast agent for PAT, in a rat in vivo [43]. Two of the PAT
The laser beam is expanded and homogenized to provide rela- images of the cerebral cortex of the rat brain are presented in
tively uniform incident fluence, which is less than 10 mJ/cm2 on Fig. 6(a) and (b), where the color bars are identical. With the

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.
WANG: TUTORIAL ON PHOTOACOUSTIC MICROSCOPY AND COMPUTED TOMOGRAPHY 177

TABLE I Hence, the photoacoustic signal (p0 ) is proportional to the

COMPARISON OF HIGH-RESOLUTION OPTICAL IMAGING MODALITIES
optical absorption coefficient. If we focus on the variation in the
local absorption coefficient, we have ∆p0 /p0 =∆µa /µa , where
∆ indicates a small variation in the modified variable. Here, the
effect of ∆µa on ∆p0 through F is neglected. Therefore, any
fractional change in the optical absorption coefficient translates
into an equal amount of fractional change in the photoacoustic
signal, which means a relative sensitivity of unity.
By contrast, the signal in confocal microscopy can be modeled
as
exogenous contrast agent in the blood, the optical absorption of
Icf ∝ Rb exp[−2(µa + µs )zp ] (30)
the blood at 800 nm was increased, and the contrast between the
vessels and the background brain tissues was enhanced. The dif- where Rb denotes the percentage of the back-scattered light to
ferential image in Fig. 6(c) depicts the distribution of differential be received by the detector, µs denotes the optical scattering
optical absorption in the rat brain due to the exogenous contrast coefficient, zp denotes the length of a pixel, and the factor 2
agent. The optical absorptions in the major blood vessels were indicates the effect of the round-trip ballistic light propagation.
averaged to represent the relative concentration of nanoshells in The effect of the variation in the absorption coefficient can be
the circulatory system of the rat [Fig. 6(d)]. After each of the expressed as
three administrations, the optical absorption of blood increased.
The first administration increased the absorption of blood by (−2µa zp )∆µa
∆Icf /Icf = . (31)
∼30%, whereas the three administrations together increased the µa
absorption by ∼63%. For typical pixel length zp = 1 µm and hemoglobin ab-
Then, the optical absorption decreased owing to clearance of sorption coefficient µa = 4, 300, and 3000 cm−1 at 800, 560
the nanoshells from the blood. (Q-band), and 420 nm (Soret band), respectively, the sensi-
After the first administration of the nanoshells, the con- tivities Scf = 2µa zp are 8 × 10−4 , 6 × 10−2 , 6 × 10−1 , respec-
centration of nanoshells in the blood is estimated to be tively. Unless the Soret band is chosen, the sensitivity to optical
∼1010 nanoshells/mL. Since the in-plane spatial resolution of absorption of confocal microscopy is significantly lower than
this PAT system is ∼60 µm and the diameters of the corti- that of photoacoustic imaging. However, the operation at the
cal blood vessels under study are <100 µm, the number of Soret band limits the penetration severely because the penetra-
nanoshells in the 60 µm × 60 µm × 100 µm resolvable volume tion depth is ∼3 µm in blood.
is ∼3600, which represents an estimate of the sensitivity of this In comparison to confocal microscopy, two-photon mi-
PAT system to nanoshells. croscopy has half the sensitivity to the one-photon optical ab-
sorption coefficient because the two-photon excitation involves
VII. COMPARISON WITH OTHER HIGH-RESOLUTION OPTICAL only one-way ballistic optical propagation and the two-photon
IMAGING MODALITIES fluorescence detection can tolerate multiple scattering. Interest-
Although several commercially available pure optical imag- ingly, optical coherence tomography also has half the sensitivity
ing modalities, including confocal microscopy, two-photon mi- because the electric field instead of the intensity is detected in-
croscopy, and optical coherence tomography, have been highly terferometrically. As a result, the sensitivity becomes 4 × 10−3
successful, none of these technologies can provide penetration around the typically used 800 nm wavelength with zp = 10 µm.
beyond ∼1 mm into scattering biological tissues because they
are based on ballistic and quasi-ballistic photons. Heretofore, VIII. LIMITATIONS OF PHOTOACOUSTIC IMAGING
there has been a void in the high-resolution optical imaging be-
There are two important depth penetration limits for optical
yond this penetration limit. Photoacoustic tomography, which
imaging. The 1 mm optical penetration limit is approximately
combines high ultrasonic resolution and strong optical contrast
correlated with the optical transport mean free path, representing
in a single modality, has broken through this penetration limi-
the depth of the quasi-ballistic regime in biological tissue [32].
tation and filled this void. Table I compares the various high-
We refer to this limit as the soft limit for high-resolution optical
resolution optical imaging modalities. The various modalities
imaging. Photoacoustic imaging is able to break through this
complement each other in both penetration and contrast, and
soft limit. Another limit exists at ∼50 mm depth, which equals
therefore, will coexist. While photoacoustic imaging is sensi-
roughly ten times the 1/e optical penetration depth in the near-
tive to optical absorption, it is insensitive to optical scattering
infrared window (∼700–800 nm wavelength). Penetration to
or fluorescence.
this limit in the diffusive regime requires approximately 43 dB
Photoacoustic imaging is more sensitive to optical absorption
one-way energy decay. We refer to this limit as the hard limit for
than that of the other modalities. Since Ae = µa F , where µa
optical imaging although it does not represent a theoretical one.
denotes the optical absorption coefficient and F denotes the
Beyond this limit, the optical energy is low and even diffuse
optical fluence (in joules per centimeter square), (8) becomes
photons are few. Diffuse optical tomography and photoacoustic
p0 = Γµa F. (29) tomography have reached this limit [44], [45]. However, if we

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.
178 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 14, NO. 1, JANUARY/FEBRUARY 2008

image tissue such as the human breast from both sides, a 10 cm is limited by the penetration depth of diffuse light rather than
thickness can be potentially imaged, which is adequate for many ballistic light. On the other hand, the spatial resolution can be
biomedical applications. One approach to overcome the optical scaled down to <10 µm, where the imaging depth is limited
penetration limitation is to adopt microwaves for photoacoustic by the penetration depth of high-frequency ultrasound. 3) PAI
excitation [5], [6]. In this case, the technology is referred to as provides high sensitivity to optical absorption. 4) PAI provides
microwave-induced acoustic (thermoacoustic) imaging. functional imaging based on physiologically specific endoge-
nous optical absorption contrasts. 5) PAI provides molecular or
reporter gene imaging based on exogenous contrasts. 6) PAI can
IX. SUMMARY AND DISCUSSION potentially be constructed to image in real time. 7) PAI provides
Several differences between the earlier described two repre- speckle-free images. 8) PAI is safe for human subjects, and it is
sentative implementations of photoacoustic imaging are sum- ready for clinical application.
marized here. First, the confocal PAM operates in reflection
(backward) mode, whereas the reconstruction-based PAT works REFERENCES
in orthogonal mode. Second, the confocal PAM provides lateral [1] A. G. Bell, “On the production and reproduction of sound by light,” Amer.
resolution using spherical focusing of the ultrasonic transducer J. Sci., vol. 20, pp. 305–324, 1880.
and the axial resolution using the time resolution of the trans- [2] A. A. Karabutov, N. B. Podymova, and V. S. Letokhov, “Time-resolved
laser optoacoustic tomography of inhomogeneous media,” Appl. Phys.
ducer. In contrast, the circular-scanning PAT derives in-plane B-Lasers Opt., vol. 63, pp. 545–563, 1996.
resolution from reconstruction and out-of-plane resolution from [3] C. G. A. Hoelen, F. F. M. de Mul, R. Pongers, and A. Dekker, “Three-
the aperture effect or cylindrical focusing of the ultrasonic trans- dimensional photoacoustic imaging of blood vessels in tissue,” Opt. Lett.,
vol. 23, pp. 648–650, 1998.
ducer. To some extent, the reconstruction can be thought of as a [4] R. O. Esenaliev, A. A. Karabutov, and A. A. Oraevsky, “Sensitivity of
numerical acoustic lens, or the acoustic lens can be thought of laser opto-acoustic imaging in detection of small deeply embedded tu-
as hardware reconstruction. Third, the confocal PAM is based mors,” IEEE J. Sel. Topics Quantum Electron., vol. 5, no. 4, pp. 981–988,
Jul./Aug. 1999.
on a scanning single-element ultrasonic transducer, whereas the [5] R. A. Kruger, D. R. Reinecke, and G. A. Kruger, “Thermoacoustic
reconstruction-based PAT can be implemented with an array of computed tomography—Technical considerations,” Med. Phys., vol. 26,
ultrasonic transducers for faster data acquisition. It should be pp. 1832–1837, 1999.
[6] L.-H. V. Wang, X. Zhao, H. Sun, and G. Ku, “Microwave-induced acoustic
mentioned that the photoacoustic imaging can be implemented imaging of biological tissues,” Rev. Sci. Instrum., vol. 70, pp. 3744–3748,
in various other forms depending on the applications in mind. 1999.
Common to both PAM and PAT, the pulsewidth of the laser [7] A. A. Karabutov, E. V. Savateeva, N. B. Podymova, and A. A. Oraevsky,
“Backward mode detection of laser-induced wide-band ultrasonic tran-
must be carefully chosen. As demonstrated with the PAM sys- sients with optoacoustic transducer,” J. Appl. Phys., vol. 87, pp. 2003–
tem, 3 mm penetration with a depth-to-resolution ratio more 2014, 2000.
than 100 can be achieved at 50 MHz. To produce photoacoustic [8] G. Ku and L.-H. V. Wang, “Scanning microwave-induced thermoacoustic
tomography: Signal, resolution, and contrast,” Med. Phys., vol. 28, pp. 4–
signals around this frequency, one should choose a laser with a 10, 2001.
pulsewidth less than 1/50 µs or 20 ns. A Q-switched laser sat- [9] K. P. Kostli, D. Frauchiger, J. J. Niederhauser, G. Paltauf, H. P. Weber,
isfies this requirement. Mode-locked lasers can produce pulses and M. Frenz, “Optoacoustic imaging using a three-dimensional recon-
struction algorithm,” IEEE J. Sel. Topics Quantum Electron., vol. 7, no. 6,
of picosecond or femtosecond durations at usually lower pulse pp. 918–923, Nov./Dec. 2001.
energy. Such short durations are unnecessary for the 3 mm pen- [10] K. P. Kostli, M. Frenz, H. P. Weber, G. Paltauf, and H. Schmidt-Kloiber,
etration system but can be useful for smaller penetration and “Optoacoustic tomography: Time-gated measurement of pressure distri-
butions and image reconstruction,” Appl. Opt., vol. 40, pp. 3800–3809,
higher resolution systems. 2001.
The spectral range of the laser must be carefully selected [11] G. Ku and L.-H. V. Wang, “Scanning microwave-induced thermoacoustic
as well. For a 3 mm penetration, the Q-band (∼560 nm) of tomography: Signal, resolution, and contrast,” Med. Phys., vol. 28, pp. 4–
10, 2001.
hemoglobin is ideal for imaging of hemoglobin because of the [12] M. H. Xu, G. Ku, and L.-H. V. Wang, “Microwave-induced thermoacoustic
moderately high optical absorption of hemoglobin in this spec- tomography using multi-sector scanning,” Med. Phys., vol. 28, pp. 1958–
tral region. The high optical absorption provides strong pho- 1963, 2001.
[13] Y. Xu and L.-H. V. Wang, “Signal processing in scanning thermoacoustic
toacoustic signals and high signal-to-background ratio while tomography in biological tissues,” Med. Phys., vol. 28, pp. 1519–1524,
the ANSI safety limit is satisfied. At the same time, the opti- 2001.
cal penetration is sufficient to reach a few millimeters depth. [14] G. Paltauf, J. A. Viator, S. A. Prahl, and S. L. Jacques, “Iterative re-
construction algorithm for optoacoustic imaging,” J. Acoust. Soc. Amer.,
For deeper (e.g., 30 mm) penetration, the near-infrared win- vol. 112, pp. 1536–1544, 2002.
dow around 700–800 nm should be used instead. However, the [15] M. H. Xu and L.-H. V. Wang, “Time-domain reconstruction for thermoa-
photoacoustic signal becomes lower and typically requires av- coustic tomography in a spherical geometry,” IEEE Trans. Med. Imag.,
vol. 21, no. 7, pp. 814–822, Jul. 2002.
eraging to enhance the SNR. [16] Y. Xu, D. Z. Feng, and L.-H. V. Wang, “Exact frequency-domain recon-
In summary, PAI has the following capabilities. 1) PAI breaks struction for thermoacoustic tomography—I: Planar geometry,” IEEE
through the current fundamental depth limit of high-resolution Trans. Med. Imag., vol. 21, no. 7, pp. 823–828, Jul. 2002.
[17] Y. Xu, M. H. Xu, and L.-H. V. Wang, “Exact frequency-domain reconstruc-
optical imaging modalities while maintaining the depth-to- tion for thermoacoustic tomography—II: Cylindrical geometry,” IEEE
resolution ratio greater than 100. 2) The depth-to-resolution Trans. Med. Imag., vol. 21, no. 7, pp. 829–833, Jul. 2002.
ratio remains approximately constant while the imaging depth [18] V. G. Andreev, A. A. Karabutov, and A. A. Oraevsky, “Detection of
ultrawide-band ultrasound pulses in optoacoustic tomography,” IEEE
and spatial resolution are scaled. On the one hand, the imag- Trans. Ultrason., Ferroelectr., Freq. Control, vol. 50, no. 10, pp. 1383–
ing depth can be scaled up to potentially centimeters, which 1390, Oct. 2003.

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.
WANG: TUTORIAL ON PHOTOACOUSTIC MICROSCOPY AND COMPUTED TOMOGRAPHY 179

[19] D. Finch, S. K. Patch, and R. Rakesh, “Determining a function from its [38] G. A. D. Briggs, Acoustic Microscopy. Oxford, U.K.: Clarendon, 1992.
mean values over a family of spheres,” Siam J. Math. Anal., vol. 35, [39] American National Standard for Safe Use of Lasers, Laser Inst. Amer.,
pp. 1213–1240, 2003. ANSI Standard Z136.1-2000, NY, 2000.
[20] A. A. Karabutov, E. V. Savateeva, and A. A. Oraevsky, “Optoacoustic [40] H. F. Zhang, K. Maslov, and L.-H. V. Wang, “In vivo imaging of sub-
tomography: New modality of laser diagnostic systems,” Laser Phys., cutaneous structures using functional photoacoustic microscopy,” Nat.
vol. 13, pp. 711–723, 2003. Protoc., vol. 2, pp. 797–804, 2007.
[21] K. P. Kostli and P. C. Beard, “Two-dimensional photoacoustic imaging [41] M. H. Xu and L.-H. V. Wang, “Universal back-projection algorithm for
by use of Fourier-transform image reconstruction and a detector with an photoacoustic computed tomography,” Phys. Rev. E, vol. 71, pp. 016706-
anisotropic response,” Appl. Opt., vol. 42, pp. 1899–1908, 2003. 1–016706-7, 2005.
[22] X. D. Wang, Y. J. Pang, G. Ku, G. Stoica, and L.-H. V. Wang, “Three- [42] X. D. Wang, Y. J. Pang, G. Ku, X. Y. Xie, G. Stoica, and L.-H. V. Wang,
dimensional laser-induced photoacoustic tomography of mouse brain with “Noninvasive laser-induced photoacoustic tomography for structural and
the skin and skull intact,” Opt. Lett., vol. 28, pp. 1739–1741, 2003. functional In vivo imaging of the brain,” Nat. Biotechnol., vol. 21, pp. 803–
[23] M. H. Xu and L.-H. V. Wang, “Analytic explanation of spatial resolu- 806, 2003.
tion related to bandwidth and detector aperture size in thermoacoustic [43] Y. W. Wang, X. Y. Xie, X. D. Wang, G. Ku, K. L. Gill, D. P. O’Neal,
or photoacoustic reconstruction,” Phys. Rev. E, vol. 67, pp. 056605-1– G. Stoica, and L.-H. V. Wang, “Photoacoustic tomography of a nanoshell
056605-15, 2003. contrast agent in the In vivo rat brain,” Nano Lett., vol. 4, pp. 1689–1692,
[24] J. A. Copland, M. Eghtedari, V. L. Popov, N. Kotov, N. Mamedova, 2004.
M. Motamedi, and A. A. Oraevsky, “Bioconjugated gold nanoparticles as [44] G. Ku and L.-H. V. Wang, “Deeply penetrating photoacoustic tomography
a molecular based contrast agent: Implications for imaging of deep tumors in biological tissues enhanced with an optical contrast agent,” Opt. Lett.,
using optoacoustic tomography,” Mol. Imag. Biol., vol. 6, pp. 341–349, vol. 30, pp. 507–509, 2005.
2004. [45] G. Ku, B. D. Fornage, X. Jin, M. H. Xu, K. K. Hunt, and L.-H. V. Wang,
[25] M. Haltmeier, O. Scherzer, P. Burgholzer, and G. Paltauf, “Thermoacous- “Thermoacoustic and photoacoustic tomography of thick biological tis-
tic computed tomography with large planar receivers,” Inverse Probl., sues toward breast imaging,” Technol. Cancer Res. Treat., vol. 4, pp. 559–
vol. 20, pp. 1663–1673, 2004. 565, 2005.
[26] G. Ku, X. D. Wang, G. Stoica, and L.-H. V. Wang, “Multiple-bandwidth
photoacoustic tomography,” Phys. Med. Biol., vol. 49, pp. 1329–1338,
2004.
[27] Y. Xu and L.-H. V. Wang, “Time reversal and its application to tomography
with diffracting sources,” Phys. Rev. Lett., vol. 92, pp. 033902-1–033902-
4, 2004.
[28] B. T. Cox and P. C. Beard, “Fast calculation of pulsed photoacoustic
fields in fluids using k-space methods,” J. Acoust. Soc. Amer., vol. 117,
pp. 3616–3627, 2005.
[29] G. Ku and L.-H. V. Wang, “Deeply penetrating photoacoustic tomography Lihong V. Wang (M’96–SM’00–F’06) received the Ph.D. degree in electrical
in biological tissues enhanced with an optical contrast agent,” Opt. Lett., engineering from Rice University, Houston, TX, in 1992.
vol. 30, pp. 507–509, 2005. He is currently the Gene K. Beare Distinguished Professor in the Depart-
[30] G. Ku, X. D. Wang, X. Y. Xie, G. Stoica, and L.-H. V. Wang, “Imaging of ment of Biomedical Engineering, Washington University, St. Louis, MO, where
tumor angiogenesis in rat brains in vivo by photoacoustic tomography,” he is also the Directer of the Optical Imaging Laboratory. He is the author
Appl. Opt., vol. 44, pp. 770–775, 2005. or coauthor of more than 135 peer-reviewed journal articles published in var-
[31] J. Zhang, M. A. Anastasio, X. C. Pan, and L.-H. V. Wang, “Weighted ious international journals. He is the author of the first textbooks in biomed-
expectation maximization reconstruction algorithms for thermoacoustic ical optics entitled Biomedical Optics: Principles and Imaging. He invented
tomography,” IEEE Trans. Med. Imag., vol. 24, no. 6, pp. 817–820, Jun. or discovered the dark-field confocal photoacoustic microscopy, photoacous-
2005. tic Doppler sensing, focused scanning microwave-induced thermoacoustic to-
[32] L.-H. V. Wang and H. Wu, Biomedical Optics: Principles and Imaging. mography, exact reconstruction algorithms for photoacoustic or thermoacoustic
Hoboken, NJ: Wiley, 2007. tomography, frequency-swept ultrasound-modulated optical tomography, sono-
[33] M. Xu and L.-H. V. Wang, “Photoacoustic imaging in biomedicine,” Rev. luminescence tomography, Mueller-matrix optical coherence tomography, and
Sci. Instrum., vol. 77, pp. 041101-1–041101-22, 2006. oblique-incidence reflectometry. His Monte Carlo model of photon transport in
[34] V. E. Gusev and A. A. Karabutov, Laser Optoacoustics. New York: scattering media has been used worldwide as a standard tool. He serves on the
American Institute of Physics, 1993. editorial boards for the Journal of Biomedical Optics and the Applied Optics.
[35] P. M. Morse and H. Feshbach, Methods of Theoretical Physics. New Dr. Wang is the Chair of the International Biomedical Optics Society. He is
York: McGraw-Hill, 1953. the recipient of the National Institutes of Health (NIH) FIRST Award, National
[36] K. Maslov, G. Stoica, and L.-H. V. Wang, “In vivo dark-field reflection- Science Foundation (NSF) CAREER Award, and the Outstanding Young Scien-
mode photoacoustic microscopy,” Opt. Lett., vol. 30, pp. 625–627, 2005. tist Award sponsored by the Johnson & Johnson Medical, Inc., and the Houston
[37] H. F. Zhang, K. Maslov, G. Stoica, and L.-H. V. Wang, “Functional photoa- Society for Engineering in Medicine and Biology. He is also a Fellow of the
coustic microscopy for high-resolution and noninvasive in vivo imaging,” American Institute for Medical and Biological Engineering, the Optical Society
Nat. Biotechnol., vol. 24, pp. 848–851, 2006. of America, and the Society of Photo-Optical Instrumentation Engineers.

Authorized licensed use limited to: WASHINGTON UNIVERSITY LIBRARIES. Downloaded on December 22, 2008 at 16:26 from IEEE Xplore. Restrictions apply.