Effects of acoustic radiation force and shear waves for absorption and stiffness sensing in ultrasound modulated optical tomography

Ultrasound-modulated optical tomography (UOT) combines optical contrast with ultrasound spatial resolution and has great potential for soft tissue functional imaging. One current problem with this technique is the weak optical modulation signal, primarily due to strong optical scattering in diffuse media and minimal acoustically induced modulation. The acoustic radiation force (ARF) can create large particle displacements in tissue and has been shown to be able to improve optical modulation signals. However, shear wave propagation induced by the ARF can be a significant source of nonlocal optical modulation which may reduce UOT spatial resolution and contrast. In this paper, the time evolution of shear waves was examined on tissue mimicking-phantoms exposed to 5 MHz ultrasound and 532 nm optical radiation and measured with a CCD camera. It has been demonstrated that by generating an ARF with an acoustic burst and adjusting both the timing and the exposure time of the CCD measurement, optical contrast and spatial resolution can be improved by ~110% and ~40% respectively when using the ARF rather than 5 MHz ultrasound alone. Furthermore, it has been demonstrated that this technique simultaneously detects both optical and mechanical contrast in the medium and the optical and mechanical contrast can be distinguished by adjusting the CCD exposure time. ©2011 Optical Society of America OCIS codes: (170.1065) Acousto-optics; (170.7170) Ultrasound. References and links 1. M. X. Tang, D. S. Elson, R. Li, C. Dunsby, R. J. Eckersley, and P. N. T. Wells, ―Photoacoustics, thermoacoustics, and acousto-optics for biomedical imaging,‖ Proc. Inst. Mech. Eng. H 224(2 H2), 291–306 (2010). 2. L. V. Wang, ―Ultrasound-mediated biophotonic imaging: a review of acousto-optical tomography and photoacoustic tomography,‖ Dis. Markers 19(2-3), 123–138 (2003-2004). 3. V. F. Humphrey, ―Ultrasound and matter—Physical interactions,‖ Prog. Biophys. Mol. Biol. 93(1–3), 195–211 (2007). 4. M. Fatemi, and J. F. Greenleaf, ―Probing the dynamics of tissue at low frequencies with the radiation force of ultrasound,‖ Phys. Med. Biol. 45(6), 1449–1464 (2000). 5. S. G. Chen, M. Fatemi, R. Kinnick, and J. F. Greenleaf, ―Comparison of stress field forming methods for vibroacoustography,‖ IEEE Trans. Ultrason. Ferroelectr. Freq. Control 51(3), 313–321 (2004). 6. R. J. Zemp, C. Kim, and L. V. Wang, ―Ultrasound-modulated optical tomography with intense acoustic bursts,‖ Appl. Opt. 46(10), 1615–1623 (2007). 7. C. Kim, R. J. Zemp, and L. V. Wang, ―Intense acoustic bursts as a signal-enhancement mechanism in ultrasoundmodulated optical tomography,‖ Opt. Lett. 31(16), 2423–2425 (2006). 8. R. Li, L. P. Song, D. S. Elson, and M. X. Tang, ―Parallel detection of amplitude-modulated, ultrasoundmodulated optical signals,‖ Opt. Lett. 35(15), 2633–2635 (2010). #138509 $15.00 USD Received 21 Dec 2010; revised 14 Feb 2011; accepted 15 Feb 2011; published 1 Apr 2011 (C) 2011 OSA 11 April 2011 / Vol. 19, No. 8 / OPTICS EXPRESS 7299 9. K. Daoudi, A. C. Boccara, and E. Bossy, ―Detection and discrimination of optical absorption and shear stiffness at depth in tissue-mimicking phantoms by transient optoelastography,‖ Appl. Phys. Lett. 94(15), 154103 (2009). 10. H. J. van Staveren, C. J. M. Moes, J. van Marie, S. A. Prahl, and M. J. C. van Gemert, ―Light scattering in Intralipid-10% in the wavelength range of 400-1100 nm,‖ Appl. Opt. 30(31), 4507–4514 (1991). 11. T. J. Hall, M. Bilgen, M. F. Insana, and T. A. Krouskop, ―Phantom materials for elastography,‖ IEEE Trans. Ultrason. Ferroelectr. Freq. Control 44(6), 1355–1365 (1997). 12. E. Bossy, A. R. Funke, K. Daoudi, A. C. Boccara, M. Tanter, and M. Fink, ―Transient optoelastography in optically diffusive media,‖ Appl. Phys. Lett. 90(17), 174111 (2007). 13. S. J. Kirkpatrick, D. D. Duncan, and E. M. Wells-Gray, ―Detrimental effects of speckle-pixel size matching in laser speckle contrast imaging,‖ Opt. Lett. 33(24), 2886–2888 (2008).


Introduction
Ultrasound-modulated optical tomography (UOT) (also called acousto-optic imaging, or AOI) is a hybrid technique which combines optical contrast and ultrasound resolution based on the detection of photons modulated by ultrasound waves.Reviews on this subject can be found in [1,2].Two main mechanisms for ultrasound modulation of light have been identified.The first is based on variations of the optical phase in response to ultrasound induced particle displacement.The second is based on variations of the optical phase in response to ultrasonic modulation of the refractive index [2].UOT has shown some potential for early cancer detection, functional imaging and molecular imaging.However, the technique is currently limited due to the weak modulation signal strength which consequently yields a low Signalto-Noise Ratio (SNR).This is because the deformation of the scattering medium induced by the ultrasound wave is very small when the ultrasound is in the megahertz frequency range.Particle displacements induced by this deformation are normally in the range of tens of nm [3].At the same time, most of the photons arriving at the detector will not have passed through the focused ultrasound region and will be unmodulated.Consequently, the intensity of modulated light is very small compared with the background un-modulated light.One potential way to increase the SNR of UOT is to increase the ultrasound induced particle displacement, either by increasing the ultrasound amplitude or by using the acoustic radiation force (ARF).
ARF is generated by changes in the spatial energy density of an acoustic field.When ultrasound is attenuated due to scattering, reflection, or absorption, energy and momentum are transferred to the medium which results in the application of a force in the direction of wave propagation [4].The magnitude of this force is proportional to the tissue attenuation coefficient [5].The particle displacements induced by ARF are essentially uni-polar and typically have amplitudes of several micrometers occurring in about a millisecond.The motion is initiated predominantly at the ultrasound focus, and then propagates as a shear wave perpendicular to the direction of the ultrasound [6].
It has been demonstrated in UOT that ARF can be used to boost ultrasound modulated optical (UO) signals [6,7].In [6] a slight improvement in UO signal was reported, using 1 MHz ultrasound.At this frequency the UO signal is likely to be dominated by the pure ultrasound modulation because the attenuation of tissue at low ultrasound frequency is relatively low and so is the ARF.At higher ultrasound frequency, e.g. 5 MHz as used in the present study, the pure ultrasound modulation signal would be low, but the higher amplitude of the ARF due to increased attenuation at 5 MHz could potentially boost the UO signal significantly.Furthermore, because higher frequency ultrasound has a smaller focal area, improved spatial resolution may be achieved.A recent study in [8] has shown larger improvements in UO signal using ARF by amplitude modulated (AM) ultrasound.However, the optical measurements were made long after the AM ultrasound started and UO signals due to acoustic radiation force may be delocalized due to the shear wave propagation.Nonlocal UO signals can reduce UOT spatial resolution and degrade UOT contrast.
A study by Daoudi et al. [9] has implemented a technique called optoelastography.The experiment was specifically designed to detect shear wave propagation and is not sensitive to acousto-optic modulation.In their study they demonstrated the detection and discrimination of optical and mechanical properties in phantoms based on the spatial correlation between successive optical speckle patterns taken over time by a CMOS camera.Although the initial results shown in the study are encouraging, no imaging or scanning was performed and it is not clear what kind of spatial resolution and sensitivity can be achieved by this method.
There were two main objectives of this present study.One was to study the time evolution of shear wave effects on UO signals and the imaging spatial resolution in detail in laboratory phantoms and to optimise the imaging process.The other objective was to demonstrate the detection and separation of optical contrast (optical absorption) and mechanical contrast (shear stiffness) by scanning an inhomogeneous phantom with UOT and ARF.

Experimental setup
The experimental setup, similar to that described in earlier work [8], is shown in Fig. 1.A focused ultrasound transducer generated a 5 MHz ultrasound wave with a lateral focal width of 0.8 mm and a length of 10 mm at a 46 mm working distance.An RF power amplifier (ENI, Inc., 240L, Rochester, NY, USA), with a linear gain of 50 dB between 10 kHz and 12 MHz, amplified the signals driving the transducer.The ultrasound focus was positioned 20 mm inside the phantom along the Z axis.Two function generators (Agilent 33220A, South Queensferry, West Lothian, UK) were used to create the AM waveform which after power amplification drove the ultrasound transducer.Function generator 1 generated a low kHz sinusoidal burst which modulated the amplitude of the 5 MHz CW waveform generated by function generator 2.
A diode-pumped CW green laser (Excelsior 532, Newport Inc, Irvine, CA, USA, 532 nm, 100 mW) was expanded to 10 mm diameter and was incident on the glass wall of a 60 mm thick water tank containing the phantom.Light that was transmitted through the phantom, generating speckle patterns caused by the random scattering of light in the phantom.The speckle pattern was detected by a CCD camera (Qimaging Retiga EXi, Surrey, BC, Canada) with 1392×1040 pixels and an iris was used to ensure that the speckle size was comparable with that of at least two CCD pixels (pixel size 6.45 μm×6.45 μm).At full resolution, the camera ran at 15 Hz frame rate and the exposure time was adjusted from 0.25 to 8 ms for different experiments.
There were two configurations of the experimental setup.In the first, a motion stage (LG Motion Ltd, UK) moved the ultrasound transducer perpendicular to the direction of the laser beam (Fig. 1(a)) for testing the evolution of the shear wave propagation, and in the second, it moved the water tank (Fig. 1(b)), while the focal point of the ultrasound transducer was always aligned with the center of the laser beam.The second setup (Fig. 1(b)) was used to scan the phantom while keeping the light intensity within the ultrasound modulation volume relatively constant.

Phantom construction
A homogeneous phantom (93.5 mm × 43 mm × 20 mm) was made with 0.8% agar and 0.4% intralipid (volumetric concentration) with a calculated Young's modulus of approximately 17 kPa and a reduced optical scattering coefficient of about 5 cm 1 [10].The phantom was used to study the effects on measured UO signal by the ARF and the propagation of shear wave.Furthermore, since the amplitude of ARF is dependent on the mechanical properties of the phantom, two more homogeneous phantoms were made with different agar concentrations to examine the effect of phantom mechanical properties on measured UO signal.These consisted of 1% and 1.4% agar to yield calculated Young's moduli of approximately 25 kPa and 42.3 kPa respectively [11].The higher concentration of agar provided shear stiffness contrast, while keeping the compression modulus (and consequently acoustic impedance) constant.
Three inhomogeneous phantoms were also created (Fig. 2(a-c)) with 0.8% agar (mass concentration) in water and 0.4% intralipid (volumetric concentration).The phantom in Fig. 2(a) contained a cylindrical optical absorber made of India ink with 6 mm diameter, which had the same agar (stiffness) and intralipid concentration (i.e. the same optical scattering property) as the surrounding material.The phantom in Fig. 2(b) contained two cylindrical inclusions, one of which contained an optical absorber consisting of India ink with 3.5 mm diameter and had the same agar (stiffness) and intralipid concentrations (optical scattering property) as the phantom.The second inclusion was made with 1.5% agar in water and 0.4% intralipid, which had the same optical scattering coefficient but a higher shear stiffness than the bulk phantom.The phantom in Fig. 2(c) contained two optical inclusions.Both of them had similar optical properties but the one on the left had the same stiffness as the bulk phantom whilst the one on the right was stiffer (made with 1.5% agar in water) than the bulk phantom.For all experiments, the acoustic wave was applied parallel to the cylindrical inclusions (Z direction) while the optical beam was applied perpendicular (Y direction).

Calculation of speckle pattern contrast
Following previous work [8], we calculated the contrast parameter, C US , of each acquired speckle pattern image in the presence of the US beam using the equation C = σ/<I>, where σ and <I> stand for standard deviation and mean of the speckle intensity respectively.We also acquired an image of the speckle pattern in the absence of the US beam, providing us with a second contrast parameter C no-US .Finally, the change in speckle pattern contrast due to the US field alone, ∆C= C US -C no-US , was calculated.

Experiments
The optical signal was modulated using bursts of AM US.The US frequency was 5 MHz.At this frequency the ultrasound attenuation and hence the amplitude of ARF is expected to be much higher than that of 1 MHz used in [6], potentially improving the UO signal amplitude.At the same time 5 MHz is still low enough for the ultrasound to penetrate through at least a few centimeters of tissue.The AM envelope was 250 Hz as used in [8].Different CCD exposure times were chosen based on the study by Zemp et al. [6], where the theories of ARF induced shear waves were described, and the effect of CCD exposure time on UO signals was explained.It was suggested that a CCD exposure time of 0.2 ms is long enough to capture the effect of ultrasound on UO signals with reasonable signal to noise ratio, but is short enough to exclude any ARF effects as ARF causes much slower particle movements.A longer CCD exposure time was necessary to include the effect of the ARF, and 2 ms was chosen according to the time scale of the ultrasound burst used and the velocity of the shear wave propagation.These parameters were also consistent with the experiments reported in [12].Another parameter that can vary is the CCD trigger delay time.By adding such a delay time the UO signal captures different phases of the effects induced by the ARF and shear wave.If the delay time is such that the ARF induced local tissue displacement is captured but the shear wave has not propagated far, the UO signal can be improved without compromising spatial resolution.

Evolution of UO signals after the acoustic burst
The evolution of UO signals after the acoustic burst was studied in a homogeneous phantom to examine the effect of the ARF and the resulting shear wave propagation.The experimental setup was described in Fig. 1(a).UO signals were recorded at different times with respect to the beginning of a 4 ms acoustic burst.The start time was varied from the beginning of the acoustic burst to as much as 20 ms later.The recordings were made with a step size of 0.5 ms under different US and camera integration time conditions.First, the ultrasound was focused on the optical axis and measurements were made with the CCD exposure time at 0.2 ms and 2 ms (Fig. 3(a)).Next the ultrasound focus was translated within the homogeneous phantom perpendicular to the US and optical axes (Fig. 3(b~c)).Measurements were made at five points as the position of the US focus was displaced in steps of 5 mm (Point 1 displacement = 0 mm, Point 5 displacement = 20 mm) away from the optical axis with the CCD exposure time at 0.2 ms (Fig. 3(b)) and 2 ms (Fig. 3(c)).Furthermore, at one US focus position 20 mm from the optical axis the number of AM ultrasound cycles in the acoustic burst was varied from 1 to 4 cycles (Fig. 3(d)).

Effect of shear wave propagation on UOT spatial resolution
Measurements were made by scanning an inhomogeneous phantom such that the inclusion with modified optical properties passed through the US focus and optical axis (Fig. 2 (a)).

Effect of phantom mechanical properties on UOT signals
To study the effect of the phantom mechanical properties on the UOT signal, the three homogeneous phantoms with different shear stiffness and similar acoustic impedance were subjected to one AM ultrasound cycle.UO signals were measured on these three phantoms with two CCD exposure times 0.2 and 2 ms.As shorter exposure times cannot capture UO signals due to the ARF, it is expected that the difference in phantom shear stiffness can only be identified with longer exposure times.

Detection and separation of optical and mechanical contrast
An inhomogeneous phantom (see Fig. 2(b)) containing two inclusions, one being optically different and the other mechanically different, was scanned in the lateral direction with a 1D computer controlled motion stage.One AM ultrasound cycle was used and the CCD exposure time was set to 0.2 ms and 2 ms.In this experiment the time interval between the start of the acoustic burst and beginning of the CCD exposure was also varied in order to study the temporal effects of the ARF and shear wave on image resolution.
A second inhomogeneous phantom (see Fig. 2(c)), which had two inclusions with similar optical absorption properties but different mechanical properties, was also scanned in the lateral direction.One AM ultrasound cycle was used and the CCD exposure time was set to 0.2 ms and 2 ms.In this experiment the time interval between the start of the acoustic burst and beginning of the CCD exposure was also varied in order to differentiate the different mechanical properties between the two inclusions for similar optical absorption conditions.

Evolution of UO signals after the acoustic burst
Figure 3(a) shows the temporal evolution of the UO signal in order to observe the effect of shear wave propagation, and the results were measured when the focal point of the ultrasound transducer was aligned with the centre of the optical detection area.The result measured with 2 ms exposure time was much bigger than that measured by 0.2 ms, as the UO signals were enhanced by the ARF and shear wave propagation.This was confirmed when the ultrasound transducer focal point was translated away from the optical axis.For 0.2 ms CCD exposure time, shear wave propagation did not significantly affect the optical measurement, and the amplitude of the UO signal therefore decreased due to the lower degree of overlap between the US focus and the optical detection volume, while the temporal position of the peak did not change (Fig. 3(b)).In this paper, we used a 1-cycle AM ultrasound burst (250 Hz AM frequency and 5 MHz carrier frequency) lasting 4 ms and hence the amplitude of ultrasound signal reached a peak at 2 ms.Therefore the UOT signal is highest at 2 ms for the measurements with 0.2 ms CCD exposure time (Fig. 3(a) red line and Fig. 3(b)), which only captures pure ultrasound modulated signals.For 2 ms CCD exposure time, the position of the peak also depends on the effect of ARF.In this case the results in Fig. 3(a) (blue line) suggest that the combined effect of ultrasound modulation and ARF is greatest between 2 ms and 4 ms after the start of AM ultrasound burst.For 2 ms exposure time (Fig. 3(c)) both the amplitude and the peak of the detected UO signal changed, consistent with the time taken for the shear wave to propagate into the optical detection volume.This experiment not only confirms that longer exposure time can detect slow tissue movements due to ARF and shear wave propagation but also enables the estimation of the shear wave propagation speed in the medium, through knowledge of the time when the measured UO signal peak arrives and the distance between the ultrasound focus and the centre of the optical detection area.In this example, the time difference of the UO peak between positions 4 and 5 (separated by 5 mm) was ~2 ms, giving an approximate shear wave velocity of 2.5 mm/ms.This is close to the result (~2.8 mm/ms) measured independently using a SSI scanner (Aixplorer, SuperSonicImagine).It should be noted that the time intervals between the neighboring peaks in Fig. 3(c) are not constant.There are two possible reasons: firstly the sampling density of CCD delay time is very coarse (1 ms) on the x-axis and hence the estimation of the position of the peaks is not accurate.Secondly, besides ARF the peaks are also affected by the pure ultrasound modulation.For the first two positions, the ultrasound focus area is still close enough to the center of the laser beam that the pure ultrasound (5 MHz) modulation is also affecting the signal (as confirmed in Fig. 3(b)) and biases the location of the peaks.
Figure 3(d) shows the UO signal using different numbers of AM cycles in the burst when the ultrasound focus was 20 mm away from the centre of the optical detection area.The results suggest that the UO signal of multiple AM cycles is the sum of the time shifted UO signal of each individual AM cycle.In this case multiple shear waves from the multiple AM cycles pass through the optical detection volume (approximately a cone of 10-30 mm in diameter) and the effect on the UO signal from each of these shear waves is combined.These results again provide evidence that the longer CCD exposure time (2 ms) contains information on the slow tissue movements caused by ARF and shear wave propagation.Furthermore, the results also suggest a possible saturation in UO signal.

Effect of shear wave propagation on UOT spatial resolution
Figure 4(a) shows the effect of the ARF and shear wave propagation for two regions of an inhomogeneous phantom (absorbing and non-absorbing, Fig. 2(a)) for two CCD exposure times.For this experiment, a 6 mm diameter absorbing inclusion was used.The red dotted and solid lines were measured in a homogeneous non-absorbing area using 0.2 ms and 2 ms CCD exposure times and again shows that the signal strength for the 2 ms CCD exposure is much higher due to the sensitivity to ARF induced particle displacement and shear wave propagation.The blue lines were measured when the ultrasound focal area was within the 6 mm diameter absorbing cylinder.For the shorter CCD exposure time (dotted lines), only the amplitude is different between the absorbing and non-absorbing areas.With longer CCD exposure times (solid lines), significantly lower signals can initially be observed (0~5 ms) for the absorbing area, but the difference becomes minimal after 5 ms.This is because the UO signals caused by the ARF and the shear wave were partly absorbed between 0 and 5 ms.After 5 ms the shear wave had completely propagated outside of the absorption area, and the signal therefore becomes comparable to that measured in homogeneous area (red solid line).It should be noted that the shear wave itself has a certain length.Therefore for the bulk of the shear wave to propagate out of the absorption area, the shear wave front needs to travel further than the size of the absorption area.Figure 4(b) and (c) show the spatial profiles of UO signals when the 3.5 mm optical absorber within an inhomogeneous phantom (Fig. 2(a)) was scanned with various CCD delay times and exposure times.Measurements were made with a CCD exposure time of 0.2 ms and a 2 ms delay time following the acoustic burst (line L1 in Fig. 4 (b) and (c)).It can be seen that the UO signal was reduced and hence presents a negative peak when the absorbing inclusion was passed through the US beam, as expected.Additional measurements were made at different delay times following the acoustic burst with a CCD exposure time of 2 ms (lines L2-5 in Fig. 4(b) and (c)).These latter measurements were made to identify the optimum measurement time and to evaluate any measurement degradation from shear wave effects.The results show that the spatial profile of the UO signal is strongly dependent on the delay time at which the measurements were made.With longer CCD exposure time (2 ms), measurements with short delay times are expected to mainly detect the effects from the ARF at the acoustic focus as the shear wave has not propagated significantly, while at longer delay times the signal would largely pick up the effects of shear wave propagation.The full width half maximum (FWHM) of the peak was 4.5 mm with the ARF detected with long CCD exposure time (L3 in Fig. 4(b) and (c)) and the FWHM was 7.5 mm for the traditional non-ARF signals measured with the short CCD exposure time (L1 in Fig. 4(b) and (c)).This represents an improvement of at least 40% in spatial resolution, given that the actual diameter of the inclusion was 3.5 mm.The differences in ∆C between the peak and the background for L3 and L1 in Fig. 4 (b) and (c) were ~0.11 and ~0.053, respectively.This represents a 110% improvement in image contrast for the ARF modulated signals (L3) than the pure ultrasound modulated signals (L1).The signal amplitude and spatial resolution peaked when the measurements were made 1 ms after the acoustic burst started.At longer delays the peak in the UO signal profile indicates that the absorber starts to become blurred since the shear wave had propagated out of the absorbing inclusion at these times.

Effect of phantom mechanical properties on UOT signals
By using the ARF in UOT, we have also shown that mechanical information about the phantom may also be obtained.Figure 5(a) shows the results measured as a function of CCD delay time on three homogeneous phantoms with different stiffness but similar acoustic impedance using 0.2 ms CCD exposure time.No difference was observed between the phantoms.However, when using 2 ms CCD exposure times the UO signal amplitude for the softest phantom (Young's modulus ~17 kPa) is much larger than the other two stiffer phantoms (see Fig. 5(b)).This is likely because the short CCD exposure time only captures the pure ultrasound modulation which is insensitive to the stiffness of the medium, while with the longer CCD exposure time the effects of the ARF and shear wave propagation, which depend on the mechanical properties of the phantom, can be observed.

Detection and separation of optical and mechanical contrast
To demonstrate the spatial resolution of this interaction, Fig. 6(b) shows the UO signal profile across the inhomogeneous phantom, which includes two inclusions (Fig. 2(b)): an optical absorber (on the left with same stiffness as background) and a stiffer object (on the right with the same optical absorption as background).Firstly measurements were made with a CCD exposure time of 0.2 ms at a delay time following the acoustic burst of 2 ms (red line).Additional measurements were made with a CCD exposure time of 2 ms at a delay time of 1 ms following the acoustic burst.The results show two troughs when the 2 ms (longer) CCD exposure time was used but only one trough when the 0.2 ms CCD exposure time was used.This demonstrates that measurements with longer CCD exposure time are able to capture both optical and mechanical contrast within the object with this system.To distinguish optical contrast from mechanical contrast, the phantom with two optical inclusions which have similar optical absorption coefficients but different stiffnesses, as displayed in Fig. 2(c), was scanned.The results (Fig. 7(b)) show two troughs with similar amplitude when the shorter CCD exposure time (0.2 ms) was used.However, the amplitude of contrast difference changed with the delay time for the longer CCD exposure time (2 ms).With a 1 ms delay time, the two dips still have similar magnitude because the acoustic radiation force induced optical modulation at this time was still mostly absorbed within the optical absorbers.With a 3 ms delay time, the trough of the stiffer inclusion is deeper than that of the other inclusion.This is likely because 3 ms after the ultrasound burst was launched, the shear wave front had partly propagated outside the optical absorption area.A lower shear wave amplitude was generated in the stiffer inclusion on the right so the trough was lower.With a 4 ms delay time, the dip on the left disappeared, but the dip on the right remained, although with a small peak in the middle.This is likely because between 4~6 ms, the shear wave fronts totally propagate out of the optical absorbers.
In this work all the phantoms were left for at least 12 hours before experiments and the size of the optical absorbers was always measured just after the experiments.This was due to the fact that optical dye was found to diffuse to the surrounding areas once the phantoms were made.Therefore the actual spatial profiles of optical properties for the optical absorbers are likely to be smooth and Gaussian shaped.This explains the shape of the spatial profiles for the optical absorbers in Fig. 4(b and c), Fig. 6 and Fig.The one on the left has the same stiffness with the background.The one on the right is stiffer than the bulk of the phantom.T in the legend stands for the measurement delay time following an acoustic burst.

Further discussion
The results demonstrate that UO signals can be enhanced by as much as 110% by the ARF which induces large tissue displacement but also generates shear waves.By controlling the timing of the optical measurements, the shear wave effect can be minimized to achieve both improved signal amplitude and good image resolution.By controlling the exposure time of the optical measurements, the optical and mechanical properties of the phantom can be distinguished.
The results also show that the spatial resolution of UOT can be greatly improved by the ARF, by at least 40% in our study.This is an interesting finding and is likely due to the fact that the ARF is created during a nonlinear process and the spatial beam profile of this force is much smaller than the ultrasound beam itself in both axial and lateral direction.Studies by Fatami [4] have demonstrated a resolution of 0.7 mm when using this radiation force for acoustic imaging.On the other hand, the beam profile of our 5 MHz ultrasound transducer is quite large and would certainly cover part of the cylindrical optical inclusion even when the acoustic focus is not within the inclusion.
Zemp et al. [6] have made UO measurements with an intense ultrasound burst at a much lower frequency (1 MHz).However, they did not compare the image resolution when measurements were made at different time points, nor did they investigate the ability of imaging both optical and mechanical contrast.Compared with their results on UO signals, our signals are higher when using a longer CCD exposure time (2 ms) and lower when using a shorter CCD exposure time (0.2 ms).The possible reasons for this are as follows: first, they used a 1 MHz ultrasound transducer, which is expected to generate a smaller ARF than the 5 MHz transducer used in this study.Second, we used a phantom with an optical reduced scattering coefficient of 5 cm 1 , which for the shorter CCD exposure time (0.2 ms) will give a lower modulated optical signal than the phantom used by Zemp with a reduced scattering coefficient of 9.2 cm 1 .Pure ultrasound modulation seems to be much less sensitive to the stiffness of the phantoms than ARF modulation.Thirdly, we used 2 pixels per speckle rather than 1 pixel per speckle which increases the contrast of the speckle pattern [13].It should be noted that when we used similar parameters as Zemp's study, very similar results were obtained.
The phantoms used in this study were made with only agar powder and intralipid and no additional acoustic scatterers were added, hence the acoustic attenuation of the phantoms was relatively low.Given that the amplitude of the ARF is determined by how much the acoustic energy is attenuated, adding additional scatterers into the phantom is likely to increase the amplitude of ARF and further improve the UOT signals.

Conclusions and discussion
We have demonstrated that our UOT system can observe shear wave propagation in phantoms.By adjusting the CCD exposure time, both the optical and mechanical properties of the imaged phantoms can be detected and distinguished.By using proper timing for CCD data acquisition, acoustic radiation force induced optical modulation can enhance UO signals by 110% and at the same time improve imaging spatial resolution by 40% in this study.The shear wave effect, which was clearly observed in some of the experiments, can be minimized.This provides a potentially very exciting tool for noninvasively imaging both optical and mechanical properties at centimeter depth with a resolution potentially at the scale of a millimeter or even less.

Fig. 2 .
Fig. 2. Three inhomogeneous phantoms.(a) The black cylinder represents the cylindrical inclusion containing India ink.(b) The black cylinder represents the cylindrical inclusion containing India ink, whilst the grey cylinder represents the inclusion with modified shear stiffness.(c) The two cylindrical inclusions contain the same amount of India ink.The one on the left has the same mechanical property as the bulk phantom, while the one on the right is stiffer than the bulk phantom.

Fig. 3 .
Fig. 3. (a) Contrast difference versus CCD trigger delay time for a 250 Hz-AM-US burst with 0.2 ms and 2 ms CCD exposure time.(b-c) Contrast difference for different positions of ultrasound focal point with 0.2 ms (b) and 2 ms (c) CCD exposure time.(d) Contrast difference for AM bursts with different number of cycles when ultrasound focal point was 20 mm away from the centre of the optical detection area.

Fig. 4 .
Fig. 4. (a) Contrast difference versus CCD trigger delay time for a 250 Hz-AM US burst with 0.2 ms and 2 ms CCD exposure time in a homogeneous area (blue line) and an absorbing area (red line) (b) Comparison of the 1-D spatial profile of an optical absorber obtained with a 0.2 ms CCD exposure time measured 2 ms after launching the acoustic burst (L1) with 1-D profiles with a 2 ms CCD exposure time measured at varying delay times after launching the acoustic burst (L2-L5, results of CCD delay time greater than 3 ms not shown in this graph).(c) a plot of the same 1-D profiles (L1-l5) as those shown in (b), plus L6-L10 which were obtained with CCD delay times between 4 and 8 ms.

Fig. 5 .
Fig. 5. Contrast difference versus CCD trigger delay time for a 250 Hz-AM US burst with 0.2 ms (a) and 2 ms (b) CCD exposure time in different phantoms

Fig. 6 .Fig. 7 .
Fig. 6.(a) The phantom picture taken right after the experiments.The diameter of the optical inclusion (the one on the left) is ~3.5 mm.The diameter of the stiffer inclusion (the one with a red-dash circle on the right) is ~8 mm.(b) 1-D profiles of an inhomogeneous phantom measured with various CCD exposure times and CCD delay times.There are two inclusions inside the phantom.The one on the left has added India ink for optical absorption but has the same stiffness with the background.The one on the right is stiffer than the bulk of the phantom but has a similar optical absorption as the background.