Acousto-optic imaging in liquids: a step towards in-vivo measurements

Acousto-optic imaging in liquids : a step towards in-vivo

measurementsPedro Santosa, Mi hael Atlana, Benoît C. Forgeta, Emmanuel Bossya, A. Claude Bo araa andMi hel GrossbaLaboratoire d'Optique, ESPCI, CNRS UPRA0005,Université Pierre et Marie Curie, 10 rue Vauquelin 75231 Paris edex 05. Fran ebLaboratoire Kastler-Brossel, Département de Physique de l'E ole Normale Supérieure,UMR 8552 (ENS, CNRS, UMPC), 10 rue Lhomond 75231 Paris edex 05. Fran eABSTRACTThe aim of this paper is to show that we an perform a ousto-opti al signal a quisition of one datapoint (orvoxel of a 3D image) in a very short time (2 - 4 ms), in order to over ome the spe kle de orrelation e�e t.To demonstrate this, we have performed experiments in in dynami s attering media su h as liquids. We willshow that we an work with pulsed wave ultrasound, to redu e the sound irradiation duration in order to be ompatible with safety limits. These are signi� ant steps towards in-vivo experiments.Keywords: Medi al and biologi al imaging, Spe kle interferometry, Turbid media, Ultrasound, Holographi interferometry 1. INTRODUCTIONA ousto-opti imaging1–4 aims at obtaining images of opti al ontrast in entimeter thi k biologi al tissues withthe millimeter spatial resolution of ultrasound imaging. It is based on the modulation of the photon travel pathswithin the area of intera tion with the fo used ultrasound beam. This pro ess is often referred to as "photontagging". The dete tion of tagged photons is di� ult : the signal is weak, spatially in oherent and sin e thedete tion s hemes are usually based on a modulation of the spe kle intensity they require that this patternremain orrelated throughout the whole measurement time. In the ase of in-vivo measurements, blood �owlimits this orrelation time to roughly 1 ms.5, 6 Last but not least, the level of applied laser light and ultrasoundmust omply with safety regulations.The heterodyne parallel spe kle dete tion s heme we have developed allows to dete t tagged photons withoptimal (shot noise limited) sensitivity.7, 8 Using either a pulse or a pseudo random phase modulation of theultrasound and the referen e beam it is possible to obtain a ousto-opti images in dynami s attering media atspeeds (a few ms per voxel) ompatible with future appli ation to in-vivo imaging.2. HETERODYNE PARALLEL SPECKLE DETECTIONThe experimental setup is represented in �gure 1. The setup itself as well as the experimental methodology ofdigital holography and heterodyne dete tion applied to AO imaging have been des ribed previously7, 9 and wewill only brie�y re all them here. As seen in �gure 1, the setup is basi ally a Ma h-Zender interferometer inwhi h the the referen e beam is frequen y shifted by a ousto-opti shifters (modulators). This is done in orderto ensure that the stati interferen e pattern re orded by the CCD or CMOS amera results from the intera tionof the referen e �eld and the so- alled �tagged-photons�2 oming from the di�using sample whi h have beenFurther author information: (Send orresponden e to Benoît C. Forget)Benoît C. Forget : E-mail: forget�optique.esp i.fr, Telephone: +33 (0)1 40 79 45 90Mi hael Atlan : E-mail: atlan�optique.esp i.frFrançois Ramaz : E-mail: ramaz�optique.esp i.frA. Claude Bo ara : E-mail: bo ara�optique.esp i.frMi hel Gross : E-mail: gross�lkb.ens.fr

hal-0

0261

656,

ver

sion

1 -

14 M

ar 2

008

Author manuscript, published in "The Seventh Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics,United States (2006)"

DOI : 10.1117/12.641676

frequen y shifted by the same amount amount through the a ousto-opti e�e t. To a hieve better dete tion thereferen e is frequen y shifted by an extra amount in order to perform parallel lo k-in dete tion10 with the CCDor CMOS amera. To further improve the signal to noise, a spatial �lter (slit) is introdu ed at the output sideof the di�using sample and the referen e beam is shifted to perform o�-axis holography.11

Figure 1. Setup. Near-infrared light (wavelength λL = 780 nm, opti al frequen y fL) is provided by a CW, singleaxial mode 400 mW output power titanium:sapphire laser (Coherent, MBR 110). The light path is split into a referen e(lo al os illator, LO) and an obje t arm, by a beam splitter. In the obje t arm, a set of lenses (not shown) expands thebeam illuminating the sample. The sample is immersed in a transparent water tank, set on a 1D displa ement stage.Transmitted light passes through a spatial �lter (SF) made of a 3 × 100 mm re tangular diaphragm. Two a ousto-opti modulators (AOM1,2: Crystal Te hnologies, di�ra tion e� ien y: 50%) are pla ed in the referen e arm, to shift the LOopti al frequen y fLO = fL + fAOM1 − fAOM2, where fAOM1,2 are the frequen ies of signals sent to AOM1,2 . A 10 mmfo al length lens is pla ed in the referen e arm in order to reate a slightly o�-axis (θ ≈ 1◦ tilt angle) virtual sour e pointin the SF plane. A 10 bit, 1 Megapixel (square pixels, pixel size: d = 17.5 µm) CMOS amera (HSS4, LaVision) is setat a distan e L = 40 m from SF, fa ing the aperture, re ording the interferen e pattern of light from both arms, atframerate fC . An a ousti transdu er (Panametri s A us an A395S-SU, fo al length: 68 mm) provides the ultrasoni pressure wave at the frequen y fA = 2.25 MHz, 0.25 MPa at fo al point.3. PULSED AND PSEUDO RANDOM MODULATION TECHNIQUESTo address the problem of spatial resolution along the ultrasoni axis z (axial resolution), Wang et al. developedfrequen y-swept AO imaging with a monodete tor.12, 13 Forget et al.14 have extended this hirp te hnique toPSD by re ording a sequen e of CCD images while sweeping the frequen y of the US. These hirp te hniquesrequire that the spe kle remains orrelated throughout the duration of the frequen y hirp (several se onds), andare thus in ompatible with in vivo tomography, be ause the spe kle looses its memory in a time ( 1ms) mu hshorter that the hirp duration (∼ 10s).

hal-0

0261

656,

ver

sion

1 -

14 M

ar 2

008

sample

reference light

(LO beam)

ultrasoundlight pulse (LP)

vsü

c ü

acoustic pulse (AP)

detection

(i) (ii) (iii)

x

y

z

Figure 2. Strobos opi dete tion of the a ousti pulse AP (pulse length vsτ where vs ≃ 1500m/s is sound velo ity) bythe LO beam light pulse LP (pulse length cτ where c is light velo ity). (i) the AP propagates in the sample while the LObeam is turned o�. (ii) AP and LP overlap within the AO volume of interest. Heterodyne ampli� ation and dete tion ofultrasound-tagged photons o urs. (iii) the LO beam is turned o� until the next AP rea hes the AO volume.To get high axial resolution in vivo, Lev and Sfez5, 15 use ontinuous-wave (CW) illumination and pulsed-wave(PW) ultrasound. However, the SNR is intrinsi ally low be ause they work with a single dete tor of a low opti althroughput. We have adapted our parallel dete tion te hnique to a ommodate pulsed-wave ultrasound.9 Theprin iple is des ribed in �gure 2.The idea is to gate one of the opti al beam in order to �ash the propagating a ousti pulse at a spe i� lo ation along axis y (de�ned in �gure 2). To better illustrate the te hnique, in �gure 2 the obje t beam isgated, but for experiments it is more pra ti al to gate the referen e beam. The interferen e pattern is present(and re orded by the CCD or CMOS amera) only for a short time orresponding to the position of the a ousti pulse at this time in the media. By introdu ing a ontrollable delay between the laun h of the a ousti and thereferen e light beam pulses it is possible to s an the positions along the propagation axis.The axial resolution is learly determined by the onvolution produ t of the two time gates (a ousti al andopti al). As in an be seen in ref.,9 in reasing the pulse duration degrades the spatial resolution resolution (aswell as the ontrast whi h is intrinsi ally linked to the resolution). This has led us to propose a new s heme toobtain axial resolution based on pseudo random phase modulation.16Consider again the interferen e pattern re orded by the amera, expressed as the produ t of the obje t �eldand the onjugate referen e �eld, but this time we repla e the gated sine phase modulation indu ed by the apseudo random phase modulation frnd(t) :I = O × R∗ = I0e

jfrnd(t−z/v)e−jfrnd(t−τ) (1)We must now take into a ount that the amera will integrate, during one frame time tcam, this signal :Icam =

∫ tcam

0

I0ejfrnd(t−z/v)e−jfrnd(t−τ)dt (2)As mentioned frnd is a pseudo random fun tion, therefore so are exp(jfrnd(t − z/v)) and exp(jfrnd(t − τ)).The results of the integration in eq. 2 an be interpreted as the auto orrelation fun tion of the pseudo randomfun tion exp(jfrnd(t − z/v)).If exp(jfrnd(t − z/v)) is onsidered a purely random fun tion, then this auto orrelation is delta fun tionmeaning that the signal re orded by the amera, Icam is zero ex ept when τ = z/v. Our pseudo random signal

hal-0

0261

656,

ver

sion

1 -

14 M

ar 2

008

is generating simply by shifting the phase of the US and the opti al referen e beam modulations ba k and forthfrom 0 to π. In this ase the auto orrelation is not a delta fun tion but still a very narrow one. As seen in �gure3, even for a duration more than two orders of magnitude greater (1 ms ompared to 3 µs), the auto orrelationfun tion is narrower and therefore the spatial resolution is 3 times better with a pseudo random modulation thanwith a pulsed sine modulation.Figure 3. Numeri al al ulation of the auto orrelation of (left) a 1 ms pseudo random signal, ( enter) a 3 µs sine pulseand (right) a 5 µs sine pulseOn e again, by adjusting the delay τ between the a ousti and referen e beam signals, we an �sele t� aspe i� position along the US propagation axis.4. EXPERIMENTA 50 mm thi k aquarium is �lled with s attering liquid : a diluted solution of intralipid 10%. This solution is thenfurther diluted to vary the opti al s ttering properties of the liquid. Measurements of the spe tral broadeningof the di�use light using the methodology dis ribed in ref.6 as shown are in the range of the kHz, whi h is omparable to su h mesurements made in vivo.6, 15 The in lusion is a small latex pou h a few milimiters indiameter �lled with the same solution dyed with bla k ink.Experimental s ans along the US propagation axis using the pulsed a ousti and referen e beam te hniqueare shown in �gures 4 and 5. The signal is plotted in normalized units. As des ribed in ref7 our te hniqueallows us th measure the signal and the shot noise simultaneously. We de�ne our normalized signal as (signal -shot noise)/shot noise. Figure 4 ompares results on two dilutions (3% and 5 %) and shows in ea h asde twos ans along the US propagation axis : one over the in lusion ( ir les) and one away from the in lusion (squares).The experiment is repeated (8 times for ea h s an) to improve the signal to noise ratio. The in lusion is learlyvisible even for a small numbe of a umulations. Figure 5 show results on further diluted liquids. As expe tedthe in lusion is more easily seen in the most diluted solution.Experimental results using the pseudo random modulation te hnique are shown in �gure 6. The dilution is5% and the in lusion is learly seen in the s an on the left and the 2D image on the right. The normalized signallevel is roughly 5 times higher than the one obtained with the pulse te hnique, the resolution omparble and thesignal to noise ratio is mu h better, with a single s an.5. CONCLUSIONThese results are an important step towards in vivo measurement in whi h e�e ts of de orrelation of the spe kleby movement of, or in, the biologi al tissues is important. The omparison of the two experimental te hniquesmust be pushed further determine whi h o�ers the beast ompromise between signal level and resolution whileremaining in the safety limits for the amplitude and duration of the US pulses. The setup an and must also beadapted for the on urrent a quisition of the signal at multiple wavelengths, adding spe tros opi informationto the images. ACKNOWLEDGMENTSThis work has been supported by the Région Ile-de-Fran e, as part of the Can er�ple d'Ile de Fran e. We wouldalso like to a knowledge the support of the Institut National du Can er.

hal-0

0261

656,

ver

sion

1 -

14 M

ar 2

008

Figure 4. Experimental s ans along the US propagation axis using the pulse te hnique. Results are shown for s an overand away from the in lusion for 2 dilutions of the s attering liquid. Averaging the signal improves the signal to noiseration

hal-0

0261

656,

ver

sion

1 -

14 M

ar 2

008

0 50

0.08

0.1

0.12

transverse axis (mm)

C = 5%, 2 accumulations

0 50

0.08

0.1

0.12

transverse axis (mm)

C = 5%, 4 accumulations

0 50

0.08

0.1

0.12

transverse axis (mm)

C = 5%, 8 accumulations

inclusion

no inclusion

0 500.05

0.06

0.07

0.08

transverse axis (mm)

C = 3%

0 50

0.05

0.1

transverse axis (mm)

C = 1.5%

0 500

1

2

3

transverse axis (mm)

C = 0.75%

Figure 5. Experimental s ans along the US propagation axis using the pulse te hnique. The in lusion is learly moreeasily distinguished as the solution is more diluted, thus less s attering.

hal-0

0261

656,

ver

sion

1 -

14 M

ar 2

008

0 10 20 30 40 50 60 70 800.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65Cut along x

Nor

mal

ized

sig

nal

Ultrasound propagation axis(mm)

Random phase modulation

x axis (mm)

Ultr

asou

nd p

ropa

gatio

n ax

is

10 20 30

0

10

20

30

40

50

60

70Figure 6. Experimental results using the pseudo random modulation te hnique.REFERENCES1. L. H. Wang, S. L. Ja ques, and X. M. Zhao, �Continuous-wave ultrasoni modulation of s attered laser lightto image obje ts in turbid media,� Opt. Lett. 20, p. 629, 1995.2. W. Leutz and G. Maret, �Ultrasoni modulation of multiply s attered light,� Physi a B 204(14), 1995.3. L.-H. Wang, �Me hanisms of ultrasoni modulation of multiply s attered oherent light: an analyti model,�Phys. Rev. Lett. 87, pp. 043903�1, 2001.4. Kempe, M. Larionov, D. Zaslavsky, and A. Z. Gena k, �A ousto-opti tomography with multiple s atteredlight,� JOSA 14, pp. 1151�1158, 1997.5. A. Lev and B. Sfez, �in vivo demonstration of ultrasound-modulated light te hnique,� J. Opt. So . Am.A 20, pp. 2347�2354, de 2003.6. M. Gross, P. Goy, B. C. Forget, M. Atlan, F. Ramaz, A. C. Bo ara, and A. K. Dunn, �Heterodynedete tion of multiply s attered mono hromati light with a multipixel dete tor,� Opt. Lett. 30, pp. 1357�1359, june 2005.7. M. Gross, P. Goy, and M. Al-Koussa, �Shot-noise dete tion of ultrasound-tagged photons in ultrasound-modulated opti al imaging,� Opt. Lett. 28, pp. 2482�84, 15 De . 2003.8. M. Atlan, B. C. Forget, F. Ramaz, A. C. Bo ara, and M. Gross, �Pulsed a ousto-opti imaging with adigital holography s heme,� in Photons Plus Ultrasound: Imaging and Sensing 2005, A. A. Oraevsky andL. V. Wang, eds., Po . SPIE 5697, pp. 209�219, 2005.9. M. Atlan, B. C. Forget, F. Ramaz, A. C. Bo ara, and M. Gross, �Pulsed a ousto-opti imaging in dynami s attering media with heterodyne parallel spe kle dete tion,� Opt. Lett. 30, pp. 1360�1362, june 2005.10. S. Lévêque, A. C. Bo ara, M. Lebe , and H. Saint-Jalmes, �Ultrasoni tagging of photon paths in s atteringmedia: parallel spe kle modulation pro essing,� Opt. Lett. 24(3), p. 181, 1999.11. F. Le ler , L. Collot, and M. Gross, �Numeri al heterodyne holography with two-dimensional photodete torarrays,� Opt. Lett. 25, p. 716, 2000.12. L. V. Wang and G. Ku, �Frequen y-swept ultrasound-modulated opti al tomography of s attering media,�Opt. Lett. , 15 June 1998.13. Gang Yao, Shuliang Jiao, and Lihong V. Wang, �Frequen y-swept ultrasound-modulated opti al tomographyin biologi al tissue by use of parrallel dete tion,� Opt. Lett. 25, p. 734, 15 May 2000.14. B. C. Forget, F. Ramaz, M. Atlan, J. Selb, and A. C. Bo ara, �High ontrast FFT a ousto-opti al tomog-raphy of phantom tissues with a frequen y hirp modulation of the ultrasound,� Appl. Opt. 42, p. 1379, 1Mar. 2003.15. A. Lev and B. Sfez, �Pulsed ultrasound-modulated light tomography,� Opt. Lett. 28, pp. 1549�51, 1Sept. 2003.

hal-0

0261

656,

ver

sion

1 -

14 M

ar 2

008

16. P. Santos, M. Atlan, B. C. Forget, F. Ramaz, A. C. Bo ara, and M. Gross, �A ousto-optiq imaging witha digital holography s heme : new s heme to obtain axial resolution,� in Novel Opti al Instrumentation forBiomedi al Appli ation II, C. Depeusinge, ed., Pro . SPIE 5864, 2005.ha

l-002

6165

6, v

ersi

on 1

- 14

Mar

200

8