Interferometric fiber optic sensors for biomedical applications of optoacoustic imaging
Upcoming SlideShare
Loading in...5
×
 

Interferometric fiber optic sensors for biomedical applications of optoacoustic imaging

on

  • 1,685 views

We present a non-metallic interferometric silica optical fiber ultrasonic wideband sensor for optoacoustic imaging applications. The ultrasonic sensitivity of this sensor has been characterized over ...

We present a non-metallic interferometric silica optical fiber ultrasonic wideband sensor for optoacoustic imaging applications. The ultrasonic sensitivity of this sensor has been characterized over the frequency range from 1 to 10 MHz. A comparative analysis has been carried out between this sensor and an array of piezoelectric transducers using optoacoustic signals generated from an optical absorbent embedded in a tissue mimicking phantom. Also, a two dimensional reconstructed image of the phantom using the fiber interferometric sensor is presented and compared to the image obtained using the Laser Optoacoustic Imaging System, LOIS-64B. The feasibility of our fiber optic based sensor for wideband ultrasonic detection is demonstrated.

Statistics

Views

Total Views
1,685
Views on SlideShare
1,674
Embed Views
11

Actions

Likes
0
Downloads
38
Comments
0

1 Embed 11

http://www.linkedin.com 11

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

Interferometric fiber optic sensors for biomedical applications of optoacoustic imaging Interferometric fiber optic sensors for biomedical applications of optoacoustic imaging Document Transcript

  • J. Biophotonics 4, No. 3, 184–192 (2011) / DOI 10.1002/jbio.201000096 Journal of BIOPHOTONICSFULL ARTICLEInterferometric fiber optic sensors for biomedicalapplications of optoacoustic imagingHoracio Lamela *; 1 , Daniel Gallego 1, Rebeca Gutierrez 1 , and Alexander Oraevsky 21 Universidad Carlos III de Madrid, 28911 Leganes, Madrid, Spain2 TomoWave Laboratories, Inc. Houston, Texas 77057, USAReceived 8 September 2010, revised 16 December 2010, accepted 22 December 2010Published online 18 January 2011Key words: fiber optic sensors, ultrasound, optoacoustic imaging, interferometry We present a non-metallic interferometric silica optical fiber ultrasonic wideband sensor for optoacoustic imag- ing applications. The ultrasonic sensitivity of this sensor has been characterized over the frequency range from 1 (a) to 10 MHz. A comparative analysis has been carried out between this sensor and an array of piezoelectric trans- ducers using optoacoustic signals generated from an op- (b) (c) tical absorbent embedded in a tissue mimicking phan- tom. Also, a two dimensional reconstructed image of the (a) Diagram showing location of the embedded object in phantom using the fiber interferometric sensor is pre- the PVCP phantom. (b) Optoacoustic image obtained sented and compared to the image obtained using the from LOIS utilizing an array of 64 PVDF transducers. Laser Optoacoustic Imaging System, LOIS-64B. The fea- (c) Optoacoustic image reconstructed from fiber optic sibility of our fiber optic based sensor for wideband ul- sensor signals. trasonic detection is demonstrated.1. Introduction visible or near-infrared wavelengths range. The opti- cal absorption of the laser pulse by the differentOptoacoustic tomography [1, 2] is a promising non- tissue structures produces an instant distribution ofinvasive non-ionizing imaging technique that can be heat sources and, as a result of the thermoelasticused to visualize biological soft tissue. It combines effect, broadband ultrasonic pulses are generated.the advantages resulting from high optical absorp- These ultrasonic waves propagate to the surface oftion contrast in biological tissues with the excep- the tissue where they are detected using ultrasonictional spatial resolution of ultrasound imaging tech- transducers. The typical frequency spectrum of theniques. In recent years this imaging method has optoacoustic signals generated at a depth of severalbeen used in several biomedical applications such as centimeters that reach the surface after suffer ultra-visualization of blood vessels and the measurement sonic attenuation of the soft tissue covers a wideof blood oxygenation [3], detection of tumors in range from $100 kHz to $10 MHz. The desirablebreast tissue [4–6], small animal imaging [7] and characteristics for ultrasonic transducers used in thefunctional imaging [8]. The optoacoustic technique is detection of optoacoustic signals and for high spatialbased on the irradiation of the tissue surface with resolution image reconstruction are high sensitivitylaser pulses of a few nanoseconds operating in the and a broadband nature.* Corresponding author: e-mail: horacio@ing.uc3m.es, Phone: +34 91 624 9476, Fax: +34 91 624 9430# 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
  • FULL ARTICLEJ. Biophotonics 4, No. 3 (2011) 185 Traditionally, detection technology used in con- an array of piezoelectric transducers made fromventional ultrasonic imaging is based on piezoelec- PVDF [14] in Section 5. Finally, in Section 6, conclu-tric transducers, which are highly sensitive but have sions and future work are presented.a narrow bandwidth resulting from their resonantnature. However the optoacoustic signals generatedrequire broadband detectors to image the differentsizes of absorption regions within the body. The de- 2. Principle of fiber optic sensor operationtectors based on thin piezoelectric polymer films,like polyvinylidene fluoride (PVDF), can be made The operation of an optical fiber interferometric sen-sensitive over an ultrawide-band using appropriate sor is based on the transduction of the ultrasonicbacking and front matching materials. However, wave by a phase shift of the light traveling throughtheir sensitivity decreases as their size is reduced. the fiber. By means of a Mach-Zehnder interferom-This presents a problem when detecting high ultra- eter the phase shift of light is transformed into asonic frequencies where both the small thickness of variation in the optical intensity. Finally, the opticalthe detector, required for high axial resolution, and signal is converted to a voltage signal using a photo-the small width of the transducer element, required diode and a transimpedance amplifier. Consideringfor high lateral resolution and improved image fi- all these factors, the total sensitivity of the detector,delity, reduce its sensitivity. Another drawback to i.e. the relation between the output voltage and thepiezoelectric sensors, related to their electrical nat- acoustic pressure, may be expressed asure, is that they are not immune to electromagneticinterference. DV DV DI Df ¼ ð1Þ Over the past 30 years, optical detection of ultra- DP DI Df DPsound has been studied as an alternative to piezo- where P is the acoustic pressure amplitude, f is theelectric technology. Clear distinctions can be made phase of the light, I is light intensity at the interfe-between two types of ultrasound optical sensors, rometer output and V is the voltage at the output ofthose that monitor pressure induced displacements the transimpedance amplifier.of a membrane or resonant optical cavity; and others In general, an acoustic wave that is incident onthat are based on a pressure induced index refrac- an optical fiber induces strain, thus producing a re-tion variation in or around the sensor material. In fractive index variation that modifies the phase ofthe first group, the following sensors can be found: light, given by Dj ¼ kðn Dl þ l DnÞ, here k is the la-etalons [9], fiber Bragg gratings [10], and dielectric ser wave number, l is the length of the sensing seg-multilayer interference filters [11]. Intrinsic fiber ment andnis the effective refractive index of the op-optic interferometric sensors [12] form part of the tical fiber. At ultrasonic frequencies, the inducedsecond group. All these optical sensors, contrary to strain, localized on a fiber segment of length l, canpiezoelectric transducers, are not affected by exter- be considered axially constrained, thus the phasenal electromagnetic disturbances or other artifacts shift induced is mainly governed by the strain-opticlike electrical noise and thermal signals produced effect taking into account in l Dn factor. Within theby the direct laser pulse illumination. In particular, regime d ( la, where d is the optical fiber diameterthe fabrication of intrinsic fiber optic interferometric and la is the acoustic wavelength, it can be assumedsensors is straightforward and involves the use of that the acoustic strain has radial symmetry, thuslow cost materials. Moreover the sensitivity of these there is no birefringence induced. The effective re-sensors can be improved by appropriate folding or fractive index perturbation is linear and proportionalcoiling of the fiber [13] which increases the surface to the pressure wave magnitude, Pð~ tÞ: r;area that interacts with the acoustic field. In this work the principle of operation of the Dnð~ tÞ ¼ aPð~ tÞj~¼~ l; r; r l ð2Þultrasonic optical fiber interferometric sensor is in-troduced in Section 2. We present, in Section 3, the Where a is a proportional constant that depends oncharacterization of the acoustic sensitivity of the op- the acoustic coupling conditions, Young’s modulus,tical fiber used to develop the sensor for the fre- the Poisson ratio and the strain-optic tensor of thequency range from 100 KHz to 10 MHz. Also, in this optical fiber material; and ~ is the position in the lsection, we compare the optoacoustic signals de- fiber. The total phase induced is the result of the in-tected using this fiber optic interferometric sensor to tegration of refractive index perturbations along thethose obtained using a PVDF transducer. In Sec- optical fiber.tion 4, a summary of the back projection algorithm Ð DfðtÞ ¼ k Dnðl; tÞ dl ð3Þand wavelet processing used to reconstruct opto-acoustic images are both introduced. We compare an The phase shift can be increased using a longer fiberimage obtained from a phantom mimicking soft tis- segment exposed to the acoustic wave. However, thesues by this system with other image obtained using use of large fiber arrangements at frequencies ofwww.biophotonics-journal.org # 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
  • Journal ofBIOPHOTONICS186 H. Lamela et al.: Interferometric fiber optic sensors for biomedical applications of optoacoustic imaging I tain constant sensitivity and linearity. This can be achieved by using a phase stabilization device [15] that fixes the operating point to the region of the maximum slope in the transfer function (i.e. quadra- ture point). In this position of quadrature, if the ΔI acoustically induced phase shift is small, Df ( 1, the interferometric sensitivity is
  • DI
  • pffiffiffiffiffiffiffiffiffi
  • ¼ 2 Ir Im
  • Df
  • In the case of Ir ¼ Im, the interferometric sensitivity φ is simplified to jDI=Dfj ¼ 2Im. -2π - 3π -π - π 0 π π 3π 2π 2 2 2 2 Δφ φ0Figure 1 Mach-Zenhder interferometric transfer function 3. Optical fiber sensor designand response of the interferometer to a perturbationaround the quadrature point. 3.1 Sensitivity characterization1 MHz cause a temporal average of the ultrasonic of the fiber optic sensorsignal, this is because the acoustic wavelength isclose to 1.5 mm. Thus, regarding the sensitivity at The sensitivity per unit length is a key parameter forultrasonic frequencies there is a trade-off between designing the optical fiber arrangement. To evaluatefiber length and temporal average. it we measured the phase shift induced in the optical The relation between the intensity of the light at fiber using ultrasonic waves generated from a cali-the output of the Mach-Zehnder interferometer and brated piezoceramic (PZT) emitter. The characte-the phase difference between both arms, which is re- rization setup is presented in Figure 2a. A Mach-presented in Figure 1, is given by Zehnder interferometer was used to obtain the pffiffiffiffiffiffiffiffiffi phase induced on a segment of the optical fiber un-I ¼ Ir þ Im þ 2 Ir Im cos Df ð4Þ der test. Water was used to ensure consistent and re-where Ir and Im are the intensities of the reference peatability acoustic coupling between the emitterand measurement arms respectively. It has been as- and the optical fiber segment. In the reference arm asumed that the coherence length is much longer than phase modulator, formed from a fiber coil wrappedthe path difference between both arms and that the around a cylindrical PZT, has been employed tolight at their output has the same polarization state. obtain the interferometric visibility at the time ofTo use the interferometer arrangement as a sensor measurement. The output of the interferometer isit is necessary to establish an operating point to ob- measured using an avalanche photodiode module Digital Oscilloscope He-Ne Laser 1.5 Squarewave 120 Normalized phase shift Pulser/Receiver Phase induced (mrad) 1.25 TRG out TRG in 100 PZT Emitter 1 80 Polarizarion 0.75 Controller APD 60 0.5 40 0.25 Polarizarion 20 PZT phase Controller 0 modulator Wave 20 40 60 80 100 120 0 1 2 3 4 5 6 7 8 9 10 11 12 Generator TRG in Peak Pressure Amplitude (kPa) Frequency (MHz) (a) (b) (c)Figure 2 (a) Experimental setup for sensitivity characterization of the optical fiber sensor. The light from a He-Ne laser(632.8 nm) is introduced in the fiber optic Mach-Zehnder interferometer. The polarization controllers are used to guaran-tee equal state of polarization at the output of both arms maximizing the visibility of the interference. The calibrated PZTemitter and the optical fiber are inside a tank filled with water to ensure a consistent acoustic coupling. The PZT phasemodulator is used to register the visibility of the interference at the same time of the ultrasonic signal measurement in orderto precisely recover the optical phase induced. (b) Relation between 1 MHz pulse amplitude pressure and phase shift inducedin the single mode SOF. (c) Normalized frequency response of fiber optic sensor to normally incident ultrasonic wave.# 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.biophotonics-journal.org
  • FULL ARTICLEJ. Biophotonics 4, No. 3 (2011) 187(Hamamatsu C5331) and monitored using a digital active sensor length is 100 mm and only correspondsoscilloscope. The bandwidth of this optical detector to the contact area between the phantom and theis 100 MHz, with a lower cut off frequency of 4 kHz. coil, given that ultrasonic waves cannot reach theThe signal captured from the oscilloscope is sent to a rest of the fiber.computer for post processing to obtain the phase. We have compared the optoacoustic signal detec-Immersion transducers from Panametrics with cen- tion using the the optical fiber sensor to those fromtral resonant frequencies of 1 MHz and 10 MHz a wideband ultrasonic PVDF transducer. The optoa-were used as ultrasonic emitters. These transducers coustic signal was produced irradiating a large cy-were excited using a square pulse generator (Pana- lindrical phantom [16] (dia. ¼ 14.5 cm, h. ¼ 8.5 cm)metrics 5077PR). The amplitude of the square pulse made of poly(vinyl-chloride) plastisol (PVCP) withwas varied between 100 V and 400 V in steps of optical pulses generated from a Nd-YAG laser100 V. The oscilloscope and the wave generator (Quantel ULTRA). The phantom mimics optical andwhich feed the phase modulator were synchronized acoustic properties of biological soft tissues has anwith the square pulse generator. optical absorption coefficient ma ¼ 0.12 cmÀ1, an ef- In our experiments we have used a single-mode fective scattering coefficient of m0s ¼ 5.4 cmÀ1 and ansilica optical fiber (SOF) (SCSM-633-HP, Stocker- anisotropy factor of g ¼ 0.8 at 1064 nm. EmbeddedYale Inc., Salem, NH, USA). We have measured at approximately 20 mm below the surface of theultrasonic signals which have been produced by the phantom is a bar shaped optically absorbing objectPZT emitters using a single-mode SOF. Figure 2b re- of dimensions 220.7 cm3 with ma ¼ 0.58 cmÀ1. Thepresents the phase induced in the single-mode SOF Nd-YAG laser emits pulses at the wavelength offrom the 1 MHz ultrasonic PZT emitter for voltages 1064 nm with 75 mJ of energy, temporal duration ofranging from 100 V to 400 V. These phase signals 6 ns and a repetition rate of 10 Hz.have been obtained by demodulation of the inter- Figure 3a shows the fiber optic Mach-Zehnder in-ferometric signals and have been digitally filtered terferometer sensor experimental setup. The lightusing a high pass filter with a cutoff frequency of from a He-Ne laser, which is linearly polarized and0.3 MHz to remove the reference arm signal. The emits at a wavelength of 632.8 nm, is divided using ameasured frequency response of the fiber sensor to 50/50 optical beam splitter and coupled into two si-a perpendicularly incident ultrasound is shown in lica optical fibers (i.e. the reference and measure-Figure 2c, having a bandwidth that exceeds 10 MHz. ment arm of the fiber optic interferometer). The op-It has been deconvolved from the response of the tical fiber sensor is in the measurement arm of thesystem to the ultrasonic waves generated using the interferometer. This sensor is placed on the surface10 MHz transducer. This transducer was previously of the phantom and has a contact area of 0.5 cm2calibrated with a 100 MHz wideband ultrasonic cali- Â 0.5 cm2. In the reference arm the phase modulatorbrated sensor (WAT-13, Fairway Medical Technolo- uses a feedback loop to stabilize the interferometergies, Houston, TX, USA). at the quadrature point thus compensating tempe- At the working frequency of 1 MHz the acoustic rature drifts and low frequency vibrations. The in-sensitivity of the SOF is measured to be 0.95 Æ terference visibility is optimized by matching the0.03 mrad/kPa and at 10 MHz is 0.87 Æ 0.11 mrad/ polarization at the output of each fiber arm usingkPa. The acoustic beam diameter was close to 3.4 mm polarization controllers. The interferential opticalat the measurement distance, which is shorter than signal is measured using the same APD module andthe length of the segment under test, thus the sensi- digital oscilloscope as before.tivity at 1 MHz per unit length is 0.28 mrad/kPa/mm. The phantom lies over an array of 64 PVDFThe sensor with an active fiber length of 100 mm wideband detectors arranged in a concave arc (Fig-and an interferometric system with a resolution of ure 3b and c, this forms part of an optoacoustic5 mrad has a noise equivalent pressure (NEP) of probe supplied as a component of the Laser Opto-0.18 kPa at 1 MHz. acoustic Imaging System (LOIS, Fairway Medical Technologies, Houston, TX, USA) [14]. Each sensor of the array system has a sensitivity of 1.66 mV/Pa at a frequency of 1.5 MHz after two stages of 20 dB amplification. The fiber optic sensor is placed directly3.2 Detection of optoacoustic generated opposite the PVDF transducer which is located in the signals center of the arc shaped array. The absorber inside the phantom is located approximately equidistantThe ultrasonic transducer designed for optoacoustic between both electronic and optical sensors.generation signals is composed of a single layer coil Figure 4a depicts a typical optoacoustic signal de-with 20 loops of single mode optical fiber at 633 nm. tected from an absorbing object with a rectangularThe sensor is placed on the surface of the medium bar shape and positioned at an angle to a sensorbeing imaged with a small acoustic contact area. The channel of the PVDF array. The voltage signal ob-www.biophotonics-journal.org # 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
  • Journal of BIOPHOTONICS188 H. Lamela et al.: Interferometric fiber optic sensors for biomedical applications of optoacoustic imagingHe-Ne Laser Intrinsic optical fiber sensor Sensor active area Phantom Laser illumination area PZT phase modulator Stabilization Control Phantom Polarization APD Controller TRG Out Targets Nd-Yag Laser TRG In LOIS Channel 30 Digital (a) Oscilloscope (b) (c)Figure 3 (a) Optoacoustic experimental setup for comparison of fiber opic and piezoelectric transducers. (b) Schematicdiagram showing cylindrical phantom placed in a hemi-cylindrical cup of an arcshaped array of PVDF transducers and afiber optic sensor located on the top. (c) Photograph of a cylindrical PVCP phantom placed at the top of the cylindricalphantom with a fiber optic sensor attached at the top. The fiber optic sensor has a contact area of 5 Â 5 mm. The piezo-electric elements of the PVDF array have dimensions of 0.1 Â 2 Â 20 mm.LOIS Channel 30 1 0.1 Amplitud e (kPa) Amplitud e (kPa) 0.5 0.05 Pressure Pressure 0 0 -0.5 -0.05 -1 -0.1 0 10 20 30 40 50 60 70 80 90 10 15 20 25 30 35 40 Time (µs) (a) Time (µs) (c) 1 0.2 Amplit ude (kPa)SOF Sensor Amplit ude (kPa) 0.5 0.1 Pressu re Pressu re 0 0 -0.5 -0.1 -1 -0.2 0 10 20 30 40 50 60 70 80 90 10 15 20 25 30 35 40 Time (µs) Time (µs) (b) (d)Figure 4 (a) Optoacoustic signal received by channel 30 of PZT array expressed in pressure units. (b) Signal detected byan extrinsic fiber optic sensor at the same time but on the opposite side of the phantom. (c) and (d) magnification of thenoise that correspond to the marked areas in the Figures (a) and (b) respectively.tained was deconvolved to pressure using the known noise levels for the PVDF transducer is about 50 Paimpulse response of the PVDF detector [14]. Fig- and in the case of the optical fiber sensor is abouture 4b shows the interferometric signal of the intrin- 100 Pa. In the former case the electronic bandwidthsic fiber optic sensor. Comparing Figures (a) and (b), is limited to 2.5 MHz, however in the last case it isa similar pulse shape and time of flight as detected limited by the digital oscilloscope low pass filter towith two different sensors can be observed. Note 20 MHz.that the absorbing object is close to the center of thephantom, but its position is not necessarily sym-metric with respect to the two ultrasonic sensors.The temporal width of the pulse is related to the 4. Optoacoustic image reconstructionspatial dimension of the absorber in the direction algorithmperpendicular to the sensor surface. A sharp nega-tive pulse at the origin in the fiber optic sensor sig-nal was due to the scattered light from the Nd : YAG 4.1 Optoacoustic tomography (OAT)laser pulses incident on the APD photodetector.This pulse can be used as a trigger to accurately syn- In optoacoustic excitation, if the laser pulse is shortchronize an array of optoacoustic sensors. Figure 4c enough so that the the heat conduction is negligibleand d show the noise characteristic that the sensors during the laser pulse (thermal confinement) andpresent during the optoacoustic acquisition. The the stress waves are longer than the duration of the# 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.biophotonics-journal.org
  • FULL ARTICLEJ. Biophotonics 4, No. 3 (2011) 189pulse travelling through the characteristic dimension measured pressure transients for the three types ofof the heated region (stress confinement), then the imaging geometries: planar, cylindrical and spheri-initial pressure distribution is proportional to the op- cal [18]. Nevertheless, in practical applications, thetical absorption coefficient and the optical fluence. measurement surfaces are finite and partially encloseThe initial pressure distribution excited by a laser the region under test, thus the OA signals cannot bepulse L(t) in a random and heterogeneous optically collected from all directions. As a consequence ofabsorbing and scattering tissue equals po ð~0 Þ. Assum- r facing an incomplete data problem, the resulting OAing a delta-pulse excitation (i.e., L(t) ¼ d(t)), the op- image is not exact.toacoustic pressure pd ð~ tÞ at position ~ and time t r; r The exact back-projection algorithm can be writ-initiated by source po ð~0 Þ can be computed using the r ten in a discrete form as:following equation [17]: P N  P N  ððð    r 0 po ð~ Þ ¼ j~0 À~ j r ri DWi b ~; t ¼ vs ri DWi ð7Þ 1 @ 1 j~0 À ~ r rjpd ð~ tÞ ¼ r; po ð~0 Þ d t À r d~0 r i¼1 i¼1 4pv2 @t vs t s vs 0 where po ð~ Þ is the initial pressure distribution to be r ð5Þ reconstructed, N is the total number of detection po-where vs (in mm/us) is the speed of sound in tissue. sitions, bð~; tÞ is the back-projection term related to ri PThe Eq. (5) essentially computes the acoustic pres- the signal pð~; tÞ and the ratio DWi = ðDWi Þ is the risure at position ~ and time t ¼ j~0 À ~j=vs integrat- ro r r solid angle weighting factor, which compensates theing the initial pressure distribution po ð~0 Þ over the r reconstruction distortion resulting from the limitedsurface of a sphere of radius j~0 À ~j (see Figure 5). r r view.For a finite pulse duration, pð~ tÞ is defined by the r; The back-projection term, bð~; tÞ, can be calcu- rifollowing convolution [17]: lated as follows: Ð þ1 @pð~; tÞ ripð~ tÞ ¼ r; Lðt À tÞ pd ð~ tÞ dt r; ð6Þ bð~; tÞ ¼ 2pð~; tÞ À 2t ri ri ð8Þ @t À1 The solid-angle element related to the measurementwhere pd ð~ tÞ is the impulse-heated optoacoustic pres- r; at position ~ depends on the detection geometry risure and L(t) is the temporal profile of the laser pulse. and obeys: The acoustic pressure at position r at time t can  be calculated integrating the initial pressure distri- DSi ~0 À ~ r ri DWi ¼ ~is 0 n ð9Þbution po ð~0 Þ on a spherical surface with radius r j~0 À ~j2 r ri j~ À ~j r rij~0 À ~j ¼ vs t. r r Therefore, the goal of OAT is to invert Eq. (5) to where ~is is a unitary vector normal to the measure- nrecover the initial pressure distribution po ð~0 Þ that r ment surface pointing to the source.requires computer-based reconstruction algorithms. The exact BP formula can be computed using the following steps: (1) calculate the back-projection term bð~; tÞ for each OA signal pð~; tÞ, (2) back-pro- ri ri ject it over spherical shells (radial BP), (3) sum over4.2 Exact time-domain back-projection all the projections, and (4) normalize each pixel ofalgorithm the reconstructed volume with the summation of DWi of each back-projection step. The exact BP formula in practical applicationsXu and Wang have derived an exact time-domain produces images with excellent resolution (sharpback-projection reconstruction formula based on boundaries) but poor contrast. As a solution to this problem, some approximate time-domain filtered back-projection reconstruction algorithms have also been reported [14]. p(r,t) r’-r 4.3 Wavelet-filtered time-domain r’ r back-projection algorithm p0 (r’) In order to sharpen object boundaries while simulta- O neously preserving high contrast of the reconstructed objects, a wavelet transform implementation using aFigure 5 Schematic diagram of the optoacoustic signal gen- wavelet family resembling the theoretical N-shapederation. OA signal has been used [14]. The wavelet transformwww.biophotonics-journal.org # 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
  • Journal of BIOPHOTONICS190 H. Lamela et al.: Interferometric fiber optic sensors for biomedical applications of optoacoustic imaging g1 µ1 g2 µ2 g3 µ3 g4 µ4p(r,t) g5 µ5 pw(r,t) g6 µ6 g7 µ7 g8 µ8 g9 µ9 (a) 100 200 300 400 500 600 700 800 900 1000 (b)Figure 6 (a) Schematic diagram of the wavelet-based bank of filters; (b) nine scales of the 3-rd derivative of the Gaussianwavelet covering a range of frequencies from k ¼ 4 to 1024 digital samples: g1(t), g2(t), . . ., g9(t).has been established in signal processing as an ex- an OPO laser (Vibrant 355, OPOTEK Inc., Carls-ceptional tool for localization of the specified signal bad, CA, USA) tuned to 480 nm which delivers, at aprofiles [19]. Images are obtained first using multi- distance of 125 cm, an elliptical laser spot with aresolution wavelet-based signal processing on the major axis of 2.4 cm and minor axis of 0.8 cm and aOA signals and subsequently using a radial BP algo- fluence of 20 mJ/cm2. The second optoacoustic im-rithm. age was reconstructed using signals acquired from Nine scales of the same wavelet covering a wide the LOIS system detected using 64 piezoelectricrange of frequencies from k ¼ 4 to 1024 digital sam- PVDF transducers which occupy the total angularples are generated: g1(t), g2(t), . . ., g9(t) (Figure 6). aperture of 174 . In this case, the phantom was illu-Oraevsky et al. used the 3-rd derivative of the Gaus- minated using a Q-switched Nd : YAG. In both casessian wavelet family as the wavelet is similar to the we have used the radial back-projection algorithmtheoretical N-shaped OA signal to be enhanced after for the reconstruction of the optoacoustic imagessignal processing. [20], reviewed in the previous section. Digitized op- The filtered OA signal pw ð~ tÞ can be written as: r; toacoustic signals detected by our optical fiber sen- sor were filtered using a band-pass filter which has a P 9pw ð~ tÞ ¼ r; mHF gi ðtÞ Ã pð~ tÞ i r; ð10Þ cut off frequency of 30 kHz and 2 MHz. These sig- i¼1 nals were registered without any temporal averaging.where gi(t) is one of the nine wavelet scales and mHF The optoacoustic signals, detected using the LOIS,is a weighting factor designed to enhance high fre- were filtered using a 9 scale wavelet filter that simul-quencies components of the signal. Here, we use: taneously converts the bipolar pressure signals to monopolar signals of absorbed energy [14]. Figure 7mHF ¼ ð1024; 512; 256; 128; 64; 32; 16; 8; 4Þ ð11Þ shows the position of the absorbing object in the phantom (Figure 7a) and the reconstructed imagesThus, the filtered back-projection formula consists of using both systems, Figure 7b depicts a LOIS imagethree steps: (1) compute pw ð~ tÞ for each OA signal r;pð~ tÞ, (2) perform a radial back-projection of each r;filtered signal, and (3) sum over all the projections.5. Optoacoustic image reconstruction (a)We have reconstructed optoacoustic images from thepressure signals measured by the optical fiber sensorand PVDF transducers from the LOIS array. In or- (b) (c)der to reconstruct a two-dimensional image, the fiber Figure 7 (a) Diagram showing location of the embeddedoptic sensor was placed in contact with the upper object in the PVCP phantom. (b) Optoacoustic image ob-part of the PVCP phantom, which was rotated to 58 tained from LOIS utilizing an array of 64 PVDF transdu-different positions, forming a total scanned array cers. (c) Optoacoustic image reconstructed from fiber opticaperture of 178 . The phantom was illuminated using sensor signals.# 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.biophotonics-journal.org
  • FULL ARTICLEJ. Biophotonics 4, No. 3 (2011) 191Table 1 Acoustic sensitivity for 1 and 10 MHz emitters of Horacio Lamela receivedsingle mode SOF, single mode POF and GIPOF-50. the Industrial EngineeringFrequency SM-SOF SM-POF GIPOF50 degree from the Universi-(MHz) (mrad/kPa) (mrad/kPa) (mrad/kPa) dad Politecnica de Madrid ´ (UPM) in 1980, the “Di- 1 0.95 Æ 0.03 13.07 Æ 0.28 13.81 Æ 0.18 plome d’Etudes Approfon-10 0.87 Æ 0.11 4.07 Æ 0.16 3.98 Æ 0.13 dies” (DEA) from the Uni- versity of Paris XI in 1981, and the “Docteur-Ingenieur” degree in optical interfero-and Figure 7c depicts an image reconstructed using metry from the Conserva-the signals obtained from the fiber optic sensor. The toire d’Arts et Metiers ofoptoacoustic images generated by both systems are Paris, 1985, France. From 1985 to 1987, he was with thein good agreement when considering the dimensions, Massachusetts Institute of Technology, Cambridge, as ashape and location of the embedded object. Also, Postdoctoral Fellow with the Electrical Engineeringbased on the shape of resolved corners, both systems and Computer Sciences Department and Visiting Scho-possess similar spatial resolution. lar at the Research Laboratory of Electronics. He is presently a full Professor with the Departamento de Tecnologia Electronica and leader of the Optoelectro- nics and Laser Technology Group (GOTL) at Universi-6. Conclusion dad Carlos III de Madrid, Spain. His research interests are high-speed semiconductor lasers dynamics and pi- cosecond gain-switching diode lasers, dual mode diodeAn ultrasonic sensor based on an optical fiber inter- lasers for mmW and THz generation, laser optoacous-ferometer has been designed, developed and cali- tics with gold nanoparticles for spectroscopy and bio-brated. The system is capable of detecting wideband medical applications and interferometric fiber opticultrasound signals over the frequency range from 0.1 sensor for optoacoustic biomedical imaging.to 10 MHz and has a noise equivalent pressure of$100 Pa. The optoacoustic sensitivity can be furtherimproved by selecting a more sensitive fiber opticmaterial [21]. By employing a polymer optical fiber Daniel Gallego received the(POF) in place of the silica optical fiber as the sens- B.Sc. degree in physics anding element, we have measured an increase in ultra- the M.Sc. degree in opto-sonic sensitivity by more than 5 times over all the electronics from the Univer-sensor bandwidth (Table 1). Based on these new re- sidad de Santiago de Com-sults, we expect to improve the constrast and detec- postela (USC), Spain intion capability of the optoacoustic imaging system by 2000 and 2002 respectively.using these polymer optical fibers. In 2004 he joined to the We have reconstructed a cross sectional image of GOTL group from Universi-an absorbing object within a high scattering media dad Carlos III de Madrid tousing our optical fiber sensor in a semi-circular scan- study the Bragg gratings asning configuration. Results are comparable to those ultrafast mode-locking diodeobtained with a system based on broadband piezo- laser pulse compressor de-electric transducers. The experiments in the breast vices in the framework oftissue mimicking phantom attempt to demonstrate the European project IST-the feasibility of the interferometric fiber optic sen- MONOPLA. There he ob-sors developed for biomedical applications, particu- tained the DEA in electronic, electric and automaticlarly in optoacoustics imaging. Further work will be engineering. He is currently working toward the Ph. D. degree in the same group. His research interest in-dedicated to improve the sensitivity of the fiber optic cludes optical sensors for detecting ultrasound, fibersensors by using polymer optical fiber and by in- optic interferometry, optoacoustic signal generation,creasing the density of the sensor active area, and optoacoustic imaging reconstruction algorithms andimplement wavelet filtering for the optoacoustic sig- biomedical imaging.nal processing. This will result in an improved signal-to-noise ratio and, in turn, contrast of optoacousticimages. Finally, it is essential to improve the scan-ning system, this will reduce data collection time.Also, it is projected that future experimental trialswill be performed using a parallel multi-channel in-terferometric sensor system.www.biophotonics-journal.org # 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
  • Journal ofBIOPHOTONICS192 H. Lamela et al.: Interferometric fiber optic sensors for biomedical applications of optoacoustic imaging Rebeca Gutierrez received ´ References the Telecommunication En- gineering degree from the [1] A. A. Oraevsky, S. L. Jacques, and F. K. Tittel, Proc. Universidad Carlos III de SPIE 1882, 86–101 (1993). Madrid, Spain, in 2008. In [2] L. H. V. Wang, Photoacoustic Imaging and Spectro- 2007 she joined the GOTL scopy (CRC Press, 2009). as a trainee researcher, [3] J. Laufer, D. Delpy, C. Elwell, and P. Beard, Phys. where she did her master’s Med. Biol. 52, 141–168 (2007). thesis project in image re- [4] A. A. Oraevsky, E. V. Savateeva, S. V. Solomatin, construction algorithms for A. A. Karabutov, V. G. Andreev, Z. Gatalica, T. Kha- optoacoustic medical imag- mapirad, and P. M. Henrichs, Proc. SPIE 4618, 81–94 ing. She joined Argongra in (2002). 2009 and is currently work- [5] S. Manohar, A. Kharine, J. C. G. Van Hespen, W. Steen- ing in the R&D Department bergen, and T. G. Van Leeuwen, Phys. Med. Biol. 50, as a research engineer in projects related with satellite 2543–2557 (2005). image processing and design, development and integra- [6] S. Manohar, S. E. Vaartjes, J. C. G. Van Hespen, J. M. tion of geographical information systems. Since 2010, Klaase, F. M. Van Den Engh, W. Steenbergen, and she is also working toward the M.Sc. and Ph.D. degree T. G. Van Leeuwen, Opt. Express 15, 12277–12285 at the Universidad Politecnica de Madrid (UPM). Her ´ (2007). research interests include remote sensing systems, image [7] X. D. Wang, Y. J. Pang, G. Ku, X. Y. Xie, G. Stoica, and processing, computer vision and data mining. L. H. V. Wang, Nat. Biotechnol. 21, 803–806 (2003). [8] H. F. Zhang, K. Maslov, M. Sivaramakrishnan, G. Stoi- ca, and L. H. V. Wang, Appl. Phys. Lett. 90, 053901 (2007). Alexander A. Oraevsky re- [9] E. Zhang, J. Laufer, and P. Beard, Appl. Optics 47, ceived his initial training in 561–577 (2008). physics and mathematics [10] P. Fomitchov and S. Krishnaswamy, Opt. Eng. 42, from the Moscow Physical 956–963 (2003). and Engineering Institute in [11] V. Wilkens and C. Koch, Opt. Lett. 24, 1026–1028 Moscow, Russia and ob- (1999). tained a doctorate in laser [12] R. De Paula, J. H. Cole, and J. A. Bucaro, J. Light- spectroscopy and laser bio- wave Technol. 1, 390–393 (1983). physics from the USSR [13] H. Wen, D. G. Wiesler, A. Tveten, B. Danver, and Academy of Sciences. He A. Dandridge, Ultrason. Imaging 20, 103–112 (1998). began his pioneering re- [14] S. A. Ermilov, T. Khamapirad, A. Conjusteau, M. H. search in the field of opto- Leonard, R. Lacewell, K. Mehta, T. Miller, and A. A. acoustic imaging, sensing and monitoring in 1988. In Oraevsky, J. Biomed. Opt. 14, 024007-1–14 (2009). 1992, as Whitaker Fellow, he joined the faculty at Rice [15] D. A. Jackson, R. Priest, A. Dandridge, and A. B. Tve- University. Prior to joining Fairway Medical Technolo- ten, Appl. Optics 19, 2926–2929 (1980). gies as Vice-President of Research and Development, [16] G. M. Spirou, A. A. Oraevsky, I. A. Vitkin, and W. M. he was the Director of the Optoacoustic Imaging and Whelan, Phys. Med. Biol. 50, N141–N153 (2005). Spectroscopy Laboratory at the University of Texas [17] L. V. Wang, IEEE J. Sel. Top. Quantum Electron. 14(1), Medical Branch in Galveston, TX and an Assistant 171–179 (2008). Professor at the Department of Ophthalmology and Vi- [18] M. Xu and L. V. Wang, Phys. Rev. E Stat Phys Plas- sual Sciences. Presently he holds an adjunct Professor mas Fluids Relat. Interdiscip. Topics 71, 016706 position at the Biomedical Engineering Department of (2005). the University of Houston. [19] G. Strang and T. Nguyen, Wavelets and filter banks, (Wellesley-Cambridge Press, USA, 1996). [20] R. A. Kruger, D. R. Reinecke, and G. A. Kruger, Med. Phys. 26, 1832–1837 (1999). [21] D. C. Gallego and H. Lamela, Opt. Lett. 34, 1807– 1809 (2009).# 2011 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.biophotonics-journal.org