Acousto-optic imaging

Project goal:

“…Exploration of the possibilities and limitations of the acousto-optic method; development of instrument prototype for quantization of chemical components...”

Introduction into acousto-optic effect

acousto-optic effect
Figure 1: Acousto-Optic effect. Focused ultrasound pressure field changes optical path 1) by changing the refractive index, 2) by displacing the scattering particles

Photons travel through a turbid media on a random paths, thus it is not possible to specify neither pathlength nor location of the path in a media. However, it is possible to overcome this issue by combining acoustics and light. Ultrasound (US) scatters in tissue much less than light, so we can use it for the localization in the tissue. A small ultrasound burst of 1µs in length is illuminated with a 1µs LASER pulse. During this 1µs illumination time, the US package is almost stationary in the tissue, and modulates photons travelling through this US pressure field. From the all the light that travels through the whole sample only a small part is modulated by the ultrasound. The modulation takes place by 1: the change of refractive index caused by compression of the ultrasound pressure (Raman-Nath effect) and 2: the scatter particles are slightly vibrating with the US frequency changing the optical path. Where scattering on the vibrating particles causes a Doppler shift with the US frequency. Concluding, the ultrasound changes the optical path, and frequency modulated the light, this small fraction of modulated light posses’ information of local fluence, because the bulk non-modulated light is ignored. Measurement of local absorption is done by measuring the modulated photon intensity which travelled through the AO tagged volume. No absorption causes full fluence to pass the ultrasound volume and maximum AO tagged intensity is measured. When absorption increases, local fluence dropped, and from the AO tagged volume less fluence is exiting the AO tagged volume, dropping the modulated intensity.

Experimental setup

Experimental setup
Figure 2: Schematics of the acousto-optical setup in transmission geometry.

AFG1&2: dual-channel programmable function generator (Tektronix AFG3102), P: MOSFET pulser drives transducer. UST: 5 MHz US transducer, L: CW LASER Ti:Sa 750nm 700mW (Coherent MBR-EL pumped by a Verdi V6), AOM: Acousto-Optical Modulator converting CW to pulse (Neos 23080), A1: Beam dump for non-deflected light, A2: spatial filter, BS2: Beam sampler, AOM1&2: Acousto-Optical Modulator as frequency shifter. CCD: camera (Basler A102f), BS50: non polarizing beam splitter cube.

Record interference pattern.

The detection scheme is designed to detect the US Doppler shifted photons. The use of multipoint detector gives a large SNR and van measure fast in the order of milliseconds. In the case when no ultrasound is applied, the Mach-Zehnder interferometer configuration causes a static interference pattern on the CCD sensor. Where, the interference pattern is a mix of the light passing A2 and the reference arm. This constructs a stationary interference pattern, because the light frequency in both mixing arms is identical. When ultrasound is applied a portion of the LASER frequency is frequency modulated with a discrete US frequency. A small part of the fluence from Aperture A2 contains a different frequency that the reference arm. The ultrasound modulated light doesn’t stationary interfere with the reference arm, but it has a beat frequency of the two sidebands +US Hz and –US Hz. The reference arm is tuned-in one of the acousto-optical sidebands by adding two Acousto-Optical Modulators in the reference arm. The applied drive frequency on AOM1 & AOM2 is miss-matched to create a beat frequency of the sideband. Now the reference wave is locked-in only the ultrasound tagged photons construct stationary interference on the CCD sensor. The bulk non-tagged light has a frequency mismatch where this is recorded as a DC offset, where the interference fringes lie on top containing the information.

Heterodyne amplification for shot noise limited AO tagged signal detection.

The all over light intensity that acousto-optically is tagged is very low (photon counting level). Where optimal detection is preferable to detect this low number of photons, simply increasing exposure time of the camera is not possible due to the Brownian motion of the tissue. This randomizing of the phase allows a max exposure time in the order < 10ms. For this reason it’s necessary to detect AO tagged photos most efficient. The AO tagged fluence from aperture A2 is low and the CCD cameras digitizer noise influences the measured AO tagged intensity. To eliminate the CCD camera noise’s influence, heterodyne amplification is used as optical amplification of the AO tagged intensity from the aperture A2 with the much stronger reference wave.


Here is ICCD the intensity measured at a CCD pixel, where the sum of IS (sample fluence) + IRef (reference fluence) is partly measured as a DC intensity. However formula is recorded as a stationary interference pattern, due to the lock-in into the US Doppler shifted photons, where IS is amplified by IRef helping to increase to interference patter out amplitude well above CCD digitizing units and readout noise. The signal to noise level is now mainly determined by the shot noise of the fluence from the sample.

Holographic image restoration.

The interference pattern that is recorded on the CCD surface is the final result of interference of a diverging reference beam and aperture A2. To simplify this, the aperture A2 could be represented as a 2D array of point sources. Point source 1 interferences with a certain angle with the reference wave on the CCD surface, and interference is constructed with a fringe distance 1. Point source 2 interference with a different angle causing an interference with fringe distance 2. The measured interference pattern exists of a 2D collection of periodic signals. Referring to ordinary Fourier transformation a periodic signal is represented as single peak in the frequency domain. Apply this on the interference pattern; the periodic fringe distance 1 will point after Fourier transformation back to the spatial frequency, representing point source 1.



A: Black card with 5x5mm aperture A2 in front of scattering phantom.

B: 2D FFT reconstruction of the measured interference pattern.

Figure 3: Example of image reconstruction

Ultrasound axial scan

For measuring the Acousto-Optical tagged intensity, we measure the intensity increase on the bright square when ultrasound power increase intensity in the square also increases. For sufficient resolution we use ultrasound burst of 1µs in length. Local light fluence is measured, by initiating an ultrasound burst that travels along the acoustical axe and illuminates the phantom with a 1µs pulse with a certain delay time referred to the initial US burst. By recording a series of interference patter, with different delay times of the laser, we could probe local light fluence along the axial axe. Plotting the intensity of the aperture in function of ultrasound position in mm. (Figure 4), Maximum sensitivity is set 22.5mm, which is the focus position of the transducer.

Figure 4: Example of a typical AO tagged signal along the axial axe, represented as SNR, note that the SNR is determined by exposure time, and thus shot noise limited.

Image reconstruction

It is possible to make an axial scan in depth, we expand the system with a lateral XY scan, imaging of hidden absorbers.

map_of_mua copy
Figure 5: Cross section of a reconstructed cuvette with 2 optical absorbing tubes, placed in a scattering turbid phantom


Scientific staff:

Rob P.H. Kooyman
Ton G. van Leeuwen

PHD student:

Aliaksandr Bratchenia


Robert Molenaar

Publications in peer-reviewed journals:

“Feasibility of quantitative determination of local optical absorbances in tissue-mimicking phantoms using acousto-optic sensing,” by A. Bratchenia, R. Molenaar, and R. P. H. Kooyman, Appl. Phys. Lett. 92, 113901-(1-3) (2008).

“Millimeter-resolution acousto-optic quantitative imaging in a tissue model system,” by A. Bratchenia, R. Molenaar, T.G. van Leeuwen, and R. P. H. Kooyman, J. Biomed. Optic. 14, 034031 (2009).

Conference Proceedings:

“Acousto-optic spectroscopy as a tool for quantitative determination of chemical compounds in tissue: a model study” by A. Bratchenia, R. Molenaar, R. P. H. Kooyman, Proceedings Vol. 6437 of Photons Plus Ultrasound: Imaging and Sensing 2007: The Eighth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics, Alexander A. Oraevsky; Lihong V. Wang, Editors, 64371P

“Application of intense ultrasound bursts for quantitative acousto-optic sensing”, by A. Bratchenia, R. Molenaar, R. P. H. Kooyman, [6856-36], Proceedings Vol. 6856 of Photons Plus Ultrasound: Imaging and Sensing 2008: The Ningth Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics, Alexander A. Oraevsky; Lihong V. Wang, Editors, 685611

“Three-dimensional acousto-optic mapping using planar scanning with ultrasound bursts”, by A. Bratchenia, R. Molenaar, R. P. H. Kooyman, Proceedings Vol. 7177 of Photons Plus Ultrasound: Imaging and Sensing 2009: The 10th Conference on Biomedical Thermoacoustics, Optoacoustics, and Acousto-optics, Alexander A. Oraevsky; Lihong V. Wang, Editors, 71771H

“Quantitative acousto-optic imaging in tissue-mimicking phantoms”, by R. Molenaar, A. Bratchenia, R. P. H. Kooyman, Proceedings Vol. 7265: Medical Imaging 2009: Ultrasonic Imaging and Signal Processing, Stephen A. McAleavey; Jan D'hooge, Editors, 72650M