January 15, 1998 / Vol. 23, No. 2 / OPTICS LETTERS 103

Direct-to-video holographic readout in quantum wellsfor three-dimensional imaging through turbid media

R. Jones, N. P. Barry, S. C. W. Hyde, and P. M. W. French

Femtosecond Optics Group, Department of Physics, Imperial College of Science, Technology and Medicine,Prince Consort Road, London SW7 2BZ, UK

K. W. Kwolek and D. D. Nolte

Department of Physics, Purdue University, West Lafayette, Indiana 47907-1396

M. R. Melloch

School of Engineering, Purdue University, West Lafayette, Indiana 47907-1396

Received September 15, 1997

Customized photorefractive quantum-well devices have been developed for real-time video acquisition ofcoherence-gated, three-dimensional images in turbid media. Large-field-of-view holographic imaging withdirect video capture is now possible. We have evaluated the role of intensity-limited device performancein Fourier-plane and image-plane holography in such devices and, using near-infrared light, have imagedthrough turbid phantoms of 13 mean free paths’ scattering depth with 50-mm transverse and 60-mm depthresolution. 1998 Optical Society of America

OCIS codes: 090.0090, 160.5320, 270.0270, 190.0190.

Fast recyclable holographic media have a wide rangeof real-time applications such as interferometry, imagecorrelation, and three-dimensional (3-D) imaging, in-cluding through turbid media, such as biological tissue,by means of coherence gating (see, e.g., Refs. 1 and2). Photorefractive multiple-quantum-well (MQW)devices satisfy many of the requirements for suchrecording media, such as high sensitivities3 and fast re-sponse times.4 In this Letter we report true real-timeholographic imaging, recording directly to videotape,using significantly improved transverse f ield pho-torefractive Al0.1Ga0.9AsyGaAs MQW devices thatwere grown at Purdue University.5 Both image- andFourier-plane recording geometries have been inves-tigated to optimize imaging performance. We expectthat the rapid image acquisition and 3-D sectioningcapability of this technology will find application inmicroscopy and in real-time imaging technologiesincluding in vivo biomedical imaging.

Both image-plane and Fourier-plane recordinggeometries were investigated. Photorefractive quan-tum-well devices must perform below maximum allow-able intensities, which directly constrain holographicimaging fidelity, especially when one is working withthe high dynamic range of Fourier-plane imaging. Weshow here, for the f irst time to our knowledge, thatimage-plane imaging is superior when one is workingwith intensity-limited holograms.

The experimental configuration used to record–reconstruct holograms in a MQW device by means ofthe Franz–Keldysh effect6 is shown in Fig. 1. LensL1 was used to relay the image beam onto the MQWdevice, where it interfered with the reference beam.Depending on the position of L1, either a Fourier-planeor a direct-image hologram was produced. Intensitiesof 4 and 2.5 mWycm2, for the image and referencebeams, respectively, were used at the MQW device. A

0146-9592/98/020103-03$10.00/0

sinusoidal ac field of peak amplitude 65 kVycm andfrequency 3 kHz was applied to the MQW well. Anac rather than a dc voltage was applied to avoid thefield inhomogeneity often seen with dc f ields,7 whichcan distort the holographic images. The hologramswere written at 830 nm with a mode-locked Ti:sapphirelaser. The holograms were read out in the f irstdiffracted order onto a conventional CCD camera by a4-mwycm2-intensity probe beam from a diode-pumpedCr:LiSAF laser tuned to the MQW device’s exciton peaks,850 nmd.

With the apparatus configured for image-planeholography, a U.S. Air Force (USAF) test chart wasused to measure the transverse resolution of theholographic system. This resolution was determinedto be 19 mm, for which the corresponding bars are

Fig. 1. Schematic of the holographic imaging sys-tem: L’s, lenses; PBS, polarizing beam splitter; ND’s,neutral-density f ilters.

1998 Optical Society of America

104 OPTICS LETTERS / Vol. 23, No. 2 / January 15, 1998

shown in Fig. 2(a). This image has been correctedfor static background noise by subtraction of the lightfield recorded in the absence of a hologram (whicharises mainly from light scattered from fixed inhomo-geneities in the MQW device surface). Depth-resolved3-D imaging was demonstrated with a 3-D test objectthat comprised a set of cylindrical steps separatedfrom one another by 100 mm. Figure 2(b) shows acomputer-generated reconstruction of this object ob-tained by adjustment of the delay in the reference armand recording of a hologram of each individual layer.These depth-resolved images were then combinedby use of rendering software. The depth resolutionwas 60 mm (limited by the coherence length of theTi:sapphire laser).

These new MQW devices exhibited a greatly im-proved optical quality across their 2-mm apertures,which, together with the improved uniformity of theapplied ac f ield, increased the field of view by 16times compared with our previously published re-sults.1 This improved optical quality and the fast[submillisecond (Ref. 1)] response times of these MQWdevices allowed us to record useful depth-resolvedimages directly from the CCD camera to a conven-tional video cassette recorder without any backgroundsubtraction or other signal processing, as shown inFig. 3(a), which exhibits a transverse resolution of19 mm. An image of the background noise couldalso be recorded onto video and subsequently sub-tracted in postprocessing, resulting in the image shownin Fig. 3(b). No significant difference can be seenwhen one compares the background-subtracted, frame-grabbed image of Fig. 2(a) and the image shown inFig. 3(b). The truly real-time depth-resolved imagingillustrated in Fig. 3(a) is possible because no computa-tional signal processing is required, as is the case for,e.g., electronic holography.8

Fourier-plane holography was investigated as ameans of increasing the f ield of view, which, in theimage-plane geometry, was limited by the apertureof the MQW device. Figure 4a shows the image of aUSAF test chart transmitted through the MQW device,and Fig. 4(b) shows its corresponding Fourier-planeintensity distribution that was incident upon theMQW device. Figure 4(c) shows the image of 100-mmbars obtained by reconstruction of a Fourier-planehologram recorded in the MQW device, and Fig. 4(d)shows the Fourier plane of this reconstructed image.The reason for the reduced image quality of Fig. 4(c)compared with Fig. 4(a) can be understood by com-parison of the Fourier planes shown in Figs. 4(b)and 4(d): The lowest spatial frequencies have notbeen recorded in the hologram, so the reconstructedimage has effectively been spatially filtered. Whereasspatial filtering may sometimes be exploited to provideedge enhancement,9 for our application this imagedistortion is a disadvantage.

The spatial f iltering occurs in the Fourier-planeholography because it is not possible to achieve auniform fringe modulation depth across the holo-gram, owing to the large difference in the intensityof the lower- compared with the higher-spatial-frequency components. For the images shown in

Fig. 4 the reference intensity was matched to that ofthe higher-spatial-frequeny components. Matchingthe high intensity of the low spatial frequencies

Fig. 2. (a) Holographic reconstruction of a USAF testchart showing 50-mm bars. (b) Computer reconstructionof the 3-D test object.

Fig. 3. (a) Holographic reconstruction of a USAF testchart recorded directly to video. (b), (a) With backgroundnoise subtracted.

Fig. 4. (a) Image and (b) Fourier transform of a USAF testchart viewed through a MQW device. (c) Image plane, (d)Fourier-plane holographic reconstruction.

January 15, 1998 / Vol. 23, No. 2 / OPTICS LETTERS 105

Fig. 5. (a) Direct image through 10 MFP’s (double pass)of scattering media. Holographic reconstructions of (b) aUSAF test chart through 13 MFP’s, showing 50-mm bars,and (c) a 3-D test object showing 100-mm depth resolutionthrough 10 MFP’s.

required an incident power on the MQW devices of,1 W, which exceeds the damage threshold of thedevice. When the intensity of the image signal wasreduced to match an acceptable reference beam inten-sity, the reconstructed image was extremely weak andthe higher spatial frequencies were not recorded. Asimilar effect is seen in Stark geometry MQW photore-fractive devices.10 In the image-plane geometry thisproblem is not encountered because the intensity acrossthe image is roughly uniform.

The existence of a maximum allowable intensity,combined with the need to use the highest intensitiesfor coherence tomography to get the largest penetra-tion through the scattering medium, appears to makeFourier-plane holography less useful for this specificapplication. On the other hand, image-plane hologra-phy has the disadvantage that imperfections in the de-vice will be imaged directly to the image plane of thecamera. A compromise between these two extremesmay be the most useful expedient, and forming thehologram between the image plane and the Fourierplane may give more-uniform illumination than theFourier plane, while keeping device imperfections outof the image plane of video.

Holographic imaging can be applied through turbidmedia because it discriminates against scattered back-ground light as a coherence gate: Any photons whosepath lengths exceed that of the ballistic signal by morethan the coherence length will not write any interfer-ence fringes. Thus, when the hologram is read out,essentially only the unscattered (nondegraded) signalis reconstructed. We investigated this conclusion byusing an image-plane geometry for which a scatter-ing cell containing an adjustable suspension of polysty-rene microspheres (0.46-mm diameter, g ­ 0.72) wasplaced before our test object (point ƒ in Fig. 1). Fig-ure 5(a) shows that the USAF test chart cannot be di-rectly viewed through a solution of 10 mean free paths’(MFP’s) scattering depth (in double pass) because ofthe high amount of scattered background light. Usingthe holographic imaging system described above, weobtained a reconstructed image of the 50-mm test barsthrough a scattering depth of as many as 13 MFP’s,as shown in Fig. 5(b). 3-D imaging through 10 MFP’s

(double pass) of scattering medium was demonstratedwith the same 3-D test object as before, and a typical3-D reconstructed is shown in Fig. 5(c). The depthresolution was again 60 mm.

In conclusion, we have demonstrated a real-timeholographic imaging system that provides ,50-mmdepth and transverse resolution. This technique usestime-gated holography in conjunction with photorefrac-tive MQW devices to give whole field two-dimensionalimage acquisition without the need for pixel-by-pixeltransverse scanning. Direct, depth-resolved, imageacquisition to a video recorder has been demonstrated.Image-plane holography has been determined to bethe optimum experimental arrangement, in spite of thelimited f ield of view, because the large intensity vari-ation across the Fourier plane leads to spatial filter-ing and its associated image degradation. We appliedthis system to image through as many as 13 MFP’s ofscattering medium, with 50-mm transverse and 60-mmdepth resolution, by means of coherence gating.

We anticipate many applications for rapid high-resolution 3-D imaging, including 3-D microscopy, e.g.,of living subjects, motion analysis, process monitoring,and as an input device for 3-D animation. To theseends, further improvements in the field of view and op-tical quality are anticipated as the technology matures.We note that holograms can be written in the MQW de-vices at any wavelength shorter than that of the excitonpeak; therefore it is possible to acquire spectrally re-solved or color 3-D images with this technology. Thispossibility is the subject of ongoing research.

Funding for this research was provided by theUK Engineering and Physical Sciences ResearchCouncil (EPSRC). R. Jones acknowledges an EPSRCstudentship. We also acknowledge funding from theDefence Evaluation and Research Agency, UK.

References

1. R. Jones, S. C. W. Hyde, M. J. Lynn, N. P. Barry, J. C.Dainty, P. M. W. French, K. M. Kwolek, D. D. Nolte,and M. R. Melloch, Appl. Phys. Lett. 69, 1837 (1996).

2. S. C. W. Hyde, N. P. Barry, R. Jones, J. C. Dainty, P. M.W. French, M. B. Klein, and B. A. Wechsler, Opt. Lett.20, 1331 (1995).

3. D. D. Nolte, D. H. Olson, G. E. Doran, W. H. Knox, andA. M. Glass, J. Opt. Soc. Am. B 7, 2217 (1990).

4. A. Partovi, A. M. Glass, T. H. Chiu, and D. T. H. Liu,Opt. Lett. 18, 906 (1993).

5. K. M. Kwolek, M. R. Melloch, and D. D. Nolte, Appl.Phys. Lett. 65, 385 (1994).

6. Q. Wang, R. M. Brubaker, D. D. Nolte, and M. R.Melloch, J. Opt. Soc. Am. B 9, 1616 (1992).

7. D. D. Nolte, N. P. Chen, M. R. Melloch, C. Montemagno,and N. M. Haegel, Appl. Phys. Lett. 68, 72 (1996).

8. H. Chen, Y. Chen, D. Dilworth, E. Leith, J. Lopez, andJ. Valdmanis, Opt. Lett. 16, 487 (1991).

9. J. Feinberg, Opt. Lett. 5, 330 (1980).10. W. S. Rabinovich, S. R. Bowman, R. Mahon, G. Beadie,

C. L. Adler, D. S. Katzer, and K. Ikossi-Anastasiou,J. Opt. Soc. Am. B 13, 2235 (1996).