We present a novel low-coherence interferometer configuration, equipped with acousto-optic deflectors that can be used to simultaneously acquire up to eight time domain optical coherence tomography en-face images. The capabilities of the configuration are evaluated in terms of depth resolution, signal to noise ratio and crosstalk. Then the configuration is employed to demonstrate simultaneous en-face optical coherence tomography imaging at five different depths in a specimen of armadillidium vulgare.
© 2013 OSA
Optical coherence tomography (OCT) is a noninvasive, noncontact imaging modality that has the potential to produce high-resolution cross-sectional images of biological tissue [1, 2]. It has been reported that conventional Time Domain (TD)-OCT can provide high transversal resolution, real-time transverse (en-face) constant depth images . Although Spectral Domain (SD)-OCT is currently regarded as superior in terms of acquisition rate or signal to noise ratio, only TD-OCT can provide real-time constant depth images . To exploit this TD-OCT advantage to the higher extent, techniques that allow simultaneous imaging at several depths have been proposed. In a first attempt, two en-face images from a post mortem human eye were collected with a configuration that was based on two Michelson interferometers each equipped with a mirror vibrating at a different frequency . However, even after some conceptual improvements  the number of imaging channels was limited by increasingly larger dispersion in each additional channel. In a different approach, recirculating loops in both reference and sample arms of a low coherence interferometer were used . More than 10 channels were obtained, however the sensitivity from one channel to the next decreased by more than 4 dB. In addition, it seemed inappropriate to create multiple delayed replicas of the signal from the object, that already contains multiple signals originating from different depths.
In this paper, a different approach is considered for the simultaneous imaging of a sample at multiple depths, that can ensure constant sensitivity amongst the channels. Multiple delays are created in the reference arm only, avoiding replication of the object beam. The optical configuration proposed is based on a combination of acousto-optic modulators with stepped optical delays. Such combinations of acousto-optic modulators with optical delays have been initially reported in the practice of phased array antenna signals . Acousto-optic modulators have also been reported in the practice of OCT  and low coherence interferometry , to create stable high frequency carrier for the interference signals. A combination of an acousto-optic modulator, operating as an acousto-optic deflector (AOD), with a stepped mirror array was used to provide high speed stepped axial scanning . A pair of AODs were also used in conjunction with optical delay lines, in a configuration originally developed for time-delay based photonic signal processing applications . A similar principle was used for space-division optical switching in telecommunications .
Novel functionality for pairs of AODs interleaved with optical delays is demonstrated here, by simultaneously driving the AODs with several radio frequency (RF) signals, of different frequency. The AODs and the optical delays are incorporated into a multiplexer block placed in the reference arm. The first AOD splits the incident beam into N directions and a different optical delay is introduced into each split beam. Then all the beams are recombined using the second AOD. To compensate for dispersion , a similar pair of AODs is used in the object arm, where no delays are used between the AODs. Such a configuration allows interrogation of N optical path difference values simultaneously. The configuration is supplementary equipped with a pair of transversal scanners, to achieve optical coherence tomography functionality. For each pixel in transversal section of an object investigated, pixel chosen by the pair of transversal scanners, interference signals are produced simultaneously from backscattered signals returned from N depths in the object.
2.1 Optical set-up
The schematic drawing of a TD-OCT system capable of simultaneously acquiring several images from multiple depths is depicted in Fig. 1 .
Light from a Superlum SLD-371-HP-DIL-SM-PD superluminescent diode emitting at 840 nm central wavelength with a 50 nm FWHM spectral bandwidth is amplified by an Optical Power Booster Superlum 850-50. The resulting amplified optical power is 23 mW with a central wavelength of 850 nm and ~15 nm FWHM spectral bandwidth. The light from the booster then is inserted into a 50/50 single mode directional coupler that divides light into an object and a reference arm. The polarization of light is controlled using Newport F-POL-IL polarization controllers, POLC. The light output of the first coupler is collimated with two microscope objectives MO1, MO2. All microscope objectives in Fig. 1 are 10x OFR LLO-4-18-NIR.
The light in the reference arm then passes through 46080-1-.85-LTD Gooch and Housego acousto-optic deflectors, AOD21 and AOD22. The deflectors are operated with a Gooch and Housego digital frequency synthesizer MSD040-150-0.8ADS2-A5H-8X1. The frequency synthesizer is controlled via a computer and can be tuned within 40 - 150 MHz range in 1 kHz increments. It can generate maximum N = 8 channels operating simultaneously with up to 200 mW of RF power delivered per channel. The light passing through the AOD21 can be diffracted into a number of beams, from 1 to 8. Two plano-convex BK7 singlet lenses L1 and L2 with focal length f = 1 m are placed at the focal length distance from the deflectors. A multiple optical delay element is inserted between the lenses so that each diffracted beam passes through a different glass ‘step’ as depicted in Fig. 2 . At the insertion point, all the beams are parallel and each beam is slightly converging. The lens L2 converges the fan of diffracted beams on the AOD22, driven by the same set of N RF signals, of identical frequencies as applied to AOD21. Light levels on each diffracted beam can be digitally controlled in each of the 8 channels. After AOD22, all beams in the first diffracted order are collinear and therefore they can be easily injected into a second 50/50 single mode optical fiber coupler using the MO4.
To match components for dispersion, the same combination of deflectors and lenses is used in the object arm, however both AOD11 and AOD12 are driven at a single frequency of 60 MHz. Additionally, after AOD22 in the reference arm, a dispersion compensator rod, DCR, made of SF6 glass is used to compensate for dispersion that is introduced by the lenses and beamsplitter in the object arm.
After the AODs in both arms, the first-order diffracted beam is selected only by using a pinhole conveniently placed (not shown). In the object arm, light is directed by a mirror M to a thick 50/50 beamsplitter plate, BS, which reflects the object beam to a pair of galvanometer scanners, in an XY scanning head, XYSH. The X-scanner determines the line in the final image and is driven with a triangular signal at 500 Hz, generated by a generator TX. The frame is completed by scanning the line vertically using the Y-scanner driven by a sawtooth signal at 2 Hz, provided by a generator TY. The beam is focused on the object using an achromatic doublet lens, AL of f = 30 mm. The reflected light propagates back along the same path till the BS, wherefrom is launched via MO3 into the other input of the second 50/50 single mode fiber coupler, where it interferes with the light from the reference arm. The interference signal is photodetected with a 80 MHz New Focus 1807 balanced photodetector receiver. The RF spectrum of the photodetected signal is analysed using an HP 8590A spectrum analyzer. This is also used to sequentially produce an en-face image for each depth. To this goal, the spectrum analyzer is tuned on each carrier corresponding to each imaging channel, its span is set to zero and its detection bandwidth adjusted to the level of the image bandwidth, at 300 kHz. In this way, the spectrum analyser operates as a mixer followed by a low pass filter. The signal from the spectrum analyzer is sent to a digitizer placed inside the PC, a card NI PXI-5124. This digitizes the signal and in conjunction with the TTL signals from the signal generators, TX, TY, driving the two transversal scanners, synthesises an en-face OCT image on the PC display. By tuning the HP spectrum analyser to each carrier, an en-face OCT image is generated for each depth selected. The multiple carriers of different frequencies are produced simultaneously and are modulated in intensity with the interference signal corresponding to the depth selected by each individual delay step in the multiple optical delay element. The system presented is simultaneous in generation of multiple carriers for several optical delays, but due to the limitations in the display capability, a single image can be displayed at any given time. Full functionality of simultaneous display of images can be accomplished with 8 similar spectrum analysers tuned each on a different frequency. Equivalently, 8 such RF filters can be assembled on a specialized digitizer.
All AODs used perform up-shifting the frequency of the incoming beam. The AODs in the object arm are driven at 60 MHz and the AODs in the reference arm are driven simultaneously at fp = 64 MHz + (p - 1)5 MHz, with p = 1 to 8. This gives: 64, 69, 74, 79, 84, 89, 94 and 99 MHz. Then, the frequency of the beating photodetected signal, νp, results as 2x(64, 69, 74, 79, 84, 89, 94, 99) – 2x60 = 8, 18, 28, 38, 48, 58, 68 and 78 MHz.
2.2 Delay method
If a laser beam is launched into an acousto optic crystal at the Bragg angle, most of the incident light will be diffracted into the first order of diffraction, m = 1. The Bragg angle for the specific acoustic and optical wavelengths can be calculated using Eq. (1).Eq. (2).
Although Eqs. (1) and (2) indicate that the direction of the diffracted beam can be easily manipulated by controlling the operating frequency, f, it is also obvious that they add certain restrictions on the laser bandwidth that can be used when operating the device on several frequencies simultaneously. Two important implications to the functionality implemented here will be examined: variation of optimal angle of incidence and diffraction angles for different operating frequencies and optical beam spectral bandwidths.
First, we will simplify the case by making a few assumptions to reduce the analysis to a case closer to that used in the experimental set-up. In this example we will operate an acousto-optic deflector at the eight acoustic frequencies mentioned above, starting from 64 MHz with a step of 5 MHz for each subsequent frequency. We will assume that the AOD is made of tellurium dioxide that exhibits a sound speed v = 4260 m/s . As an additional constraint, the deflected beams should be spatially separated to avoid crosstalk between channels. This imposes restrictions in terms of the angle of diffraction for the first order diffracted beams.
After diffraction of the incident beam into multiple diffracted beams, it is important to secure non-overlapping diffracted optical beams, to allow insertion of different optical elements in the paths of the multiple diffracted beams. By using Eq. (2) and simple trigonometry, a formula that allows evaluation of the lateral distance between beams from two adjacent channels, produced by RF signals of frequency fp + 1 and fp, where p = 1 to p = 7 was derived and presented in Eq. (3).
For instance, using Eq. (3) for a typical 840 nm SLD and for two beams diffracted by signals of 94 MHz and 99 MHz, the two beams do not overlap as long as the optical bandwidth is less than 43 nm. It should be noticed that the bandwidth value for non-overlap depends on the central wavelength used and is larger for longer wavelengths. Special care should be taken when choosing the bandwidth of the light source. As the technique, requires having spatially non-overlapping beams, to maintain a reasonable footprint of a system, the diameter of the optical beam must be also considered.
Before applying all 8 signals simultaneously to the two AODs, the multiplexer is investigated for three cases, where in each case, two signals of different frequency are applied only. In the first case, the AODs are operated at 64 MHz and 69 MHz, (p = 1) in the second case at 79 and 84 MHz (p = 4) and in the third case at 94 MHz and 99 MHz (p = 7). All three cases are modeled for a FWHM of the optical bandwidth of the injected optical signal of 0 nm, 20 nm, 30 nm and 40 nm. The graphs in Fig. 3 are obtained using Eq. (3) and demonstrate the effect of optical bandwidth on the allowed lateral distance, Δx, between adjacent beams.
Figure 3 demonstrates that there is a tradeoff between source bandwidth and the footprint of the system. The larger the optical bandwidth, the wider the fan of diffraction, therefore, less are the gaps Δx between adjacent channels, which can be only compensated by a larger l.
Figure 3 shows that for all the examples chosen there is no overlap of the beams. However, these graphs are obtained for d = 0 therefore for a given l, Δx from the graph in Fig. 3 should be interpreted as determining the maximum value d allowed for the beam diameter. The top black bold line, in Fig. 3, for an optical source with an infinitesimally small bandwidth gives also the distance between the beam centers of the two adjacent beams. In this case the beam retains the same diameter when diffracted by Bragg gratings of different periods.
The delay element requires an adequate spacing between beam centers (larger than the beam diameter of the diffracted beams, which depends on both the incident beam diameter and the combined effect of diffraction with the bandwidth of the optical source). Additionally, the edges of the individual delay elements are not perfectly cut and therefore some extra clearance is required between the beam FWHM boundaries (of at least 0.1 mm). A trade-off obviously exists between the separation of the diffracted beams and the footprint of the overall system layout.
Figure 3 shows for example that at l = 1 m distance from the deflector, for an optical bandwidth of 30 nm, Δx varies from 0.3 mm to 0.5 mm depending on the set of frequencies used. This suggests that the incident beam diameter can be increased up to the same value, of 0.3 – 0.45 mm. For the same l = 1 m, the bold dark curve which represents the distance between the beam centers, gives 1mm. These numerical values are compatible with delay elements of 1 mm width.
In order to be able to generate well separated en-face OCT slices, the differential delay to be introduced should be equal or larger than the coherence length of the optical source used. Multiple optical delays can be implemented by stacking several glass plates in a stair like fashion as depicted in Fig. 2. In order to achieve small delays, thin borosilicate glass BK7 microscope coverslips of average thickness ~170 μm are used. Counting on an index of refraction n = 1.51, each such coverslip exhibits a differential optical delay of δ = 43 μm.
3.1 RF spectrum of the photodetected signal
In the following, results are presented obtained by driving the two AODs in the object arm at 60 MHz. The AOD21 and AOD22 in the reference arm were driven at 64 MHz + (p-1)5 MHz, with p = 1 to 8, the channel order. The interference information from different depths is carried by signal of frequencies equal to: νp = 8 MHz + (p-1)10 MHz, as shown above.
To evaluate the performance of the configuration, a mirror was initially used as object. Figure 4 presents the RF spectrum of the photodetected signal. The AOD21 and AOD22 are driven with all 8 RF signals simultaneously. The RF modulation power applied to each channel then is experimentally adjusted so that the peak amplitude of the interference signal would be approximately the same for each individual imaging channel from optical path differences stepped by (p-1)δ, with p = 1 to 8, where δ is the optical thickness of a single step in the multiple optical delay element. In the left in Fig. 4(a), the RF signal analyzer spectrum is shown when the optical path difference is adjusted on 3δ, in which case the carrier frequency is νp = 38 MHz. In addition to the main peak at 38 MHz, intermodulation products either side of it exist, however they do not deteriorate image quality. Apart from electrical crosstalk originating from the frequency synthesizers and elastic crosstalk occurring in the Bragg cell, there is also optical crosstalk. This is due to light from the adjacent channels that goes through the same optical delay segment. This limits the dynamic range to 30 dB. In principle, optical crosstalk can be reduced by improving the optical design of the system.
Then, spectra for each pδ were obtained, by adjusting the reference path length using a translation stage to move the launcher holding the fiber input after AOD22. In this way, individual spectra for all 8 channels were obtained and then superposed in Fig. 4(b). The compound spectrum in Fig. 4(b) is equivalent to that obtained from a volumetric sample consisting in 8 parallel layers interspaced by δ/2 that each returns the same amount of light.
3.2 En face OCT imaging
Figure 5 shows the autocorrelation function of the system, measured using a mirror as object, by axially displacing the launcher holding the fiber input after AOD22. For a spectral FWHM of 15 nm, the autocorrelation function FWHM should have been 40 μm, however 47 μm was obtained (the depth resolution is half of these values, as moving the input fiber to the balanced coupler modifies a single trip of the wave along the reference path). This shows that the system still retains some uncompensated dispersion.
The system with 5 channels simultaneously driven (at 64, 69, 74, 79, 84 MHz) was used to image a sample of armadillidium vulgare, also known as a common pill woodlouse. The RF modulation intensity in each channel in the reference arm was adjusted so that OCT image intensities from all depths exhibited similar strength.
The images in Fig. 6 are presented on logarithmic scale. The imaged area (width × height) in Fig. 6(a) is 1.7 x 1.9 ( ± 0.1) mm2. Each original image consists in 500x500 pixels (500 pixels along 500 lines in the frame). Elimination of flyback at the line edges reduced the images to 441 pixels laterally. For a line scan of 500 Hz triangular excitation made from two ramps, each of 1 ms, an exposure time for each pixel in transversal section of 2 μs results. This is the time interval for acquiring simultaneously all N pixels in all N en-face OCT images for each position of the object beam, as controlled by the signals applied to the X, Y galvoscanners. The incident optical power on the sample was 3 mW.
Figure 6(a) shows up to three adjacent body segments of armadillidium vulgare and its shell structure. To reduce the image noise, an averaging filter that replaces the signal for each pixel with the average signal of its 3x3 neighbors is applied.
Then, the system with all 8 channels driven (at 64, 69, 74, 79, 84, 89, 94, 99 MHz) was used to image a tilted 10 pence coin. Figure 6(b), of 2.3 x 2.6 ( ± 0.1) mm2 shows the compound image obtained by superposing all 8 images. Due to the tilt of the coin, the image in each channel collects signal from a different part of the coin. The strength of the signal in the 94 and 99 MHz channels (given by the two bottom stripes) are multiplied by a factor 1.4 and 1.7 respectively to compensate for the responsivity loss of the photodetector at higher frequency signals.
The combinations of an AOD and of pairs of AODs with delays have been suggested before as commented in the introduction. Two different solutions have been reported: (i) a configuration using a single AOD and employing the zero and 1st order of diffraction to produce a bulk interferometer  and (ii) a configuration with two AODs [10,11]. Here the solution with two AODs, (ii), has been adapted, more amenable to be used in transmission, which was implemented by incorporating two such AODs in each interferometer arm of a hybrid interferometer. The second configuration, (ii), is also compatible with balance detection. Such a configuration also allows separation of the path where optical signal processing takes place from the object path, where signal is weak. In Fig. 1, the losses due to the AODs in the object arm can in principle be compensated by increasing the optical power from the source and the weak signal from the object does not need to traverse the AOD, as required by configuration (i). Securing low losses in the object arm when imaging biological subjects, especially the eye, is crucial for good signal to noise ratio.
In contrast to , [10–12], where a single signal of a varying frequency was applied sequentially at any given time, here multiple RF signals are applied simultaneously to drive the pair of AODs placed in the reference arm. Applying a single driving signal to the AOD presents the advantage that all optical power is diffracted by a single diffraction grating. Applying several signals of different frequencies to the AOD leads to division of optical power into multiple gratings, with a grating created by each applied RF signal. Therefore, the sensitivity drops with the increase in the number of channels. However, in principle, the same sensitivity value can be targeted with this configuration for all channels, as a major improvement from . The sensitivity was evaluated using a mirror behind a round trip attenuator of 34 dB and a single RF signal applied to the multiplexer, in which case 22 dB signal over the noise level was obtained, for the optimum RF excitation. With a bandwidth of 100 kHz in the low pass filter of the analyser, this corresponds to a sensitivity of 106 dB in 1 Hz. By applying 8 signals to the AOD21 and AOD22, the sensitivity is expected to be reduced, as the reference power in each channel decreases by N2 = 82 = 64 times in comparison to the power when only a single RF signal was applied. Consequently, the interference strength on each channel is expected to decrease by N = 8. Additionally, the simultaneous application of N signals, creates the acoustooptic Bragg cell intermodulation products. Minimizing the optical power losses due to intermodulation products requires the reduction of the amplitudes of the signals driving the AODs. As a further complication, the AODs responded differently to different frequency of the driving signal. Therefore, taking these criteria into consideration, the optimum set of RF powers when exciting all 8 channels, was experimentally found at each RF frequency, as follows, at 64MHz: 0.089W, at 69MHz: 0.098W, at 74 MHz: 0.069W, at 79MHz: 0.053W, at 84MHz: 0.020W, at 89MHz: 0.018W, at 94 MHz: 0.052 and at 99 MHz: 0.14W.
Further improvement of this multiplexing technology will be only possible by refinement of the crystal technology, and electrode excitation to allow significant RF driving powers while still maintaining spurious-free dynamic rage suitable for imaging.
The method presented exhibits similarities to the SD-OCT, where the depth is encoded on the frequency of the photodetected signal. The larger the optical path difference (OPD) in the interferometer, the larger the number of peaks and troughs in the channeled spectrum at the interferometer output. By FFT of the read-out signal of the channeled spectrum, the frequency of the output signal is proportional to the OPD, ie the OPD is encoded on frequency. A main difference is that SD-OCT can easily achieve more than 250 pixels in depth (for instance, when using a linear camera spectrometer with 500 pixels covered by the channeled spectrum, 250 cycles modulation are resolvable and each cycle in the number of cycles in the channeled spectrum corresponds to a coherence length). Here, the number of pixels in depth is limited to 8 only. It is expected that further work on the multiplexing technology will allow the application of more RF signals to AODs, with less crosstalk amongst the adjacent channels and optimized efficiency in using the light, enabling the imaging with more than 8 channels. Even after aforementioned developments, the method presented here will exhibit the same disadvantage as SD-OCT in terms of acquiring all depth under the same focus. However, this method does not exhibit the other two disadvantages of the SD-OCT, namely the decay of sensitivity with depth and mirror terms. The sensitivity here was almost the same on each channel and is expected to be the same even for an increase in the number of channels to larger numbers than 8. This is because each channel still operates like a TD-OCT, whose sensitivity does not depend on the OPD.
Regarding the mirror terms, SD-OCT suffers from OPD sign ambiguity. Therefore, in SD-OCT, the top of the object needs to be placed slightly away from the position where OPD = 0, otherwise, if OPD = 0 is placed inside the object, the same modulation is produced in the channeled spectrum for scattering points placed symmetrically around OPD = 0 and the part of the image corresponding to positive OPD is superposed to the part of the image corresponding to negative OPD.
The method presented here does not suffer from this phenomenon. The frequency of carriers is strictly dependent on the diffracted beam used on each delay step, hence, each depth in the object determines a unique frequency value.
This property, of unique allocation of frequency to an OPD value illustrates the potential of the method to be extended to SD-OCT, where it could ensure mirror terms free demodulation and extension of the axial range . This possibility is the subject of a future report, where a swept source would replace the broadband optical source used here. In such a case, the demand for a high number of channels is significantly reduced, as two or three carriers may suffice to cover the most objects that are currently being investigated in practice of OCT as they usually do not exceed the thickness of a few millimeters.
When compared with the current level of SD-OCT technology, this method is sufficiently competitive in terms of the time required to produce an en-face image. The en-face images presented here are made from 500 lines of 500 pixels, acquired in 0.5 s. Considering a commercial swept source at 100 kHz or a linear camera at 100 kHz, collection of 500 x 500 A-scans would require 2.5 s, without considering the time for data calibration and computational slicing of the volume created.
This method is only exceeded in the time required to produce an en-face image by extremely fast spectral OCT systems operating over 1 MHz line rate , which can acquire the same volume as in the example above in 0.25 s. However, the method presented here exhibits the advantage that any movement of the object while it is being imaged, distorts all the en-face images originating from all the depths collected, at the same time. This is not the case for SD-OCT, based on collection of A-scans. Dealing with a similar distortion in all en-face images may allow a simpler correction procedure for movement effects.
A system capable to perform en-face OCT imaging at eight depths simultaneously was presented. The system was used to generate five OCT images from an armadillidium vulgare specimen and eight simultaneous images of a metal surface of a tilted 10 pence coin. RF coding of optical path difference allows simultaneous imaging and decoding of interference signals from different depths by using a single photodetector receiver. The main advantage of such an approach is the possibility to secure similar sensitivity across all channels by adjusting the RF powers of the driving signals applied to the AODs. As an additional advantage, low differential dispersion between adjacent channels is achievable by fabricating the optical delays in the multiplexer using a similar material to that of the object investigated.
Because the division of power amongst channels is confined to the reference arm only, higher number of channels may be possible with progress in the power of white light spatially coherent sources. If such high power optical sources become available, then the increase in optical source power can be compensated by decreasing the efficiency of the 1st coupler towards the object arm, to maintain the same safety power level on the object, while transferring more proportion of the power towards the reference beam, subject of division to an increased number of channels.
A further more immediate improvement should be in displaying of all N images simultaneously. For the moment, all N = 8 channels are acquired in parallel, but a single RF filter was available, made from a spectrum analyzer which needed to be sequentially tuned to each carrier to obtain the corresponding en-face OCT image. Procedures are currently devised to store a full frame of the RF compounded photodetected signal, which will allow digital filtering of all N carriers in parallel and by doing so, extension of the technology to moving organs.
The research leading to these results has received funding from the European Research Council under the European Union's Seventh Framework Programme, Advanced Grant agreement 'COGATIMABIO', No: 249889. A. Podoleanu is also supported by the NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology. The authors thank Dr. Chris Pannell and Dr. Warren Seale from Gooch & Housego for valuable discussions on the technology of AODs.
References and links
1. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R. Hee, T. Flotte, K. Gregory, C. A. Puliafito, and J. G. Fujimoto, “Optical coherence tomography,” Science 254(5035), 1178–1181 (1991). [CrossRef]
2. B. J. Vakoc, D. Fukumura, R. K. Jain, and B. E. Bouma, “Cancer imaging by optical coherence tomography: preclinical progress and clinical potential,” Nat. Rev. Cancer 12(5), 363–368 (2012). [CrossRef]
3. A. Podoleanu, “Route to OCT from OFS at university of Kent,” Photonic Sens 1(2), 166–186 (2011). [CrossRef]
5. A. G. Podoleanu, G. M. Dobre, D. J. Webb, and D. A. Jackson, “Simultaneous en-face imaging of two layers in the human retina by low-coherence reflectometry,” Opt. Lett. 22(13), 1039–1041 (1997). [CrossRef]
6. A. G. Podoleanu, J. A. Rogers, R. C. Cucu, D. A. Jackson, B. Wacogne, H. Porte, and T. Gharbi, “Simultaneous low coherence interferometry imaging at two depths using an integrated optic modulator,” Opt. Commun. 191(1-2), 21–30 (2001). [CrossRef]
7. L. Neagu, A. Bradu, L. Ma, J. W. Bloor, and A. G. Podoleanu, “Multiple-depth en face optical coherence tomography using active recirculation loops,” Opt. Lett. 35(13), 2296–2298 (2010). [CrossRef]
9. C. Hitzenberger, P. Trost, P. W. Lo, and Q. Zhou, “Three-dimensional imaging of the human retina by high-speed optical coherence tomography,” Opt. Express 11(21), 2753–2761 (2003). [CrossRef]
11. N. A. Riza, “Acousto-optically switched optical delay lines,” Opt. Commun. 145(1-6), 15–20 (1998). [CrossRef]
12. H. C. Ho, E. H. Young, and W. Seale, “Microwave frequency translation with multiple bragg cells,” P Soc Photo-Opt Ins 1703, 37-42 (1992).
13. A. P. Goutzoulis, D. R. Pape, and S. V. Kulakov, Design and fabrication of acousto-optic devices (Marcel Dekker, 1994).
15. T. Klein, W. Wieser, C. M. Eigenwillig, B. R. Biedermann, and R. Huber, “Megahertz OCT for ultrawide-field retinal imaging with a 1050 nm Fourier domain mode-locked laser,” Opt. Express 19(4), 3044–3062 (2011). [CrossRef]