Balanced detection for spectral domain optical coherence tomography

Wen-Chuan Kuo; Chih-Ming Lai; Yi-Shiang Huang; Cheng-Yi Chang; Yue-Ming Kuo

doi:10.1364/OE.21.019280

1. Introduction

Today, optical coherence tomography (OCT) has become a powerful tool for non-invasive in vivo tomographic imaging in transparent and turbid specimens [1]. In particular, the development of new broad bandwidth light sources has led directly to ultrahigh resolution OCT (UHR OCT). For example, the combination of superluminescent diodes with different central wavelengths into one broadband sources [2] and the use of mode-locked solid state lasers [3–5] or supercontinuum light source [6–9] have all been proposed in the literature. The subcellular axial resolution (1–5 μm) now available from these advances has made it possible to obtain in vivo OCT tomograms close to the level of histology, which holds enormous promise for early cancer detection and the assessment of tissue pathologies [10,11].

The earliest implementation of UHR OCT involved the mechanical scanning of a reference mirror to perform A-scans in the time domain (TD) [4]. However, in a time domain setup, the faster scanning speed and longer depth range inevitably lower the system sensitivity. Therefore, high frame rate imaging is difficult to realize in time domain-based UHR OCT. Over the last few years, an alternative OCT system has been proposed that is based on Fourier domain (FD) construction. The main advantage of Fourier domain OCT (FDOCT) is that one full depth scan is obtained synchronously, without the need for moving parts in the reference arm. Thus, FDOCT allows imaging at rates higher by an order of magnitude than in TDOCT [12]. FDOCT can generally be categorized into swept source-based SSOCT, which uses a fast wavelength tunable laser, and spectral radar-based SDOCT, which uses a fast line scan sensor spectrometer [13, 14]. The full axial image can be retrieved through a Fourier transform of the spectral interferogram, with the speeds limited by the electronic capture rate of the imaging sensor, currently greater than 300 kHz for both FDOCT approaches [15, 16].

Balanced detection (BD) is a commonly used detection method in many optics experiments with the need for increased signal-to-noise (SNR) [17]. In comparison to the unbalanced configuration, the BD configuration in TDOCT can yield a SNR in excess of 6 dB since the signal current is double that of the single detection and also because of the additional capability of balanced reception to suppress relative intensity noise (RIN) [18, 19]. For fast TDOCT systems, the use of the balanced configuration can result in a better than 10 times improvement in the SNR [20]. The use and advantages of BD in OCT have been amply demonstrated in TDOCT and SSOCT setups. However, to the best of our knowledge, this technique has yet to be adopted in spectrometer-based UHR SDOCT systems. Although shot-noise-limited detection is always to be expected in SDOCT for large exposure time, with decreasing of exposure time and increasing of reference power, RIN may be the dominant contribution [5, 21]. In this study, we evaluate the detection performance of balanced operation in high speed SDOCT. Carefully calibrated high speed line scan cameras were used in two separate spectrometer units. In vivo imaging in both transparent and turbid specimens was demonstrated in order to experimentally confirm the performance enhancement of the BD method compared to the conventional single detection operation in high speed SDOCT.

2. Theory

The sensitivity figure of merit employed for gauging OCT performance is the signal-to-noise ratio (SNR) quantity. In conventional SDOCT, the interference signal $I (k)$ is in principle

I (k) = S (k) (1 + \int_{- \infty}^{\infty} a (z) \cos (2 k n z) d z + A C t e r m s)

where

S (k)

is the spectral intensity distribution of the light source, a(z) is the scattering amplitude of the elementary waves versus depth, and n is the refractive index of the scatterer. The first term is a constant dc spectral term from the light source. The second term includes the total number of useful interference signal photons which encode the depth information of the object. Since only the real part of the complex spectral density is detected in SDOCT, the result is that the signal is given by

\frac{η^{2} τ^{2}}{{(h ν)}^{2}} P_{s a m} P_{r e f}

where

P_{s a m}

and

P_{r e f}

are the collected sample and reference power respectively.

η

is the quantum efficiency of the spectrometer-detector combination,

τ

is the total signal collection time, and

h ν

is the energy of a single photon. The third, autocorrelation (AC) term describes the mutual interference of all elementary waves and leads to low frequency noise. This contribution of the sample AC noise can be reduced by increasing the reference arm power with respect to the signal.

For a simple unbalanced configuration, the total noise in the SDOCT system after the Fourier transform includes the CCD receiver noise (i.e. readout and dark noise), shot noise, and the random intensity noise (RIN). These overall noise in electrons squared per readout cycle and per sensing element is given by

σ_{n o i s e}^{2} = σ_{r + d}^{2} + \frac{η τ}{h ν^{}} P_{r e f} + \frac{η^{2} τ}{{(h ν)}^{2}} P_{r e f}^{2} τ_{c o h}

where

τ_{c o h} = \sqrt{\frac{2 \ln 2}{π}} \frac{λ_{0}^{2}}{c δ λ}

is the coherence time, c is the speed of light,

λ_{0}

is the central wavelength,

δ λ

is the bandwidth of the light source.

The SNR is defined as $S N R = 〈 S_{^{O C T}}^{2} 〉 / σ_{n o i s e}^{2}$ [22]. In most applications the reference reflectivity is much larger than that of the sample arm, so the condition of $P_{s a m}$ << $P_{r e f}$ was assumed. Using this approximation, the SNR may be written as

S N R = 10 \times \log (\frac{(\frac{η τ}{h ν}) P_{s a m}}{[1 + \frac{η}{h ν} P_{r e f} τ_{c o h}] + \frac{h ν}{η τ} \frac{σ_{r + d}^{2}}{P_{r e f}}}) (unit : dB)

Under optimal SNR performance, shot noise dominates both the readout noise and RIN. Thus, the SNR in the SDOCT system has been shown to be [12]:

S N R_{s h} = 10 \times \log [\frac{η P_{s a m} τ}{h ν}] (unit : dB)

However, large bandwidth sources can achieve powers in excess of 50 mW and when launched into a photonic crystal fiber or tapered fiber, can exhibit linewidths in excess of 200 nm [6–9]. Due to the high power, the RIN may dominates the shot noise when the ratio of

\frac{η P_{r e f}}{h ν} τ_{c o h}

is larger than one. Thus, the shot noise limited SNR (SNR_sh) in the SDOCT system will then not be achieved.

However, if the balanced configuration is employed, two spectra with $π$ phase shift difference are generated simultaneously. Thus, a differential spectral interferogram can be calculated as

I_{d i f f} (k) \approx 2 S (k) \int_{- \infty}^{\infty} a (z) \cos (2 k n z) d z

Following from the above equation, one can see that the dc and AC terms cancel out each other and there is a two fold increase of the useful signal. With regards to the noise, it has been shown that in the BD configuration, although the shot noise in each of the detectors of the balanced receiver is independent and their variances are summed up, the contribution of the RIN can be largely minimized. Therefore, if the balance detection was considered ideal, high SNR values compared with the standard single detection SDOCT configuration are expected, the SNR may be written as

S N R_{B D} = 10 \times \log (\frac{(\frac{η τ}{h ν}) P_{s a m}}{1 + \frac{h ν}{η τ} \frac{σ_{r + d}^{2}}{P_{r e f}}}) (unit : dB)

Figure 1 shows theoretical SNR as a function of reference arm total power. A SNR comparison between the BD and non-BD setups (i.e. the difference between Eq. (4) and Eq. (7)) with the value for our SDOCT system was calculated. Also shown are the signal to receiver noise ratio (SNR_re), signal to shot noise ratio (SNR_sh), and signal to RIN ratio (SNR_RIN). One can see that SDOCT systems may operate at an SNR below the shot noise limited performance at higher source powers. At these higher source powers, although attenuating the reference beam to reduce excess photon noise is a method to leave primarily shot noise, attenuating the reference beam is wasteful of photons, and especially for short exposure time, increasing reference arm power can elevate signal level moves away from the fixed noise level of the image sensor. Thus, BD would be expedient for canceling excess photon noise and other fluctuations in light source intensity, sometimes referred to collectively as RIN, thus maintaining the shot noise limit SNR especially in ultrafast SDOCT.

Fig. 1 Calculated SNR as a function of reference arm power for standard single detection scheme (non-BD) and BD setups. Also shown are the signal to receiver noise ratio (SNR_re), signal to shot noise ratio (SNR_sh), and signal to RIN ratio (SNR_RIN). Different integration time indicates different CCD saturation power. (CCD well depth = 800ke-, $σ_{r + d} = 456 e^{-}$ , $λ_{0}$ = 860nm, $δ λ_{FWHM}$ = 100nm, $η$ = 0.35, sample power fixed to 0.4 $μW$ )

Download Full Size | PDF

For the shortest integration time of the camera (20 $μ s$ ) in our SDOCT system, a pixel well will become saturated at 800,000e (corresponding to a total power of 109 μW). At that time, the largest SNR difference between the non-BD and BD setups will be 5.03 dB. However, with the advances in high speed line scan CCD, integration times of 5 $μ s$ or even lower are now easily attainable, resulting in larger than 7 dB enhancement in the SNR compared to the unbalanced configuration. In this paper, we experimentally evaluate the expected performance enhancement of the BD method in an ultrafast (i.e. ≤ 25 $μ s$ integration time) UHR SDOCT as compared with the standard single detection method.

3. Method

As shown in Fig. 2, the UHR BD-SDOCT system is constructed by placing a circulator (ENACO communications company, CIR-PI-850-FC/APC) in front of a fiber optic Michelson interferometer. A supercontinuum laser (NKT Photonics, Denmark) is applied as a broadband light source. The emission spectrum of the source is shaped by optical filtering to achieve a median wavelength of 860 nm and spectral bandwidth of 200 nm. A coupler (Thorlabs, WBC 2X2 50/50) with a coupling ratio of 50/50 is arranged to split light power into 50% for the reference arm, and 50% for the sensing arm. Inside the coupler, the field that is reflected experiences a $90 °$ phase shift with respect to the transmitted field. Thus, after the recombination of the sample and reference beams, the outputs from the coupler and circulator consist of two different interference signals, which are distinguished by a phase shift of $π$ . The two dispersed spectrums were formed after two transmission gratings (Wasatch Photonics) and recorded with two 4096 element line scan camera (AWAIBA Lda, DR-4k-500). The quantum efficiency $η$ of the spectrometer-CCD combination was 35%, which was determined by the ratio of the optical power detected by the line scan camera and the power at the detection arm fiber tip. The maximum digital value (4096) corresponded to 800,000 electrons. The spectral data are transferred to a computer via a CameraLink and a high speed frame grabber board (National Instruments, USA). Two line scan sensors and the galvanometer scanners are synchronized. If line scan sensors operated in the shortest integration time (20 $μ s$ ) in our SDOCT system, high imaging speed with up to 50 kHz A-scan rate can be achieved.

Fig. 2 Schematic of the UHR BD-SDOCT system. CO, collimator; DM, dichroic mirror filter; L1 and L2, lens; G, grating; PC, polarization controller; OBJ, objective lens; GS, galvanometer scanner mirror.

Download Full Size | PDF

After data acquisition, several post processing steps were performed. Since the BD-SDOCT system requires two identical spectrometers to record two output spectrums that are $180 °$ out of phase, pre-calibration is required. In the first step, the average of 1000 reference arm spectra was obtained as the background spectrum of the two channels. In each channel, every spectrum captured during the imaging was then divided by its background spectrum. The resulting spectrum was multiplied by a Gaussian to reshape the spectrum. This procedure removes fixed-pattern noise in the image. In the second step, the overlap spectrum between these two spectrometers was extracted from each individual spectrum. Then, a resampling from $λ$ to a specified k space was performed on each spectrum to ensure point-to-point correspondence in the k space. This procedure ensures the spectrum to be linearly interpolated in k-space and accurately recovered to the depth profile [23]. Finally, followed the subtraction of one spectrum from the other channel, an inverse Fourier transform was necessary. In this manner, in the BD-SDOCT system, the use of the BD operation inherently doubles the signal and minimizes the common background noise.

4. System performance

To examine directly the SNR performance of the BD and non-BD schemes, a mirror and an attenuator (neutral density filter) were used to mimic a sample. The power reflected by the weak reflector measured at the fiber tip in the detection arm was 3 nW (per sensing element). The detection arm was then connected to the spectrometers, and spectra were acquired at a speed of 25 $μs$ per spectrum. In conventional SDOCT (i.e. non-BD scheme), only the single spectrometer output in Fig. 2 was calculated. Figure 3(a) is the initial interferogram obtained by the single spectrometer. We noticed that instead of a high frequency modulation from the interference of the refection in sample arm and reference arm, the envelope of the interferogram also has a distinct oscillation profile, which may due to the etalon effect in the CCD sensor. Since this low frequency oscillation of envelope will cause fixed-pattern noise and side lobes in the image, multiplication of the spectrum with a Gaussian was applied to calibrate the spectrum shape.

Fig. 3 (a) interference fringes detected by the single spectrometer. (b) interference fringes detected by channel 1, channel 2, and the BD scheme (i.e. signals in channel 1 were subtracted from channel 2). (c) The zoomed-in signals in the narrow spectral interval exhibited a π phase difference between the two spectra. (d) The depth resolved signals after FFT from the BD-SDOCT (red curve) and standard single detection SDOCT (black curve) respectively.

Download Full Size | PDF

After we had divided the original spectrum in each channel by its background spectrum and multiplied it by a Gaussian function, the overlap spectrum between these two spectrometers was extracted from each individual spectrum and re-sampled from $λ$ to a specified k space. Thereafter, the BD signal was calculated by subtracting one spectrum from the other channel. As shown in Fig. 3(b) and the enlarged picture in Fig. 3(c), we can see that there is an apparent π phase shift between these two simultaneously obtained interferograms (i.e. channel 1 and channel 2). Finally, the FFT result after subtraction is shown in Fig. 3(d), where the peak at zero is due to the autocorrelation of the source spectrum. The picture in Fig. 3(d) is clear proof of the dc suppression (~30 dB) and SNR enhancement (~3.5 dB) benefits in the BD scheme compared to that in single channel detection.

Figure 4 shows the measured SNR at an exposure time of 22 µs at 860 nm with a variable optical power in the reference arm and a fixed optical power of 0.4 $μW$ returning from the sample arm to the corresponding spectrometer. The power in the reference arm was varied by means of an adjustable mechanical aperture. For low optical powers in the reference arm, the SNR enhances with increasing reference arm power because the elevated signal level moves away from the fixed noise level of the image sensor. For higher optical powers, the SNR stagnates and even drops off when the excess noise of the light source becomes the dominant noise process. If BD was considered ideal, the excess photon noise was entirely canceled, except for a component of the excess photon noise remains which is called beat noise [24]. If only this noise term is considered, high SNR values are expected. In practice, based on our current result, a residual noncanceled output of excess photon noise may existed (i.e ~40%). This may because for values of high reference power, the mismatch in the elements of the balanced receiver may contribute more noise. Deviations from the ideal BD regime of operation may arise owing to the differences between the two spectrometers in terms of gain, noise, electronics circuit processing the two signals, as well as the differences between the two incident optical signals in different polarization or spectrum contents. However, the better SNR of BD SDOCT system as compared to conventional single channel SDOCT setups can still be observed.

Fig. 4 Measured SNR for different reference arm power settings with a fixed sample arm power of 0.4 $μW$ returning to the spectrometers for standard single detection scheme (non-BD) and BD setups.

Download Full Size | PDF

Figures 5 (a) and 5(b) show the respective depth-dependent decay (log scale) for different detection schemes. The measured ratio of the A-scan peak height to the noise floor (between 0 and 1.2 mm, excluding the dc noise) corresponding to Figs. 5(a) and 5(b) are shown in Fig. 5(c). For the peak at 0.2 mm, a SNR of 50 dB was obtained using the BD dual-detection scheme. This value is of the same order as the theoretically expected shot noise limited sensitivity of 50.6 dB where $η$ is 0.35, $τ$ is 25 $μs$ and $P_{s a m}$ is 3 nW (per sensing element) in Eq. (5). The damping of the ND filters amounting to −40 dB needed to be added to this value to obtain the sensitivity of the system. Thus, a sensitivity of 90 dB was obtained for the peak at 0.2 mm using the BD operation. Due to the finite spectrometer resolution, a depth dependent sensitivity loss was observed and this effect was more severe in the BD setup. However, compared to conventional single detection SDOCT for sample reflective signals within 1 mm, sensitivity above 80.5 dB and a SNR improvement of maxima ~3.5 dB can be achieved using the BD method. Figure 5(d) shows the measured free-space axial resolution at the −3dB point of the A-scan peaks in Figs. 5(a) and 5(b), which were in the range 3.5–4.5 $μm$ and 3.8–4.4 $μm$ , respectively. These values are consistent with the theoretically predicted axial resolution of ~3.8 $μm$ in air, as determined by a combination of the light source bandwidth, spectrometer bandwidth range, and CCD pixel number.

Fig. 5 Depth dependent decay in (a) standard single detection UHR SDOCT and in (b) UHR BD-SDOCT. (c) Depth dependent SNR measured by using the BD and single detection method. (d) Measured free-space axial resolution by using BD and single detection method.

Download Full Size | PDF

A tape was then used as a multilayer sample to demonstrate the ability of BD-based system to remove the AC term. B-scan OCT images taken by a conventional single detection method is shown in Fig. 6(a). Partial dc and AC noise still appear despite the use of background subtraction. Figure 6(b) shows the results from BD SDOCT method, where dc and AC noise are largely suppressed. Since in our proposed method, A-scans are calculated as a combination of two phase-opposed interferometric spectra acquired simultaneously, this system not only suppresses artifacts due to AC noise but also doubles the signal of interest. In this manner, more layers of tape [Fig. 6(b) compared to 6(a)] can be observed using our proposed UHR BD-SDOCT method. Another example shown in Figs. 6(c) and 6(d) demonstrates how the mutual interference of backscattered light between the cornea and the iris (indicated by “AC”) in living fish eye tissue can also be suppressed.

Fig. 6 OCT images of a multilayer tape and a living fish eye acquired with (a) and (c) conventional single detection UHR SDOCT, and with (b) and (d) dual detection UHR BD-SDOCT. The autocorrelation term from the mutual interference between multilayer of a tape and between cornea and the iris of the fish eye are indicated by AC.

Download Full Size | PDF

Figure 7 compares OCT images of an anterior segment in a living fish eye acquired with conventional single detection UHR SDOCT and UHR BD-SDOCT. The lens and posterior pigmented layer were shown in Fig. 6(b) using the UHR BD-SDOCT system, which achieves larger dynamic range performance than the single detection method [Fig. 7(a)]. The higher magnification image of the rectangular part in Fig. 7(b) demonstrates UHR performance (<3 $μm$ axial and ~6 $μm$ lateral resolution in biological tissue) by clearly showing the detailed structures of a cornea, including the cornea epithelium layer, Bowman’s membrane (BM), and several collagen fibrils (F) within the cornea stroma. Since a low dynamic range can be tolerated in transparent tissue such as the eye, but not in highly scattering tissue, we compare the detection performance of an adult (non-transparent) living fish head using conventional single channel UHR SDOCT [Fig. 8(b)] and UHR BD-SDOCT [Fig. 8(c)]. Top to bottom 2D images show five different positions separated from each other by 300 $μm$ . For comparison, Fig. 8(a) demonstrates the results of the conventional single detection SDOCT method without performing background subtraction processing. Here, it is evident that a large fixed pattern noise, including dc and several horizontal lines, seriously degrades image quality. Tissue structures, including the olfactory bulb and the brain, are observed more clearly in Fig. 8(c) than in Figs. 8(a) or 8(b).

Fig. 7 OCT images of the anterior segment in a living fish eye acquired with (a) conventional single detection UHR SDOCT, and with (b) dual detection UHR BD-SDOCT. (c) The higher magnification image of the rectangular part in (b) shows the detail structures in the cornea. BM: Bowman’s membrane, F: collagen fibrils, E: cornea endothelium.

Download Full Size | PDF

Fig. 8 OCT images of a living fish head in different positions acquired with (a) conventional single detection UHR SDOCT, (b) conventional single detection UHR SDOCT with background subtraction, and (c) dual detection UHR BD-SDOCT.

Download Full Size | PDF

Since more photons from the object contribute to the interference signal, the dynamic range becomes higher, which leads to deeper penetration in nontransparent tissue. Therefore, time integration is a simple and widely used method to improve the dynamic range in SDOCT. However, the disadvantage of performing OCT with signal integration is that any phase fluctuations due to vibrations in the system or sample during the integration period cause blurring and averaging of the modulation depth on the array detector, which results in a distortion in the detection performance. Therefore, BD provides a power-efficient method, which decreases the required exposure time and thus is helpful for enabling ultrafast UHR SDOCT imaging. This performance allows in vivo three-dimensional (3D) imaging or high temporal resolution imaging with high image quality possible. In order to validate this advantage, Fig. 9 shows a 3D human nail fold imaging in vivo. The finger is laterally scanned from the nail (left side) over the nail fold (right side). Image dimensions are $2mm \times 2mm \times 1mm$ in X, Y, Z directions (i.e. including 250k A-scans). Index matching gel (G) was applied to reduce the surface reflected signal. Within all the tomograms, stratum corneum (S), epidermis (E), basement membrane (BM), papillary dermis (PD) and dermis (D) can be clearly distinguished. The blood vessels (BV) and papillary capillary loops (PCL) within the dermis are illustrated with high contrast. We also noticed that capillary morphology is “reverse U-shaped” or “hairpin-like” in appearance that is consistent with the description in the report [25] by using nail fold capillaroscope. However, nail fold capillaroscope (NFC) is a contact method and it can only provide two-dimensional (2D) images. The imaging penetration depth of NFC is also limited, so that only nail proximal region can be observed. Because abnormal microangiopathy of nail fold often occurs in systemic rheumatic diseases, especially in scleroderma, Raynaud's phenomenon and related conditions [26], UHR BD-SDOCT system allows non-contact, noninvasive, and fast analysis of the 3D capillary morphology and microcirculation of nailfold that may help to efficiently indicate the systemic inflammatory diseases.

Fig. 9 in vivo UHR BD-SDOCT imaging of a human nail fold. (a) Schematic diagram, (b) 3D reconstruction with 2D demonstration images showing X-Z, X-Y, and Y-Z planes. Image dimensions are $2 mm \times 2mm \times 1mm$ . G-gel; S-stratum corneum; E-Epidermis; D-Dermis; BV-Blood vessel; PCL-papillary capillary loops; BM- basement membrane; PD-papillary dermis; HD-hypodermis; SA-small artery; SV-small vein

Download Full Size | PDF

5. Discussion and conclusion

UHR OCT imaging performed by using supercontinuum laser source has been demonstrated based on BD-TDOCT setup. Up to 103 dB sensitivity in 25.4 Hz A-scan rate [7] or 98 dB sensitivity in 50 Hz A-scan rate [8] were reported respectively. However, in TDOCT setup, the faster mechanical scanning, the lower the system sensitivity, so that real time imaging is difficult to realize. Since no depth scanning is necessary in FDOCT, the data acquisition speed is much higher than in TDOCT. However, due to limited sweeping range of now available fast sweeping laser source, achieving UHR imaging (i.e. axial resolution below 3 μm) by using BD-SSOCT setup is difficult. P. Cimalla et. al [9] demonstrates that UHR SDOCT imaging using supercontinuum laser and standard single detection scheme, a sensitivity of 94 dB at an A-scan rate of about 12 kHz was reported.

In this study, the advantages of employing balanced dual detection in high speed UHR SDOCT, wherein the bandwidth of the source is greater to 200 nm and the exposure time is less than 25 $μs$ (corresponding to an A-scan rate of ~40 kHz), are presented using two separate spectrometer units. In order to achieve these improvements, careful calibration procedures in the two channels are required. This calibration step only needs to be performed once before the actual imaging. Our experiments clearly demonstrate that the proposed UHR BD-SDOCT yields a SNR improvement as compared to the standard UHR SDOCT using a single detection scheme in both transparent and highly scattering media. With the advances in high speed line scan CCD, much lower integration times are now easily attainable, which can lead to larger than 5 dB enhancement in the SNR compared to the unbalanced configuration. Thus, although the unbalanced configuration with the optimized reference-power alternative seems attractive in terms of performance and cost, we show that, for fast imaging, the BD configuration may be helpful. However, based on our current data, a ~40% residual noncanceled output of excess photon noise still existed. This may because the differences between the two spectrometers. Besides, this SNR improved performance is largely degraded along the imaging depth, since an optimum $π$ phase shift between the two individual spectrometers is hard to achieve for interferograms with high modulation frequencies (i.e. the interference signal originates from greater depths). As a result, further modifications of the signal processing, such as the addition of autocalibration method [27] for wavelength assignment of each spectral element for the two spectrometers, may be need to further maintain the accurate $π$ phase shift in each imaging depth.

In summary, the feasibility and necessity of using BD in high speed SDOCT was first experimentally confirmed in this study by showing its advantages in high speed SDOCT. First, BD provides a practical approach for suppressing dc background noise from the light source spectrum and the mutual interference of backscattered light (as indicated by “AC” in Fig. 6) between each interface in a multilayer surface. Even in a highly scattering medium, high speed BD-SDOCT system can also provide dc suppression and SNR enhancement. This is because BD is helpful for eliminating RIN and achieving shot noise limited detection. Second, BD aids efficient power usage, which is particularly suitable for use in limited power systems (e.g., for eye diagnosis), or for use in highly scattering tissue wherein the attenuation of power with depth is large. By applying the BD scheme to SDOCT, the required exposure time for integrating sufficient photons can be decreased, which is useful to enable ultrafast in vivo imaging with shot noise limited detection. This performance allows fast in vivo 3D tissue visualization (as demonstrated in Fig. 9) or the study of tissue dynamic property with high temporal resolution.

Acknowledgments

This work was supported by grants from the National Science Council (grant no. NSC 99-2112-M-010-002-MY3) of Taiwan.

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] [PubMed]

2. T. H. Ko, D. C. Adler, J. G. Fujimoto, D. Mamedov, V. Prokhorov, V. Shidlovski, and S. Yakubovich, “Ultrahigh resolution optical coherence tomography imaging with a broadband superluminescent diode light source,” Opt. Express 12(10), 2112–2119 (2004). [CrossRef] [PubMed]

3. B. E. Bouma, G. J. Tearney, I. P. Bilinsky, B. Golubovic, and J. G. Fujimoto, “Self-phase-modulated Kerr-lens mode-locked Cr:forsterite laser source for optical coherence tomography,” Opt. Lett. 21(22), 1839–1841 (1996). [CrossRef] [PubMed]

4. W. Drexler, U. Morgner, F. X. Kärtner, C. Pitris, S. A. Boppart, X. D. Li, E. P. Ippen, and J. G. Fujimoto, “In vivo ultrahigh-resolution optical coherence tomography,” Opt. Lett. 24(17), 1221–1223 (1999). [CrossRef] [PubMed]

5. R. A. Leitgeb, W. Drexler, A. Unterhuber, B. Hermann, T. Bajraszewski, T. Le, A. Stingl, and A. Fercher, “Ultrahigh resolution Fourier domain optical coherence tomography,” Opt. Express 12(10), 2156–2165 (2004). [CrossRef] [PubMed]

6. B. Povazay, K. Bizheva, A. Unterhuber, B. Hermann, H. Sattmann, A. F. Fercher, W. Drexler, A. Apolonski, W. J. Wadsworth, J. C. Knight, P. St. J. Russell, M. Vetterlein, and E. Scherzer, “Submicrometer axial resolution optical coherence tomography,” Opt. Lett. 27(20), 1800–1802 (2002). [CrossRef] [PubMed]

7. A. Aguirre, N. Nishizawa, J. Fujimoto, W. Seitz, M. Lederer, and D. Kopf, “Continuum generation in a novel photonic crystal fiber for ultrahigh resolution optical coherence tomography at 800 nm and 1300 nm,” Opt. Express 14(3), 1145–1160 (2006). [CrossRef] [PubMed]

8. F. Spöler, S. Kray, P. Grychtol, B. Hermes, J. Bornemann, M. Först, and H. Kurz, “Simultaneous dual-band ultra-high resolution optical coherence tomography,” Opt. Express 15(17), 10832–10841 (2007). [CrossRef] [PubMed]

9. P. Cimalla, J. Walther, M. Mehner, M. Cuevas, and E. Koch, “Simultaneous dual-band optical coherence tomography in the spectral domain for high resolution in vivo imaging,” Opt. Express 17(22), 19486–19500 (2009). [CrossRef] [PubMed]

10. W. Drexler, “Ultrahigh-resolution optical coherence tomography,” J. Biomed. Opt. 9(1), 47–74 (2004). [CrossRef] [PubMed]

11. L. Liu, J. A. Gardecki, S. K. Nadkarni, J. D. Toussaint, Y. Yagi, B. E. Bouma, and G. J. Tearney, “Imaging the subcellular structure of human coronary atherosclerosis using micro-optical coherence tomography,” Nat. Med. 17(8), 1010–1014 (2011). [CrossRef] [PubMed]

12. N. Nassif, B. Cense, B. H. Park, S. H. Yun, T. C. Chen, B. E. Bouma, G. J. Tearney, and J. F. de Boer, “In vivo human retinal imaging by ultrahigh-speed spectral domain optical coherence tomography,” Opt. Lett. 29(5), 480–482 (2004). [CrossRef] [PubMed]

13. S. R. Chinn, E. A. Swanson, and J. G. Fujimoto, “Optical coherence tomography using a frequency-tunable optical source,” Opt. Lett. 22(5), 340–342 (1997). [CrossRef] [PubMed]

14. M. A. Choma, M. V. Sarunic, C. H. Yang, and J. A. Izatt, “Sensitivity advantage of swept source and Fourier domain optical coherence tomography,” Opt. Express 11(18), 2183–2189 (2003). [CrossRef] [PubMed]

15. R. Huber, D. C. Adler, and J. G. Fujimoto, “Buffered Fourier domain mode locking: Unidirectional swept laser sources for optical coherence tomography imaging at 370,000 lines/s,” Opt. Lett. 31(20), 2975–2977 (2006). [CrossRef] [PubMed]

16. B. Potsaid, I. Gorczynska, V. J. Srinivasan, Y. Chen, J. Jiang, A. Cable, and J. G. Fujimoto, “Ultrahigh speed spectral / Fourier domain OCT ophthalmic imaging at 70,000 to 312,500 axial scans per second,” Opt. Express 16(19), 15149–15169 (2008). [CrossRef] [PubMed]

17. G. L. Abbas, V. W. S. Chan, and T. K. Yee, “A dual-detector optical heterodyne receiver for local oscillator noise suppression,” J. Lightwave Technol. 3(5), 1110–1122 (1985). [CrossRef]

18. A. M. Rollins and J. A. Izatt, “Optimal interferometer designs for optical coherence tomography,” Opt. Lett. 24(21), 1484–1486 (1999). [CrossRef] [PubMed]

19. K. Takada, “Noise in optical low-coherence reflectometry,” IEEE J. Quantum Electron. 34(7), 1098–1108 (1998). [CrossRef]

20. A. G. Podoleanu, “Unbalanced versus balanced operation in an optical coherence tomography system,” Appl. Opt. 39(1), 173–182 (2000). [CrossRef] [PubMed]

21. T. Mitsui, “Dynamic range of optical reflectometry with spectral interferometry,” Jpn. J. Appl. Phys. 38(Part 1, No. 10), 6133–6137 (1999). [CrossRef]

22. R. Leitgeb, C. K. Hitzenberger, and A. F. Fercher, “Performance of fourier domain vs. time domain optical coherence tomography,” Opt. Express 11(8), 889–894 (2003). [CrossRef] [PubMed]

23. E. G. Ötzinger, M. Pircher, and C. K. Hitzenberger, “High speed spectral domain polarization sensitive optical coherence tomography of the human retina,” Opt. Express 13, 10217–10229 (2005).

24. R. C. Haskell, D. Liao, A. E. Pivonka, T. L. Bell, B. R. Haberle, B. M. Hoeling, and D. C. Petersen, “Role of beat noise in limiting the sensitivity of optical coherence tomography,” J. Opt. Soc. Am. A 23(11), 2747–2755 (2006). [CrossRef] [PubMed]

25. M. Cutolo, A. Sulli, M. E. Secchi, S. Paolino, and C. Pizzorni, “Nailfold capillaroscopy is useful for the diagnosis and follow-up of autoimmune rheumatic diseases. A future tool for the analysis of microvascular heart involvement?” Rheumatology (Oxford) 45(Suppl 4), iv43–iv46 (2006). [CrossRef] [PubMed]

26. H. A. Moneib, S. A. M. Salem, D. G. Aly, H. T. M. Khedr, H. A. Wafaey, and H. E. Hassan, “Assessment of serum vascular endothelial growth factor and nail fold capillaroscopy changes in systemic lupus erythematosus with and without cutaneous manifestations,” J. Dermatol. 39(1), 52–57 (2012). [CrossRef] [PubMed]

27. M. Mujat, B. H. Park, B. Cense, T. C. Chen, and J. F. de Boer, “Autocalibration of spectral-domain optical coherence tomography spectrometers for in vivo quantitative retinal nerve fiber layer birefringence determination,” J. Biomed. Opt. 12(4), 041205 (2007). [CrossRef] [PubMed]

Balanced detection for spectral domain optical coherence tomography

Abstract

1. Introduction

2. Theory

3. Method

4. System performance

5. Discussion and conclusion

Acknowledgments

References and links

Cited By

Figures (9)

Equations (7)

Optics Express