Open Access
Add to BibSonomyAbstract
Although high optical illumination power is favored in optical coherence tomography (OCT) for better signal-to-noise ratio, optical power is often limited by a damaged threshold for biomedical living tissues and autocorrelation signals observed in tomograms. In order to improve signal sensitivity without increasing the optical illumination power, a spectrally sampled multi-wavelength light source is proposed for the OCT system. A fiber Sagnac comb filter was used to spectrally sample the output of a continuous spectral light source. Point spread function analysis shows that the spectrally sampled OCT has an almost 50% dynamic range improvement in comparison with a conventional continuous spectral light source OCT for the same average optical power of 6 mW.
©2008 Optical Society of America
1. Background and objective
Optical coherence tomography (OCT) is a very attractive biological imaging technique for obtaining noninvasive high-resolution images that use a low-coherence interferometer to acquire a micron-scale cross-sectional image [1,2]. Originally, time-domain (TD) OCT was proposed based on the optical interference signal variation of a broadband light source along a time-dependent path-length difference [3]. In recent years, Fourier domain (FD) OCT has been intensively developed as a prospective technique owing to its many advantages over conventional TD-OCT, such as higher acquisition speed without a mechanical delay line and a higher signal-to-noise ratio of over 20 dB [4–6]. FD-OCT can be also divided into swept-source (SS) OCT [7,8] and spectral domain (SD) OCT [9–11]. The former method uses a high-speed wavelength-swept laser source and a photo detector. The latter technique is built with a broadband light source and line-scan charge-coupled devices (CCDs) followed by high-performance spectrometers. There have been considerable developments on the broadband light source in both the TD-OCT and SD-OCT methods, whereas a wavelength-swept laser source is used for the SS-OCT.
For the OCT imaging technique, high detection sensitivity of the optical interference signal is required to provide sufficient penetration depth for the accurate diagnosis and quantitative evaluation of tissue properties. A greater level of optical power at the sample is expected to ensure improvement of detection sensitivity, but the clinical limitation concerning the maximum available level of exposure power is determined according to the American National Standards Institute (ANSI) laser safety standard for continuous exposure to image biological samples [12]. Since the OCT method uses a biomedical living tissue sample as its measurement target, unlike other optical measurement methods, there is a dilemma in that higher optical power is expected at the sample, but the maximum exposure power into the sample is limited. The power of light at the sample is also limited due to cross-correlation terms, which are the results of a cross-correlation of waves originating from a number of reflecting layers within the sample. Since an upper limit on useful exposures can be estimated to avoid parasitic cross-correlation terms in SD-OCT, it is necessary to find the optimal conditions of optical power and exposure time below the ANSI safety limit [13].
In order to solve these problems by increasing the signal sensitivity without raising the optical illumination power, a spectrally sampled multi-wavelength source is proposed. The suggested light source is cascaded with a fiber Sagnac comb filter, which is incorporated with a series of polarization-maintaining fibers (PMF), a fiber coupler, and polarization controllers (PC). We show by experiment that the proposed multi-wavelength source in a center wavelength of 1.3 µm using an all-fiber Sagnac loop interferometer provides an advanced dynamic range over the conventional continuous spectral source. A recent study on the optical frequency comb (OFC) source has demonstrated that the imaging depth problem caused by the finite number of CCD pixels for a SD-OCT can be overcome [14]. However, the conventional fiber Fabry-Perot interferometer (FPI) for an optical frequency comb still has a few critical drawbacks, including its relatively high cost, high insertion loss, and fixed channel spacing. In the proposed system, the fiber Sagnac comb filter shows superior performance properties, such as low price, low insertion loss, and tunable channel spacing [15,16].
2. Study design and methods
The proposed SD-OCT system configuration is shown in Fig. 1. The light source is a cascaded broadband light source of around 1.3 µm. A superluminescent diode (SLD) is connected by a semiconductor optical amplifier (SOA) with an equal spectrum on the SLD to amplify it. The light source at port 1 has a center wavelength at ~1305 nm, a full width at half-maximum (FWHM) of 35 nm, and a maximum output power of 15 mW. The employed SOA is driven by a 200 mA current source, providing a small signal gain of 20 dB at 1300 nm. The SOA has a saturation output power of 14 dBm and a polarization sensitivity of ~0.5 dB.
In this research, a fiber Sagnac comb filter based on a PMF is used to generate a multi-wavelength spectrum in region I and is cascaded with a broadband light source [15]. The comb channel spacing, Δλ, is inversely proportional to the length and birefringence of the PMF as follows:
where λ is the operating wavelength, Δneo is the birefringence, and L is the length of the PMF [17,18]. For the tuning of the comb channel space, a PC of λ/2 can be used additionally to rotate the fast axis angles of the adjacent PMF fiber segments with respect to the plane of the Sagnac loop and to each other [15]. When the effective length of the PMF is 30 m and the birefringence is 0.00038, each peak is separated by 0.145 nm in a multi-wavelength spectrum centered at 1.3 µm. An isolator is used to block the partial reflection light from the fiber Sagnac comb filter into the SOA. This fiber Sagnac comb filter is placed behind the isolator.
Fig. 1. Configuration of the SD-OCT system using a multi-wavelength source based on a fiber Sagnac comb filter. SLD, superluminescent diode; SOA, semiconductor optical amplifier; PMF, polarization-maintaining fiber; PC, polarization controller; TDC, tunable directional coupler; ND, neutral density filter; GM, galvanometer mirror.
For a comparison of OCT performance depending on optical source types such as an unfiltered continuous spectral light source and a filtered light source through the fiber Sagnac comb filter, port 1 is directly connected to the Mach-Zehnder interferometer of region II to implement the conventional SD-OCT and, alternatively, port 2 is used to compare it with the SD-OCT using a multi-wavelength light source. In region II, a tunable directional coupler, a 50/50 coupler, and circulators are used to build the Mach-Zehnder interferometer of the SD-OCT. According to the reflectivity of the sample, the optical power of the incident beam at the sample can be freely controlled through the tunable directional coupler. The sample path is scanned laterally by a galvano mirror and focused by an f=40 mm objective lens. The reflected beams from both arms of the sample and mirror are combined with a 50/50 coupler after passing through the circulators. The collimated beam, 8 mm in diameter, passes through a diffraction grating (d=1200 mm-1) and exposed to an InGasAs CCD (Ophir-Spiricon Inc.) of 648 pixels with an achromatic lens (f=150 mm) to detect the optical interference signal in the spectral domain. Because this CCD camera is limited in its ability to obtain the deep depth image of a sample owing to its limited pixel number and resolution, an optical spectrum analyzer (OSA: AQ6317B produced by ANDO Electronic. Co.) is also used in the position of a spectrometer to obtain most of the OCT experimental data in this research. There is a strong relationship between the data from the CCD and those from the OSA, in principle. The OSA shows much higher resolution and broader bandwidth in comparison with the CCD, but it requires a relatively longer acquisition time. The acquired data from the OSA are transferred in real time to a computer through a shielded general-purpose interface bus cable (GPIB produced by National Instruments). The maximum GPIB transfer rate is more than 7.2 Mb/s. The acquired data are processed and visualized by the customized software written with LabView®.
Fig. 2. Experimental measurements of the spectra of (a) the continuous spectral source of 11 mW, (b) the multi-wavelength source of 6 mW, and (c) the continuous spectral source of 6 mW. The inset shows that the spectra shown for source (a) is almost equal to the peak envelope of source (b) when measured within the narrow span band of 5 nm.
The spectra of each source are presented with a mid-resolution mode measurement of 0.1 nm of the OSA in Fig. 2. Three source types, (a) continuous spectral source of 11 mW, (b) multi-wavelength source of 6 mW, and (c) continuous spectral source of 6 mW, are applied to the Michelson interferometer of the SD-OCT. First, the high intensity source of (a) is measured from port 1 and the output port is fusion-spliced to the fiber Sagnac comb filter to measure the multi-wavelength spectrum of (b) at port 2, as shown in Fig. 1. Due to the partially reflected ratio of the Sagnac interferometer, the optical power at port 1 of 11 mW is reduced to 6 mW at port 2, as shown in Fig. 1. It appears that the envelope of the multi-wavelength spectrum is distorted from the original continuous spectral shape. Using the wide span band of the 50 nm mode of the OSA, all of the multi-peaks with 0.145 nm spacing cannot be distinguished with a clear, 100% contrast. However, as shown in the inset, when measured for the narrow-band span of 5 nm, the peak envelope of the (b) source is found to be almost equal with the spectra of the (a) source, and the contrast of multi-peaks is almost 100%.
For the purpose of comparison, the third spectrum (c) of Fig. 2 is also measured at port 1 when the optical intensity of the continuous spectral source is reduced to 6 mW, which is the same as the average power of source (b). When comparing spectra (b) and (c) of Fig. 2, both sources have the same value of optical output power, i.e. 6 mW, but the signal envelope of spectrum (b) is much higher than that of spectrum (c). These combinations of various types of light sources can be extended for a suitable experimental comparison for an effective SD-OCT imaging system.
Fig. 3. Experimental measurements of the interference fringe spectra of the OCT at an optical path length difference of 150 µm between two mirrors using (a) the continuous spectral source of 11 mW and (b) the multi-wavelength source of 6 mW.
With the two different light sources of spectra (a) and (b), the interference fringe is measured with the SD-OCT configuration with a single mirror reflection at the sample location. Figures 3(a) and 3(b) show the interference fringe spectra at the optical path length difference of 150 µm using the source spectra (a) and (b) of Fig. 2, respectively. The additional fine interference pattern of Fig. 3(b) is induced from the multiple interactions between many channel peaks owing to the fiber Sagnac comb filter. These phenomena will be discussed in detail in Section 3. When we simply compare the spectra (a) and (b) of Fig. 3, it is observed that the two sources have similar signal sensitivities above the noise level because of the same envelope spectra. From the experimental results, we can preliminarily determine that there is no great difference between the SNRs of the fringe spectra for both sources in order to obtain the optical path length difference information. This is a noticeable result because the optical output power of the spectrum (a) source is almost twice that of the spectrum (b) source. Noisy deformation patterns are unexpectedly monitored, especially in Fig. 3(a). In general OCT studies, a smooth cosine-like interference signal is expected for a single-surface sample. We believe that this spiky noise spectrum is due to the surrounding noises, such as external perturbation and periodic vibration, during a slow scanning time of 360 ms for the OSA to detect over 50 nm.
Fig. 4. Point spread functions at various optical length differences for (a) the continuous spectral source of 11 mW, (b) the multi-wavelength source of 6 mW, and (c) the continuous spectral source of 6 mW.
In order to analyze the changes in dynamic range from the three types of optical sources, the point spread functions (PSFs) for variable lengths of the reference arm are presented in Fig. 4. Upon increasing the optical path length difference between the two mirrors up to 2 mm, the interference fringe is experimentally measured with the SD-OCT configuration at each sample location. The converted PSFs show the exact value of the optical path length differences between the two mirrors. For the reference of the highest signal peak at the optical length of 250 µm of source (a), the normalized amplitude relations are simultaneously plotted for the three types of optical sources along the optical length up to 2 mm. Since our system uses a tunable directional coupler with an arbitrary optimal splitting ratio and an OSA with a large limit of optical saturation power, it is not easy to define the overall sensitivity value of this SD-OCT system. However, the dynamic ranges of the logarithm-scaled PSFs at the optical length of 250 µm are monitored as ~32 dB, 30 dB, and 28 dB for this OCT system of the three types of optical sources, (a), (b), and (c), respectively.
Comparing the dynamic ranges of source (a) and source (b), the amplitudes of the PSFs from source (b) at various depths are almost 70% of those from source (a). When we recall that the optical power of source (b) is almost 55% of that of source (a), it can be concluded that the multi-wavelength source has an advantage in improving the dynamic range compared with the continuous spectral source. This conclusion can be proved also by examining the other comparison made between sources (b) and (c). As shown in Fig. 4, the amplitude of the PSFs from source (b) is almost 50% higher than those from source (c), in spite of the fact that these two sources have the same average value of optical power. Therefore, this result leads to a similar conclusion regarding the advantage of the multi-wavelength source in improving the dynamic range of the OCT system.
3. Theoretical analysis
It can be concluded from the experimental results of Fig. 4 that a spectrally-sampled multi-wavelength source is helpful in improving dynamic range compared with that of a continuous spectral source. However, from the qualitative point of view, it would be meaningful to know why the amplitudes of the PSFs from source (b) are no higher than 70% of those from source (a), and why the amplitude of the PSFs from source (b) are just 50% higher than those from source (c). This phenomenon can be explained by the power loss of the fast Fourier transform (FFT) signal of the fiber Sagnac comb filter. We will start this explanation and theoretical analysis of a power budget from the simplification of two cascaded interferometers, namely the Sagnac and Mach-Zehnder types.
The block diagram of Fig. 5 shows the simplified total signal flow of the experimental setup with the Sagnac loop interferometer for comb filtering and the Mach-Zehnder interferometer for OCT, respectively.
In region I of Fig. 1, the output signal of the proposed fiber Sagnac comb filter is constructed by the interference relationship between a clockwise (CW) direction wave and a counterclockwise (CCW) wave. The input wave is split at the coupler, which is assumed to be lossless. The power coupling ratio at the coupler does not have a perfect 50/50 ratio over the broadband source. Therefore, the output signal has not only DC components but also AC components. In that case, the interference signal, like the interference signal of a general interferometer, can be defined as
where p CW and p CCW represent the optical power rotating in a CW and a CCW direction within the Sagnac loop, respectively, while k is the wave number. The interference term of the output signal has the wavelength spacing of the peaks, which is defined by both the birefringence and the length of the PMF within the Sagnac loop [15,17]. We consider here the well-balanced case, for simplicity, where α is defined by the product relationship between a CW and a CCW direction. Therefore, the output signal can be simply represented by
In region II, the output signal represents the general OCT interference signal. For the well-balanced case between the reference and the sample arm with a mirror, the power illuminated from the reference arm and the sample arm are the same and the reflectivity from the two arms are also the same. In this case, the following equation can be similarly given as
where β is defined by the product relationship between the reference and the sample arm signal, while ΔL is the optical path length difference between the reference and the sample arm.
From Eqs. (3) and (4), the final output signal can be simply expressed as
where s(k) is the spectral property of the broadband source.
Base on the modulation theory of a high frequency carrier, the discrete Fourier transform (DFT) of Eq. (5) yielded that an SD-OCT signal is given by
where ζSag and ζOCT, which are the center frequencies of the fiber Sagnac comb filter and the OCT interferometer, respectively. Their inverse values correspond to 2/Δn eo L and 1/ΔL, respectively. The symbol * represents the operator of convolution.
Since there are two peaks corresponding to the positive and negative frequency components, we select the only positive frequency term in Eq. (6).
We assume here that the value ζSag is larger than that of ζOCT. The first term is the DC signal of the optical source, the second term is the OCT interferometer signal (ζOCT), and the third term is the fiber Sagnac interferometer signal (ζSag). The fourth and fifth terms are the OCT interferometer signals carried by the fiber Sagnac interferometer frequency (ζSag).
In order to analyze experimentally the high-frequency modulation effect of the fiber Sagnac interferometer as shown in the above theoretical analysis, we repeat the measurement with the high resolution mode of the OSA. The maximal imaging depth range is inversely proportional to the spectrometer resolution [19]. The conversion of the OCT fringe spectra between two mirrors up to 2 mm difference is repeated as used already for plotting Fig. 4, but we can obtain additional signals for the optical length range between 2 mm and 8 mm to analyze a deeper imaging range. Figure 6 shows the PSFs for (a) the continuous spectral source of 11 mW and (b) the multi-wavelength source of 6 mW, respectively. Upon increasing the imaging depth, as in Fig. 6(a), there is no peak except in the range of (i). This means that most of the distributed spectral power is used for the OCT images only. Unlike Fig. 4(b), the high-frequency terms are generated in Fig. 6(b) for the image depth range of 4 mm to 7.5 mm, such as (ii), (iii), and (iv).
In comparing Eq. (7) with Fig. 6(b), the second term of the OCT interferometer signal (ζOCT) corresponds to the range of (i), and the third term of the fiber Sagnac interferometer signal (ζ Sag) is in the range of (ii). The fourth term of the OCT interferometer signals carried by the higher fiber Sagnac interferometer frequency (ζSag + ζOCT) is in the range of (iv). Similarly, the fifth term of the OCT interferometer signals carried by the lower fiber Sagnac interferometer frequency (ζSag - ζOCT) is in the range of (iii). For a clear division between ranges (i) and (iii), we have increased the difference of an optical path only up to 2 mm, which is selected to be the maximum depth range in this experiment. If the difference of an optical path is more than 3 mm in Fig. 6(b), the range of (i) will move to the right on the optical length scale in Fig. 6(b) and the range of (iii) will expanded to the left, which may cause overlapped spectra.
From this power distribution analysis, it can be proved that the high-frequency terms resulting from the fiber Sagnac comb filter cause a slight sensitivity degradation of the OCT imaging signal (b) in Fig. 4. Since the power loss due to the fiber Sagnac comb filter is relatively smaller than the signal power of the OCT interferometer, we can also clearly know that the amplitudes of the PSFs from multi-wavelength light source (b) are higher than those from continuous spectral source (c) for the same optical power sources, as shown in Fig. 4.
Fig. 6. Point spread functions from high-resolution measurements for (a) the continuous spectral source of 11 mW and (b) the multi-wavelength source of 6 mW.
4. Experiment result
We made images of various samples to demonstrate the improvement effect of the dynamic range by reducing the spectral density of the optical source. Using the experimental setup of the SD-OCT, we can image a typical depth-resolving sample, which consists of four slide glasses with a thickness of 150 µm, for the three types of input sources in Fig. 2. For the optical source with 11 mW, the optical power of the incident beam in the sample was measured to be 3.4 mW, which is determined by the power-splitting ratio of the tunable directional coupler and the insertion losses of the in-path components. For a demonstration of imaging depth, cross-sectional images of up to three or four slide glasses are preferred. Thus, optimal illumination of optical power is controlled by replacing high-attenuation films with various types of opaqueness acetate polymer, which have the thickness of 60 µm and the refractive index of ~1.34 at a 1.3 µm wavelength range.
The required time to detect the interference signal using the OSA is 620 ms in the high-resolution mode of 0.05 nm. Though this OSA spectrometer has a very slow acquisition time, it is meaningful to analyze quantitatively the influence of the multi-wavelength source on the OCT imaging performance. However, for practical purposes, the speed of the scan can be greatly increased by using a generally-used CCD spectrometer. A series of images of 1024 depth points was continuously acquired. The image was composed of 512 axial x 256 transverse pixels. At small depths below 0.5 mm, the free-space axial resolution was experimentally measured to be approximately 24 µm, which is close to the 21 µm calculated from the source bandwidth. The mismatch of the axial resolution between the measured value and the theoretical value was caused by a calibration error of the frequency linearity, the non-Gaussian spectrum of the optical source, and the dispersion mismatch in the sample and reference arm.
Fig. 7. OCT imaging of the four slide glasses under attenuation film for (a) the continuous spectral source of 11 mW, (b) the multi-wavelength source of 6 mW, and (c) the continuous spectral source of 6 mW, respectively. The scale bar represents the relative log-scaled intensity.
As shown in Fig. 7, a few air gaps can be distinguished between the slide glasses within the tomograms. The transmission ratio of the upper attenuation film is carefully selected to show effectively the critical difference in maximum image depth by changing the type of optical input sources. It is clear that Fig. 7(a), using a continuous spectral source of 11 mW, shows all four slide glasses clearly. However, in the case of the proposed multi-wavelength source with 6 mW, the upper layer of the fourth slide glass is imaged in Fig. 7(b) but its lower layer is not detected, which means that only three slide glasses are clearly distinguished. In Fig. 7(c), using the continuous spectral source of 6 mW, both the upper and lower layers of the fourth slide glass are no longer detected and only three glasses can be distinguished. These results were already expected from the 2 dB differences of dynamic range among the three groups of logarithm-scaled PSF data of Fig. 4.
Fig. 8. OCT imaging of the an ex-vivo rat eye using (a) the multi-wavelength source of 6 mW and (b) the continuous spectral source of 6 mW, respectively. The scale bar represents the relative log-scaled intensity.
The next experiment of OCT imaging is shown with an ex vivo rat eye as shown in Fig. 8. The two images are obtained by using two optical sources, a multi-wavelength source of 6 mW and a continuous spectral source of 6 mW, respectively. As shown in both images, the cornea, iris, and lens can be distinguished clearly. While this sample is not suitable for counting the multiple depth layers in the deeper region, unlike Fig. 7, we can determine that the background noise density in the upper region of Fig. 8(a) is relatively smaller than that of Fig. 8(b). Since both images are separately normalized in a logarithm scale for the reference of the highest signal peak in each data sequence, the smaller region of the same noise level means that this image [Fig. 8 (a)] is obtained from the relatively higher amplitude of the PSF signal. These results are also supported from the single mirror PSF data of Fig. 4 and the four slide glass images of Fig. 7. It means that, for the same optical power of a conventional continuous spectral light source, the multi-wavelength light source generates a higher dynamic range signal for better OCT imaging with deeper depth region (Fig. 7) and clearance to noise (Fig. 8). Though the 2 dB difference of dynamic range may not seriously improve most OCT images in a logarithm scale, it will still be meaningful to maximize the sensitivity of the OCT system from limited power by adding a low-cost fiber Sagnac comb filter only.
5. Conclusion
We propose and demonstrate a novel spacing-tunable multi-wavelength source to achieve both a lower total exposure intensity and higher detection sensitivity simultaneously for a SD-OCT. The spectrally sampled light source is built with a simple fiber Sagnac comb filter based on a flexible length of the PMF. We experimentally demonstrate the imaging comparison of the conventional SD-OCT system using three types of sources: a continuous spectral source of 11 mW, a multi-wavelength source of 6 mW, and a continuous spectral source of 6 mW. We show by experiment and analysis that a uniform and stable multi-wavelength spectral distribution provides a superior dynamic range compared with the conventional continuous spectral light source for the same averaged optical power.
Acknowledgments
This work was partly supported by the Basic Research Program of the Korea Science, Engineering Foundation (KOSEF) funded by the Korean government (MOEST) R01-2007-000-20553-0, and by the Ministry of Knowledge and Economy of Korea through the Ultrashort Quantum Beam Facility Program.
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, 1178–1181 (1991). [CrossRef] [PubMed]
2. A. F. Fercher, W. Drexler, C. K. Hitzenberger, and T. Lasser, “Optical coherence tomography—principles and applications,” Rep. Prog. Phys. 66, 239–303 (2003). [CrossRef]
3. G. J. Tearney, B. E. Bouma, and J. G. Fujimoto, “High-speed phase- and group-delay scanning with a grating-based phase control delay line,” Opt. Lett. 22, 1811–1813 (1997). [CrossRef]
4. R. Leitgeb, C. K. Hitzenberger, and A. F. Fercher, “Performance of Fourier domain vs. time domain optical coherence tomography,” Opt. Express 11, 889–894 (2003). [CrossRef] [PubMed]
5. J. F. de Boer, B. Cense, B. H. Park, M. C. Pierce, G. J. Tearney, and B. E. Bouma, “Improved signal-to-noise ratio in spectral-domain compared with time-domain optical coherence tomography,” Opt. Lett. 28, 2067–2069 (2003). [CrossRef] [PubMed]
6. M. A. Choma, M. V. Sarunic, C. Yang, and J. Izatt, “Sensitivity advantage of swept source and Fourier domain optical coherence tomography,” Opt. Express 11, 2183–2189 (2003). [CrossRef] [PubMed]
7. S. H. Yun, C. Boudoux, G. J. Tearney, and B. E. Bouma, “High-speed wavelength-swept semiconductor laser with a polygon-scanner-based wavelength filter,” Opt. Lett. 28, 1981–1983 (2003). [CrossRef] [PubMed]
8. S. W. Lee, C. S. Kim, and B. M. Kim, “External-line cavity wavelength-swept source at 850 nm for optical coherence tomography,” IEEE Photon. Technol. Lett. 19, 176–178 (2007). [CrossRef]
9. 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, 480–482 (2004). [CrossRef] [PubMed]
10. M. Wojtkowski, R. Leitgeb, A. Kowalczyk, T. Bajraszewski, and A. F. Fercher, “In vivo human retinal imaging by Fourier domain optical coherence tomography,” J. Biomed. Opt. 7, 457–463 (2002). [CrossRef] [PubMed]
11. R. A. Leitgeb, W. Drexler, A. Unterhuber, B. Hermann, T. Bajraszewski, T. Le, A. Stingl, and A. F. Fercher, “Ultrahigh resolution Fourier domain optical coherence tomography,” Opt. Express 12, 2156–2165 (2004). [CrossRef] [PubMed]
12. American National Standards Institute, “American National Standard for Safe Use of Lasers,” ANSI Z 136-1 (2000).
13. A. Szkulmowska, M. Wojtkowski, I. Gorczynska, T. Bajraszewski, M. Szkulmowski, P. Targowski, A. Kowalczyk, and J. J. Kaluzny, “Coherent noise-free ophthalmic imaging by spectral optical coherence tomography,” J. Phys. D: Appl. Phys. 38, 2606–2611 (2005). [CrossRef]
14. T. Bajraszewski, M. Wojtkowski, M. Szkulmowski, A. Szkulmowska, R. Huber, and A. Kowalczyk, “Improved spectral optical coherence tomography using optical frequency comb,” Opt. Express 16, 4163–4176 (2008). [CrossRef] [PubMed]
15. C. S. Kim and Jin U. Kang, “Multi-wavelength switching of Raman fiber ring laser incorporating composite PMF Lyot-Sagnac filter,” Appl. Opt. 43, 3151–3157 (2004). [CrossRef] [PubMed]
16. E. J. Jung, J.-S. Park, M. Y. Jeong, C. S. Kim, T. J. Eom, V. A. Tougbaev, B.-A. Yu, W. Shin, D.-K. Ko, and J.- H. Lee, “Multi-wavelength source for the lower exposure intensity of spectral OCT,” Proc. SPIE 6849–10, (2008).
17. C. S. Kim, B. Choi, J. S. Nelson, Q. Li, P. Z. Dashti, and H. P. Lee, “Compensation of polarization-dependent loss in transmission fiber gratings by use of a Sagnac loop interferometer,” Opt. Lett. 30, 20–22 (2005). [CrossRef] [PubMed]
18. A. Yu and A. S. Siddiqui, “Optical modulators using fibre optic Sagnac interferometers,” IEE Proc. Optoelectron. 141, 1–7 (1994). [CrossRef]
19. G. Hausler and M. W. Lindner, “Coherence radar and spectral radar—new tools for dermatological diagnosis,” J. Biomed. Opt. 3, 21–31 (1998). [CrossRef]
