1. Introduction

Optical coherence tomography (OCT) is one of the most rapidly advancing medical imaging modalities that can provide noninvasive, high-resolution cross-sectional images and 3-D volumetric images of biological tissues in vivo [1, 2]. Recently, Fourier-domain OCT (FD-OCT), which involves the use of either a rapidly tuned wavelength-swept laser (swept-source OCT, FD/SS-OCT) or a spectrometer (spectral-domain OCT, FD/SD-OCT), has attracted considerable attention owing to its dramatically improved sensitivity and higher imaging speed than those of a conventional time-domain system [2–5]. Each technique has been reported to have several advantages over the other. It is well known that FD/SS-OCT can provide an increased imaging depth of over 5 mm, a lower sensitivity roll-off, and a higher acquisition speed of a few hundred kilohertz or even up to a few megahertz because of the development of narrowband optical frequency-swept lasers with broad tuning ranges based on the Fourier-domain mode-locking (FDML) method or a high-speed polygon scanner [6–8]. FD/SD-OCT can easily achieve a higher axial resolution of below 3 μm in tissues because large-bandwidth light sources (e.g., a superluminescent diode and a Ti:sapphire laser) in the 800- or 1000-nm region are commercially available [9–12]. In addition, FD/SD-OCT can obtain cross-sectional images at high speeds with an acquisition speed of up to 312 kHz by using a subset of the total number of camera pixels [13]. Recently, FD/SD-OCT with a 500 kHz line rate has been presented using interlaced detection based on two high-speed cameras [14].

OCT is a powerful diagnostic tool in several areas of medicine such as ophthalmology, intravascular cardiology, skin, dentistry, bronchology, and so on. In particular, the diagnosis and treatment monitoring of retinal diseases such as retinal pigment epithelium detachment, retinal layer thickness changes, macular degeneration, glaucoma, macular edema, and diabetic retinopathy have been the major driving force, as no other technology can compete with the OCT technology because of the impossibility of taking a biopsy of the retina [15–19]. In addition, it has been extensively used for examining the anterior segment of the human eye, including diagnoses of corneal disorders such as keratoconus, scarring, and dystrophy [20–22]. Anterior segment imaging using OCT is also useful for monitoring the risk of angle-closure glaucoma [23].

Even with the remarkable progress made recently in OCT technologies, it is still difficult to examine the shape and dimensions of segments of the entire eye for diagnoses of myopia, presbyopia, and cataract [24]. A light beam cannot be simultaneously focused on both the anterior segment and the retina because of the refractive power of the former. In addition, the imaging depth is not large enough to cover the entire eye even if a complex conjugate removal technique is applied. Therefore, some physicians obtain images of the anterior segment and the retina using separate commercial ophthalmic OCT systems (e.g., Visante and Cirrus OCT of Carl Zeiss Meditec). Others use a switchable OCT system (e.g., Spectralis of Heidelberg Engineering), which enables a physician to choose imaging areas by introducing additional optics. It is expensive to use two different OCT systems for anterior segment and retina imaging. On the other hand, with a switchable OCT system, it is necessary to add or remove optics from the sample arm, adjust the length of the reference arm, and re-compensate for any mismatched dispersion between the sample and the reference arms. Recently, OCT systems that can simultaneously image both the retina and the anterior segment have been introduced [24–27]. Dai et al. demonstrated the possibility of simultaneous, dual depth imaging by FD/SD-OCT using two beams, two interferometers, and two spectrometers [24]. Dhalla et al. introduced FD/SS-OCT using coherence revival and polarization encoding techniques for a similar purpose [26].

In this paper, we present a novel ophthalmic FD/SD-OCT system that employs dual-illumination with two orthogonal polarizations in the sample arm and interlaced detection using a single spectrometer and an ultrafast 1 × 2 optical switch. For simultaneous imaging of the human anterior segment and retina, the A-scan rate is reduced to 35,000 Hz, which is half the full line rate of a CCD camera. This optical switch can be set to deliver light from only the anterior segment or the retina at the full A-scan rate of 70,000 Hz. We successfully applied a similar optical switch to polarization-sensitive FD/SD-OCT for the interlaced detection of two orthogonally polarized lights in the detector arm [28]. In addition, the anterior segment was imaged using the off-pivot complex conjugate removal technique, and the retina was imaged at a conventional imaging depth [29–31]. This system was successfully applied to image the human eye in vivo, the results of which are presented.

2. Experimental setup

2.1 System configuration

A schematic of our ophthalmic FD/SD-OCT system with dual illumination and interlaced detection is depicted in Fig. 1 . It has been modified from our previous version, which used beam splitters instead of polarization beam splitters in the sample arm [25]. We used a broadband superluminescent diode (SLD; Superlum Ltd., Ireland) with a center wavelength of 830 nm, full width at half maximum (FWHM, Δλ) of 64 nm, and total output power of 25 mW, which can create a theoretical axial resolution of 4.7 μm in air, corresponding to 3.5 μm in tissues (n = 1.35). The beam from the SLD is coupled to two fiber-based Michelson interferometers (80:20 fiber couplers; FC2, FC3; SLT, Inc., Korea) via a 50:50 fiber coupler (FC1; Thorlabs, Inc., USA). A fiber optic isolator (Thorlabs, Inc., USA) is employed between the SLD and the FC1 to exclude back reflection. Each interferometer has its own reference and sample arms. The light in each interferometer is split between the sample (20%) and the reference (80%) arms. These two interferometers are built to have two different optical path lengths, so that a sample can be imaged at any two different axial locations within the axial eye length (AEL ≈25 mm). The two beams in the sample arms are combined using a polarization beam splitter (PBS1), from which two orthogonally polarized lights proceed to the sample via different paths, as shown in Fig. 1. The horizontal component (p-pol.) is used to image the retina, and the vertical component (s-pol.) is used to image the anterior segment. The horizontally polarized light is collimated by passing through the L1 and L3 lenses and is focused into the retina. The vertically polarized light travels a different optical path, as reflected by PBS2, and is guided by two mirrors to PBS3. An additional lens (L2) is inserted into the vertical path and, along with L2, L3 focuses the vertically polarized light into the cornea. L1, L2, and L3 have the same focal length of 75 mm. In the reference arm, we employ a similar strategy utilizing three PBS-cubes to match the polarization states and the dispersion between the reference and the sample arms. In addition, focusing achromatic lenses and a dispersion compensation unit (BK7 prisms pairs; Edmund Optics, USA) are used in front of the reference mirrors to minimize dispersion. The prism pairs are used only in the reference arm for retina imaging. The polarized lights reflected from the sample and reference arms are recombined to generate spectral interference at FC2 and FC3. Two spectral interference signals are then delivered to the 1 × 2 optical switch (Boston Applied Technologies Inc., USA), which can be set to transmit them into a spectrometer in turns. This optical switch, which is made of an electro-optic (EO) material, is a solid-state switch operating in free-space architecture without any moving parts. This optical switch operates in the region of 830 ± 50 nm at up to 1 MHz.

 

Fig. 1 Schematic of dual imaging SD-OCT system at 830 nm designed to acquire in vivo images of the anterior segment and the retina of the eye simultaneously. FC: fiber coupler, PC: polarization controller, CL: collimation lens, PBS: polarization beam splitter, NDF: neutral density filter, DC: dispersion compensation unit, L1–L3: lens.

Download Full Size | PDF

The spectrometer comprises a transmission diffraction grating (1,800 lines/mm; Wasatch Photonics, USA), an achromatic doublet lens (f = 75 mm; Thorlabs, USA), and a line-scan camera (spL2048-140k; Basler AG, Germany). The line-scan camera, which is of the complementary metal-oxide semiconductor (CMOS) type, has 2048 pixels, with a maximum line rate of 70 kHz. The imaging depth range without complex artifact removal was designed to be approximately 3.3 mm in air, which was calculated from the spectral resolution (δλ = 0.052 nm) of the spectrometer.

To synchronize all of the devices (two-dimensional galvanometer, optical switch, and camera) of the system, an analog-output board with 4 channels (PCI-6711; National Instruments, USA) is used. As shown in Fig. 2 , synchronized signals are generated for the fast-axis galvanometer (triangular, red), line-scan camera (TTL, blue), and optical switch (TTL, green) [28]. These three signals are triggered by the internal clock of an analog-output board and an acquisition start command in software. The optical switch can be operated in two different imaging modes: a simultaneous imaging mode (both the anterior segment and the retina are imaged simultaneously) and a single imaging mode (only the anterior segment or the retina is imaged). These modes can be chosen as desired by a simple click on the graphical user interface (GUI) that we developed. When the channel for the optical switch generates 5.0 V, spectral interference from the anterior segment is transmitted to the spectrometer. On the other hand, when the channel for the optical switch generates 0.0 V, the spectrometer detects interference signals from the retina. To simultaneously obtain images of both the anterior segment and the retina, two TTL signals of 70 kHz and 35 kHz are supplied to the line triggers of the camera and the optical switch, respectively, as shown in Figs. 2(a) and 2(b). Although the maximum exposure time is 12.9 μs at a camera speed of 70 kHz, the exposure time was set to be 12 μs because the rising and falling time of TTL signals for the camera are slightly faster than those of TTL signals for the optical switch, as shown in Fig. 2 (b). On the other hand, to obtain only the anterior segment image, a fixed voltage of 5 V, as shown in Fig. 2(c), is supplied to the optical switch. In addition, Fig. 2(d) shows a fixed voltage of 0 V being supplied to the optical switch to obtain only the retinal image. Each A-line acquisition occurs at the rising edge of the camera signal. The acquisition speed in the simultaneous imaging mode is 35 kHz, which is half the maximum speed of the camera; the full imaging acquisition speed of 70 kHz can be achieved during the single imaging mode.

 

Fig. 2 Synchronized signals for line-scan camera (blue), optical switch (green), and fast-axis galvanometer (red). (a) Simultaneous imaging mode; (b) magnified signals for the camera and optical switch in simultaneous imaging mode, where each interference signal from the anterior segment and the retina is transferred to the camera by the optical switch for interlaced detection; (c) single imaging mode for the anterior segment; and (d) single imaging mode for the retina.

Download Full Size | PDF

In imaging the anterior segment, we adopted the off-pivot complex conjugate removal algorithm [29–31] to increase the imaging depth to over 5 mm. The 2D galvanometer in the sample arm was mounted on an x-y-z translation stage to adjust the position of the light beam on the fast-axis mirror and thus generate a uniform phase shift between adjacent A-scans, which results in a constant carrier frequency in the spatial domain. Finally, we could achieve an imaging depth of around 6 mm for the anterior segment.

The zero-delay position and the focal point for the anterior segment were set to the middle plane between the cornea and the top surface of the iris so that we could visualize both structures. The zero-delay position for the retina images was located just in front of the nerve fiber layer.

2.2 Signal processing

Data sets from each interferometer detected by the camera were digitized using a frame grabber (PCIe−1429; National Instruments Inc., USA) with 12-bit resolution and transferred to a computer for signal processing. In constructing the anterior segment image, a complex conjugate removal algorithm based on the Hilbert transform was used to suppress the mirror artifacts and to double the imaging depth [29–31]. In brief, the acquired 2D spectral data from the off-pivot configuration were transformed to a complex spectral interferogram using the Hilbert transform along the x-direction for each wavelength, after removing the fixed pattern noise and DC term by subtracting the averaged spectral data. Then, it is followed by conventional FD/SD-OCT data processing. Wavelength calibration between the camera pixel and the wavelength domains was carried out using several fiber Bragg gratings (FBGs) [32]. For retinal images, the conventional FD/SD-OCT process was performed without the complex conjugate removal algorithm because retinal tomograms do not need to be in the full-range domain. Finally, a Wiener filter and a median filter with 3 × 3 matrices were applied to reduce the white and speckle noises in all images.

3. Results

3.1 System performance

To evaluate the performance of the proposed FD/SD-OCT, we measured the sensitivity and axial resolution as a function of depth. The optical powers in the sample arm were set to be ~0.68 mW for retina imaging and ~1.0 mW for anterior segment imaging, such that the total power in front of the cornea was ~1.68 mW, which meets the safety requirements set by the American National Standards Institute (ANSI) Z136.1 limits [33]. These powers could be individually adjusted in any of the sample arms using polarization controllers (PCs). We also controlled the PCs in the reference arms to generate and obtain the maximized interference signals. A gold-coated reflecting mirror and a 20-dB neutral density (ND) filter were placed in the sample arm. The ND filter was inserted to reduce the amount of light reflected from the mirror. To measure the sensitivity and axial resolution in the retinal path, we added an additional doublet lens (f = 30 mm) as an eye model.

Figure 3 shows the measured sensitivities in the retinal (Fig. 3(a)) and anterior segment imaging paths (Fig. 3(b)) as a function of depth. The measured depth range (~20 dB) for retina images is approximately 3.0 mm. On the other hand, the imaging depth for the anterior segment was doubled, so that a larger portion of the anterior segment, including the entire cornea, could be accommodated. A gold-coated mirror placed as a sample was moved from 120 μm to 3.0 mm away from the zero-depth location in 400-μm intervals. The results were added to the attenuation constant of 40 dB (double the attenuation value of the ND filter because of the double path configuration). The highest sensitivities in the anterior segment and retinal paths were measured to be approximately 91.5 dB and 89.5 dB, respectively, at a depth of 120 μm when the exposure time of the camera was set to be 12 μs. The theoretical sensitivities of our OCT system were calculated to be approximately 102.7 dB and 101.3 dB in the anterior segment path and the retina path, respectively. These numbers and the corresponding difference of 1.67 dB were calculated based on the power difference between the two paths. The discrepancy between the calculated and the measured sensitivity values is approximately 11 dB. We estimated that the sum of the electrical noise in the camera and the relative intensity noise (RIN) of the source spectrum was approximately 3 dB. In addition, losses in the optical components in the sample paths, the fiber couplers themselves, and the coupling between a fiber coupler (or optical switch) and free space optics was approximately 7 dB. The optical loss of two imaging paths via hardware can be considered to be similar. We could observe that the sensitivities in both imaging paths decreased down to ~10 dB within a depth range of ~1.8 mm. The R-numbers of the sensitivity roll-off for both paths were similar, being approximately 5.55 dB/mm, because we used the same spectrometer. These values are similar to those reported previously for FD/SD-OCT systems in other studies [13, 14, 34, 35].

 

Fig. 3 Sensitivity curves of retinal (a) and anterior segment imaging paths (b) as a function of depth. The anterior segment imaging depth was increased by the complex conjugate removal algorithm and was double the retinal imaging depth.

Download Full Size | PDF

Figures 4(a) and 4(b) show the point spread function for the axial resolution and the changes in the axial resolution as a function of depth, respectively. The axial resolution was measured to be ~6.7 μm in air, which corresponds to ~4.9 μm in tissues (n = 1.35) at an imaging depth position of 120 μm from the zero-depth location. The measured axial resolution was higher than the theoretical value because the Nyquist sampling theorem between the image depth and the spectrum full bandwidth (λ_full) detected by the spectrometer was not satisfied. The interval between the image pixels (δz = 2⋅z_max/N, N: total pixel number of camera) in the axial direction of an OCT image should be smaller than half the theoretical axial resolution to satisfy the Nyquist sampling theorem [12, 36]. Therefore, λ_full detected by the spectrometer should be defined as λ_full ≥ π⋅Δλ/(2⋅ln2) = 2.26618Δλ [12, 36]. Using our SLD with an FWHM of 64 nm, λ_full should be as large as 145 nm to achieve the theoretical axial resolution. In our system, which uses a spectrometer with 2048 camera pixels, a λ_full value of approximately 112 nm was used because an imaging depth range of at least 3.0 mm was required for anterior segment imaging. As shown in Fig. 4(b), the axial resolution degraded to ~9.5 μm in air within the acceptable imaging depth (3.0 mm). The degradation of axial resolution in the depth direction is caused because the interpolation process is quite sensitive to small errors at high-frequency regions [5]. In summary, all of the parameters for each imaging path were optimized by considering the trade-offs among the imaging depth, axial resolution, and lateral field of view.

 

Fig. 4 Measured point spread function of axial resolution at 120 μm from zero-depth location (a) and changes in axial resolution as function of depth (b).

Download Full Size | PDF

3.2 In vivo images of anterior segment and retina of a human volunteer’s eye

To demonstrate the capability of the proposed simultaneous imaging system, we acquired in vivo images from the anterior and the retinal regions of a human volunteer’s eye. Figure 5 shows the anterior segment images with the mirror artifacts (Fig. 5(a)) and without the artifacts (Fig. 5(b)) by using the complex conjugate removal algorithm. As shown clearly, the positive and negative images are overlapped in Fig. 5(a), whereas the mirror image is sufficiently suppressed to display the entire anterior chamber of a human eye from the cornea to the frontal surface of the lens (red arrow). However, as shown in Fig. 5, the artifacts near the DC zone have not been fully suppressed. The major causes include saturation of the probe beam in highly reflecting surfaces and the camera readout noise [29, 30]. In addition, the complex conjugation removal algorithm used in this study is based on constant phase shifts between adjacent A-lines. The sample motion in the axial direction may introduce undesirable phase shift [30]. Each image was resized after one with 2048 (axial) × 1500 (lateral) pixels was obtained. The lateral scan range is approximately 12.5 mm.

 

Fig. 5 Acquired images of the anterior segment of a human volunteer’s eye with mirror artifacts (a) and without mirror artifacts (b) corrected using off-pivot based complex conjugate removal implementation. The images consist of 2048 (axial) × 1500 (lateral) pixels. The image size is 6.0 mm (axial) × 12.5 mm (lateral).

Download Full Size | PDF

Figure 6 shows images of the anterior segment and the retina that were obtained simultaneously. The anterior segment (left) and retinal (right) images consist of 2048 (axial) × 1500 (lateral) pixels and 1024 (axial) × 1500 (lateral) pixels, respectively. The microstructures of both parts of the eye are shown in Fig. 6. The lateral size of the retina image is in the range of 5–6 mm, which is smaller than that of the anterior image, even though the pixel numbers and scan angles of the images are the same. We discuss this issue in Section 4. Media 1 is a real-time movie that shows both images with 1500 A-lines. Although the theoretical maximum frame rate is 23 fps, it displays at lower frame rate of ~4 fps because of the large computational load. We have to process the complex artifact removal and the standard SD-OCT algorithms for a large amount of data (1500 lateral × 2048 axial × 2), and in doing to, we have not yet employed multithreaded or GPU methods. Careful observation reveals that both images fluctuate in the same manner because they are obtained simultaneously.

 

Fig. 6 Simultaneous images of the anterior segment and the retina (Media 1). The anterior segment and retinal images consist of 2048 (axial) × 1500 (lateral) pixels and 1024 (axial) × 1500 (lateral) pixels, respectively. The image size is 6.0 mm (axial) × 14 mm (lateral) for the anterior segment and 3.0 mm (axial) × 5.6 mm (lateral) for the retina.

Download Full Size | PDF

We can easily choose between two different types of imaging modes (simultaneous imaging mode and single imaging mode), as mentioned previously. In many cases, the single imaging mode would be desirable because the image acquisition speed is doubled. In our case, the 70 kHz A-scan rate can be fully utilized instead of 35 kHz. In addition, the computational load can be dedicated to obtaining more detailed information from either one of the eye parts.

Figure 7 shows a series of retinal images that were taken after the single imaging mode for the retina was chosen. Only a part of the entire image is shown in this figure, which has a size of 600 (axial) × 1500 (lateral) pixels. The lateral scan range in the retinal image shown in Fig. 6 is relatively small compared to that in the image shown in Fig. 7 because we adjusted the scan angle of the galvanometer to the lateral dimension of the anterior segment. We could increase it during a single imaging mode, as shown in Fig. 7. The major retinal layers around the macula (Fig. 7(a)) and the optic disk (Fig. 7(b)) are observed. In Fig. 7(c), which shows a magnified view of the red box in Fig. 7 (a), in particular, the external limiting membrane (ELM) and three layers are distinguished, including the photoreceptor inner and outer segment junction (IS/OS), photoreceptor outer segments (PR OS), and retinal pigment epithelium (RPE).

 

Fig. 7 Images of human retina (single imaging mode) around the macular (a) and the optic disk (b) with ROI of 600 (axial) × 1500 (lateral) pixels taken at an acquisition speed of 70 kHz; (c) magnification of the apex region of the macular (red box). The image size of each image is 3.0 mm (axial) × 7.8 mm (lateral). NFL: nerve fiber layer, IPL: inner plexiform layer, INL: inner nuclear layer, OPL: outer plexiform layer, ONL: outer nuclear layer, ELM: external limiting membrane, IS/OS: photoreceptor inner segment/outer segment junction, PR OS: photoreceptor outer segments, RPE: retinal pigment epithelium, and CH: choroid

Download Full Size | PDF

4. Discussion

In this paper, we propose a new FD/SD-OCT design that uses dual illumination with two orthogonally polarized lights and interlaced detection with a single spectrometer and a high-speed optical switch. It was demonstrated that this system could simultaneously obtain in vivo images of the anterior segment and the retina of a healthy human eye. Our proposed FD/SD-OCT design has several advantages. First, a single device can acquire images from multiple regions without changing any optical part of the hardware, which is highly desirable because it eliminates the need for dispersion compensation and adjustments in the optical alignment and optical path length to image different regions. It is inevitable, though that the reference position of the retinal path should be adjusted slightly for each patient after the anterior segment is positioned properly in the imaging domain because the axial eye length of each patient is different. Second, a user can easily switch between the two regions or between the simultaneous imaging mode and one of the individual regions by simply clicking on the GUI program. Therefore, the relative motion and the axial length between the two regions inside a single organ can be estimated, and the user can focus on a single region of interest with the full A-scan speed. The final advantage of our system compared to the FD/SD-OCT system with two interferometers and two spectrometers proposed by Dai et al. is that it can be set up inexpensively because we used a single SLD, single spectrometer, and an inexpensive optical switch.

The quality of our anterior segment images needs to be improved further. Currently, the zero-delay depth is set near the mid-plane of the entire anterior segment without the lens. Because the frontal lens surface is located near the lowest sensitivity zone, it is barely visible. In addition, the imaging depth of nearly 6.0 mm after the full-range expansion is still not sufficient to image the entire anterior segment, including the eye lens. In eye clinics, it is critical to investigate the lens opacity, lens thickness, and other pathological changes. We expect that the imaging depth can be increased to over 12.0 mm, with an axial resolution similar to that of our current system, if a line-scan camera with 4096 pixels is used instead of one with 2048 pixels.

As shown in Fig. 6, during the simultaneous imaging mode, the lateral dimension of the retinal image is smaller than that of the anterior segment image even if the scan angle and the number of pixels are the same. This is caused by the difference between the refractive powers of the sample lens (L3) and an eye lens. The lateral dimension can be easily adjusted to any size either by increasing the total number of A-scans or by increasing the lateral distance between adjacent A-scan lines. Caution needs to be taken when imaging the anterior part because a complex artifact removal algorithm needs to be employed to extend the axial domain. Increasing the distance between adjacent A-scan lines will shift the corresponding phase difference, which necessitates repositioning the beam on the B-scanning galvano-mirror. In some cases in which the scanning pivotal point cannot be placed precisely at the iris, it may be necessary to administer a drug for pupil dilation so that a wider angle of incidence can be ensured [24] to increase the scan area of the retina. The scan area on the retina can also be increased by employing an ocular lens such as stock plano-convex spherical lenses [37].

The physical dimensions of various eye parts could not be specified in this manuscript due to the well-known optical distortion that is mainly induced by refraction on the ocular surface [38, 39]. We could solve this problem by using software to compensate for the corneal refractive index [24]. Then, our system could be used for more diverse clinical applications such as surgical intervention during intraocular implantation, accurate estimation of the optical artifacts originating from various eye parts, and ocular growth diagnosis.

As mentioned previously, the A-scan rate of our FD/SD-OCT for simultaneous imaging is 35 kHz, which is half the maximum speed (70 kHz) of the camera. Compared to the speeds of other commercially available products, this speed is sufficiently fast and can be used for routine diagnosis. However, the current A-scan rate is still slow compared to that of FD/SS-OCT reported previously in another study [26]. A higher acquisition speed is highly desirable to obtain 3D volumetric images without motion artifacts caused by involuntary eye movement. One of the solutions may be devoting half of the total number of camera pixels to the anterior part and the other half to the retina within a single spectrometer [40, 41]. In this manner, the full speed of the camera could be used for the A-scan rates of both images.

5. Conclusions

In this paper, we propose a new FD/SD-OCT design based on a single spectrometer with dual illumination and interlaced detection at 830 nm that could simultaneously provide retinal and corneal tomograms. For dual illumination, two orthogonal polarization components were used so that both parallel and focused beams could be incident onto an eye simultaneously. In the detector arm, a single spectrometer is illuminated via a customized ultrafast optical switch for interlaced detection. Using a customized GUI makes it possible to easily control the optical switch to choose either the simultaneous imaging or the single imaging mode. The axial resolution of our FD/SD-OCT system was measured to be ~6.7 μm in air, which corresponds to 4.9 μm in tissues (n = 1.35). The sensitivities in the anterior segment and the retina parts were measured to be approximately 91.5 dB and 89.5 dB, respectively, when the acquisition speed was 35,000 A-scans per second and the depth position was close to 120 μm from the zero-depth. By using our spectrometer, we could obtain an imaging depth of 3.0 mm. For anterior segment imaging, we implemented the off-pivot complex conjugate removal technique to double the imaging depth. Our system was successfully used to obtain various OCT images of a healthy volunteer’s eye.