Open Access
Add to BibSonomyAbstract
We present an ultra-thin fiber-body endoscopy probe for optical coherence tomography (OCT) which is based on a stepwise transitional core (STC) fiber. In a minimalistic design, our probe was made of spliced specialty fibers that could be directly used for beam probing optics without using a lens. In our probe, the OCT light delivered through a single-mode fiber was efficiently expanded to a large mode field of 24 μm diameter for a low beam divergence. The size of our probe was 85 μm in the probe’s diameter while operated in a 160-μm thick protective tubing. Through theoretical and experimental analyses, our probe was found to exhibit various attractive features in terms of compactness, flexibility and reliability along with its excellent fabrication simplicity.
© 2015 Optical Society of America
1. Introduction
Compact, robust and reliable designs of endoscopic probes or imaging catheters have long demanded in various fields of biomedical optics. The size does matter for the imaging probes in getting better accessibility to deeply located regions by a minimally invasive manner. Compactness also helps improving the operational flexibility for easily navigating through the curved channels. In the endoscopy of optical coherence tomography (OCT), miniaturization of the probing optics has been promoted by the operation principle of OCT in part. It can perform the axial scans with no mechanical means added to the side of the imaging probe. This allows the OCT probe to easily acquire multi-dimensional images and tomograms with preferably minimum complexity of very small imaging optics. Diverse designs of OCT endoscopy probes have been developed so far by utilizing various technologies of passive optics assemblies [1–10] or those of active actuation mechanisms [11,12]. However, even a simple design, such as a side-viewing probe made of a rod lens and an angled reflector [1–3], still needs delicate fabrication in reality. The dimensions and alignments may sensitively affect the performance by the nature of the bulk optics.
For miniaturization and structural robustness, some fully fiber-optic designs of OCT endoscopy probes have been proposed in which imaging probes consist of wholly optical fibers [4–8]. In fabrication, a fusion splicer can be used in robustly and automatically integrating the glassy components. In this way, a single-mode fiber (SMF) can equip a ball-shaped lens formed from a fiber [2,6]. Or a graded-index multimode fiber can be spliced to an SMF as a small GRIN lens component [7,8]. All those designs provide easy ways of making miniaturized robust imaging probes. Owing to the inherent compactness, they were found to be useful in the applications such as a needle probe where a micro-thin probe is highly demanded for minimal invasions [7–9]. However, the bulk-optic complexity is still present in those fiber-body probes. Geometric factors such as a curvature radius of the fiber lens, length of the GRIN component or a spacer of glass gap are still sensitive design parameters which must be precisely measured and controlled in the fabrication stages. On the other hand, a novel lens-free fiber-body OCT probe was recently proposed in a radically different scheme [4]. This design takes advantage of a large-core fiber (LCF) for optimized output beam characteristics instead of using a conventional lens component. It was found interesting that the role of a lens could be mimicked by the fiber-optic mechanism. The advantages of the lens-free fiber-optic probe include simple fabrication and ultimate miniaturization. Indeed, the integration of the components is far less sensitive on the fabrication factors, except for the splice alignments that are automatically served by a fiber splicer. The thickness of the probe can be scaled down to the diameter of a necessary core as low as 30 μm. This type of probes can develop to versatile imaging tools for those attractive features. Still, further in-depth research and design optimizations are desired on the lens-free probe design to be finally adopted as a practical endoscopy tool.
In this report, we present further improvement of the lens-free fiber-body endoscopy probe with systematic analysis on its design parameters, particularly in emphasis of an enhanced effective imaging range and insertion loss. In this research, we introduced a stepwise transitional core (STC) fiber as a useful design element for the imaging probe. In spite of its minimalistic configuration, simply implemented by spliced fibers, our STC fiber could efficiently manipulate the waveguiding properties for optimized beam probing. In our experiment, a large transition of the mode field size was achieved at an enlargement factor of 2.7 with a little insertion loss of ~1 dB. A sufficiently long effective imaging range of ~0.9 mm was obtained with a beam diameter of 24 μm. Those features were all obtained by our ultra-thin probe whose diameter was only 85 μm. According to the best of our knowledge, this is the thinnest among the endoscopy imaging probes ever reported. The optical and mechanical characteristics of our probe were found to be promising for its compactness, flexibility and reliability along with the wonderful easiness of fabrication.
2. Probe design
The structure of our STC fiber probe is illustrated in Fig. 1. The OCT light is delivered to the endoscopy probe through a delivery fiber of SMF. Our OCT probe consists of three parts after the SMF: an STC section, imaging fiber (IF) section and a reflector. The purpose of the STC section is to make a smooth transition of the fiber’s core size to that of the imaging fiber for a better coupling efficiency. By transitional fibers (TFs), the mode field of the guided light is expanded in a stepwise manner. The imaging fiber consequently produces a lowly diverging beam that is required for practical lens-free OCT imaging. The reflector finally gives the side-viewing characteristic to be scanned by an external means. In our probe, the declined end surface of the imaging fiber directly makes an angled reflector plane. The reflector angle, denoted by ϕ, is defined by the angle of the normal vector to the fiber axis. Its surface reflects the guided light through the total internal reflection when ϕ exceeds the critical angle of 44°. The resultant beam is deflected by 2⋅ϕ toward a nearly perpendicular direction.
2.1 Imaging fiber
The guiding property of an optical fiber is characterized by the normalized frequency, better known as V-number, which is defined by
where a is the core radius, λ is the wavelength of light, NA is the nominal numerical aperture defined by the refractive index contrast, n1 is the index of the core, and Δ is the index contrast [13]. Here, the core-cladding index contrast Δ is the ratio of the refractive index difference, formally defined bywhere n1 and n0 are refractive indices of the core and cladding, respectively. Note that Δ is usually very low, less than 2% for an ordinary silica fiber. The index difference between the core and cladding is below 0.03 with n0 = 1.45 at λ = 1.31 μm. The number of the guided modes depends on the V-number. It supports a single mode of the fundamental when V<2.405 is taken for a simple step-index fiber. In designation of the linearly polarized (LP) modes, high-order modes (HOMs) other than the fundamental LP01 get supported in the order of LP11, LP21, LP02, etc. as the V-number increases. Once supported, the amount of optical power carried by each mode is determined by the condition of excitation or light launching given in the input side of the fiber.The fundamental mode well resembles a Gaussian beam in the mode field distribution. In the same way of the Gaussian beam, the mode field diameter (MFD) of a fiber mode can be measured by the beam diameter of points where the intensities decrease to 1/e2 of the peak. The beam at the fiber end is diffracted accordingly along the axis of z with a Rayleigh range, zR, given by
where n is the refractive index of the medium [14]. Here, half of the MFD is the beam radius w0, measured at the beam waist. Within the Rayleigh range, i.e., −zR<z< + zR, the intensity drops no more than half of the peak at the focus.Larger the beam size is, wider the Rayleigh range gets to produce a more steadily diffracted beam at the output. As a rule of thumb for a Gaussian-like beam, the effective imaging range of OCT is said to be −2⋅zR<z< + 2⋅zR around the focus. We will take this definition of the effective imaging range, in which a transverse resolution is maintained no more than 2.23 of the minimum resolution at the focus. A region outside the effective imaging range is imaged at a significantly degraded local sensitivity and transverse resolution depending on the distance from the waist. With a typical transverse resolution of OCT given by 2⋅w0 = 10~20 μm, the full length of the effective imaging range is estimated to be 0.34 mm to 1.35 mm by Eq. (3) in typical tissues (n = 1.4) at λ = 1.3 μm. On the other hand, an optical fiber gives the effective imaging range of 0<z< + 2⋅zR when directly used for beam probing without help of any lens. Here, it is worth noting that, for the case of the lens-based imaging, the full range of 4⋅zR can be utilized only if the back focal length gives a sufficient working distance larger than 2⋅zR. For an ultra-thin probe of a lens diameter smaller than 100 μm, this requirement is not trivial. The fiber-body probes previously reported usually have working distances shorter than 0.5 mm, considerably affecting the actual imaging ranges of the probes [6–8].
So conveniently, the use of LCF for an imaging fiber eliminates the need for an objective lens. It can simplify the probe structure and the fabrication. This benefit is achieved by the inherent property of the LCF’s light guidance with a large core size. The optical characteristic is reliably intrinsic to the LCF not to be easily interfered by other extrinsic implementation factors. Taking a targeted effective imaging range to be ~1 mm for practical OCT imaging, an optimal beam diameter (2⋅w0) is calculated to be ~24 μm by Eq. (3) with n = 1.4 and λ = 1.3 μm. The corresponding core diameter is, thus, ~30 μm. In our research, an LCF of 2⋅a = 29 μm and Δ = 0.11% was selected. Its MFD was approximately three times bigger than that of the conventional SMF which met our demand on imaging fiber. As a shortcoming, our LCF supported multiple modes as V = 4.8 given by the large core size.
The beam divergence characteristic of our imaging fiber was investigated at a popular OCT wavelength of λ = 1.3 μm. While the LCF was radiating the guided light out of the fiber through its normally cleaved end, the intensity distribution of the diffracted beam was measured by using a translated high-NA fiber. This probing fiber had a small core that acted as a circular detection aperture, smaller than 4 μm in diameter. This measurement was performed along a lateral axis that passed through the center of the beam. For the purpose of sole excitation of the fundamental mode, a fiber-coil mode filter was used to fully suppress the HOMs in experiment [15].
Figure 2 shows the beam intensity distributions measured in the lateral direction of x (a), and the increase of the beam width along the axial direction of z (b), respectively. The coordinate setting of x and z is described by the schematic given in the left-hand side. Each plot in Fig. 2(a) came from the intensity measurements performed at an axial position in air. Notice that it is displayed with a horizontal offset for better visibility in Fig. 2(a). It is clearly observed that the diffracted fields resembled Gaussian intensity distributions with beam widths that depended on the distance, z.
Fig. 2 Beam intensity distributions measured at the output of our LCF along the lateral direction (a), and the increase of the beam width (FWHM) along the axial direction (b), respectively.
In Fig. 2(b), the increase of the beam width is plotted with square dots of the measurement result. Here, the beam widths were found in the full widths at half maxima (FWHM) of the measured intensity distributions. By a curve fitting, the most similar beam expansion curve of a Gaussian beam was found and plotted by the solid curve of Fig. 2(b). The beam width of the corresponding Gaussian beam was 2⋅w0 = 23.6 μm. The Rayleigh range was zR = 0.33 mm for the Gaussian beam. Thus, it could support an effective imaging range of ~0.92 mm for typical tissues of n = 1.4. This property matched our requirement for OCT endoscopy. Compared to the standard SMF, our LCF exhibited 7-fold extension in Rayleigh range owing to the MFD increased 2.7 times. To the preliminary study of the lens-free fiber probe reported earlier [4], the imaging range was increased by × 1.65.
2.2 STC fiber
With our optimal LCF, the OCT probe needed a means for efficient coupling with the SMF delivery. The MFD mismatch was so large between the SMF and the LCF that the direct connection would lose most of the signal power. The loss is not only to waste the optical power but it may also excite the LCF’s HOMs in a troubling manner. By taking a Gaussian beam approximation of the fundamental modes, the loss of a direct splice can be easily estimated. The spatial overlap integral of the normalized Gaussian mode fields yields a comprehensive estimation on the coupling efficiency. When two different fibers are spliced together with no lateral offset and with no angular mismatch, the splice loss measured in dB scale, LdB, is obtained by
where η is the coupling efficiency of the fiber joint that consists of two different fibers whose MFDs are 2⋅w1 = 2⋅w0 and 2⋅w2 = 2⋅αw0, respectively, with an MFD enlargement factor of α [16]. For the standard SMF and our LCF, the splice loss was estimated to be 3.7 dB for α = 2.7 by Eq. (4). This would give a sensitivity penalty of 7.4 dB in the round trip of the OCT signals. Furthermore, a real splice may produce a higher loss partially due to the increased sensitivity on the angular mismatch of the splice. We observed the loss was higher than 4 dB for the direct SMF-LCF splices in experiment.In order to enhance the coupling efficiency, various methods of transitional core structures can be considered. Figure 3 compares those fiber-optic structures with schematic diagrams. The tapered fiber and thermally expanded core (TEC) structures are the well-established techniques of core geometry manipulations. In a fiber taper, the core size gradually shrinks at the same rate of the cladding diameter reduction [17–19]. Splicing the tapered LCF to the SMF can make a very low-loss transitional structure with matched mode fields. The reduced cladding diameter is, however, a critical drawback when used in an OCT probe unless the tapered section can be properly reinforced.
Fig. 3 Various fiber-optic structures of core size transition: fiber taper, thermally expanded core (TEC) and stepwise transitional core (STC).
On the other hand, the TEC fiber is fabricated by thermally excited diffusion of the core dopants [20]. Continuous transition of the core size can be obtained by a gradual variation of the temperature or the heating time profile. The dopant diffusion speed is so low in silica glass that it would take a long time in getting a large core diameter well above 20 μm. Moreover, the light guidance of the TEC fiber is excessively weak for a largely expanded core [15]. Since the index contrast, Δ, is proportional to the density of the core dopants, an enlargement in core size by diffusion results in the decrease in index contrast. For example, a core diameter enlargement of 10 μm to 30 μm in TEC may reduce its index contrast to ~10%, making core-cladding index difference too small, far below 10−3. It gives an extremely weak guidance, becoming too vulnerable to bend losses.
Our STC fiber has a discrete taper-like core geometry. By splicing a series of transitional fibers with incrementally increasing core size, an efficient transition of MFD is achieved in a simple fabrication method. Accepting a slight increase in insertion loss due to the remaining MFD mismatches, our STC structure is still advantageous in terms of mechanical robustness and fabrication simplicity. One more advantageous feature is found in the index adjustability that the ordinary tapered fiber does not provide. The index contrast can be optimized in the spliced sections by simply selecting suitable fibers. This gives a better chance for optimizing the V-number profile. Note that a trade-off relation exists between the number of fiber modes and the bending sensitivity in determining the core size and the index contrast. A sufficient index contrast is required for fiber-optic flexibility. At the same time, a small number of modes are preferred for effectively single-mode guidance. The optimal design must have the index contrast moderately decreased as the core is enlarged in the STC structure.
In this study, a simple formula of the splice loss estimation for our STC was derived under the Gaussian beam approximation. Let us suppose that the total enlargement factor of α is obtained by a plurality of fiber splices with uniform MFD mismatch. The number of splices is denoted by N with a step enlargement factor of αn = α1/N for integer n when the number of the transitional fibers is given by (N−1). The total loss is a sum of the splice losses. From Eq. (4), the loss of the STC in total is derived to
Figure 4 plots the STC loss with the number of steps, N, for given MFD enlargement factors of α that Eq. (5) predicted. It is clear that the loss gradually decreases to zero as N increases. Compared to the direct splicing, i.e. the case of N = 1 given in Fig. 4, the loss is significantly reduced, even when the number of steps is not so high.In our STC probe design, the total enlargement factor was α = 2.7. For this case, the approximate formula of Eq. (5) suggested that increasing N does not help much in improving the coupling efficiency when N>5. Because the misalignment-induced loss was thought to be ~0.2 dB for each large-core splice, the additional gain in coupling efficiency became negligible or even negative for N>5. We chose N = 3 for our STC probe in an economic design. Then, the core diameter was enlarged by a ratio of ~1.5 at each step. Equation (5) showed that an STC loss would be 1.40 dB or 28% in lost portion.
2.3 Numerical simulation on fiber modes
Our probe was made of LCF, SMF and two kinds of TFs for N = 3. All the fibers used in our probe are listed in Table 1. Notice that the two largest fibers of LCF and TF1 were multimode fibers because of the V-numbers that exceeded 2.4. For better understanding the characteristics of our STC-based OCT probe, a series of numerical calculations were carried out on the guided modes of the STC fiber. The theory of weakly guiding step-index fibers was used for the mode analysis [13]. The scalar-wave solution was found for each LP mode with an effective index, group delay and the mode field pattern. They gave better predictions on the splice loss, the signal ghosts and the bending loss.
Table 1. Specification of the fibers used in our STC probe.
Figure 5 shows the calculated normalized propagation constant (a), and the normalized group delay (b) as functions of V for the modes of LP01, LP11, LP21 and LP02. Figure 5 also shows the intensity patterns of the mode fields on the fiber plane (c). The vertical dashed lines in Fig. 5(a) and 5(b) indicate the lines of V = 4.8 that is the case of our LCF imaging fiber. The guiding properties of a fiber mode can be represented by the effective index and the group index in more intuitive ways. Those index-based properties can be converted from the normalized propagation constant, b, and the normalized group delay, γ, respectively [13]. Here, the effective index, neff, is found by
from the normalized propagation constant given in Fig. 5(a). The effective index neff approaches to the cladding index of n0 as b decreases to zero for its guidance becoming weaker. On the other hand, the group delay of signal is related to the group index of the mode, Ng, which is found bywhere N0 is the group index of the cladding material, N1 is that of the core material, and γ is the normalized group delay, respectively [13].Fig. 5 Normalized propagation constant (a), and the normalized group delay (b) as functions of V for LP01, LP11, LP21 and LP02 modes; with the intensity patterns of the mode fields (c).
The inter-modal delay, δt, is the relative trip-time difference between the LP01 mode and other HOM propagating in a fiber of length L. As a function of δγ, which is the difference in γ between the modes, the inter-modal delay is expressed by
where c is the speed of light, and δNg is the difference of the group index, respectively. For most fibers, we may take in approximation. Figure 5(b) suggests that δγ is no more than 0.3 for LP02 or for any other HOM in the vicinity of V = 4.8. It means a centimeter-long LCF can induce a relative delay of <16 fs for the HOMs with respect to the fundamental mode. This time difference is converted to <3.7 × λ in delay length, which is shorter than the coherence length of a typical OCT light source. So, the HOM-involved effect can be ignored when short sections of LCF are used in the STC probe.Coupling efficiencies were calculated with the mode field distributions in a more exact sense [16]. For each splice, the power coupling factor, η, was obtained by the overlap integral of the normalized mode fields as
where E1 and E2 are real-valued mode field distributions of the fibers given with the spatial coordinates, x and y. The insertion loss of LP01 through the last two splices (TF2-TF1-LCF) in our STC structure was calculated to be 0.85 dB, slightly lower than what Eq. (5) suggested. The actual loss of SMF-TF2 splice was much lower than the theories. Taking the experimentally measured splice loss for the SMF-TF2 joint, the total STC loss was anticipated to be ~1.0 dB if perfectly aligned in splicing. It is a little lower than the loss estimated by Eq. (5). For the splice of SMF and TF2 with smaller cores, an optimized splicing condition of excessive arc intensity could reduce the loss by making more gradual interface [21]. This efficiency gain must be hardly expected for the large core splices.The coupling of LP01 of TF2 to LP02 of LCF was also calculated with the calculated mode field distributions. A considerable level of coupling efficiency, 10.7%, was obtained so that the power ratio of LP02 to LP01 was 0.12 (−9.2 dB) inside the LCF section. The other HOMs were not to be excited in theory due to the asymmetric mode fields. But they could be weakly coupled because of imperfect splices in practice. The total HOM power was thought to be, at worst, 20% of the fundamental LP01 power. This is relatively small but not surely negligible due to the wide dynamic range of OCT imaging.
Such an inter-modal coupling can excite the HOM and, consequently, degrades the OCT images with multiple signal paths. The delayed OCT signal would produce a faint replica of the original image contents, slightly translated along the A-line axis. However, thinking of the small relative delays and the low magnitudes, the HOMs could make weak spurious peaks with negligible delays in an OCT A-line if a short section of HOM-supporting fibers were used in the STC. Then, the ghost image would be buried by the sidelobes of the main image contents becoming unnoticeable in effect.
In this context, the length of the HOM-supporting fibers is an important parameter to avoid the imaging ghosting effect. Let us assume that the axial point spread function (APSF) of the OCT system resembles a nice Gaussian function with a full width at quarter maximum (FWQM or 6-dB width), denoted by δz. The actual APSF usually consists of distinct lobes with a higher level of sidelobe amplitudes than those of the Gaussian APSF. The power carried by an HOM-involved path should be lower than that of the main contents to make its ghost practically invisible. To be under the Gaussian APSF envelope, it should be lower than −6 dB for a length delay c⋅δt, smaller than 1.0 × δz. For larger delays, the HOM power should be less than −20 dB for c⋅δt <1.8 × δz, and −40 dB for c⋅δt <2.6 × δz. Notice that those rough criteria were made in that the axis of the OCT A-line was halved from the actual optical delays.
A typical OCT system operating at the 1.3-μm band has δz = 5 μm for its axial resolution in air. When our HOM-supporting fiber section had a short length of L<10 mm, those conditions were surely fulfilled for the signal component that traveled through the LP02 mode in one way taking δγ<0.3 of Eq. (8). Twice passing through the LP02-involved path in round-trip manner might double the relative delay but also doubled the coupling efficiency to be −18.4 dB, still below the sidelobes of the APSF. The effects of the other HOMs were thought to be no stronger than those of the LP02 mode of the LCF, thinking of the lower coupling efficiencies and the relative delays of δγ<0.3 still kept for the others. Based on those estimations, we chose L = 0.7 cm for each section of the HOM-supporting fibers in fabricating the STC structure which could surely eliminate the HOM-involved effects.
2.4 Curvature durability
Based on the guiding properties calculated for the fundamental modes, optical flexibility of the fibers used in our STC probe was examined in theory. The curvature durability is measured by the minimum bend radius limited by the acceptable curvature loss. Because our STC probe consists of fiber-optic components throughout the whole probe body, any part can be curved with minimal effects on the imaging performance. At a moderate curvature radius, a thin fiber dominantly experiences a macrobend loss also known as curvature loss. The classic formula that Marcuse originally derived under an approximation of non-deformed field distributions [15,22,23] was used to predict the macrobend loss. Figure 6 shows the calculated macrobend losses of our LCF, TF1 and the standard SMF with bend radius, R, for their fundamental modes. It showed, in all the cases, the bend loss becomes negligibly small when R>3 cm. Notice that the actual loss may exceed those simple estimations, partially because the deformed mode fields of the curved fibers may considerably reduce the coupling efficiencies at the fiber splices. The curvature loss can only give the upper bound of the additional loss that an STC probe produces under a curvature.
We also evaluated a potentially alternative approach of using a specialty SMF of a large core or a TEC fiber that originated from the standard SMF. A large-core single-mode (SM) section could be used instead of the LCF imaging fiber. A core diameter of 2⋅a = 29 μm and an extremely low index contrast, Δ = 0.029%, were considered in making V = 2.41 for the SM condition at λ = 1.3 μm. As anticipated by the weak guiding property, this fiber was found to exhibit a very high curvature loss >7.4 dB/cm even for a moderately bent section of R = 5 cm. This result well justified the need for sufficient index contrasts for the fibers used for the STC in spite of the potential multimode signal transmissions.
3. STC probe fabrication and performance test
Based on the theoretical and experimental analyses given in the previous section, an OCT endoscopy probe was implemented in an easy fabrication method. Various performance factors were examined to prove the advantageous features of our probe for practical applications.
3.1 Fabrication
Our STC-based endoscopy probe was fabricated by utilizing common fiber-optic instruments of a sliding-blade cleaver and a fusion splicer based on electric arc discharge. All the non-standard fibers (TF1, TF2 and LCF given in Table 1) were obtained from commercially available specialty fibers (LiekkiTM Passive Fibers, nLight Photonics). For thick fibers, the cladding diameter was reduced to 125 μm by wet etching with hydrofluoric acid in a preprocessing stage. The fibers were spliced one by one for TF1, TF2 and LCF after cleaving the fiber at length L = 0.7 cm for each. In splicing, the arc intensity was set to be higher than usual for a minimized splice loss. After making an STC fiber, the cladding diameter was further reduced from 125 μm to 85 μm by etching a stripped fiber section of 25 cm length. In the final stage, our endoscopy probe was fabricated to equip an angled reflector plane at the end. The reflector was polished in a fiber holder with diamond-lapping films. For the purpose of protection, the fabricated STC endoscopy probe was inserted and kept in a silica-polyimide micro-tubing. It was basically a silica capillary tube with an outer coating layer of polyimide resin which reinforced the mechanical strength. According to the manufacturer’s datasheet, the inner and outer diameters of the micro-tubing were 100 μm and 160 μm, respectively, including the 12-μm thick polymer coating on the silica capillary.
Figure 7 shows the microscopic pictures of the fabricated STC probe (a), and the coordinate setting of x, y and z that will be used in the following descriptions (b). The final STC fiber probe was very thin, only 85 μm and 160 μm for the diameters of the probe body and the protection micro-tubing, respectively. The mechanical flexibility was fairly comparable to the optical fibers of the standard 125-μm diameter. The stripped fiber section including the STC probe body, 25 cm long in total, did not equip fiber’s polymer coating. But it was still shielded by the silica-polyimide micro-tubing for safe operations. The beam coming through the angled reflector was nearly perpendicular to the fiber axis but was slightly declined by ~5° to the perpendicular direction. The axis of this declined beam will be denoted by z. The lateral axis perpendicular to the fiber axis will be denoted by x while the remaining axis of the rectangular coordinates will be y in the following.
Fig. 7 Microscopic pictures of the fabricated STC probe (a), and the coordinate setting of x, y and z (b).
3.2 Insertion loss and imaging range
The insertion loss of the STC probe was evaluated in experiment. The power loss including that of the connector was measured to be 1.3 dB at the outside of the protective tubing by using an optical power meter. After applying a bend to our STC probe in the tubing, the additional loss was found to be <0.2 dB for a bend radius R = 3 cm and became ~0.7 dB for a tighter bend of R = 2 cm. These results proved that our STC probe maintained a low insertion loss with a good optical flexibility. However, such a simple power measurement could not distinguish the loss of the fundamental mode from the HOM’s although it gives a rough estimation on the power losses of our STC probe. An exact measurement of modal powers required a special means such as a long LCF that could discriminate the modes by different group delays [17] or by different bend loss characteristics [15]. Unfortunately, the maximum utilizable length of our LCF was ~20 cm for each segment.
For an alternative measurement method of the LP01 mode loss with a short section of LCF, we fabricated an STC fiber pair that was a series of spliced fibers that contained two symmetrically paired STC structures: SMF-TF2-TF1-LCF-TF1-TF2-SMF in a doubled STC arrangement. The lengths of the two TF1 and TF2 sections were 0.7 cm for each whereas the LCF was 10 cm long. The incident incoherent light experienced the two STC structures in series that doubled the insertion loss. The loss was measured by a power meter at the other end. The power of the light that passed through the HOM-involved paths in the STC pair must have been so low in the final output. Its transmission efficiency was thought to be lower than a few percents for the twice inter-modally coupled power. Based on the theoretical estimation given in Section 2, the LP01-LP02-LP01 path involved with the LCF’s HOM might produce a loss of ~20 dB while the wholly LP01 path exhibited a much lower loss of ~2 dB. Therefore, most of the detected optical power could be regarded as the signal given through the LP01-mode path. The full loss of the STC pair was measured to be 2.4 dB at λ = 1.3 μm. The effective insertion loss of a single STC structure was found to be 1.2 dB. This reasonably matched the theoretical estimations for the LP01 loss given in the previous section.
The beam characteristics were experimentally investigated for the output of our STC probe. We used an incoherent light source of a semiconductor optical amplifier (SOA) operating at λ = 1.3 μm and a microscopic image system that equipped an infrared image sensor. Figure 8 shows the measured intensity patterns on the xy-planes for various distance, D, when the probe was immersed in water (ne = 1.32) and index-matching gel (ne = 1.45). Here, D was the distance from the center of the probe measured along the beam axis of z. Note that the nearly Gaussian-like intensity distributions observed at D = 0.10 mm proved the dominance of the LP01 mode power. For a very thin side-view endoscopy probe, the curved surface of the probe body or the protective micro-tubing effectively could act as cylindrical lenses [4,7]. The resulting beam became asymmetric in the lateral intensity distributions due to the significant astigmatism. The beam shapes at a distance depended on the refractive index of the surrounding medium, ne, by this cause.
Fig. 8 Measured intensity patterns on the xy-planes for various distance, D, when the probe was immersed in water (ne = 1.32) and index-matching gel (ne = 1.45). The black square on the right-hand side shows the dimensions of the beam size.
Figure 9 shows the beam widths measured in FWHM from the acquired intensity patterns. The beam width was dependent on the lateral directions of measurement as well as influenced by the environmental index of ne. The overall diffraction characteristics were noticeably different from those of a symmetric Gaussian beam. In any case given in Fig. 9, the FWHM beam width was maintained below 30 μm for a distance D≤0.9 mm that was the effective imaging range (2⋅zR) estimated by the Gaussian approximation. The maximum 1/e2-intensity beam width was 51 μm in the same range giving an average lateral resolution of ~30 μm (1/e2 resolutions) for OCT. This resolving power in the transverse axes was lower than the usual lens-based OCT imaging. As suggested by Eq. (3), it was inevitable that our fiber-based probe exhibited a slightly poorer transverse resolution lowered by a factor of 21/2 = 1.41 for a given effective imaging range. However, one should note that the actually achievable imaging range of a lens-based ultra-thin probe can be considerably shorter than 4⋅zR. The back focal length of a micro-lens hardly exceeds 2⋅zR when the diameter is smaller than 80 μm. Thus, the real loss of our probe in transverse resolution was thought to be 20~30% compared to the realizable counterpart of the lens-based probe in the same probe diameter.
Fig. 9 Beam width expansion as a function of distance, D, for the case of ne = 1.32 (a), and for that of ne = 1.45 (b). The circular dots represent the measurements along the x-axis while the triangular dots plot the measurements along the y-axis.
3.3 OCT imaging test
OCT imaging tests were performed with a spectral-domain OCT (SD-OCT) system based on a broadband light source and a spectrometer. The center wavelength of the light source was 1310 nm with a 3-dB spectral width of 80 nm. The full axial range of imaging was 2.3 mm in air. At first, a human fingertip was imaged with our STC probe and SD-OCT system. 2D OCT images were acquired by scanning our STC probe in a manual push-pull manner. For comparison, the fingertip was also imaged with the same SD-OCT system that equipped a galvo-mirror scanner and an objective lens which had a transverse resolution of 12 μm at the focus. Figure 10 shows the OCT image obtained with our STC probe (a), and the one with the objective lens (b). For the case of our STC probe, the sample fingertip was contacted by the protective tubing while the gap between them was filled with water. The overall image quality of our STC probe was comparable to that of the OCT imaging performed with an objective lens. The layered structures including the sweat ducts (arrowed by SD in each image) were clearly visualized. In Fig. 10(a), the horizontal straight lines (arrowed by L) were from the probe’s internal back-reflections that occurred at the STC probe end, the inner and outer interfaces of the protective capillary. Because of the beam axis nearly normal to the capillary’s interfaces, excessively high-amplitude signals were caused there and resulted in annoying autocorrelation artifact of imaging ghosts, as indicated by AC in Fig. 10(a). This minor problem would not be emphasized if the reflector angle were optimally set off from 45°.
Fig. 10 OCT images of human fingertips obtained with our STC probe (a), and with a conventional objective lens (b).
Another imaging test was performed for rabbit’s lung sample ex vivo. Our STC probe was carefully inserted into rabbit’s airway through the trachea deeply to the bronchial tree. Due to the small size, our endoscopy probe could easily reach a terminal bronchiole with minimal invasions. Figure 11 shows the OCT image of the airway of the lung (a), and that of the trachea region (b). Note that the lower 60% areas of the OCT images were cropped to remove the AC ghosts. In Fig. 11(a), the alveoli of the rabbit’s lung appeared as black areas in the image. Various features of the trachea’s inner wall were also acquired successfully with our STC probe as seen in Fig. 11(b).
Fig. 11 OCT ex-vivo image of the airway in the rabbit’s lung (a) and that of the trachea region (b), acquired by our STC probe.
In our imaging tests, our probe was scanned manually in a push-pull operation. A simple handset device was equipped to safely hold the delivery fiber in motion. A linear scan could be applied during OCT imaging by moving the holder of the deliver fiber which was located ~30 cm far from the end of the probe. Note that a typical side-view probe is operated by primary rotational scans using an external rotator. A secondary pull-back scan can be given for 3D tomograms in this case. The major difficulty of operation is on the scan lags and mechanical frictions routinely observed in a side-view probe passively operated by delivery of external torques or linear forces. In many cases, the probe body is too massive compared to the delivery part. Or it is too loosely bound in a protective tubing in concern of the mechanical resistance concentrated on the probe’s face. As a consequence, prompt motions are hardly obtained for such a probe. Fast rotations and slow pull-backs are the preferred scan strategy for this reason. In contrast, our STC probe has its body no different from the other delivery part. They are all the optical fibers of the same mechanical properties and dimensions. The probe is stiffly self-sustaining in a microscopic viewpoint while maintaining the macroscopic flexibility. The compact head of our probe can be easily moved at a very prompt response with little lags in linear motions as well as in rotations. This characteristic might be helpful in reducing the complexity of the probe scanning mechanism and also that of the image reconstructions. Linear scans in priority may be advantageous in some cases, e.g., briefly inspecting a tubular object. It does not require heavy components such as a fiber-optic rotary joint and even a motor. By taking advantage of our probe’s ultra-thin geometry, multiple probes can be integrated which acquire OCT images in different viewing angles in a multiplexed imaging strategy. In manual scanning, various tracking schemes have been developed for free-hand operations, which might be useful when combined to our imaging probe [24–26].
4. Conclusion
In this research, the STC fiber-optic endoscopy probe was introduced for a very compact and flexible OCT endoscopy tool. Our STC probe utilized a fiber-optic scheme for engineering the output beam characteristic and the efficient light coupling, required for practical OCT signal probing micro-optics. By taking advantage of a large-core fiber (LCF) and stepwise transitional core (STC) structure, a fully fiber-based design was presented. The fabrication was very simple with suitable specialty fibers. It was nothing but a series of cleaving and splicing plus polishing the end surface. No precise control of dimensions and alignments was added to cleaving-and-splicing processes. In this research, we fabricated an STC probe of diameter 85 μm. Including the protection tubing, the total thickness was 160 μm, still thinner than the standard diameter of bare polymer-coated optical fibers. The minimum achievable size is not optically constrained but limited by the mechanical reliability. A probe as thin as 30 μm is surely feasible for specialized applications where the probe’s intrinsic strength does not matter.
In this research, various characteristics of the STC structure and the STC-based endoscopy probe were investigated in theory and in experiment. After all, the overall performance of our STC probe was found to be very attractive. The effective insertion loss was as low as 1.3 dB. The HOM-involved artifact was not observed due to the optimal coupling and length in our STC structure. The curvature loss was found to be so low for moderate bends of radii larger than 3 cm proving our probe was optically flexible as well. At an acceptable sacrifice of the transverse resolutions, a sufficiently long effective imaging range was obtained which was measured to be ~0.9 mm in tissue.
The application area suited to our STC probe is diverse. A very compact endoscopy probe can reach deeper regions with minimal invasions. A needle probe is a good example. It can easily cooperate with other imaging modality. Because of the fabrication simplicity and reproducibility, our probe is suitable for multiplexed A-line acquisitions by which multiple points of a sample are imaged or monitored simultaneously with no mechanical part in use [27]. The inexpensiveness of our probe may help single usage to reduce the cross-contamination issues. Out of the imaging applications, the technology developed in this research will be useful in optical sensing and fiber-optic components. The STC fiber can be used for a versatile mode field converter replacing tapered fibers or TEC fibers for better mass-productiviy.
Acknowledgment
This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF-2014R1A1A1006183).
References and links
1. Z. Yaqoob, J. Wu, E. J. McDowell, X. Heng, and C. Yang, “Methods and application areas of endoscopic optical coherence tomography,” J. Biomed. Opt. 11(6), 063001 (2006). [CrossRef] [PubMed]
2. B. H. Lee, E. J. Min, and Y. H. Kim, “Fiber-based optical coherence tomography for biomedical imaging, sensing, and precision measurements,” Opt. Fiber Technol. Part B 19(6), 729–740 (2013). [CrossRef]
3. J. Jing, J. Zhang, A. C. Loy, B. J. F. Wong, and Z. Chen, “High-speed upper-airway imaging using full-range optical coherence tomography,” J. Biomed. Opt. 17(11), 110507 (2012). [CrossRef] [PubMed]
4. S. Moon, Z. Piao, C.-S. Kim, and Z. Chen, “Lens-free endoscopy probe for optical coherence tomography,” Opt. Lett. 38(12), 2014–2016 (2013). [CrossRef] [PubMed]
5. U. Sharma and J. U. Kang, “Common-path optical coherence tomography with side-viewing bare fiber probe for endoscopic optical coherence tomography,” Rev. Sci. Instrum. 78(11), 113102 (2007). [CrossRef] [PubMed]
6. K. M. Tan, M. Shishkov, A. Chee, M. B. Applegate, B. E. Bouma, and M. J. Suter, “Flexible transbronchial optical frequency domain imaging smart needle for biopsy guidance,” Biomed. Opt. Express 3(8), 1947–1954 (2012). [CrossRef] [PubMed]
7. D. Lorenser, X. Yang, R. W. Kirk, B. C. Quirk, R. A. McLaughlin, and D. D. Sampson, “Ultrathin side-viewing needle probe for optical coherence tomography,” Opt. Lett. 36(19), 3894–3896 (2011). [CrossRef] [PubMed]
8. X. Yang, D. Lorenser, R. A. McLaughlin, R. W. Kirk, M. Edmond, M. C. Simpson, M. D. Grounds, and D. D. Sampson, “Imaging deep skeletal muscle structure using a high-sensitivity ultrathin side-viewing optical coherence tomography needle probe,” Biomed. Opt. Express 5(1), 136–148 (2013). [PubMed]
9. X. Li, C. Chudoba, T. Ko, C. Pitris, and J. G. Fujimoto, “Imaging needle for optical coherence tomography,” Opt. Lett. 25(20), 1520–1522 (2000). [CrossRef] [PubMed]
10. J. Wu, M. Conry, C. Gu, F. Wang, Z. Yaqoob, and C. Yang, “Paired-angle-rotation scanning optical coherence tomography forward-imaging probe,” Opt. Lett. 31(9), 1265–1267 (2006). [CrossRef] [PubMed]
11. P. H. Tran, D. S. Mukai, M. Brenner, and Z. Chen, “In vivo endoscopic optical coherence tomography by use of a rotational microelectromechanical system probe,” Opt. Lett. 29(11), 1236–1238 (2004). [CrossRef] [PubMed]
12. S. Moon, S.-W. Lee, M. Rubinstein, B. J. F. Wong, and Z. Chen, “Semi-resonant operation of a fiber-cantilever piezotube scanner for stable optical coherence tomography endoscope imaging,” Opt. Express 18(20), 21183–21197 (2010). [CrossRef] [PubMed]
13. K. Okamoto, Fundamentals of Optical Waveguides (Academic Press, 2000), Chap. 3.
14. B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics (Wiley Interscience, 1991), Chap. 3.
15. S. Moon and Z. Chen, “Mode-filtered large-core fiber for optical coherence tomography,” Appl. Opt. 51(34), 8262–8270 (2012). [CrossRef] [PubMed]
16. D. Marcuse, “Loss analysis of single-mode fiber splices,” Bell Syst. Tech. J. 56(5), 703–718 (1977). [CrossRef]
17. S. Moon and D. Y. Kim, “Effective single-mode transmission at wavelengths shorter than the cutoff wavelength of an optical fiber,” IEEE Photon. Technol. Lett. 17(12), 2604–2606 (2005). [CrossRef]
18. S. Moon, G. Liu, and Z. Chen, “Mode-filtered large-core fiber for short-pulse delivery with reduced nonlinear effects,” Opt. Lett. 36(17), 3362–3364 (2011). [CrossRef] [PubMed]
19. A. Ishikura, Y. Kato, and M. Miyauchi, “Taper splice method for single-mode fibers,” Appl. Opt. 25(19), 3460–3465 (1986). [CrossRef] [PubMed]
20. M. Kihara, M. Matsumoto, T. Haibara, and S. Tomita, “Characteristics of thermally expanded core fiber,” J. Lightwave Technol. 14(10), 2209–2214 (1996). [CrossRef]
21. J. T. Krause, W. A. Reed, and K. L. Walker, “Splice loss of single-mode fiber as related to fusion time, temperature, and index profile alteration,” J. Lightwave Technol. 4(7), 837–840 (1986). [CrossRef]
22. D. Marcuse, “Curvature loss formula for optical fibers,” J. Opt. Soc. Am. 66(3), 216–220 (1976). [CrossRef]
23. D. Marcuse, “Field deformation and loss caused by curvature of optical fibers,” J. Opt. Soc. Am. 66(4), 311–320 (1976). [CrossRef]
24. J. Ren, J. Wu, E. J. McDowell, and C. Yang, “Manual-scanning optical coherence tomography probe based on position tracking,” Opt. Lett. 34(21), 3400–3402 (2009). [CrossRef] [PubMed]
25. A. Ahmad, S. G. Adie, E. J. Chaney, U. Sharma, and S. A. Boppart, “Cross-correlation-based image acquisition technique for manually-scanned optical coherence tomography,” Opt. Express 17(10), 8125–8136 (2009). [CrossRef] [PubMed]
26. B. Y. Yeo, R. A. McLaughlin, R. W. Kirk, and D. D. Sampson, “Enabling freehand lateral scanning of optical coherence tomography needle probes with a magnetic tracking system,” Biomed. Opt. Express 3(7), 1565–1578 (2012). [CrossRef] [PubMed]
27. H. Y. Lee, T. Marvdashti, L. Duan, S. A. Khan, and A. K. Ellerbee, “Scalable multiplexing for parallel imaging with interleaved optical coherence tomography,” Biomed. Opt. Express 5(9), 3192–3203 (2014). [CrossRef] [PubMed]
