Two-photon endoscopy based on a gradient-index lens has been widely utilized for studying cellular behaviors in deep-lying tissues with minimal invasiveness in vivo. Although the efficient collection of emitted light is critical to attain high-contrast spatiotemporal information, the intrinsic low numerical aperture of the endoscopic probe poses a physical limitation. We report a simple solution to overcome this limit by incorporating a reflective waveguide ensheathing the endoscopic probe, which improves the collection efficiency by approximately two-fold. We describe its principle, fabrication procedure, optical characterization, and utilities in biological tissues.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
By providing subcellular spatial resolution in scattering biological media, two-photon microscopy has been widely adapted as a gold-standard tool to study cellular processes in living animals [1,2]. Although a longer excitation light ameliorates the scattering issue, the penetration depth is still limited typically to <1 mm in most biological tissues. Among the various solutions available, a miniature endoscopy has gained particular attention because it provides a versatile access to any internal organs through their natural orifices (e.g.. colon, esophagus) or minimally invasive surgical openings [3–7]. Gradient-index (GRIN) lenses are typically used to construct endoscopic probes because of its excellent optical performance in a small form factor and ease of integration into conventional microscopy. The major utility of two-photon endoscopy includes studies on deep brain structures with fluorescence and muscle contractility with second-harmonic generation (SHG) [8,9].
Continued efforts have been conducted to optimize its optical performance, especially on the collection efficiency. Two-photon excitation is remotely confined only to the focal volume by its nonlinearity; therefore, all the emitted photons, either ballistic or scattered photons, carry meaningful signal information . Thus, optic design is focused on collecting as much emitted photons as possible to attain high-contrast spatiotemporal information. The non-descanned detection with a low magnification, high numerical aperture (NA) objective lens has become the default configuration in two-photon microscopy . Strategies to further improve the collection efficiency have also been proposed, such as an additional fiber-optic collection, an objective lens with an incorporated beam splitter, and an auxiliary parabolic reflective collector [10, 12, 13]. However, these approaches are not compatible with two-photon endoscopy, where an endoscopic probe (NA ≈0.5) limits the maximum collectable photons.
Here, we describe a simple solution to overcome the limited collection efficiency in two-photon endoscopy by incorporating a reflective waveguide onto the endoscopic probe. The reflective waveguide was stably formed by the sputter deposition of a silver film onto off-the-shelf singlet GRIN probes. Then, the collection efficiency is determined by the reflective waveguide, providing significantly higher collection NA than the endoscopic probe. We demonstrate that the silver-coated probe improves the collection efficiency by approximately two-fold in practical applications, such as the fluorescent imaging of neurovascular system in the brain and the SHG-based imaging of skeletal muscles.
2. Materials and Methods
2.1 Preparation of phantom samples
For fluorescent samples, rhodamine B (208108, Sigma Aldrich) was dissolved in distilled water (0.1 mM) and 100-nm fluorescent beads (F8801, Invitrogen) were diluted to 1:1000 in 1% agarose gel. For the SHG samples, barium titanate (83689, Sigma Aldrich) was mixed with a solution of polydimethylsiloxane (PDMS; Sylgard 184, Sigma-Aldrich) with 10% w/w curing agent. After dispersing the particles, the mixture was incubated in a vacuum chamber for 1 h to remove air bubbles and was subsequently solidified at 70 °C for 2 h.
2.2 Animal preparations
Male transgenic mice (Thy1-YFP-H, Jackson Laboratory) aged eight-weeks old and male C57B6 mice were used for preparing for the brain and muscle samples, respectively. After deep anesthesia with an intraperitoneal injection of 1 mg/g urethane, the mouse was transcardially perfused with a cold saline followed by 4% paraformaldehyde. The fresh brain and muscle tissues were extracted and the brain was sliced with a vibratome (VT1000S, Leica) to 200 μm in thickness. For in vivo studies, we anesthetized a male wild-type mouse aged eight-weeks old (C57BL6J, Jackson Laboratory) by intraperitoneal injection of 0.12 mg/g zoletil (Virbac) and 0.01 mg/g rompun (Bayer Korea). After the craniotomy, 100 μl of 2.5% w/v TRITC-dextran (500kDa, Sigma Aldrich) in phosphate-buffered saline was injected intravenously through retro-orbital route for blood flow imaging. All animal experiments were performed in compliance with the institutional guidelines and approved by the subcommittee on research animal care at Sungkyunkwan University.
2.3 Two-photon endoscopy system
The two-photon microscope (IntraVital, Bruker) was composed of a femtosecond laser with a prechirping unit (Chameleon Vision II, Coherent), a dual-axis galvo-scanner (6215H, CambridgeTech), four-channel non-descanned detectors with GaAsP photomultiplier tubes, and a water immersion objective lens (25 × , 1.1 NA; Nikon). A singlet GRIN probe (1 mm in diameter, 4.38 mm in length, distal working distance of 0.25 mm in water, proximal working distance of 0.1 mm in air, 0.5 NA at 860 nm; NEM-100-25-10-860-S, GrinTech) was held by a three-axis micrometer after attaching the proximal surface to a coverslip (#1) with an optical adhesive (NOA81, Norland Products). The probe’s proximal surface was positioned to the focal plane of the microscope objective as shown in Fig. 1. The sample was positioned using a three-axis translation stage.
2.4 Image processing and data analysis
We used ImageJ (NIH) and Matlab (Mathworks) for image quantification and Prism (Graphpad) for statistical analyses. The data are presented as either mean ± standard error or box-and-whisker plots, unless otherwise indicated. We performed group comparisons with either unpaired or paired t-tests, and considered a p-value less than 0.05 to be statistically significant.
3. Results and Discussion
3.1 Concept of improved collection efficiency
GRIN lenses have spatially inhomogeneous refractive indices that vary quadratically with their radii; thus, the incoming beam follows a sinusoidal trajectory . A singlet probe is designed to have slightly less than half a sinusoid such that the point at its proximal surface can be relayed near the distal surface with the desired working distance (e.g., 250 μm in water). The excitation and emission beams follow nearly the same trajectories; therefore, the NAs for the excitation and emission are nearly identical, although some chromatic aberration exist [Figs. 1(a) and 1(b)]. This miniature 4f image-relay system can be easily integrated to a benchtop microscope by positioning its proximal surface at the image plane under the objectives [Fig. 1(c)] . The objective NA is typically matched to that of the probe to avoid coupling loss.
By adding a reflective waveguide, however, the probe collects more photons that emerge into the outside of its collection cone [Fig. 1(b)]. Consequently, the collection NA becomes higher than the excitation NA. In our endoscope with a diameter of 1 mm and a working distance of 250 μm, the collection NA is estimated to be ~1.2, which can provide up to ~5.6-fold higher collection efficiency compared to a bare probe (NA ≈0.5). Although the possibility of light guidance exists even in the bare probes through total-internal reflection at the surface (n ≈1.54), the guiding efficiency becomes compromised when interfaced with turbid biological tissues (n ≈1.38). By contrast, the reflective waveguide formed by the silver film shows superior guiding efficiency and is independent of the external medium. It is noteworthy that the collected photons lose image information, thus the enhanced collection efficiency only occurs with multiphoton endoscopy where all the scattered photons carry useful signal information.
To test the theoretical validity of our concept, we performed optical simulations using a commercial ray tracing software under the non-sequential mode (OpticStudio, Zemax). For the probe, we used a Zemax file provided by the manufacturer (GRINTech) and modified it to match our probe design, which is detailed in Section 2.3. We then placed a point source with isotropic emission at the focus, which is 250 μm away from the distal surface of the probe. A square detector (1 mm by 1 mm) covering the proximal surface of the probe was placed at 0.1 mm away from the probe’s proximal surface to mimic the non-descanned collection of the guided lights. A sphere with a radius of 10 mm encasing all the parts was used to control optical properties of the external medium (e.g. air, water, silver). For quantitative analyses, we used the Monte Carlo ray tracing with 105 analysis rays from the point source. The total hits at the detector were quantified as the relative incoherent irradiance collected by the probe, which is then used to calculate the normalized collection efficiency. The simulation result showed that the reflective probe collects over 10-fold more emission compared to the bare probe when the light guiding through the total internal reflection is ignored by setting the external refractive index higher than that of the probe (nclad > 1.5) [Figs. 1(d) and 1(e)]. Even with considering the total internal reflection surrounded by a homogeneous medium with a refractive index similar to that of a biological tissue (n ≈1.38), the reflective probe exhibited over two-fold higher collection efficiency compared to the bare probe. Therefore, in a practical condition of an endoscopic probe embedded in a biological tissue, we expect approximately a two-fold improvement in the collection efficiency.
3.2 Fabrication of a reflective-waveguiding endoscope
Multiple techniques are available to form a reflective silver film on a substrate. We initially attempted a simple solution-based reductive method, where the silver ion is reduced to form a thin reflective layer on a substrate . Using this approach, we coated the singlet lenses by immersing them in a solution of silver nitrate and dextrose. However, the reductive approach consistently resulted in a sparse coating with weak adhesion that easily delaminated even with gentle washing or careful handling.
We subsequently adapted a sputter-based method, which deposits a thin film on a substrate via plasma in vacuum . To coat only the curved side, the proximal and distal surfaces were sealed with a polyimide tape [Fig. 2(a)]. After a sputter deposition of ~16 min (150 W at 10 mTorr), the probes were fully ensheathed by a silver film with its thickness of ~200 nm [Fig. 2(b)]. We reliably obtained the silver coating for all the probes we processed (n = 23 probes), although the removal of the sealing tape sometimes damaged the coating (2 out of 23 probes). The coating was notably stronger than that formed by our previous solution-based approach. To further protect the silver coating, we additionally deposited a thin chrome layer for ~8 min (200 W at 6 mTorr). The polyimide tapes were subsequently detached. The double-layered coating provided resilience to practical applications such as gentle handling with forceps. Polymer or metal sheath may be added for attaining a higher stability for long-term usage.
To check the optical quality of the silver coating, we next measured the reflectivity of a silver-film formed on a flat slide glass. We obtained over 95% reflectivity for 450–600 nm, which was comparable to a commercial silver mirror [Fig. 2(c)]. Our sputter-based method on a cylindrical substrate may form a non-uniform thickness over the surface, but it should not compromise the optical performance as reflection occurs at the internal interface. This procedure is extendable for the mass production of the coated probes. In a single sputter deposition, multiple probes (>100) can be simultaneously coated in less than 30 min. The cost of silver material is also low (<$1 per piece). We envision that our method can be immediately applied to the GRIN industry.
3.3 Optical characterization
Considering the design of the probe, we chose a 25 × , 1.1-NA water-immersion objective lens, which covers the whole circular aperture of the probe (~1 mm in diameter) with a large collection cone. The emitted light was collected by a low-pass dichroic mirror with a cutoff at 700 nm, and captured by non-descanned detectors composed of four-channel GaAsP photomultiplier tubes.
We first tested the resolution of the bare and silver-coated probes using fluorescent beads (100 nm in diameter). We obtained a lateral resolution of ~1 μm and an axial resolution of ~10 μm with both bare and coated probes, indicating that the excitation NAs for both probes are similar and our coating procedure did not compromise the excitation capability of the GRIN probes [Figs. 3(a) and 3(b)]. We next compared the collection efficiency using a solution of Rhodamine B emitting red fluorescence, collected at 605 ± 35 nm [Fig. 3(c)]. To mimic the condition of the tissue imaging, the curved side of the bare probe was covered with PDMS with 0.1% alumina (n ≈1.4). The fluorescent solution was dipped at the tip of either the bare or coated probes and the fluorescence intensity was measured at the same excitation intensity (~2.5 mW measured at the objective back aperture; 840 nm). The fluorescent images through the endoscopes showed the typical radially decreasing intensity profile due to the off-axis optical aberrations . Compared to the bare probe, the silver-coated probe showed approximately two-fold brighter fluorescence [Fig. 3(d)]. We next tested on barium titanite nanoparticles, which efficiently generate coherent and directional SHG emissions at half the fundamental wavelength . Although individual variabilities exist in the SHG intensity among the BaTiO3 microparticles, the silver-coated probe resulted in approximately a two-fold brighter collection on average [Figs. 3(c) and 3(d)]. These results consistently show that the addition of reflective waveguides on the endoscopes significantly improve the collection efficiency by approximately two-fold.
3.4 Applications in biological tissues
To test the biological feasibility, we next compared the bare and coated probes on a fluorescent reporter mouse (Thy1-YFP-H) expressing yellow-fluorescent protein in excitatory neurons. We imaged the cortical layers in coronal sections in a z-stack with an axial spacing of 2 μm by either the bare or coated probes. As expected from the phantom studies, the silver-coated probe showed a brighter fluorescence intensity in the fluorescent neurons by ~2.6 fold [Figs. 4(a) and 4(b)]. Consistently, the SHG emission measured on striated muscles on a murine thigh also showed approximately a two-fold enhancement [Figs. 4(c)], and contrast of the sarcomeric pattern (peak-to-peak signal) was increased by ~1.5 fold. [Figs. 4(d) and 4(e)] [18,19].
Finally, we applied the reflective probe on in vivo imaging of vascular dynamics, which is a hallmark of living mammals [Fig. 5]. In an anesthetized mouse, we performed open craniotomy and subsequently injected fluorescent dextran intravenously to label the blood plasma. We took two-photon fluorescent images from the same field-of-view on the exposed cortex by the bare and coated probes. In agreement with previous in vitro and ex vivo measurements, the coated probe provided brighter images by ~1.8 fold [Fig. 5(a)]. Furthermore, the improved collection efficiency enabled robust quantification of arterial blood flow by a line-scanning method , which was often hampered by the limited signal-to-noise ratio in endoscopic imaging [Fig. 5(b)].
These results collectively suggest that the coated probes will provide immediate practical utilities in biological studies. For example, the doubled collection efficiency can provide an improved signal-to-noise ratio theoretically by ~40% (Poisson statistics) or a two-times faster acquisition at a constant excitation power . Alternatively, the same imaging quality can be obtained at a reduced excitation power by ~40%, which facilitates in reducing phototoxicity and photobleacing for longer-term observations.
We have reported that the simple integration of a reflective waveguide on a GRIN probe can improve the collection efficiency by approximately two-fold in multiphoton endoscopy. The sputter-based protocol offers an efficient and scalable fabrication with low cost. We expect the rapid adaptation of this approach especially for functional imaging on neural activities where the amount of collected signal limits the temporal resolution and signal-to-noise ratio. Strategies proposed to improve the collection efficiency for two-photon microscopy, such as an integrated beam splitter, can also be adapted for improving the collection efficiency further 
M.C. conceived the experiment. J.H., S.L., and P.C. designed and performed the experiments and data analysis. J.W. and Q.D. performed the optical simulation. K. L., J.P., and K-B. Lee fabricated the endoscopic probes. M.C., J.H., S.L, and P.C. cowrote the manuscript with inputs from all authors.
This work was supported by the Institute of Basic Science (IBS-R015-D1); by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2017R1A6A1A03015642); and by the KBRI Basic Research Program through Korea Brain Research Institute funded by the Ministry of Science and ICT (18-BR-03-02).
The authors declare competing financial interests: M.C., J.H., S.L., and P.C. are the inventors of the patent-pending technologies described herein.
References and links
2. M. Choi, S. J. J. Kwok, and S. H. Yun, “In vivo fluorescence microscopy: lessons from observing cell behavior in their native environment,” Physiology (Bethesda) 30(1), 40–49 (2015). [CrossRef] [PubMed]
4. J. K. Kim, W. M. Lee, P. Kim, M. Choi, K. Jung, S. Kim, and S. H. Yun, “Fabrication and operation of GRIN probes for in vivo fluorescence cellular imaging of internal organs in small animals,” Nat. Protoc. 7(8), 1456–1469 (2012). [CrossRef] [PubMed]
5. F. Helmchen, M. S. Fee, D. W. Tank, and W. Denk, “A Miniature Head-Mounted Neurotechnique Two-Photon Microscope. High-Resolution Brain Imaging in Freely Moving Animals,” Neuron 31(6), 903–912 (2001). [CrossRef] [PubMed]
7. J. W. Choi, J. K. Kim, M. Choi, Y. R. Kim, and S. H. Yun, “In vivo imaging of Lgr5-positive cell populations using confocal laser endomicroscopy during early colon tumorigenesis,” Endoscopy 46(12), 1110–1116 (2014). [CrossRef] [PubMed]
9. M. E. Llewellyn, R. P. J. Barretto, S. L. Delp, and M. J. Schnitzer, “Minimally invasive high-speed imaging of sarcomere contractile dynamics in mice and humans,” Nature 454(7205), 784–788 (2008). [CrossRef] [PubMed]
10. C. J. Engelbrecht, W. Göbel, and F. Helmchen, “Enhanced fluorescence signal in nonlinear microscopy through supplementary fiber-optic light collection,” Opt. Express 17(8), 6421–6435 (2009). [CrossRef] [PubMed]
13. C. A. Combs, A. V. Smirnov, J. D. Riley, A. H. Gandjbakhche, J. R. Knutson, and R. S. Balaban, “Optimization of multiphoton excitation microscopy by total emission detection using a parabolic light reflector,” J. Microsc. 228(3), 330–337 (2007). [CrossRef] [PubMed]
14. J. C. Jung, A. D. Mehta, E. Aksay, R. Stepnoski, and M. J. Schnitzer, “In vivo mammalian brain imaging using one- and two-photon fluorescence microendoscopy,” J. Neurophysiol. 92(5), 3121–3133 (2004). [CrossRef] [PubMed]
15. W. Wang, Y. Jiang, Y. Liao, M. Tian, H. Zou, and L. Zhang, “Fabrication of silver-coated silica microspheres through mussel-inspired surface functionalization,” J. Colloid Interface Sci. 358(2), 567–574 (2011). [CrossRef] [PubMed]
16. V. Jayaraman, Y. S. Lin, M. Pakala, and R. Y. Lin, “Fabrication of ultrathin metallic membranes on ceramic supports by sputter deposition,” J. Membr. Sci. 99(1), 89–100 (1995). [CrossRef]
17. D. M. Huland, C. M. Brown, S. S. Howard, D. G. Ouzounov, I. Pavlova, K. Wang, D. R. Rivera, W. W. Webb, and C. Xu, “In vivo imaging of unstained tissues using long gradient index lens multiphoton endoscopic systems,” Biomed. Opt. Express 3(5), 1077–1085 (2012). [CrossRef] [PubMed]
18. C. Odin, T. Guilbert, A. Alkilani, O. P. Boryskina, V. Fleury, and Y. Le Grand, “Collagen and myosin characterization by orientation field second harmonic microscopy,” Opt. Express 16(20), 16151–16165 (2008). [CrossRef] [PubMed]
19. C. Lefort, R. P. O’Connor, V. Blanquet, L. Magnol, H. Kano, V. Tombelaine, P. Lévêque, V. Couderc, and P. Leproux, “Multicolor multiphoton microscopy based on a nanosecond supercontinuum laser source,” J. Biophotonics 9(7), 709–714 (2016). [CrossRef] [PubMed]
20. D. Kleinfeld, P. P. Mitra, F. Helmchen, and W. Denk, “Fluctuations and stimulus-induced changes in blood flow observed in individual capillaries in layers 2 through 4 of rat neocortex,” Proc. Natl. Acad. Sci. U.S.A. 95(26), 15741–15746 (1998). [CrossRef] [PubMed]