Introduction. Common-path optical coherence tomography (CP-OCT) [1] is a well-established microscopy technique with potential applications in micro-imaging [2], endoscopy [1,3–5], needle-based OCT, [1,6,7], and surgical interventions [8–11]. For most of those applications, CP-OCT instruments rely on an optical fiber probe that transmits light into and from the sample via graded-index (GRIN) lenses [12–14], chemically etched axicons [15,16], or ball-shaped lenses [17,18]. (Note: here, we exclusively consider forward-looking probes. For side-looking probes, we refer the reader to, for instance, [19–23].) In this Letter, we introduce a new ball-lensed CP-OCT probe based on an etched optical fiber equipped with a $\approx 65\mathrm{\mu m}$ diameter barium titanate microsphere, which is mounted on the inward cone left by the etching process on the cleaved end of the fiber. Our approach yields probes that have a smaller outer diameter than any of the GRIN-lensed and ball-lensed CP-OCT systems reported so far in the literature. With respect to axicons, our probes offer a larger depth penetration and higher resolution both in air and in buffer liquid solutions, where the high refractive index of the microsphere guarantees sufficient optical contrast at the lens–liquid interface. These features may eventually enable broader utilization of CP-OCT imaging in ultrathin cavities or in applications that might require a set of closely spaced probes.

Experimental details. Our ball-lensed optical fiber probes were fabricated according to the following protocol. The cleaved end of a single-mode fiber (SMF) (Corning SMF-28) is etched for 2 h at room temperature in a buffered solution of hydrofluoric acid (Hydrofluoric acid, 48 wt.% Sigma-Aldrich, The Netherlands) diluted in deionized water (40% v/v) (see Ref. [24]). The etching bath reduces the outer diameter of the fiber and, because of the different etching rate between core and cladding, creates a truncated conical cavity in correspondence of the fiber core (see Fig. 1). The etched fiber is then fixed vertically on an $x-y-z$ stage (NanoMax-300, Thorlabs GmbH) in the field of view of an optical microscope (A-zoom2 40×, QIOPTIQ, USA), where the tip of the fiber is briefly brought to contact with a micropipette covered with fresh droplets of two-component epoxy glue (UHU plusEndfest $n=1.61$). This procedure allows one to consistently deposit a small amount of adhesive in correspondence to the cavity. The tip of the fiber is further brought to contact with a barium titanate microsphere (BTGMS-HI-4.15, Cospheric LLC), which is held in position by means of another micropipette. When the sphere enters into contact with the glue, the capillary forces of the glue automatically drag the sphere into the cavity, resulting in a high degree of reproducibility and self-alignment [25]. Finally, the fibers are left vertical for 90 min to allow the glue to cure.

 

Fig. 1. Illustration of the fabrication steps for production of the optical probe (not to scale): (I) initial cleaved fiber; (II) fiber with a truncated conical cavity after the chemical etching treatment; (III) injection of a two-component epoxy glue by means of a micropipette; (IV) positioning of the barium titanate microsphere; (V) final ball-lensed optical fiber probe.

Download Full Size | PDF

The probes were tested using a commercial spectral domain OCT system (Telesto II series, Thorlabs GmbH, Germany) in common-path mode. The system relies on a superluminescent diode (SLD, D-1300 HP, Superlum, Ireland) with a full width half-maximum (FWHM) of 85 nm and central wavelength of 1310 nm. The output from the SLD is coupled into the probe via a broadband circulator (CIRC-3-31-PB, Gould Fiber Optics, USA). The light reflected at the fiber–probe interface (i.e., the reference arm) and by the sample (i.e., the sample arm) couples back into the fiber and is sent by the circulator into a spectrum analyzer (Thorlabs 1310, Wasatch Photonics, Inc). The data collected are processed using a custom-designed program (Labview, National Instruments, USA). All OCT datasets presented in this work are shown as acquired, without any filter correction applied.

To analyze the performance of our probes, we decided to focus on the following parameters: beam profile, roll-off A-scan test, and imaging capability. The beam profile was measured in air via a scanning slit optical beam profile system (BP209-IR, Thorlabs GmbH, Munich, Germany) with an InGaAs detector and a scanning slit of 25 μm. Measurements were repeated as a function of separation along the direction of the beam propagation $d_p$, where $d_p$ was varied, via a step motor, in steps of 50 μm between $d_p=0.2\mathrm{mm}$ (the minimum value allowed by the scanning slit optical beam profile system) and $d_p=5\mathrm{mm}$. The peak intensity of the roll-off A-scan signal was measured in air and in a phosphate buffered solution (PBS) by positioning a mirror in front of the probe and varying the fiber–mirror separation $d_m$, via a step motor, from $d_m=0.1\mathrm{mm}$ to $d_m=3.5\mathrm{mm}$, in steps of 50 μm in air and 200 μm in PBS. Data were collected without applying any apodization algorithm. To demonstrate the imaging capability of the probe developed, we further carried out 2D cross-sectional OCT scanning microscopy on two samples: a human finger and a mouse brain slice. The first measurement was performed in air, while the second one was performed in PBS.

Results. The fabrication protocol, tested on 50 probes, yields optical fibers with an outer diameter equal to $73\pm 2\mathrm{\mu m}$ and a $16\pm 1\mathrm{\mu m}$ deep conical cavity, equipped with $65\pm 3\mathrm{\mu m}$ diameter spherical lenses (Fig. 2). By observing eight randomly selected probes through the optical microscope, we further found that the glue filling between the inner base of the truncated cavity and the microsphere has a length of $11\pm 1\mathrm{\mu m}$, which does not significantly depend on sphere radius. By using the Fresnel equations to describe the reflection and transmission of the incident light at fiber–glue, glue–lens, and lens–medium boundaries, one can show that, due to the small difference in refractive index between the fiber core and the glue, the reflection from the fiber–glue interface is negligible. Moreover, the reflection from the lens–medium interface is 10 and 3 times stronger than the reflection from the glue–sphere reflection in air and water, respectively. In Fig. 3 (first row) we report the profile images obtained for different values of $d_p$. The circular shape of those images confirms that the profiles are symmetric. Figure 3 (second row) also shows a typical distribution profile obtained with the scanning optical beam setup for $d_p=0.5\mathrm{mm}$. The measured profile matches well with a Gaussian curve in both directions, suggesting that the microsphere is well aligned with the SMF core. Figure 4 finally shows the FWHM of the Gaussian distribution profiles collected in the two orthogonal directions as a function of $d_p$. The values obtained from the analysis of the $x$-axis profile perfectly overlap with those obtained from the analysis of the $y$-axis profile, as expected for a symmetric beam. From the trend of the FWHM, one can also infer that the focal length of our probes is located at $d_p<0.2\mathrm{mm}$. Unfortunately, our setup does not allow us to bring the probe any closer to the detector of the scanning optical beam profiler and, therefore, cannot be used to directly measure the lateral resolution of the system.

 

Fig. 2. Scanning electron micrograph and optical microscope view of a concave cone-etched fiber, (a) and (b); and scanning electron micrograph and optical microscope of one of the completed probes fabricated for this Letter, (c) and (d).

Download Full Size | PDF

 

Fig. 3. Top images: measured beam profile images at 0.5 mm, 1.0 mm, and 1.5 mm distance from the probe end. The scale bar corresponds to 500 μm. Bottom graphs: beam intensity distribution (blue) with Gaussian fit (red) in the $x$ (horizontal) and $y$ (vertical) axes at 0.5 mm distance.

Download Full Size | PDF

 

Fig. 4. Measured FWHM plotted as a function of distance from the optical fiber probe.

Download Full Size | PDF

Figure 5a shows the results obtained with the roll-off experiment in air. As expected for a test performed at $d_m$ larger than the focal length, the averaged ($1000\times$) intensity of the A-scan signal monotonically decreases with the increase of $d_m$. Increasing $d_m$ from 0.1 mm to 3 mm, for instance, results in a drop in sensitivity from 83.38 dB to 48.68 dB. The $-6\mathrm{dB}$ roll-off point is measured around 350 μm. The roll-off data were further used to infer the signal-to-noise ratio (SNR) of our system, which was calculated as the ratio between the amplitude of the A-scan peak and the root-mean-square of the background noise measured when no reflections or backscatters were present. Figure 5b shows the value of SNR obtained with this method as a function of $d_m$. The graph emphasizes that, at close separations, the SNR values are above 50 dB. An $\mathrm{SNR}>30\mathrm{dB}$ is still achievable at an axial distance of 3 mm from the fiber tip, demonstrating the capability of our ultrathin probe to perform well even for extended depth of field. Similar results were obtained when the probes were tested in PBS, as illustrated in Figs. 5c and 5d. In this case, an SNR larger than 35 dB is still achievable for $d_m<1.8\mathrm{mm}$. The beam profile and the roll-off experiments, repeated two times with three randomly selected, different probes, show that the variability between each probe is negligible.

 

Fig. 5. Roll-off A-scan and SNR performance in air, (a) and (b), and in PBS, (c) and (d), plotted as a function of the distance between the end of the probe and the reflective mirror positioned in front of it.

Download Full Size | PDF

Figure 6 shows in vivo images of a human finger and a mouse brain slice obtained with our system in air and PBS, respectively. The image quality is comparable to that offered by commercial systems, although the penetration depth is limited to approximately 1.5 mm.

 

Fig. 6. (a) OCT B-scan images of human fingertips in air at a distance of 0.5 mm and (b) of mouse brain in PBS at a distance of 1 mm from the probe. The white arrow in the bottom image indicates the glass substrate. B-scans were obtained by translation of the sample stage.

Download Full Size | PDF

Discussion. The results reported above show that, following the protocol described above, it is possible to reproducibly fabricate thin optical fiber probes for CP-OCT imaging. The fabrication procedure is relatively simple and cost effective, and guarantees a good alignment between the core of the fiber and the focusing lens, as demonstrated by visual inspection and by the symmetry of the beam profile measured at the exit of the probe. Our roll-off measurements further prove that, in air, the intensity signal and the SNR are only 4 dB and 33 dB worse than the ones reported for larger lenses [18], respectively, and largely better than the ones obtained with axicons [14]. For distances larger than 200 μm, our probes have a lower lateral resolution than the previously reported higher NA ones, but have a much smaller diameter ($<75\mathrm{\mu m}$), which makes them ideal for in vivo imaging deep into the tissue. However, a higher lateral resolution could be obtained by using a larger sphere with higher NA and/or with a lower refractive index. Importantly, due to the high refractive index of the lens used ($n=1.95$), the performance of the system does not deteriorate when the probe is immersed in a liquid—a major advantage for applications in life science experiments.

Conclusion. We described a novel approach to fabricate ultrathin microlensed probes for CP-OCT imaging. We showed that our probes allow successful acquisition of OCT images in terms of SNR and penetration depth, in both air and liquids. Based on our findings, we believe that by tuning the main parameters of our probe (i.e., microsphere radius, refractive indices, and concave cone dimensions), it is possible to tailor our probe design to achieve the desired imaging performance for different applications. Our preliminary findings might pave the way for the development of new instruments for in-depth tissue analysis, with potential applications, among others, in the medical field, where small diameter probes might be integrated into next-generation of minimally invasive tools.