Open Access
Add to BibSonomyAbstract
Based on an advanced silicon optical bench technology with integrated MOEMS (Micro-Opto-Electro-Mechanical-System) components, a piezo-driven fiber scanner for confocal microscopy has been developed. This highly-miniaturized technology allows integration into an endoscope with a total outer probe diameter of 2.5 mm. The system features a hydraulically-driven varifocal lens providing axial confocal scanning without any translational movement of components. The demonstrated resolutions are 1.7 μm laterally and 19 μm axially.
© 2014 Optical Society of America
1. Introduction
Optical biopsy for differentiating healthy and cancerous tissue by optical methods becomes more and more method of choice since conventional biopsy is very often an exhausting procedure for the patients [1]. Due to the three-dimensional (3D) nature of tissue, optical biopsy requires 3D imaging techniques like confocal microscopy or optical coherence tomography (OCT), which are both scanning imaging methods. The integration of scan mechanisms in an endoscopic probe, however, is an elementary challenge.
Scanning microscopes basically consist of a scanning mechanism, an illumination and detection unit as well as illumination and detection optics. The illumination and detection unit can be placed outside of the body, and light is transferred to the probe via optical fibers [2]. It is favorable to use one micro-objective for both, illumination and detection, to realize a space-saving architecture. The complexity of micro-objectives range from carefully designed systems consisting of several lenses [3, 4] to simple gradient-index (GRIN) lens systems [5]. The first ones allow aberration correction and adaption to fiber properties. GRIN lenses have very small size and no spherical aberrations by nature. In our design, we have chosen a combination of a GRIN lens and a varifocal liquid-filled membrane lens which allows to match the numerical aperture (NA) of fiber and objective, and which can be used as axial scan mechanism.
Lateral scanning can be performed either inside or outside the body [2]. In the latter case, the lateral information is transmitted via a fiber bundle. Imaging resolution is restricted by the Nyquist sampling theorem and the distance between two cores in the fiber bundle. Scan pattern generation at the distal end of the endoscope can be realized by fiber scanners or micro scan-mirrors. Whereas the latter ones are suitable for sideward-looking probes [6], fiber scanners can be used in forward-looking endoscopes. A fiber scanner basically consists of the distal end of a fiber which is oscillating in a spiral mode and driven near its resonance frequency by a piezo tube [7–11] or electromagnetically [12]. We have chosen a piezo-driven fiber scanner because it is smaller and less complex in fabrication.
The axial scan can be realized by object scanning and focal spot scanning. Hydraulic suction of tissue is an example for object scanning [13]; focal spot scanning has been demonstrated by many methods. First, the entire probe could be moved stepwise in axial direction. This method shows very low precision and is difficult to handle. Changing the distance between fiber and objective, is another option [4], but requires complex mechanics and moving parts along the entire length of the endoscope. The same principle has been realized by use of multiple fibers with slightly shifted axial positions [14]. This allows multifocal imaging, but the number of axial sample points is limited by the number of fibers and thus the diameter of the endoscope increases with axial scan range. A third option is the variation of refractive power of the objective lens. Recently, an axially tunable system was introduced by Jabbour et al. [15], where an electrically tunable lens is used in a macroscopic confocal microscopy setup. Another approach employs infrared light for tuning varifocal lenses which has been tested in an endoscopic application [16]. In our approach, we realize tunability by a liquid-filled tunable membrane lens. To our best knowledge, for the first time axial scanning in confocal microscopy has been realized without any translational movement of optical components in an endoscopic probe.
In the present work, an endoscopic probe based on an advanced silicon bench technology [17] has been designed and fabricated. It combines a piezo-driven fiber scanner and a varifocal membrane lens to realize 3D scanning without movable parts. The fiber-coupled probe focuses laser light by a micro-objective consisting of a GRIN lens and a tunable membrane lens. Backscattered light is collected by the same optics and transferred via the same single mode fiber to the detector. The fiber core acts as confocal pinhole and allows the probe to capture 3D confocal images. This article describes the design, fabrication and characterization of a miniaturized 3D scanner for confocal microscopy with outer dimensions of 1.5 × 1.78 × 13.1 mm3. Figure 1 gives an overview of the fundamental design concept with all essential components.
2. Design
The silicon bench illustrated in Figure 1 houses mechnical and optical components. A piezo scan tube with an attached optical fiber builds up the scanning mechanism. Applying different voltages to the four electrodes of the piezo deflects the one side of the tube in x- and y-direction. The deflection is to small to be used as scan pattern. To reach a magnification of the pattern the driving signals are chosen to hit the resonance frequency of the optical fiber. A membrane lens in combination with a GRIN lens are used as varifocal objective. The GRIN lens delivers high NA while the refractive power of the membrane lens can be tuned by volume change of the liquid inside the lens. Therefore liquid channels are buried on the bottom of the optical bench. In this section optimization of all free design parameters is discused.
2.1. Optical design
The optical design can be defined by five specifications as stated subsequently:
- working distance: 400 – 500 μm
- axial scan range: 100 – 200 μm
- field of view (FOV): 200 μm in diameter
- lateral resolution: 2 μm
- axial resolution: 25 μm
The optical design is a straight-forward result of few boundary conditions. Firstly, the penetration depth in confocal microscopy is limited to about 200 μm due to absorption and scattering which leads to the axial scan range. The numerical aperture of the system is limited by the single mode fiber. Accordingly, there is always a tradeoff between high resolution and large field of view. Our optical performance aims to resolve cellular structures in the order of 2 μm, consequently all other parameters and distances are predefined, as listed above.
For a space-saving design, we combine a GRIN objective lens (GRINTECH) with length and diameter of 2.41 mm and 1 mm, respectively, with a self-fabricated, liquid-filled plano-convex membrane lens. The optical design benefits from both high NA of the GRIN-lens and focal tuning of the membrane lens.
In general, there are four configurations to arrange the optical elements, of which two fundamental ones are shown in Figure 2; in the other two arrangements, the plane surface of the tunable lens faces the GRIN lens. The optical design is optimized by ray tracing (Zemax), where the distances x1 and x2, the membrane curvature and the fiber deflection are varied for all configurations. The profiles of the membrane lenses have been measured by profilometry at different pressure values, and hence different curvature, and have been taken as input for the ray tracing analysis. It has been shown that the arrangements ’GRIN lens left’ (b) deliver high FOV but low resolution. The combinations ’GRIN lens right’ (a) are more suitable for high resolution approaches and turn out to be the best choice for the postulated specifications. Since the membrane lens is employed with positive pressure, the lens has to be located as illustrated in Figure 2 to ensure that the plane side faces the higher NA.
Fig. 2 Two fundamental arrangements of the optical elements which have been analyzed by ray tracing simulations. Configuration (a) yields a high resolution, configuration (b) generates a larger field of view but lower resolution.
Axial resolution Δx in confocal microscopy is given by [18]
with n, the refractive index of the immersion medium, assumed to be 1 in the following discussion, and λ the wavelength. With Equation 1 and specification e), the NA is not allowed to be smaller than 0.18 if a helium-neon laser (λ = 633nm) is used for illumination. The single mode fiber with NA = 0.13 makes a magnification of the objective greater than 1.38 for the illumination direction necessary. Considering specification c) and a maximum mechanically stable and reliable scan-diameter of the fiber of 300 μm, the magnification has to be less than 1.5 to achieve the required FOV.The lateral resolution of the system in case of confocal microscopy is given by
yielding an expected value of 1.5 μm for a numerical aperture of NA = 0.18.To fulfill the specifications, the two distances x1 and x2 between fiber and varifocal lens and between the two lenses have to be optimized (s. Figure 2). Ray tracing shows that the imaging quality increases if the distance x2 is as small as possible. To ensure free movement of the membrane, x2 was chosen to be 0.3 mm. Distance x1 is optimized with a careful look on the simulation results in Figure 3. Both axial scan range and magnification are directly related to x1. The axial scan range corresponds to the shift of the focal plane for a change of the pressure in the membrane lens from 1 to 20 kPa. For x1 = 0.7mm, the axial scan range fulfills specification b) and the NA is in the required range. In Figure 3 (b), the maximum spatial frequency which is captured by the objective with a contrast of 50 % is plotted over x1 for different deflections of the fiber. For high deflections, the frequency drops if x1 is greater than 0.8 mm because a part of the beam is truncated and does no longer enter the membrane lens, hence reducing the effective NA. In conclusion, x1 has been set to 0.7 mm.
Fig. 3 (a) Axial scan range and magnification as a function of the design parameter x1. The optical setup is illustrated in Fig. 2(a). x1 = 0.7mm results in a magnification of 1.43 and a scan range of more than 200 μm. (b) Maximum spatial frequency with a contrast of more than 50% as a function of the design parameter x1 for different fiber deflections. x1 = 0.8mm provides the highest cut-off frequency for a fiber deflection of 150 μm.
In a next step, the membrane thickness of the fluidic lens has been optimized. A thickness of 22 μm yields the best homogeneous image quality over the entire axial scan range. It can be shown that for such a thin membrane a pressure range from 0 to 14 kPa is adequate to achieve an axial scan range of 200 μm.
A final characterization of the design by Zemax determines an objective NA of 0.183 and 0.196 for pressures on the membrane lens of 0 and 14 kPa, respectively. Corresponding magnifications are 1.41 and 1.51 with respect to the fiber NA. The simulated ratio between the maximum intensity in the center of the spot and the central intensity directly behind the fiber is minimal for a pressure of 2 kPa and amounts 1.7 for the non-deflected fiber and 0.7 in case of maximum fiber deflection. For a pressure of 14 kPa, this ratio becomes maximal with values of 2.0 (non-deflected) and 1.3 (maximum deflection). Due to field curvature, the fiber deflection yields a slight axial spot displacement which is less than 15 μm for maximum fiber deflection. Table 1 briefly summarizes the most significant optical design parameters and deduced results.
Table 1. Summarization of the most significant optical design parameters and design results. n: refractive index, t: thickness, Δf : focal tuning range, M: magnification, x1: distance fiber–fluidic lens, x2: distance fluidic lens–GRIN lens, Δz: axial resolution, Δx: lateral resolution.
2.2. Mechanical design
The basic actuation principle is sketched in Figure 4 and shows how the optical fiber is deflected by a four-quadrant piezo tube.
Fig. 4 Schematic of the actuation of the piezo tube with integrated optical fiber. The protruding fiber end performs a linear, circular or spiral movement dependent on the applied voltages and phase difference.
Three major points determine the mechanical design of the fiber scanner: First, the fiber tip should be deflected by at least 150 μm; second, the ratio between maximum sampling rate of the detector and resonance frequency of the scanner should be high enough to fulfill Nyquist’s sampling theorem; the third point concerns the break resistance of the fiber.
Both break resistance and resonance frequency are directly related to the fiber length. A protruding fiber length of 3 mm leads to a resonance frequency of 11.2 kHz under consideration of Young’s modulus and volumetric mass density as mentioned below and a fiber cross-sectional diameter of 125 μm. The maximum sampling rate of the detector (Thorlabs: PDA8A/M) is 50 MHz, thus more than 4000 samples are recorded within one oscillation. The FOV diameter is 200 μm (fiber deflection is ±150 μm), hence the maximum distance between two sample points is 141 nm for the circular steady-state oscillation and fulfills the Nyquist criterion. In case of 3 mm fiber length, the minimum bending radius of the fiber is 17 mm for a fiber deflection of 150 μm which is above minimum bending radius specified by the manufacturer.
To estimate the gain between piezo movement and fiber deflection, a Young’s modulus and density of the fiber material of 73.1 GPa and 2203 kg/m3 have been assumed. Damping is described by a loss factor which is not a material property, but depends on structure and environmental conditions. In literature, one does not find fixed values for this and in particular, there are no values for microscopic glass structures. For macroscopic structures, values range from 0.6 · 10−3 to 2 · 10−3 [19]. To be on the safer side, we chose the highest value for macroscopic structures as a worst case value in the design phase; the real value has to be determined by measurements. The assumed values lead to a gain of 780, thus the piezo has to be deflected by 192 nm to fulfill the specification of ±150μm deflection. The piezo tubes, made of PIC 255, have an outer and inner diameter of 1.5 mm and 0.9 mm, respectively. The inner cylindrical surface is completely covered with one electrode, the outer surface features four electrodes covering cylindrical segments of 45° over the entire length. For a piezo length of 4 mm, an acceptable voltage of 91 V is needed to achieve the required deflection [20]. Table 2 lists the most important mechanical design parameters and derived results.
Table 2. Most relevant mechanical design parameters and design outcome. Δr: max. deflection of the piezo, Umax: maximum voltage, E: Young’s modulus, ρ: densitiy, AΔr: maximum deflection of the fiber (A is the gain), fres: calculated resonance frequency.
3. Fabrication
3.1. Si components
Almost all components of the fiber scanner, including the alignment platform, the tunable lens, and the mechanical holders for the GRIN lens and the optical fiber, are realized by standard silicon bulk micromachining. The fabrication technology and assembly procedure is described in various articles before and can be found in [17, 21, 22].
The silicon micro-bench, which is the assembly and alignment platform for all optical components, is fabricated by various wet and dry etch steps. It is a miniaturized high-precision optical bench with integrated fluidics, optics, mechanics and electrical components. This highly modular and flexible approach allows the realization of a variety of optical designs.
The tunable lens is based on a liquid-filled circular cavity in a square silicon frame with two fluid channels, sealed on one side by a Pyrex glass chip and confined on the other side by a 22 μm thin, highly elastic, and transparent PDMS (polydimethylsiloxane) membrane (Sylgard 184, Dow Corning). The clear optical aperture has a diameter of 1 mm.
3.2. Commercial components
The piezo unit is a customized commercial component. Dimensions of the tube and the design of its electrodes can be find in subsection 2.2. One end of the piezo is glued inside a KOH etched V-groove, the other side houses the optical fiber. To ensure optimal centering, the fiber is glued (Araldite® 2020) in a Si holder which has been designed as illustrated in Figure 5.
Fig. 5 (a) Photograph of single Si MEMS components fabricated for the confocal fiber scanner. (b) Photograph of the fully assembled optical micro-bench. The components were glued by a stamping process. The electrical connection was realized by reflow soldering of copper wires. The single mode fiber was aligned and fixed without any internal stress. A silicone tubing connects the internal fluid channels with an external pressure transducer.
The employed GRIN lens (GRINTECH GT-IFRL-100-010-50-CC) with dimensions as described above has a numerical aperture of NA = 0.5. The optical fiber is a single mode fiber (Thorlabs SM600) for λ = 633nm with NA = 0.13.
All fabricated Si components and the finally assembled probe are shown in Figure 5.
3.3. Fluidic actuation
For precisely controlling the axial scan by the varifocal optical system, a suitable actuation mechanism has to be designed under consideration of the required minimum axial focus displacement. The specification for the axial resolution is 25 μm, hence the focus has to be shifted in steps of less than 12.5 μm to fulfill the Nyquist criterion. To provide a slight oversampling, the minimum focal steps have been set to 5 μm.
Actuation of the varifocal lens may be performed by either pressure control [23], or temperature control [24], or by volume displacement. The latter one is the most accurate and reproducible method, as it does not depend on any variations of the membrane or fluidic channels, and it is hardly affected by environmental changes. In the present approach, we have selected volume displacement as the preferred method using a highly controlled piston actuator. A stepping motor drives via a fine thread a slide which is connected to the piston of a syringe. The achievable volume resolution is 0.7 nl, details concerning actuation are described in subsection 4.2.
4. Measurement
4.1. Mechanical performance
The piezo scanner and hence the tip of the fiber driven in resonance can be deflected in two directions. If harmonic signals are applied to the electrodes of the piezo, Lissajous figures are generated, as for example a simple circle. Because of its simplicity, this steady-state mode is used for first characterizations of the piezo scanner. To generate a scan pattern which covers the complete field of view, the amplitude of the harmonic signals is varied. By linear changes of the amplitude, a spiral pattern is generated as illustrated in Figure 6. This swinging mode is in a non-steady state and thus much more challenging to be measured. At the end of this section, a method to observe the non-steady-state response of the piezo scanner is presented.
Fig. 6 The voltage signals applied to the piezo electrodes responsible for x and y deflection are illustrated in red and blue. The scan patterns generated by the given driving signals are shown below.
As mentioned before the mechanical behavior of the fiber scanner has been analyzed first in a steady-state oscillation mode at discrete amplitudes. The measurement setup for characterizing the frequency response is shown in Figure 7. The focal spot is magnified via a telescope by a factor of 20 and imaged onto a CCD camera. Operating the camera with an exposure time much higher than the oscillation period, a circular image results. From the CCD images, the oscillation radii are determined for specific frequencies and different amplitudes. Figure 8 shows the corresponding frequency response. The resonance frequency shows a slight shift towards smaller frequencies for increasing amplitudes due to increased damping at larger fiber deflection.
Fig. 7 Schematic of the setup used for measuring fiber deflection and axial focus position. Fiber deflection is measured in a steady-state. The telescope amplifies the radius of the circular oscillation by a factor of 20.
Fig. 8 Measured frequency response of the circular oscillation of the fiber at different voltage amplitudes applied to the electrodes of the piezo tube. The resonance frequency shifts towards smaller values with increasing amplitude.
The spiral scan pattern requires voltage amplitudes ranging from 0 to 135 V. A resonance frequency at medium voltage is determined as fixed scan frequency. This results in a slight non-linear behavior of the scan radius for varying amplitudes. As illustrated in Figure 9, the highest slope is located in the medium voltage range. Since the scan frequency is fixed, the fiber is operated in resonance only at medium amplitude. Furthermore, Figure 9 shows that the actual damping factor is much higher than expected. The simulation value of 0.002 from literature results in a much higher slope. The virtual behavior is best described by a value of 0.006. Nevertheless, the expected field of view could be achieved despite a much stronger damping of 0.006 since the fiber deflection is larger in the non-steady-state mode (spiral).
Fig. 9 Radius r of measured and simulated circular oscillation as a function of applied actuation voltage for a frequency around 9.9 kHz. For the first simulation, a damping ratio of 0.002 has been assumed according to literature. Subsequent simulations show that a ratio of 0.006 much better describes the fiber rotation.
The steady-state measurements deliver basic information about the expected signal pattern. In a scanning mode, as in the present example with a spiral scan pattern, the fiber deflection as a response to the actuation amplitude is expected to show a non-linear behavior. To resolve the transient response of the fiber, sampling at at least 20 kHz (two times the expected scanning frequency) is demanded. This can be measured by position-sensitive diodes. In the context of this work, we determined the fiber response by a method which does not require any additional hardware.
To estimate the shape of the spiral pattern, reflections of a plane glass surface have been measured. The setup for this measurement is discussed in detail in subsection 4.3. Since deflection of the fiber causes a difference in optical path length, interference of light reflected at the surface and inside the probe generates a fringe pattern. Figure 10 gives two examples recorded at slightly different frequencies. For figure (a), the steady-state scan frequency has been chosen. Even small changes in the scan frequency result in a completely different pattern as illustrated in (b). The scanned field of view is increased as the larger number of fringes shows. In an ideal case, the fringes should appear as perfect circles. Deviations from an optimum circle are caused by direction-dependent damping which originates from small lacks in symmetry in the fabricated device, however the corresponding distortion can be reduced by data post processing (see subsection 4.3). In case of real biological samples, these interferences are of subordinate importance, since the method is intended to be used for fluorescence microscopy.
Fig. 10 Image of a planar reflective surface. The fringes originate from interferences between reflections from the surface of the sample and from inner reflections of the fiber scanner. The surface was imaged at two different scan frequencies. The first interference maximum is marked by a dashed black line in both images. Each ring in both images is measured with the same fiber deflection. Thus, the more rings can be seen, the larger the field of view. Image (a) was recorded with a scan frequency of 10.01 kHz equal to the resonant frequency in steady-state condition. For image (b), the scan frequency was changed to 10.09 kHz. In a non-steady state, the resonance frequency changes and leads to a maximum fiber deflection at other conditions.
4.2. Focal length tuning
For the axial focal tuning, the micro-lens was filled with an optically clear liquid (FC-40, 3M™). For the characterization of the axial focal tuning, a volume of 78 nl was initially pumped into the system to ensure a slight tension of the silicone membrane, which is not fully stretched at the beginning. After 30 s, the system had reached a steady state and the measurements were started. By moving the CCD camera on a linear stage, the current position of the focus was detected with the setup presented in Figure 7. The lens was actuated in steps of 5 nl, and the focus was detected after a waiting period of 10 s each. To prevent damage of the membrane, a maximum volume of 240 nl was chosen. The speed of the focal tuning can easily be improved by an integrated actuation scheme, as for example in [25].
As the fluidic tubes slightly expand during actuation and air bubbles cannot be totally avoided, the volume control differs from theoretical predictions. Even so, the focal length varies linearly with the actuated volume with high repeatability. Figure 11 shows the averaged focus of six volume actuations and the corresponding standard deviation as error bars. The sensitivity from a linear fit turns out to be 0.7 μm/nl with a standard deviation of 0.11 μm/nl. The focal length could be tuned over a range of 100 μm with a volume of less than 150 nl.
Fig. 11 Axial displacement of the focus as a function of the actuated volume. The focal position is the working distance behind the GRIN lens. After filling the lens and closing the vent duct, an initial volume of 78 nl is required to ensure a slight tension in the membrane. A further increase in volume of about 150 nl shifts the focus by about 100 μm towards the last facet of the GRIN lens. The error bars reflect the precision (± one standard deviation) due to 6 different sweeps.
4.3. Imaging performance
For the characterization of the optical performance and functionality, the piezo-driven fiber scanner was tested in the measurement setup illustrated in Figure 12. Light from a HeNe laser with λ = 633 nm is coupled via a 50:50 fiber coupler into the fiber scanner and focused onto the sample. The second arm of the fiber coupler is connected to a photodiode to control the laser intensity and compensate for fluctuations. The backscattered and reflected light from the sample is captured by a detector (PDA8A/M, Thorlabs) with 50 MHz bandwidth. According to the spiral movement with parametric representation x(t) = A(t)cos(ωt), y(t) = A(t)sin(ωt), the spatial information can be assigned to the time-resolved detected signal.
Fig. 12 Schematic of the measurement setup for confocal imaging with a HeNe laser as light source. The signal of the photodiode is used for compensating instabilities of the laser output power. The signal of the detector is proportional to the amount of backscattered and reflected light from the sample.
Image distortion
To assign and calibrate the lateral x and y coordinates, a well-known regular pattern is taken as target. As shown in Figure 13, the confocal image of a regular grid shows distortion which originates first from asymmetric changes in the resonance frequency at different actuation voltages during the spiral scan and second from uneven deflection of the piezo tube in x and y direction. The effect is largest in the center of the image, where a curl-like distortion can be observed; it changes to a pincushion distortion towards the edges. These distortions are strongly dependent on the applied actuation frequency, but show stable and repeatable behavior at a fixed frequency.
Fig. 13 Confocal image of a reflective grid with 1.7 μm lines and 10 μm period (left). The middle inset shows a microscope image of the grid. Comparison of both images provides information about distortion of the confocal scan. The right graph shows the intensity profile of a line grating with line widths of 2 μm at a period of 4 μm.
Actually the optical system measures voltages which correspond to stray light intensities over time. In post image processing, x- and y-coordinates are assigned to each temporal point due to
where ω is the scan frequency which is highly stable during the entire measurement and well-defined by the driving frequency. Ideally, A(t) and B(t) are linear functions with identical slopes. Non-linearities of A(t) and B(t) cause distortions in the image. To calibrate the lateral coordinates, we suggest an iterative procedure, where A(t) and B(t) can be optimized to reach highest correlation of images taken by the piezo scanner and a commercial microscope. A suitable target for this is a regular reflective grid as shown in Figure 13.Lateral resolution
The right part of Figure 13 shows a 1D cross-sectional intensity profile from the confocal image of a line grating with line widths of 2 μm at a period of 4 μm. For high deflection angles, the reflected light does not reach the aperture of the fiber due to the reflective properties of the grating lines. In case of scattering samples such as biological tissue or turbid media, photons will enter the fiber even at higher angles. In the present example, the light intensity of the scanned reflective grating decreases strongly with increasing distance from the center of the image.
The lateral resolution Δx of the scanner was determined by imaging reflective Cr line gratings on a glass substrate with decreasing line width and period. According to the design and predicted by the optical layout, the fiber scanner is able to resolve structures below 2 μm.
Axial resolution
The axial resolution of the system was determined by a series of confocal 3D images from a reflective grating with 8 μm wide lines and averaged intensity of each axial layer. The tunable membrane lens was actuated, as described in Section 4.2, to perform the axial scanning. At the beginning, the focus was placed behind the grating. Then, the focal length was reduced in steps of 3.5 μm. The measured data were fit by a Gaussian, as illustrated in Figure 14.
Fig. 14 Averaged axial intensity of a reflective surface. The measured data are approximated by a Gaussian. FWHM was determined to be 19 μm, providing a measure for the axial resolution of the confocal system.
Since the focal spot was smaller than the illuminated line of the grating, the grating worked as a reflecting surface, and full width at half maximum (FWHM) of the fit complies with the axial resolution of the system. The fiber scanner was designed for an axial resolution of 25 μm. The measurement reveals a resolution of 19.4 ± 3.1μm, which even exceeds the requirements. The uncertainties come from the precision of the actuation of the lens, as described in Section 4.2.
Depth discrimination
Depth discrimination is demonstrated with the same measurement principle as described in the previous paragraph. The target here is a 3D electroplated gold structure with two layers of different height, as shown in Figure 15. Layer 1 has a meander shape and a height of 3.5 μm, layer 2 is a flat surface with a height of 42 μm. The field of view of the confocal scans was in the range of 320 μm. During the 3D confocal scan, the focal plane was shifted by about 66.5μm in steps of 3.5μm. In image (a), the focus was adjusted on top of the thinner layer, in (b) between both layers, and in (c) on top of the thick flat layer. The intensity of both layers was integrated separately for all 20 confocal slices and plotted versus the position of the focal plane. As seen in the graphs of Figure 16, the two electroplated layers are clearly separated from each other. The graph also shows, that the peak positions of the averaged intensities highly correlate with the axial positions of the reflecting electroplated layers.
Fig. 15 Left: Microscope image of a 3D gold structure. The flat area has a thickness of 42 μm, the meander structure of 3.5 μm. Right: Confocal 2D images of the 3D structure. In image (a), the confocal plane was positioned exactly on the thinner meander layer, in (b) between both layers, and in (c) on top of the thick flat layer.
Fig. 16 The intensity of the two layers of the 3D structure from Figure 15 was integrated at 20 different focal positions (abscissa). The result clearly demonstrates the modality of depth discrimination. The focus was shifted backwards from infinity (zero pressure) towards the probe (highest actuation) by about 66.5μm in steps of 3.5μm.
5. Conclusion
By MOEMS technology, thereby implementing a tunable membrane lens and a four-quadrant piezo tube with incorporated optical fiber in an optofluidic silicon micro-bench, it was possible to realize a small endoscopic probe for confocal microscopy with a diameter of less than 2.5 mm. Resonant spiral excitation of the optical fiber yields a field of view of up to 320 μm at a working distance of 0.5 mm. The tunable lens was used to allow confocal axial scanning over a range of 100 μm without any movement of the entire endoscopic probe or the sample. A GRIN objective with high NA primarily defines the refractive power and yields a resolution of 2 μm in lateral direction. Hence, the system has the potential to resolve cellular micro-structures. The axial resolution is better than 20 μm and allows confocal 3D volumetric imaging.
Acknowledgments
This work was supported by the German Research Foundation (DFG). The article processing charge was funded by the open access publication fund of the Albert Ludwigs University Freiburg.
References and links
1. T. D. Wang and J. Van Dam, “Optical biopsy: a new frontier in endoscopic detection and diagnosis,” Clin. Gastroenterol. Hepatol. 2(9), 744–753 (2004). [CrossRef] [PubMed]
2. F. Helmchen, “Miniaturization of fluorescence microscopes using fibre optics,” Exp. Physiol. 87(6), 737–745 (2002). [CrossRef] [PubMed]
3. Y. S. Sabharwal, A. R. Rouse, L. Donaldson, M. F. Hopkins, and A. F. Gmitro, “Slit-scanning confocal microendoscope for high-resolution in vivo imaging,” Appl. Opt. 38(34), 7133–7144 (1999). [CrossRef]
4. A. R. Rouse, A. Kano, J. A. Udovich, S. M. Kroto, and A. F. Gmitro, “Design and demonstration of a miniature catheter for a confocal microendoscope,” Appl. Opt. 43(31), 5763–5771 (2004). [CrossRef] [PubMed]
5. J. Knittel, L. Schnieder, B. Buess, B. Messerschmidt, and T. Possner, “Endoscope-compatible confocal microscope using a gradient index-lens system,” Opt. Commun. 188(5–6), 267–273 (2001). [CrossRef]
6. D. L. Dickensheets and G. S. Kino, “Micromachined scanning confocal optical microscope,” Opt. Lett. 21(10), 764–766 (1996). [CrossRef] [PubMed]
7. C. M. Lee, C. J. Engelbrecht, T. D. Soper, F. Helmchen, and E. J. Seibel, “Scanning fiber endoscopy with highly flexible, 1 mm catheterscopes for wide-field, full-color imaging,” J. Biophoton. 3(5–6), 385–407 (2010). [CrossRef]
8. E. J. Seibel, Q. Y. Smithwick, C. M. Brown, and P. G. Reinhall, “Single-fiber flexible endoscope: general design for small size, high resolution, and wide field of view,” Proc. SPIE 4158, 29 (2001). [CrossRef]
9. E. J. Seibel and Q. Y. J. Smithwick, “Unique features of optical scanning, single fiber endoscopy,” Lasers Surg. Med. 30(3), 177–183 (2002). [CrossRef] [PubMed]
10. E. J. Seibel, R. S. Johnston, and C. D. Melville, “A full-color scanning fiber endoscope,” Proc. SPIE 6083, 608303 (2006). [CrossRef]
11. N. Zhang, T.-H. Tsai, O. O. Ahsen, K. Liang, H.-C. Lee, P. Xue, X. Li, and J. G. Fujimoto, “Compact piezoelectric transducer fiber scanning probe for optical coherence tomography,” Opt. Lett. 39(2), 186–188 (2014). [CrossRef] [PubMed]
12. T. Matsunaga, R. Hino, W. Makishi, M. Esashi, and Y. Haga, “Electromagnetically driven ultra-miniature single fiber scanner for high-resolution endoscopy fabricated on cylindrical substrates using MEMS process,” IEEE 23rd International Conference on Micro Electro Mechanical Systems (MEMS), 999–1002 (2010).
13. K. C. Maitland, A. M. Gillenwater, M. D. Williams, A. K. El-Naggar, M. R. Descour, and R. R. Richards-Kortum, “In vivo imaging of oral neoplasia using a miniaturized fiber optic confocal reflectance microscope,” Oral Oncol. 44(11), 1059–1066 (2008). [CrossRef] [PubMed]
14. D. R. Rivera, C. M. Brown, D. G. Ouzounov, W. W. Webb, and C. Xu, “Multifocal multiphoton endoscope,” Opt. Lett. 37(8), 1349–1351 (2012). [CrossRef] [PubMed]
15. J. M. Jabbour, B. H. Malik, C. Olsovsky, R. Cuenca, S. Cheng, J. A. Jo, Y.-S. L. Cheng, J. M. Wright, and K. C. Maitland, “Optical axial scanning in confocal microscopy using an electrically tunable lens,” Biomed. Opt. Express 5(2), 645–652 (2014). [CrossRef] [PubMed]
16. X. Zeng, C. T. Smith, J. C. Gould, C. P. Heise, and H. Jiang, “Fiber endoscopes utilizing liquid tunable-focus microlenses actuated through infrared light,” J. Microelectromech. Syst. 20(3), 583–593 (2011). [CrossRef]
17. N. Weber, H. Zappe, and A. Seifert, “A tunable optofluidic silicon optical bench,” J. Microelectromech. Syst. 21(6), 1357–1364 (2012). [CrossRef]
18. J. B. Pawley, Handbook of Biologic Confocal Microscopy (Springer, 2006). [CrossRef]
19. L. Cremer and M. Heckl, Structure-Borne Sound (Springer, 1988). [CrossRef]
20. C. J. Chen, “Electromechanical deflection of piezoelectric tubes with quartered electrodes,” Appl. Phys. Lett. 60, 132–134 (1992). [CrossRef]
21. N. Weber, H. Zappe, and A. Seifert, “Endoscopic optical probes for linear and rotational scanning,” in 2013 IEEE 26th International Conference on Micro Electro Mechanical Systems (MEMS), 1065–1068 (2013).
22. N. Weber, H. Zappe, and A. Seifert, “High-precision optical & fluidic micro-bench for endoscopic imaging,” in 2010 International Conference on Optical MEMS and Nanophotonics (OMN), 85–86 (2010).
23. P. Waibel, D. Mader, P. Liebetraut, H. Zappe, and A. Seifert, “Chromatic aberration control for tunable all-silicone membrane microlenses,” Opt. Express 19(19), 18584–18592 (2011). [CrossRef] [PubMed]
24. W. Zhang, H. Zappe, and A. Seifert, “Wafer-scale fabricated thermo-pneumatically tunable microlenses,” Light Sci. Appl. 3, e145–e150 (2014). [CrossRef]
25. P.-H. Cu-Nguyen, A. Grewe, M. Hillenbrand, S. Sinzinger, A. Seifert, and H. Zappe, “Tunable hyperchromatic lens system for confocal hyperspectral imaging,” Opt. Express 21(23), 27611–27621 (2013). [CrossRef]
