Achromatic miniature lens system for coherent Raman scattering microscopy

eScholarship provides open access, scholarly publishing services to the University of California and delivers a dynamic research platform to scholars worldwide. Abstract: We discuss the design and performance of a miniature objective lens optimized for coherent Raman scattering microscopy. The packaged lens assembly has a numerical aperture of 0.51 in water and an outer diameter of 8 mm. The lens system exhibits minimum chromatic aberrations, and produces coherent Raman scattering images with sub-micrometer lateral resolution (0.648 µ m) using near-infrared excitation pulses. We demonstrate that despite the small dimensions of the miniature objective, the performance of this lens system is comparable to standard microscope objective lenses, offering opportunities for miniaturizing coherent Raman scattering imaging probes without sacriﬁcing the image quality.


Introduction
Coherent Raman scattering (CRS) microscopy is a nonlinear optical imaging technique, including both coherent anti-Stokes Raman scattering (CARS) and stimulated Raman scattering (SRS), which provides label-free, molecular selective contrast at fast image acquisition rates. [1] A number of examples illustrate the potential of CRS for in vivo biomedical research. For instance, CRS has been used for studying lipid abnormalities in superficial tissues and in tissues accessible through minimally invasive surgery. [2][3][4][5][6][7] In addition, a potential to examine structures and topically applied drugs in skin in vivo has been shown. [8][9][10] Such studies underline the clinical potential of CRS microscopy. However, the practical translation of CRS techniques into a clinical setting is accompanied by several technological challenges. One of these challenges is the engineering of robust and small optical probes that can be directed to accessible parts of the body, including hollow tracts of the human body. Traditional microscope optical elements are not optimized for this purpose. In particular, the physical dimensions of the microscope objective lens, with diameters of typically 2 -4 cm, are incompatible with the In order to incorporate a focusing element with a high numerical aperture (NA) in small optical probes, miniaturization of optical components is required. Advances in this area can be divided into two categories: objective lens systems that use miniature multi-lens objectives and systems based on the use of gradient refractive index (GRIN) optics. Multi-lens miniature objectives have been used for linear and nonlinear imaging techniques such as confocal reflectance [11][12][13][14][15], two-photon excitation fluorescence (TPEF) [16][17][18][19][20] and CARS. [21,22] Similarly, GRIN lenses have been utilized in confocal reflectance [23], TPEF [24][25][26][27][28] and CRS [29] imaging applications.
The design criteria for miniaturizing the focusing element in CRS microscopy are particularly challenging. CRS microscopy uses two excitation beams of different wavelength, which need to be focused to the same focal spot for optimum resolution and signal generation. This implies that, relative to other nonlinear optical imaging techniques such as TPEF, the suppression of chromatic aberrations is pertinent in CRS probe development. The necessity to minimize chromatic aberrations puts particular constrains on the lens design, requiring design parameters that are currently not trivially met in GRIN lens systems. [30] In terms of suppressing chromatic distortions in practice, multi-lens systems exhibit more flexibility as additional lenses can be incorporated to reach acceptable levels of achromaticity. [30] Concurrently, the lens system needs to be designed for use in aqueous media, exhibit a high NA, feature an acceptable field of view and its physical dimensions should be compatible with the use of a small fiber-coupled probe. Despite significant advances made in this area, the performance of miniature lenses for CRS imaging has been inferior to the performance of commercially available, aberration-corrected microscope objectives. In this contribution, we discuss the design and imaging performance of a miniature multi-lens focusing system that meets the criteria mentioned above, and achromaticity in particular. By engineering the system specifically for minimum chromatic aberrations, excellent focusing parameters are obtained for both excitation beams, producing high-resolution CRS images with contrast comparable to conventional objective lenses.

Miniature objective design
We used ZEMAX software for optical design of the lens system. The basic layout of the miniature objective design follows the design of standard microscope objectives, which consist of multiple lens components. Our design consists of 6 elements including a glass coverslip, forming a lens system that is optimized for water immersion and exhibits a NA of 0.51, see Fig.  1(a). The design makes use of achromatic and aspheric surface lens elements to compensate for various aberrations. The basic arrangement uses a thick meniscus negative component oriented convex towards the positive components. The first element (L1) is a 4.5 mm diameter lens (Cat# 67595, Edmund Optics) used to collimate the beam from a point source. This first element can be removed in case the objective is illuminated with collimated beams, in which case the lens system is compatible with an infinity corrected imaging system. The second lens (L2) is a custom-made negative achromatic lens. Its main purpose is to diverge the beam and improve the NA of the system. This lens element is fabricated by JenOptik according to our design specifications. The third and fourth lenses (L3, L4) are positive lenses consisting of two cemented achromatic doublets (Cat# 49923, Edmund Optics), followed by an aspheric singlet (L5, Cat# 355390, Lightpath technologies). The cemented achromatic doublets are made from two different glass materials to correct for chromatic aberrations of the pump and Stokes beams, while the aspheric lens located near the object mainly compensates monochromatic wavefront errors such as spherical aberrations and coma. Incorporation of a single aspheric lens replaced multi- ple spherical optical lenses in the design, resulting in a more compact objective. The microscope objective is designed for use with a coverslip (L6), which protects the lens from moisture and external damage. The glass thickness of the coverslip contributes a significant amount of spherical aberrations and coma, which are compensated in the design calculations. The objective is corrected for chromatic aberrations over the spectral range from 810 nm to 1064 nm.
The objective design has a working distance of 0.872 mm. The maximum diameter of the lens system is 5.0 mm. To fix the lens positions, the elements are housed in a home-made brass sheath. Figure 1(b) shows a rendering of the components in the packaged objective. The physical length of the fully packaged objective is 19 mm and the outer diameter is 8.0 mm, see Fig. 1(c). To test the image quality of aberration corrected miniature objective, a comparison is made with an uncorrected aspheric singlet lens (Cat# 87165, Edmund Optics) and a commercial objective with a comparable NA (Plan-Neofluar 20X, 0.5 NA, Zeiss). The summary of the design specifications of the three lens systems is given in Table I.

Experimental layout
The experiments in this work were carried out with a pump wavelength of 817 nm and a Stokes wavelength of 1064 nm. The Stokes beam was derived from a Nd:vanadate laser (PicoTrain, HighQ Lasers), which produces ∼10 ps pulses at 76 MHz. We tested the performance of the lens with two different sources of the pump beam, producing either ps pulses or fs pulses. The ps pump pulses were derived from an optical parametric oscillator (Levante Emerald, APE), producing tunable 7 ps pulses in the 760-960 nm range. The fs pulses were generated by a Ti:sapphire laser (Mira-900, Coherent), which delivers 150 fs pulses tuned to 817 nm. The Nd:vanadate laser was electronically synchronized to the Ti:sapphire laser, yielding time-synchronized pump and Stokes pulses with a temporal jitter less than 0.500 ps. Imaging studies were performed by integrating the lens in an imaging setup consisting of a laser scanning system and an optical detector. Briefly, the scanning system consisted of two computer controlled galvanometric mirrors (Fluoview 300, Olympus), a scan lens (f = 70) and a tube lens (f = 180mm). Because the imaging system was infinity corrected, L1 was removed to accommodate the collimated excitation beams at the back aperture of the miniature objective. The nonlinear signal was detected both in the forward and in the epi-direction. In the forward direction, the CARS signal was captured by a condenser, filtered by a bandpass filter (650 nm, Semrock), and detected by a photomultiplier tube (R3896, Hamamatsu). In the epi-direction, the signal was separated from the incident light with a dichroic mirror (760SWP, Chroma), filter by a bandpass filter and detected with an identical photomultiplier tube. The second harmonic generation signal was detected by replacing the filter with a 400 nm bandpass filter (400 nm, Semrock). For the SRS experiments, the Stokes beam was amplitude modulated at 10 MHz with an acoustic optic modulator (Crystal Technologies). The stimulated Raman loss (SRL) signal was detected in the forward direction at the pump wavelength by a photodiode (FDS1010, Thorlabs), and the signal was demodulated with a home-built lock-in amplifier.
Point source Image

Materials
30 nm diameter Si nanoparticles (Meliorum Technologies, Inc.) were used for lateral resolution measurement. The electronic four-wave mixing signal from the Si particles was used to visualize the focal width. [31] For axial resolution measurements, 1 µm diameter polystyrene beads (Polysciences, Inc.) were used. The samples were suspended in water, sonicated and embedded in an agraose gel. To test the capability of the miniature objective to image biological tissues, we imaged a freshly excised mouse ear (SAMP-8 mice) and rabbit skin (Pathogen-free New Zealand White rabbits Western Oregon Rabbit) tissue. In addition, to measure the imaging penetration depth of the microscope objectives we utilized turkey breast tissue. The samples were mounted directly on the coverslip for the imaging studies.

Results
The primary design criteria for the miniature CRS objective lens are minimum chromatic aberration and a high numerical aperture. These design criteria enable tight focusing of the pump and Stokes beams to a near diffraction-limited focal spot. Some common metrics of image quality are: the root-mean-square (r.m.s) radius of the spot size, the modulation transfer function (MTF), the Strehl ratio and the point spread function. Figure 2 The axial chromatic error between the wavelengths is estimated as a maximum chromatic focal shift in the range of 1.29 µm, which is well below the diffraction-limited value of 3.22 µm. This microscope objective is well corrected for the on-axis aberrations, the dominating form of aberration is off-axis non-symmetric coma wavefront distortion, which increases with distance from the principal axis.  The objective performance is predicted close to the diffraction limit curve, with most deviation near the edge of the 100 µm field of view, with 38% contrast for 0.5 lp/µm. The Strehl ratio is another measurement of the objective lens performance and represents the illumination at the center of the diffraction pattern of an aberrated image, relative to that of an aberration-free image. The on axis and off axis (50 µm) Strehl ratio for the miniature objective are 0.968 and 0.86 respectively. The ratio indicates an acceptable level of wavefront aberration within the diffraction-limited image. Figure 3 summarizes the imaging performance of the lens at the two fundamental excitation wavelengths. In Fig. 3(a), a transmission image of a high-resolution United States Air force (USAF) 1951 target is shown for the 817 nm excitation wavelength. Figure 3(b) shows the intensity profile of the cross section taken at group 9, element 1 as shown by the red line in (a). The width of each line is 0.98 µm, and the separation between lines is 0.98 µm. This image test shows that the objective is capable of resolving (sub-)micrometer structures in the one-photon imaging mode. Figures 3(c,d) show comparable results for the imaging performance at 1064 nm, albeit of slightly lower resolution because of the longer wavelength. An impression of the uniformity across the FOV is given in Fig. 3(e), where a Ronchi ruling is visualized with 817 nm in transmission mode. The straightness of the lines is an indication that field distortions are relatively minor. In Fig. 3(f) the lateral deviation is depicted, showing that distortion in the image is only a few µm across the entire FOV. Note that the field distortion is a combination of  While the one-photon imaging test provides useful information about the focusing quality of the objective, the actual CRS resolution depends on the focal volumes of both excitation beams. To determine the lateral resolution of the objective in the CARS modality, we imaged 30 nm diameter Si nanoparticles embedded in an agarose gel with both beams incident on the sample. The full-width-at-half-maximum (FWHM) of the CARS excitation volume was found at 0.648 ± 0.073 µm), see Fig. 4(a). The axial resolution was determined from a three-dimensional image of 1 µm diameter polystyrene beads, acquired with a 300 nm axial step size increment. Fitting the axial profiles of the beads to a Gaussian function, an axial FWHM of 4.9 ± 0.16 µm was obtained, see Fig. 4(b). A comparison with the theoretical values of the point spread function for aberration-free system is shown in Table II.
To test the image quality of the aberration corrected miniature objective, a comparison is made with an uncorrected miniature aspheric singlet lens and a commercial objective. The miniature aspheric lens was chosen to illustrate the performance of typical off-the-shelf miniature optics used for monochromatic optical probe designs. The commercial microscope objective is an achromatic lens system with a similar NA to our miniature objective. The performance of the commercial system serves as a direct comparison with the performance of the miniature CRS objective lens. The design specifications and the experimentally determined lateral resolution of each of the lens system are summarized in Table 1. It can be seen that the chromatically corrected miniature objective lens exhibits excellent confinement and overlap of the focused beams, producing a high spatial resolution.
For the comparison of the CARS image quality over the extended field of view for each of the lens systems we imaged the distribution of 16 µm polystyrene beads. The miniature objective reveals a high quality image that appears undistorted within the field of view, as shown in Fig. 5(a). We attribute these favorable imaging properties to the reduced level of chromatic aberrations. In contrast, the conventional aspheric lens is uncorrected for aberrations and shows non-uniform image quality within the field of view, see Fig. 5(b). Such distortions can be attributed to variations in the overlap of the pump and Stokes beams in the focal plane as the beams are angularly adjusted at the back aperture of the focusing lens. We note that even at increased illumination powers the image quality of the aspheric lens remained poor. In addition, the images obtained with the miniature objective were compared with the commercial objective, shown in Fig. 5(c). The CRS image quality for the commercial lens and the miniature objective is comparable. However, as shown in Table I, the useable field of view of the commercial objective is significantly larger at 635 x 635 µm versus that of the miniature objective at 190 x 190 µm. Hence, the expansion of the FOV for the miniature objective is a future design challenge. We have performed CARS imaging studies on mouse ear tissue to test the usefulness of the miniature objective for biomedical imaging. The lasers were tuned to the CH 2 vibrational stretch of lipids. Figure 6 shows representative images acquired with the custom-made objective lens at various locations in the tissue. Several tissue components, including adipocytes in the subcutaneous layer ( Fig. 6(a,b)), stratum corneum ( Fig. 6(c)) and hair structures (Fig. 6(d)) are observed with excellent resolution and with high contrast. The image studies discussed thus far show that the miniature objective produces high contrast CARS images when detected in the forward direction. However, for tissue imaging applications, CRS signals are typically collected in the epi-direction. For generating images when detecting in the epi-direction, the objective lens needs to intercept backscattered light with a high collection efficiency. In Fig. 7(a), an epi-CARS image of a hair shaft in rabbit skin ex vivo is shown. The CARS signal from the hair is strong and appears with high contrast. In addition, the epi-detected SHG signal of the surrounding collagen is also shown, indicating that the miniature lens is suitable for multimodal nonlinear imaging of thick tissues. Furthermore, the excellent quality of the dual-color focal volume produced by the miniature objective also gives rise to strong SRS signals. In Fig. 7(b) an SRL image of adipocytes in the mouse ear is depicted. Similar to the CARS images, the SRL images are free from blurring and exhibit sub-micrometer lateral resolution.
Finally, we studied the strength of the CRS signal as a function of penetration depth in the tissue. For these measurements we have chosen turkey breast tissue cut to a thickness > 1 mm. We collected the forward CARS signal and analyzed the decay of the coherent nonresonant background as a function of focusing depth. We find a 1/e decay parameter of 0.180 mm, indicating that the miniature objective is suitable for acquiring CRS images well over 100 µm deep into the tissue.

Discussion and conclusion
Objectives are the most essential component in an optical probe design, and are the primary elements responsible for image quality. For CRS applications, the lens design needs to take suppression of chromatic aberrations explicitly into consideration. In systems where management of chromatic aberrations is minimum, such as in single elements aspheric lenses and in most GRIN lens designs, inferior CRS image quality may be expected. In our study, the aspheric lens, a common element in miniature probe systems, showed a reduced spatial resolution (1 ± 0.157 µm) and distortions within the field of view that can be ascribed to non-perfect overlap of the pump and Stokes beams. The aberrated performance of most single element miniature lenses in CRS imaging formed the motivation for the multi-element objective design discussed in this study.
A challenging aspect of optimizing the CRS objective with ray tracing methods is to achieve a balance between a high NA and low aberrations. By increasing the NA of the system, additional compensation is needed to keep the aberrations at tolerable levels. A design of a variety of lenses nullified the collective aberrations caused by individual optical elements such as chromatic and spherical aberrations. Another complication with increasing the NA is the shortening of the working distance. In this design we targeted an objective with a long working distance (0.872 mm), high NA (0.51) and a small diameter (5 mm) of the lens elements. The CRS imaging performance of the multi-element lens system is superior to single focusing elements, both in terms of spatial resolution and in uniformity of the CRS signal within the field of view. In fact, the resolution and image quality of the miniature lens is comparable to the imaging performance of commercial microscope objectives.
The imaging performance of the miniature lens shows that the high contrast images acquired in conventional CRS microscope systems can also be attained with optics only a fraction the size of regular microscope components. Better contrast images are imperative for making miniature CRS probes viable diagnostic tools, and the results presented in this study underline that such criteria can be realized with smaller sized optics. Nonetheless, while the current specifications are very much in line with the minimum requirements for a practical CRS probe design, there remains a future need for improving both the NA and the field of view of the system. Additionally, the ZEMAX computational design of the miniature objective could be optimized for chromatic aberration over a much broader spectral range including the visible wavelengths.
In short, we have designed, assembled and tested a 0.51 NA miniature objective for CRS imaging. The system is fully packaged and features an outer diameter of 8 mm. The lens sys- tem shows near diffraction-limited performance with a spatial resolution sufficient for imaging subcellular details. The miniature objective is successfully used for tissue imaging, producing images with excellent CRS contrast. This work shows that it is possible to fabricate miniature lens assemblies for CRS microscopy with imaging capabilities that are comparable to the qualities of conventional microscope optics. Such capabilities will be important for the development of small probes and endoscopes, propelling the translation of CRS imaging technologies into practical clinical tools.