ol-38-16-3103

8/12/2019 ol-38-16-3103

http://slidepdf.com/reader/full/ol-38-16-3103 1/4

Miniature varifocal objective lens for endomicroscopy

Dimitre G. Ouzounov,1,* David R. Rivera,1 Watt W. Webb,1 Julie Bentley,2 and Chris Xu1

1School of Applied and Engineering Physics, Cornell University, Ithaca, New York 14853, USA2The Institute of Optics, University of Rochester, Rochester, New York 14627, USA

*Corresponding author: [email protected]

Received May 23, 2013; revised July 8, 2013; accepted July 9, 2013;

posted July 9, 2013 (Doc. ID 190790); published August 13, 2013A miniature catadioptric lens for endoscopic imaging based on the principle of wavelength division multiplexing ispresented. We demonstrate change of the magnification and the field of view (FOV) of thelens without any mechani-cal adjustment of the optical elements. The lens provides magnifications of  ∼−1.5× at 406–750 nm and  ∼−0.2× at800 nm. The lens is used to demonstrate large-FOV (1.3 mm) reflectance imaging and high-resolution (0.57   μm)multiphoton fluorescence imaging of unstained mouse tissues. © 2013 Optical Society of America

OCIS codes:   (120.3890) Medical optics instrumentation; (170.3880) Medical and biological imaging; (180.4315)Nonlinear microscopy.

http://dx.doi.org/10.1364/OL.38.003103

 A challenge faced in the development of clinically usefulendoscopes is the desire to provide a large field of view(FOV) and high spatial resolution within a miniature

 probe. A large FOV is useful so that a clinician can

quickly survey a large tissue area, whereas high spatialresolution is necessary to resolve cellular details for ac-curate medical diagnosis. Because these two require-ments cannot be achieved simultaneously by a miniature objective lens, optical zoom is an essentialfunctionality for practical endoscopes.

To date, many clinical biopsy procedures are guided byoptical endoscopes. Although conventional wide-fieldendoscopes have large FOVs, they provide only a grossinspection of tissue morphology, which is often inad-equate for reliable diagnostics. Over the past decade, a number of new high-spatial-resolution imaging modalities(e.g., optical coherence tomography [1], multiphoton

microscopy [2,3], and confocal microscopy [4]) have beenused for endoscopic imaging [5–13]. Although each imag-ing modality derives its contrast from different physical

 phenomena, the endoscopic probes have similar architec-tures [14], including optical fiber delivery of excitationlight and collection of relevant signal, a scanningmechanism (e.g., fiber scanner, microelectromechanicalsystems), and a miniature objective lens. A variety of objective lenses have been used: simple monolithicgradient-index (GRIN) lenses [15–19], compound GRINlenses, [12,20,21], compound spherical [10,22] or aspheri-cal lenses [11,23], micromachined lenses [24], and glass or 

 plastic lenses [25,26]. These objective lenses perform suc-cessfully in terms of resolution, size, and cost, but achiev-

ing optical zoom capability in miniature endoscopes hasbeen elusive. Currently, pathologists study ex vivo biopsysamples with at least two optical magnifications (e.g., a 4× objective for viewing the tissue architecture and a 20× or  40× objective for resolving cellular details). This

 variable magnification and FOV is invaluable in helpingobtain an accurate diagnosis of a patient’s health state.While optical zoom capability is easily implemented ina conventional light microscope by switching amongmultiple objective lenses, this mechanical approach is dif-ficult to implement in a miniature endoscope due to sizelimitations. For practical implementation, both large FOV imaging and high-resolution imaging must be obtained

with the same endoscope apparatus without mechanicaladjustments of the distal endoscope parts. Chen   et al.

reported [27] a small dual-zone lens, but nearly half of the excitation power incident on the sample does not

 produce useful information. An optical endoscope that provides high-spatial-resolution imaging with opticalzoom capability has never before been demonstrated.

Traditional varifocal/zoom lenses, including catadiop-tric lenses [28–31], achieve their functionality by movingone or more of their components.

In this Letter, we present a miniature catadioptric var-ifocal lens based on the idea of separating the optical

 paths of excitation light with different wavelengths.The lens provides two magnifications and FOVs, andthe zoom operation is achieved by changing the wave-length of the excitation light without any mechanicaladjustments of the lens components.

The objective lens consists of 3 elements, has a 3 mmoutside diameter (OD) and is  ∼8 mm in length [Fig. 1(a)].The variation of magnification and FOV is enabled bythe dichroic coatings deposited at the central part of the proximal surface of element 3 and at the peripheralregion of the proximal surface of element 2 [Fig. 1(a)].The central part of the proximal surface of element 2is uncoated. Starting from the left in Fig.  1(a), the exci-tation light emerges from the delivery/scanning fiber andis directed by element 1 onto the dichroic coating at thecentral part of the proximal surface of element 3.Depending on the incident wavelength ( λi), the excitationlight is either reflected (e.g.,   λi 800  nm) to the pat-terned dichroic coating on the proximal surface of 

element 2 and then focused to the sample with highnumerical aperture (NA), or transmitted (e.g.,   λi 406 nm) and focused to the sample with low NA. There-fore, the change of magnification and FOV is achievedsimply by changing the wavelength of the excitation lightwithout any mechanical adjustment.

The high-resolution multiphoton imaging mode is de-signed (Table   1) to operate between 800 and 900 nm.

 At   λi 800 nm, the calculated full width at half-maximum (FWHM) of the lateral point spread function(PSF) is   ∼0.7   μm [Fig.   3(a)], and the Strehl ratio isapproximately 1 over the central 160   μm FOV, indi-cating diffraction-limited optical performance. The

 August 15, 2013 / Vol. 38, No. 16 / OPTICS LETTERS 3103

0146-9592/13/163103-04$15.00/0 © 2013 Optical Society of America 

8/12/2019 ol-38-16-3103

http://slidepdf.com/reader/full/ol-38-16-3103 2/4

low-magnification imaging mode (Table 1) operates be-tween 350 and 730 nm. At  λi 406 nm, the lateral reso-lution (FWHM) is  ∼4.5   μm, and the FOV is 1.3 mm. Wechose 406 nm for one-photon reflectance imaging sinceit is one of the preferred excitation wavelengths for nar-rowband imaging [32]. In both imaging modes, the focal

 planes of the miniature zoom lens have small curvatures.Such a curved image plane allows significantly better aberration correction than a planar image plane and isinconsequential for   in vivo   tissue imaging. The radiusof curvature of the image surface is   −1.09   and−2.01  mm for high-magnification and low-magnificationmodes, respectively. For   in vivo   imaging, this is not a 

 problem because tissue features are not flat. Since the plastic optical fibers (POFs) collect through nonimaging pathways, the curvature of the imaging surface will havelittle effect on signal collection. We performed toleranceanalysis by calculating the root-mean-squared (rms)wavefront error, that is, deviations from an ideal spheri-cal wavefront. The nominal error of the lens design is

0.007. We performed Monte Carlo simulations with toler-ances listed in Table 2; for 5000 runs, more than 90% of 

the trials had an rms wavefront error less than 0.05,which is below the accepted practical limit (0.07) for a diffraction-limited optical system.

The dichroic coatings on elements 2 and 3 [Figs.  1(a)and 1(b)] that enable the varifocal operation lead to non-

reciprocal propagation of the excitation and the fluores-cence light. Therefore, epicollection of the two-photonexcited fluorescence signal through the excitation path-way is inefficient. As a result, we use 10 large-core(500   μm OD) POFs (ESKA acrylic, Mitsubishi) located

 just behind element 2 [Fig.  1(d)] to collect the fluores-cence signal. The same POFs collect the scattered lightat low-magnification mode. Different locations of the im-aging planes result in different collection efficiency,which is smaller for the low-magnification mode dueto the longer working distance. However, at low magni-fication the signal level is much higher due to the linear contrast mechanism.

The optical elements were manufactured [Fig.  1(e)]and assembled by Optics Technology Inc., Pittsford,New York.

We tested the performance of this miniature objectiveby imaging a US Air Force (USAF) test target in transmis-sion. The experimental setup is shown in Fig.  2. Charac-terization and imaging in high-resolution mode is donewith a mode-locked Ti:sapphire laser (Tsunami, Spectra Physics) operating at 800 nm. For low-magnification,large-FOV imaging, we use a fiber-coupled continuouswave semiconductor laser operating at 406 nm (LP406-SF20, Thorlabs). The laser beam is raster-scanned by a 

 pair of galvo mirrors and is focused to the object planeof the zoom endoscope lens by a low-NA objective

(Olympus, 0.1 NA,4×). We analyzed the intensity line pro-file across the indicated feature [Fig. 3(a)] to quantify thelateral resolution for both modes. The intensity profile atthe edge of the feature is the edge-response function, andits derivative is the line-spread function, which corre-sponds to the cross section of the PSF. The lateral res-olution of the high-magnification mode (FWHM) is∼0.75   μm [Fig. 3(a)], which corresponds to a two-photonresolution (FWHM) of ∼0.5   μm. The side lobes of the PSF are caused by the obstruction at the center of the backaperture, which is a well-known limitation for a reflectiveobjective lens. However, these side lobes will be signifi-cantly suppressed in two-photon excitation because of 

Fig. 1. (a) Lens design layout, drawn to scale. (b) Transmissionspectrum of the patterned dichroic coating deposited on the proximal surface of element 2 (shown in the inset). The pat-terned structure of the coating is clearly seen. (c) Transmissionspectrum of the dichroic coating deposited on the proximal sur-face of element 3 (shown in the inset). (d) 3D lens layout, in-dicating the positions of the plastic fibers for signal collection.(e) Photograph of the objective lens assembly.

Table 1. Design Specifications for MiniatureZoom Objective

Parameters High-Mag Mode Low-Mag Mode

Sample space Water Air  Magnification   −0.2×   −1.47×NA 0.55 (at 800 nm) 0.075 (at 405 nm)FOV 160   μm 1.3 mmOne-photon resolution 0.7   μm 4.5   μmWavelength 800–900 nm 350–730 nm

Fig. 2. Lens characterization and imaging setup.

3104 OPTICS LETTERS / Vol. 38, No. 16 / August 15, 2013

8/12/2019 ol-38-16-3103

http://slidepdf.com/reader/full/ol-38-16-3103 3/4

the quadratic dependence on excitation intensity. TheFOV of the high-magnification mode is   ∼160  μm[Fig. 3(b)]. The output beam of the miniature zoom lensoperating in high-magnification mode has an annular 

 profile, which reduces the axial resolution [33,34]. Wecharacterized the two-photon axial resolution of thehigh-magnification mode by stepping a 500 nm Rhod-

amine B thin film through its focus. The measured FWHMof the thin film response is ∼10.5  μ m [Fig. 3(d)], which issufficient for resolving cellular layers in biological tis-sues. Using the USAF test target, the one-photon lateralresolution of the low-magnification mode was measuredto be   ∼4.5  μm. The low-magnification imaging modeachieved a large FOV of 1.3 mm [Fig.  3(c)].

The transmission of the varifocal lens is   ∼40%   and∼70% , respectively, for the high- and low-magnificationmodes. In this prototype, the only coatings depositedwere the patterned dichroic coatings on the proximal sur-faces of elements 2 and 3 (Fig.  1) needed for varifocaloperation. Antireflection coatings on the other optical

surfaces will result in significantly higher power transmission.

To demonstrate and test the imaging capabilities of thedual magnification objective, we acquired ex vivo imagesfrom various unstained mouse tissues. The two-photonfluorescence and one-photon reflected/scattered lightemitted from these tissues was epicollected using the

10 large-core POFs and detected by an ultra bialkali(R7600U-200, Hamamatsu) photomultiplier tube (PMT).The PMT output current was amplified and convertedto voltage (C7319, Hamamatsu). Prior to imaging, the ex-cised tissue samples (i.e., a segment of colon, a singlelung lobe, and a whole kidney obtained from an adultCD-1 mouse) were embedded in agarose gel andmounted on microscope slides. The slide was mountedon a three-dimensional (3D) translation stage (MP-285,Sutter). Low-magnification images were acquired first.To acquire high-magnification images, the site of interestwas placed at the center of the FOV, and then thesample was moved axially to the image plane of the

high-magnification mode. Image acquisition is performedwith the software ScanImage.Figure 4(a) shows a high-resolution two-photon image

of unstained mouse colon tissue (inner epithelial lining of the colon). Typical features (e.g., enterocytes, gobletcells, and crypts) are visible. A corresponding low-mag-nification, reflection/scattering image of mouse colon isshown in Fig.  4(b). Figure 4(c) shows a high-magnifica-tion multiphoton image of unstained mouse lung, wherecharacteristic features including alveolar walls andlumens are clearly distinguishable. The correspondinglow-magnification image is shown in Fig. 4(d). All imageswere acquired at a rate of 2 frames∕s. The average power at the sample during multiphoton imaging was  ∼25 mW.

In Fig. 5(a), we show a low-magnification image of un-stained mouse kidney. High-magnification two-photonfluorescence and second harmonic generation (SHG)images of unstained mouse kidney at the surface and∼80   μm below the surface are shown in Fig.   5(b)   andFig.  5(c), respectively. In Fig.  5(b), the fibrous compo-nents of the kidney capsule are visible. Figure 5(c) dis-

 plays an optical cross section of the proximalconvoluted tubules, which are separated by renal inter-stitium (i.e., dark, nonfluorescent spaces containingsparse amount of connective tissue).

In conclusion, we have designed, built, and character-ized a miniature endoscopic objective lens that provides

Table 2. Lens Prescription and Tolerances for Design in Fig.  1a

Elements

Radius Value,(mm)

Radius±Tolerance

(fr/mm)

Thickness Value(mm)

Thickness±Tolerance

(mm) GlassIndex

±Tolerance

Wedge/ Roll

TIRb (mm)

IrregularityTolerance(fringes)

Fiber tip 1 0.1 air  Element 1R1 infinity 5 fringe 1 0.15 silica 0.001 0.05 0.25Element 1R2   −2.218 0.05 3 0.25 air 0.05 0.25Element 2R1 3.022 0.01 1.567 0.01 silica 0.001 0.004 0.1Element 2R2 1.6 0.01 0 0.004 0.1

Element 3R1 1.6 0.01 2.269   0.0∕−0.1   NLAF2 0.001 0.25

Element 3R2 infinity 5 fringe 0.1 seawater 0.02 0.25Image   −1.092 seawater  

aImage plane values are for high-magnification mode.bTotal indicator reading.

Fig. 3. (a) Calculated (red solid line) and measured (bluedashed line) lateral PSF for the high-resolution imaging mode.Inset: Group 9 of USAF resolution target imaged using the high-resolution imaging mode at 800 nm. Inset: Intensity line profileacross the indicated feature and its derivative. (b) USAF reso-lution target imaged using the high-magnification mode( λi 800  nm). (c) USAF resolution target imaged in transmis-sion using the low-magnification mode ( λi 406  nm). (d) Axialscan of a thin Rhodamine B film showing the two-photon axialresolution (FWHM) of 10.5   μm.

 August 15, 2013 / Vol. 38, No. 16 / OPTICS LETTERS 3105

8/12/2019 ol-38-16-3103

http://slidepdf.com/reader/full/ol-38-16-3103 4/4

dual magnification and FOV without any mechanical ad- justments of its elements. The operation is based on sepa-rating the optical path for excitation light with differentwavelengths. We fully characterized the two modes of operation by imaging a USAF resolution target in trans-mission. We demonstrated the feasibility to employ thislens for dual magnification and FOV endoscopic imagingby acquiring multiphoton and reflectance images of un-stained   ex vivo  mouse tissues. Future work will includethe integration of the objective into a fully functionalendomicroscope probe and demonstrating its imagingcapabilities.

We thank members of the Xu and Webb groups for  valuable discussions. The project was supported bythe NIH Grants R01-CA133148 and R01-EB006736.

References

1. D. Huang, E. Swanson, C. Lin, J. Schuman, W. Stinson, W.Chang, M. Hee, T. Flotte, K. Gregory, C. Puliafito, and J.Fujimoto, Science 254, 1178 (1991).

2. W. Denk, J. Strickler, and W. Webb, Science 248, 73 (1990).3. W. R. Zipfel, R. M. Williams, and W. W. Webb, Nat.

Biotechnol. 21, 1369 (2003).4. M. Minsky, Scanning  10, 128 (1988).5. R. Kiesslich, M. Goetz, M. Vieth, P. R. Galle, and M. F.

Neurath, Gastroenterol. Clin. North Am.  15, 715 (2005).6. J. F. Xi, Y. Chen, Y. Zhang, K. Murari, M. Li, and X. Li, Opt.

Lett.  37, 362 (2012).7. C. M. Brown, D. Rivera, I. Pavlova, D. Ouzounov, W.

Williams, S. Mohanan, W. Webb, and C. Xu, J. Biomed.Opt.  17, 040505 (2012).

8. E. Seibel and Q. Smithwick, Laser Surg. Med.   30, 177(2002).

9. B. A. Flusberg, J. C. Lung, E. D. Cocker, E. P. Anderson, andM. J. Schnitzer, Opt. Lett.  30, 2272 (2005).

10. H. C. Bao, J. Allen, R. Pattie, R. Vance, and M. Gu, Opt. Lett.

33, 1333 (2008).11. Y. Wu, J. Xi, M. Cobb, and X. D. Li, Opt. Lett. 34, 953 (2009).12. D. Rivera, C. Brown, D. Ouzounov, I. Pavlova, D. Kobat, W.

Webb, and C. Xu, Proc. Nat ’l. Acad. Sci. USA   108, 17598(2011).

13. P. Kim, E. Chung, H. Yamashita, K. Hung, A. Mizoguchi, R.Kucherlapati, D. Fukumura, R. Jain, and S. Yun, Nat.Methods  7, 303 (2010).

14. B. A. Flusberg, E. Cocker, W. Piyawattanametha, J. Jung, E.Cheung, and M. Schnitzer, Nat. Methods  2, 941 (2005).

15. M. Myaing, D. McDonald, and X. Li, Opt. Lett.   31, 1076(2006).

16. Y. C. Wu, Y. X. Leng, J. F. Xi, and X. D. Li, Opt. Express  17,7907 (2009).

17. L. Fu, A. Jain, C. Cranfield, H. K. Xie, and M. Gu, J. Biomed.

Opt.  12, 2756102 (2007).18. D. Bird and M. Gu, Opt. Lett.  28, 1552 (2003).19. G. J. Liu, T. Xie, I. Tomov, J. Su, L. Yu, J. Zhang, B.

Tromberg, and Z. Chen, Opt. Lett.  34, 2249 (2009).20. C. J. Engelbrecht, R. S. Johnston, E. J. Seibel, and F.

Helmchen, Opt. Express 16, 5556 (2008).21. R. Barretto, B. Messerschmidt, and M. J. Schnitzer, Nat.

Methods  6, 511 (2009).22. C. Liang, K. B. Sung, R. Richards-Kortum, and M. R.

Descour, Appl. Opt.  41, 4603 (2002).23. S. M. Landau, C. Liang, R. T. Kester, T. S. Tkaczyk, and M. R.

Descour, Opt. Express  18, 4758 (2010).24. D. L. Dickensheets and G. Kino, Opt. Lett.  21 , 764 (1996).25. K. Carlson, M. Chidley, K. Sung, M. Descour, A. Gillenwater,

M. Follen, and R. Richards-Kortum, Appl. Opt.   44, 1792

(2005).26. R. T. Kester, T. C. Tkaczyk, M. R. Descour, T. Christenson,

and R. Richards-Kortum, Opt. Express  15, 2409 (2007).27. M. Chen, C. Xu, and W. Webb, Opt. Lett.  35, 2735 (2010).28. E. Ermolaeva, I. Livshits, and V. Vasiliyev, Proc. SPIE 8488,

848804 (2012).29. D. R. Shafer, Proc. SPIE  2539, 235 (1995).30. B. Shen, J. Chang, J. Ouyang, L. Zhang, and A. Wei, Proc.

SPIE   8557, 85571C (2012).31. E. Betensky, Proc. SPIE  3129, 90 (2012).32. M. A. Kara and J. Bergman, Endoscopy  38, 627 (2006).33. M. Gu, J. R. Sheppard, and H. Zhou, Optik  93, 87 (1993).34. J. R. Sheppard, Optik  48, 329 (1977).

Fig. 4. High-magnification two-photon intrinsic fluorescenceimages of unstained ex vivo mouse (a) colon and (c) lung. Lungimage is unaveraged; colon image is averaged over threeframes. In (a), enterocytes (“e”), goblet cells (“g”), and crypts(“c”) are visible. In (c), alveolar walls(“w”) and lumens(“L”) aredistinguishable. (b) Low-magnification reflection/scattering im-age of unstained mouse colon (b) and lung (d). All images areacquired at two frames per second. Average power at the sam- ple during multiphoton imaging was ∼25  mW. The white circlesin (b) and (d) indicate the approximate location of the sitesfrom which the multiphoton images shown in (a) and (c) areobtained.

Fig. 5. (a) Low-magnification reflection/scattering image of unstained mouse kidney. High-magnification two-photon intrin-sic fluorescence and SHG image of unstained   ex vivo mousekidney (b) at the surface and (c) approximately 80   μm belowthe surface. In (b), the collagen fibers of the kidney capsuleare well distinguishable. In (c), cuboidal epithelium (“CE”)and the renal interstitium (“RI”) are visible. The white circlein (a) indicates the approximate location of the site from whichthe multiphoton images shown in (b) and (c) is obtained.

3106 OPTICS LETTERS / Vol. 38, No. 16 / August 15, 2013