June 15, 2008 / Vol. 33, No. 12 / OPTICS LETTERS
1333
Fast handheld two-photon fluorescence microendoscope with a 475 m à 475 m field of view for in vivo imaging Hongchun Bao,1,3 John Allen,2 Robert Pattie,2 Rod Vance,2 and Min Gu1,3,* 1
The Center for Micro-Photonics, Faculty of Engineering and Industrial Sciences, Swinburne University of Technology, Hawthorn, Victoria 3122, Australia 2 Optiscan Pty. Ltd., 15-17 Normanby Road, Notting Hill, Victoria 3168, Australia 3 E-mail:
[email protected] *Corresponding author:
[email protected]
Received February 13, 2008; revised April 28, 2008; accepted May 5, 2008; posted May 9, 2008 (Doc. ID 92692); published June 12, 2008 A fast handheld two-photon fiber-optic fluorescence endoscope for three-dimensional (3D) in vivo cellular imaging is developed. The compact handheld probe of the two-photon endoscope can simply be placed into contact with the target tissue to reveal clear 3D surface and subsurface histological images without biopsy. The new system has, to the best of our knowledge, the largest field of view (FOV) of 475 m ⫻ 475 m and a 3D imaging volume larger than 475 m ⫻ 475 m ⫻ 250 m. A real-time two-photon fluorescence image is displayed at 0.4 mm2 / s. The lateral and axial resolutions of the two-photon fluorescence endoscope are better than 1 and 14.5 m, respectively. © 2008 Optical Society of America OCIS codes: 180.6900, 110.2350, 190.4370.
Multiphoton microendoscopy enables visualization of cellular layers and intracellular organs with deep penetration depth and provides valuable information for early diagnosis of diseases and understanding complex mechanisms of biophenomena in living animals for research [1–3]. In vivo multiphoton microendoscopy can be used to directly observe molecular mechanisms of life without the need for sacrificing animals or removing tissue from the body, which is essential for practical applications. Multiphoton endoscopy uses an optical fiber for providing a comfortable distance between a target and optical and electronic hardware and a small sized probe for imaging internal organs in vivo [3]. A single-mode fiber was first tried for two-photon endoscopy, but it was demonstrated that such an endoscopy system has a low signal level because of the low numerical aperture (NA), the chromatic dispersion, and the nonlinearity [2–7]. A photonic crystal fiber with a microelectromechanical system (MEMS) mirror has been proven to be an efficient way for 3D twophoton fluorescence endoscopy through thick tissue with a high signal level [8–12]. However the design probe has a small FOV, and the current MEMS mirror is not necessarily robust to the vibration required for practical use. Alternatively, a double-clad fiber (DCF) combined with a tubular piezoelectric actuator as a microscanner have been adopted for two-photon fluorescence endoscopy [13]. Such an endoscope with a fiber length of 1.37 m generates two-dimensional (2D) circular images with a diameter of 147 m that corresponds to an FOV of around 0.017 mm2 and a frame rate of 2.6 Hz, which displays the FOV of 0.04 mm2 / s [13]. These features have limited the two-photon endoscope in 3D imaging for practical use. We report here on a fast handheld 3D twophoton fluorescence endoscope that has a high signal level through a 2.9 m dispersion-compensated DCF. 0146-9592/08/121333-3/$15.00
The handheld probe adopts a 3D microscanner that evenly scans the fiber line by line and generates a 475 m ⫻ 475 m FOV with an area of 0.23 mm2 and a real-time display rate of 0.4 mm2 / s. This compact handheld two-photon endoscope probe is readily placed into contact with target tissue to reveal subsurface images in vivo. A schematic of the new two-photon fluorescence endoscope is shown in Fig. 1. A Ti:sapphire laser that emits 80 MHz 70 fs short pulses is used as an excitation source. Before the laser is coupled into a fiber by an objective lens L1 (Olympus A10; NA, 0.25), it is prechirped by a 400 g / mm grating pair to compensate for the chromatic dispersion of the optical fiber [14,15]. A DCF is used for delivery of the excitation light from the Ti:sapphire laser to a handheld probe and the fluorescence signal collection. The DCF has a
Fig. 1. (Color online) Schematic of the handheld twophoton fluorescence endoscope. Inset, photo of the probe. © 2008 Optical Society of America
1334
OPTICS LETTERS / Vol. 33, No. 12 / June 15, 2008
core, an inner cladding, and an outer cladding with diameters of 3.6, 105, and 250 m, respectively [13]. The end of the fiber is inserted into a waterproof, compact, and handheld probe. The inset of Fig. 1 displays a photo of the probe. Within the probe, a custom design lens (diameter, 3 mm; NA, 0.35) that consists of multiple elements to accurately correct image aberration within the FOV is used for imaging, and a 3D microscanner is used for scanning the DCF and moving the fiber and the lenses in the axial direction to complete 3D imaging with a volume larger than 475 m ⫻ 475 m ⫻ 250 m [16]. The two-photon florescence signal from the sample is collected by the DCF in the backward direction, separated from the excitation light by a dichroic mirror, coupled by an objective lens L2 (Olympus MA20; NA, 0.4 to a multimode fiber, and sent to a photomultiplier tube (PMT) after being filtered by a 3 mm thick BG18 glass filter. The signal is synchronized with the 3D microscanner by a modified image processor FIVE1 (Optiscan Pty. Ltd.) to build up the image frame by frame [16]. The DCF nonlinearity and chromatic dispersion will broaden the pulses of the excitation laser and reduce the two-photon-excited fluorescence signal from the samples for in vivo imaging. To achieve a highfluorescence signal collection with a long length of fiber, the chromatic dispersion compensation by the prechip unit is studied for different grating pair distances. Figure 2(a) shows the two-photon-excited florescence signal intensity using 1, 2, and 3 m DCFs while the incident power Pin on the sample is fixed at 7 mW. The sample is a thin layer of polymer with 4, diethylaminobenzylidene-malononitrile (DABM) as a dye. As expected, the fluorescence signal exhibits a maximum value at an optimized grating pair distance in all three cases. The maximum fluorescence intensity is slightly decreased as the fiber length increases from 1 to 3 m. However, the 3 m fiber shows high tolerance if the grating pair distance is tuned
Fig. 2. (Color online) (a) Two-photon-excited fluorescence intensity 共If兲 and PN as functions of the grating pair distance for 1 (open triangles), 2 (crosses), and 3 (horizontal bars) m fibers and a 2.9 m fiber connected with the probe (open circles). (b) Log of the two-photon-excited fluorescence intensity 共If兲 versus log of the incident excitation laser power 共Pin兲 for the 1 m fiber at the optimized grating pair distance.
away from the optimized value. This feature is caused by the nonlinearity of the fiber. To further demonstrate this point, we plot, in Fig. 2(b), the log of the two-photon fluorescence signal intensity versus the log of the incident power Pin at the optimized grating pair distance for the 1 m fiber. At the low excitation laser power, the log of the fluorescence intensity and the log of the incident power show a linear relation with a gradient of about 1.8, which is close to 2. Thus the fluorescence signal intensity is proportional to the square of the excitation laser power, and the pulse width of the output pulses does not change with the variation of the power at the low power range. However, as the excitation power increases, this relation deviates from the linear relation, because the nonlinearity of the optical fiber starts to create a nonlinear frequency chirp, and the pulses are further broadened by this nonlinearity effect. We define the incident power PN at the point where the log curve deviates from the linear relation. Figure 2(a) also shows PN as a function of the grating pair distance. It is seen that the PN is minimum at the optimized grating pair distance, since the pulse width is smaller, and thus a lower average power is required to produce the nonlinearity effect. As the grating pair distance is away from the optimized value, the pulses become wider and therefore PN increases. The minimum value of PN for the 3 m fiber drops by 57% compared with that of the 1 m fiber, because the nonlinearity increases with an increase of the fiber length. It should be pointed out that in the nonlinear region with the power larger than PN, the two-photon fluorescence signal strength could be further improved by using prenonlinearity compensation [15]. On the basis of the dependence of the optimized grating pair distance on the fiber length and considering the extra dispersion from the multielement lens of the handheld probe, we used a 2.9 m DCF together with the probe to work near an optimized condition. The two-photon fluorescence intensity and PN
Fig. 3. (Color online) (a) Intensity profile along the center of the two-photon fluorescence image of a Fluoresbrite plain yellow–green (YG) 1 m microsphere. (b) Axial response of the two-photon-excited fluorescence signal from a thin layer of DABM dye. (c)–(f) Two-photon-excited fluorescence images of 10 m diameter fluorescent microspheres with an axial step of 8 m. The image size is 475 m ⫻ 475 m. The excitation wavelength is 800 nm. The excitation power is approximately 15 mW on the sample.
June 15, 2008 / Vol. 33, No. 12 / OPTICS LETTERS
1335
Fig. 4. (Color online) (a) Two-photon fluorescence images of a rat kidney surface. (b)–(d) Rat kidney 20, 40, and 60 m deep from the surface. The image size is 475 m ⫻ 475 m.
as functions of the grating pair distance are displayed in Fig. 2(a). The optimized fluorescence intensity in this case is slightly less than that of the 3 m DCF. This is because the multiple elements of the lens in the probe result in a further pulse broadening effect owing to the propagation time difference along the different paths that could not be compensated for by the prechirp unit [17]. The lateral and axial resolutions of the two-photon microendoscope are characterized by the lateral and axial responses of the two-photon-excited fluorescence signal from Fluoresbrite plain YG 1 m microspheres and a thin layer of DABM dye, respectively. Figures 3(a) and 3(b) show the two-photon-excited intensity profile along the center of a 1 m microsphere and the axial position of the thin layer DABM dye, where the FWHM of the intensity profile is 1 and 14.5 m, respectively. Therefore the lateral and axial resolutions of the two-photon endoscope are about 1 and 14.5 m, respectively. As the dye layer thickness is not accurately known, we believe that the axial resolution is better than 14.5 m. The diffractionlimited lateral and axial resolutions of the lens with an NA of 0.35 are 1 and 10.5 m, respectively. Therefore the experimentally measured lateral and axial resolutions of the two-photon endoscope are close to the diffraction limitation. The optical sectioning ability of the two-photon fluorescence endoscope is demonstrated by a set of the two-photon fluorescence images of Fluoresbrite plain YG 10 m microspheres, as displayed in Figs. 3(c)–3(f). A male rat Sprague–Dawley (approximately 780 g in body weight) was intravenously injected with 0.3– 0.4 ml of fluorescein (1% solution diluted in saline). Fluorescein is a safe fluorophore that can be used in humans, and the amount given was approximately the same standard dosage as a 70 kg adult human, where the body weight has been adjusted. Figure 4(a) is the two-photon fluorescence image of a rat kidney surface and Figs. 4(b)–4(d) 20, 40, and 60 m depths from the surface, respectively. The excitation laser power at the animal tissue is approximately 30 mW at a wavelength of 790 nm. Figure 4 demonstrates that the developed two-photon endoscope is able to 3D image in vivo cellular structures of animal tissue.
In summary, we have demonstrated a 3D twophoton endoscope with a large FOV and a fast frame display rate, which are 13 and 10 times larger, respectively, than the currently reported results [13]. A proper chromatic dispersion compensation through a fiber length of 2.9 m provides a comfortable distance between a sample being imaged and the control unit hardware. This compact handheld two-photon endoscope probe is capable of in vivo 3D visualization at cell structures of animal tissue. References 1. W. Denk, J. H. Strickler, and W. W. Webb, Science 248, 73 (1990). 2. B. A. Flusberg, E. D. Cocker, W. Piyawattanametha, J. C. Jung, E. L. M. Cheung, and M. J. Schnitzer, Nat. Methods 12, 941 (2005). 3. L. Fu and M. Gu, J. Microsc. 226, 195 (2007). 4. F. Helmchen, M. S. Fee, D. W. Tank, and W. Denk, Neuron 31, 903 (2001). 5. D. Bird and M. Gu, Opt. Lett. 27, 1031 (2002). 6. D. Bird and M. Gu, Opt. Lett. 28, 1552 (2003). 7. L. Fu, X. Gan, and M. Gu, Opt. Lett. 30, 385 (2005). 8. L. Fu, X. Gan, and M. Gu, Opt. Express 13, 5528 (2005). 9. L. Fu, A. Jain, C. Cranfield, H. Xie, and M. Gu, J. Biomed. Opt. 12, 040501 (2007). 10. L. Fu and M. Gu, Opt. Lett. 31, 1471 (2006). 11. W. Piyawattanametha, R. P. J. Barretto, T. H. Ko, B. A. Flusberg, E. D. Cocker, H. Ra, D. Lee, and O. Solgaard, Opt. Lett. 31, 2018 (2006). 12. B. A. Flusberg, J. C. Jung, E. D. Cocker, E. P. Anderson, and M. J. Schnitzer, Opt. Lett. 30, 2272 (2005). 13. M. T. Myaing, D. J. MacDonald, and X. Li, Opt. Lett. 31, 1076 (2006). 14. S. W. Clark, F. O. Ilday, and F. W. Wise, Opt. Lett. 26, 1320 (2001). 15. M. Lelek, E. Suran, F. Louradour, A. Barthelemy, B. Viellerobe, and F. Lacombe, Opt. Express 15, 10154 (2007). 16. M. Goetz, C. Fottner, E. Schirrmacher, P. Delaney, S. Gregor, C. Schneider, D. Strand, S. Kanzler, B. Memadathil, E. Weyand, M. Holtmann, R. Schirrmacher, M. M. Weber, M. Anlauf, G. Klöppel, M. Vieth, P. R. Galle, P. Bartenstein, M. F. Neurath, and R. Kiesslich, Endoscopy 39, 350 (2007). 17. M. Gu, Advanced Optical Imaging Theory, Springer Series in Optical Sciences 75 (Springer-Verlag, 2000).