MICROSCOPY RESEARCH AND TECHNIQUE 47:210–214 (1999) Effect of a Confocal Pinhole in Two-Photon Microscopy R. GAUDERON, P. B. LUKINS, AND C. J. R. SHEPPARD* Physical Optics Department, School of Physics A28, University of Sydney, NSW 2006, Australia KEY WORDS fluorescence; second-harmonic generation; resolution ABSTRACT We investigated the effect of a finite-sized confocal pinhole on the performance of nonlinear optical microscopes based on two-photon excited fluorescence and second-harmonic generation. These techniques were implemented using a modified inverted commercial confocal microscope coupled to a femtosecond Ti:sapphire laser. Both the transverse and axial resolutions are improved when the confocal pinhole is used, albeit at the expense of the signal level. Therefore, the routine use of a confocal pinhole of optimized size is recommended for two-photon microscopy wherever the fluorescence or harmonic signals are large. Microsc. Res. Tech. 47:210–214, 1999. r 1999 Wiley-Liss, Inc. INTRODUCTION Confocal fluorescence microscopy is a well-established laser scanning technique for three-dimensional (3D) fluorescence imaging in both reflection and transmission modes (Pawley, 1989; Paddock, 1999). Normally, a visible CW laser is used together with a pinhole in front of the detector. This pinhole allows optical sectioning by preferentially detecting the light from the focal region, and also leads to an improvement in resolution by spatial filtering of the image on the pinhole. That is, the optical sectioning and hence the 3D capability of confocal microscopy arises from the design of the optical detection system. More recently, multi-photon optical microscopes have been developed (Denk et al., 1990; Gannaway and Sheppard, 1978; Hell et al., 1996; Mueller et al., 1998) based on interactions between nonlinear optical properties of the specimen and the high peak power of the (normally) pulsed illuminating laser. The nonlinear nature of this interaction means that most of the optical signal of interest is generated at the focus of the laser beam in the specimen where the intensity is at a maximum. The spatial properties of this interaction provide a useful mechanism for optical sectioning in such a microscope via a ‘‘soft-aperture’’ effect. Therefore, the optical sectioning arises from the illumination system and so a confocal pinhole is no longer required. The collection efficiency of the detection system in a nonlinear microscope is, therefore, considerably improved because the requirement for an aperture is removed. This, together with the reasonable efficiency of certain nonlinear processes, particularly two-photon excited fluorescence, means that these techniques have lead to new and useful 3D microscopies. By far the most important of these techniques is two-photon excited fluorescence imaging (TPFI), which is now a wellestablished technique with many applications in biomedical, biological, and materials sciences. Secondharmonic generation imaging (SHGI) (Gauderon et al., 1998) is another nonlinear microscopy but has the ability to image, in three dimensions, the intrinsic local second-order nonlinearity of a specimen. r 1999 WILEY-LISS, INC. The resolutions obtained in confocal and two-photon microscopies are, broadly speaking, comparable. Although the focal spot size of the longer wavelength laser light (usually, but not necessarily, in the near infrared region) used in two-photon microscopies is larger than in the visible confocal microscopy case, the point-spreadfunction (PSF) for two-photon imaging is reduced in width by a factor of 冑2 due to the spatial characteristics of the nonlinear interaction (Gu and Sheppard, 1995). The combination of good resolution comparable to confocal microscopy and the greater penetration into biological specimens by the longer wavelength radiation used in two-photon microscopies are important reasons for the current popularity of TPFI in particular. In view of the above, there is the prospect that implementation of a confocal pinhole in two-photon microscopes will provide enhanced transverse and axial resolution, albeit at the expense of signal strength. In this paper, we investigate the effects of adding a confocal pinhole to two-photon microscopes based on two-photon excited fluorescence and second-harmonic generation. MATERIALS AND METHODS The effect of the confocal pinhole on TPFI and SHGI was studied using a modified inverted Leica TCS NT confocal microscope coupled to a Coherent Mira 900 femtosecond Ti:sapphire laser (⬃ 120 fs pulsewidth, 720–860 nm) pumped by a Coherent 5W Verdi solidstate green laser. For this laser pulse width, an adjustable group velocity dispersion (GVD) precompensation arrangement was not necessary. This microscope operated in the beam-scanning mode, giving an acquisition rate of up to 4 images per second. An important capability of this system is that it can operate simultaneously in transmission and reflection modes. The system provided several detection options in the reflection mode, a range of dichroic beamsplitters and block- Contract grant sponsor: Australian Research Council. *Correspondence to: C.J.R. Sheppard, Physical Optics Department, School of Physics A28, University of Sydney, NSW 2006, Australia. E-mail: c.sheppart@ physics.usyd.edu.au Received 28 August 1999; accepted in revised form2 September 1999 TWO-PHOTON MICROSCOPY WITH A PINHOLE 211 ing filters, and a confocal pinhole whose size was adjustable using a motorized actuator. For transmission mode SHGI, the fundamental beam was excluded from the detector using a 400-nm interference filter. RESULTS AND DISCUSSION The effect of the confocal pinhole on the transverse resolution in TPFI is demonstrated in Figures 1 and 2. For a test specimen consisting of a small fluorescent bead with a diameter of 210 nm, images were acquired for four different sizes of the confocal pinhole at constant detector sensitivity (Fig. 1). The bead was imaged using excitation at 800 nm and an oil-immersion 63⫻/ 1.3 NA microscope objective that has a resolving power of ⬃260 nm. Since the beads had dimensions smaller than the diffraction limit, the resulting images represent effectively the intensity PSF of the system. For each image, a line scan through the middle of the bead is also shown. The resolution, estimated as the fullwidth half-maximum (FWHM) of the Gaussian fit to the intensity profile, was calculated for each pinhole size. Clearly, the transverse resolution is improved when the confocal pinhole is used. Theoretically, the improvement in transverse resolution using a small pinhole is approximately 冑2. In practice, we measured an apparent improvement somewhat greater than this. We attribute this to the detection of scattered radiation for larger pinhole sizes. Figure 2 demonstrates this effect for a specimen comprising an intact spinach chloroplast of width ⬃ 4 µm. Since chlorophyll has a major absorption peak near 415 nm, the laser wavelength was set at 830 nm for two-photon excitation. The fluorescence near 680 nm was imaged both without a pinhole (Fig. 2a) and with a pinhole of optimized size (Fig. 2b). Without a pinhole, there is little or no structure evident and the crosssectional intensity profile is almost Gaussian. However, with an optimum confocal pinhole size, the chloroplast periphery is better resolved, the almost step-like crosssectional intensity profile is more representative of the chloroplast morphology, and the internal structure is slightly better resolved. Fine structure of the grana stacks was not highly resolved but this is due mainly to the relatively low NA objective used (Leica dry 40⫻/ 0.75 NA). In Figure 2b, we also see that the improved resolution occurs at the expense of a decrease in the signal level. The effect of the confocal pinhole on the axial resolution in TPFI is demonstrated in Figure 3. The specimen consisted of DAPI-stained cultured onion root cells during the later stages of mitosis. Once completely separated, the separation between the two groups of chromosomes will increase and two cells will be formed (cytokinesis). Because the chromosomes are dispersed in 3D, the specimen is particularly suited for threedimensional imaging or axial imaging. It can be seen that the resolution, particularly in the axial direction, is improved by using a confocal pinhole. This is most clearly seen for the chromosomes (CH) that are in the focal plane and give stronger contrast in Figure 3b when compared with other chromosomes in the images. Fig. 1. Series of two-photon excited fluorescence images and line scans of a single fluorescent bead of diameter 210 nm for four different sizes of the confocal pinhole. a: Confocal pinhole almost fully open rd ⫽ 4 Airy units [A.U.]. b: rd ⫽ 2 A.U. c: rd ⫽ 1 A.U. d: rd ⫽ 0.5 A.U. The FWHMs are quoted relative to the case (a) without a pinhole. 212 R. GAUDERON ET AL. Fig. 2. Two-photon excited fluorescence images of an intact spinach chloroplast of width ⬃ 4 µm for (a) a confocal pinhole almost fully open and (b) an optimised pinhole size. Fig. 3. 2D optical sections of two groups of DAPI-stained onion root chromosomes in a 3D volume imaged by two-photon excited fluorescence. a: Confocal pinhole almost fully open. b: optimised confocal pinhole size. The scan size was 50 ⫻ 50 µm and the excitation was at 730 nm. Because the optical layouts of TPFI and SHGI are similar, demonstration of the SHGI technique in the modified Leica microscope and investigations of the effect of the confocal pinhole were also carried out. For these experiments, lithium triborate (LBO) specimens were used since this material has substantial local second-order nonlinearity, high optical transparency, and a high laser damage threshold. Since secondharmonic radiation is produced in the forward direction, the microscope was used in the transmission mode in order to demonstrate the technique. Figure 4 shows a 3D transmission second-harmonic image and its corresponding two-dimensional image obtained with a Zeiss brightfield microscope for a LBO crystal fragment. TWO-PHOTON MICROSCOPY WITH A PINHOLE 213 Fig. 5. SHG reflection image of LBO for a confocal pinhole almost fully open (a) and for the optimised pinhole size (b). A comparison of line scans made in the region denoted by the circle is shown for 6 different confocal pinhole sizes (c). Fig. 4. 3D SHG transmission image of a LBO crystal fragment (a) and a corresponding 2D brightfield image (b). Features such as crevasses, cracks, and microcrystals are observable in both images and correlate well. This offers the interesting possibility of relating the local second-order nonlinear coefficients of the specimen with the corresponding refractive and absorption properties. In order to investigate the effect of the confocal pinhole, SHGI had to be performed in the reflection mode. In this experiment, the specimen consisted of a thin polished plate of LBO coated on one side with aluminum. Hence, in the reflection mode, the detected second-harmonic signal arose from both back-scattered and retro-reflected second-harmonic light, the relative proportion depending on the specimen transparency. For confocal operation, the laser beam was focused close to the surface of the coating in order to guarantee minimum deviation of the second-harmonic light on its return path. A set of six second-harmonic images of the aluminum-coated LBO specimen was recorded. Figure 5a shows the image with no confocal pinhole while Figure 5b shows the image with an optimum pinhole size. This optimal size was determined by examining the line scans for features with strong intensity gradients and comparing their intensity distributions (Fig. 5c). The resolution is optimal for approximately rd ⫽ 1.5 (where rd is the pinhole size expressed in Airy units) as can best be seen for the two peaks near the centre of the line scan. Apart from changes in signal level and resolution, some alteration in the shape of the line image is apparent. We attribute this to the fact that the intensity in the plane of the pinhole is not uniform, perhaps as a result of refractive effects. We would not expect this to be as evident in two-photon fluorescence as a result of the isotropic emission. The effect of the confocal pinhole is more apparent when xz-sections are taken, as illustrated in Figure 6. Ideally, such sections should be perfectly thin, i.e., dimensionless lines. In reality, the lines spread in both directions. However, it can be observed that the image obtained using an optimised confocal pinhole (Fig. 6b) exhibits better axial resolution than the corresponding image without a pinhole (Fig. 6a). Previous theoretical studies showed that the signalto-noise ratio and the resolution of multi-photon micro- 214 R. GAUDERON ET AL. detection sensitivity. On the other hand, if the signal is large, then substantial resolution improvements can be obtained by using a confocal pinhole. We also demonstrated that both TPFI and SHGI can be easily carried out in a modified inverted commercial confocal fluorescence microscope coupled to a femtosecond laser. Equivalently, a confocal configuration could be implemented in two-photon systems. Fig. 6. Effect of the confocal pinhole size on SHG reflection xz-sections of an aluminum-coated LBO thin plate with the confocal pinhole almost fully open rd ⫽ 3.6 A.U. (a) and with an optimal pinhole size rd ⫽ 0.25 A.U. (b). The scan size was 30 ⫻ 30 µm. ACKNOWLEDGMENTS The authors thank Leica Microsystems, Coherent Scientific Pty Ltd, the Australian Key Centre for Microscopy and Microanalysis, and Dr Guy Cox. Special thanks are due to Mrs. Eleanor Kable and Mr. Greg Stevens who kindly prepared the biological specimens. REFERENCES scopes were improved by adding a confocal configuration to the detection path (Gauderon and Sheppard, 1999; Gu and Gan, 1996; Wilson and Sheppard, 1979). Our experimental results have confirmed that adding a confocal pinhole gives a further improvement in transverse and axial resolution. We found that the use of a confocal pinhole slightly improves the transverse resolution, but gives an even greater improvement in the axial resolution. This particular improvement in the axial direction is especially beneficial since the unimproved axial resolution is generally inferior to transverse resolution in laser scanning microscopes. The use of a confocal pinhole is, therefore, most effective for high-resolution 3D imaging. Unfortunately, the signal collection efficiency is reduced by using a confocal pinhole. In addition, if a confocal pinhole is not used, the emitted radiation need not be descanned so that even scattered fluorescence can be detected. If the signal levels are low or if the total laser exposure that can be tolerated by the specimen is limited, addition of a confocal pinhole may be problematic in terms of the Denk W, Strickler JH, Webb WW. 1990. Two-photon laser scanning fluorescence microscopy. Science 248:73–76. Gannaway J, Sheppard CJR. 1978. Second harmonic imaging in the scanning optical microscope. Opt Quantum Elect 10:435–439. Gauderon R, Sheppard CJR. 1999. Effect of a finite-size pinhole on noise performance in single-, two- and three-photon confocal fluorescence microscopy. Appl Optics 38:3562–3565. Gauderon R, Lukins PB, Sheppard CJR. 1998. Three-dimensional second-harmonic generation imaging with femtosecond laser pulses. Optics Lett 23:1209–1211. Gu M, Gan XS. 1996. Effect of the detector size and the fluorescence wavelength on the resolution of three- and two-photon confocal microscopy. Bioimaging 4:129–137. Gu M, Sheppard CJR. 1995. Comparison of three-dimensional imaging properties between two-photon and single-photon fluorescence microscopy. J Microsc 177:128–137. Hell SW, Bahlmann K, Schrader M, Soini M. 1996. Three-photon excitation in fluorescence microscopy. J Biomed Opt 1:71–74. Mueller M, Squier J, Wilson KR, Brakenhoff GJ. 1998. 3D microscopy of transparent objects using third-harmonic generation. J Microsc 191:266–274. Paddock SW. 1999. Confocal fluorescence microscopy: methods and protocols. New Jersey: Humana Press. 446 p. Pawley JB. 1989. Handbook of biological confocal microscopy. London: Plenum Press. 632 p. Wilson T, Sheppard CJR. 1979. Imaging and super-resolution in the harmonic microscope. Optica Acta 26:761–770.