You are on page 1of 16

ARTICLE IN PRESS

Optics and Lasers in Engineering 44 (2006) 208223

Use of confocal laser scanning microscopy (CLSM) for depthwise resolved microscaleparticle image velocimetry (m-PIV)
J.S. Parka, K.D. Kihmb,
a

Department of Mechanical Engineering, Texas A&M University, College Station, Texas 77843-3123, USA b Mechanical, Aerospace and Biomedical Engineering Department, University of Tennessee, Knoxville, TN 37996-2210, USA Available online 23 May 2005

Abstract A novel use of confocal laser scanning microscopy (CLSM) makes the truly focused eld-ofview with well-dened depthwise resolution possible for microscale particle image velocimetry (m-PIV) applications. The operating principle of the CLSM is presented using the point spread function (PSF) that describes diffracted images of extremely small particles. The implemented high-speed CLSM system using a Nipkow rotating disk is applied to measure the microscale rotating Couette ow eld conned between two parallel horizontal disks that are 180-mm apart, with the bottom one stationary and the top one rotating and seeded by 200-nm uorescent spheres. The CLSM provides much distinct particle images in comparison with the conventional wide-eld microscopy (WFM) and the measured vector proles are more concentric and accurate depicting closer to an ideal Couette ow. r 2005 Elsevier Ltd. All rights reserved.
Keywords: Confocal laser scanning microscopy (CLSM); Wide-eld microscopy (WFM); Point spread function (PSF); Microscale Couette ow; Microscale particle image velocimetry (m-PIV)

Corresponding author. Tel.: +1 865 974 5292; fax: +1 865 974 5274.

E-mail address: kkihm@utk.edu (K.D. Kihm). URL: http://www.engr.utk.edu/$minsfet. 0143-8166/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.optlaseng.2005.04.005

ARTICLE IN PRESS
J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223 209

1. Introduction Confocal laser scanning microscopy (CLSM) has many distinctive advantages compared with conventional microscopy [1]. The fundamental concept of confocal microscopy is the elimination of out-of-focus images by having a confocal pinhole thus producing a better resolution in the lateral as well as axial directions. In practice, the best lateral resolution of a confocal microscope can be as small as 0.2 mm, and the best axial resolution can be as ne as 0.5 mm [2,3]. The more striking feature of the confocal microscopy is its ability to deliver in-focus images by true means of depth wise optical slicing, and allowing the gathering of 3-D reconstructed information without the need for physical slicing of specimens. Fig. 1 illustrates the concept of a confocal pinhole. The pinhole aperture, located at the confocal point, allows the emitted uorescent light exclusively from the focal point (solid rays) to pass through the detector, and lters out the uorescent light emitted from outside the focal point (dashed rays). This spatial ltering is the key idea to enhance the optical resolutions by devising the depthwise optical slicing. In a conventional wide-eld microscopy (WFM), a specimen is completely illuminated by an excitation light, so the entire specimen is uorescing or emitting simultaneously. It is no doubt that the excitation light has the highest intensity at the focal point of the objective lens, but nonetheless, the other parts of the specimen can absorb some of the excitation light and they also uoresce. This phenomenon contributes to a background haze in the captured image. An extensive study by Sandison and Webb [4] shows that the signal-to-background ratio, with background dened as the detected light originating outside the resolution volume, obtained with a confocal microscope can be more than 100 times greater than the signal-to-background ratios available with WFM, and the optimized confocal signal-to-noise ratio can be a factor of 10 or greater than that of the conventional microscope. The pinhole limits the illumination of the specimen to one particular spot, and light must be delivered to every point of the specimen in a raster pattern. The galvanometric steering of the focal point of the traditional confocal microscopy limits its scanning speed to approximately one (1) frames-per-second (FPS), which is too slow for real-time observation of moving objects at any practical speed. The innovative use of a rotating micro-lens array [5,6], replacing the single pinhole, makes

Fig. 1. A schematic diagram for a concept of confocal pinhole.

ARTICLE IN PRESS
210 J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223

it possible for confocal microscopy to scan full-eld images at substantially higher FPS rates (Fig. 2). The essential innovation of the CLSM is the use of dual highspeed spinning disks (Fig. 2): the upper disk is a rotating scanner that consists of 20,000 micro-lenses, and the lower one is called a Nipkow disk that consists of matching 20,000 pinholes of 50-mm in diameter [7,8]. Both the incident excitation light and emitting uorescence light paths are dened by a similar optical path. The pumping light is focused by the micro-lens of the scanning disk through the pinholes on the Nipkow disk. A dichroic mirror, located between the two disks, reects the returning confocal uorescence image to the CCD for a real-time recording. As a maximum scanning rate of 360 per second with the disk rotation of 30-rps is possible, and the multiple pinholes sweep the eld-of-view, a full eld imaging of up to 120-FPS is achieved by averaging three sweeps per single eld for statistical enhancement. Further study has been done to use the high-speed confocal microscopy with a rotating scanner for advanced bio-medical applications of real-time 3-D imaging of single molecular uorescence [9]. For microuidic applications, Park et al. [10] rst successfully mapped the optically sliced velocity proles for Poiseuille ows developed in a 100 mm diameter capillary with pronounced improvement in the measurement accuracy and reducing the image ambiguity. A fully developed horizontal laminar ow was measured at several depthwise planes using the CLSM and the results showed substantial improvement in the measurement accuracy in comparison with the conventional PIV results. Park et al. [11,12] also performed a micro two-phase ow experiment that presented the uid ows around a moving bubble in a 100-mm square channel. They successfully achieved a full-eld and optically sectioned ow velocity mapping for the region ahead of the moving air bubble using the CLSM. Recently, the conventional microscale-particle image velocimetry (m-PIV) technique was

Fig. 2. A dual-disk system of high-speed confocal microscopy (Yokogawa).

ARTICLE IN PRESS
J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223 211

theoretically studied for its optical limitations in the depthwise resolution by examining a thick depth of focus and a background noise [13,14]. They attempted to improve the measurement accuracy by correcting the biased velocity eld using a correlation function depending on the numerical aperture (NA) of the objective and the particle diameter. However, given the objective NA, the fundamental limitations of the conventional WFM still remains, and such a correction cannot make the depthwise resolution of WFM anyway near that achieved by the CLSM optical slicing. In the present paper, the confocal imaging principle is discussed using the point spread function (PSF) that describes diffracted images of extremely small particles. Example ow measurements have been conducted for a rotating Couette ow conned in a 180-mm gap between the top rotational disk and the bottom stationary plane, and comparative results are presented between the CLSM m-PIV and the conventional WF epi-uorescence m-PIV measurements.

2. Confocal imaging and PSF 2.1. 3-D diffraction pattern and PSF The diffraction occurs because of the wave nature of light, i.e., the electromagnetic light waves emitted from a point source and passing through a rear aperture of objective lens are superposed constructively or destructively depending on the optical path differences of individual waves. When a ne source point in specimen is observed using an optical microscope even at a perfect focusing, the image is formed as a 2-D airy diffraction pattern transversely spread on the image plane. Although the objective lens acts to focus the waves on an innitely small point for an image formation, it is inevitable that the waves interfere with each other near the focal point to produce the diffraction pattern. Accordingly, the off-focused source point in the specimen is represented by a much widely and three-dimensionally spread diffraction pattern. The 3-D diffraction pattern is called a PSF that is oscillating laterally as well as axially. The PSF is extended symmetrically before and after the focal plane along the optical axis as well as radially around the optical axis [15]. Considering Fig. 3, a monochromatic spherical wave emitted from a point source transverse an entrance pupil of an objective lens and emerges from an exit pupil of the objective lens, and converges towards the back focal point O. The PSF, h(u,v) at an arbitrary point P(x,y,z), with the u and v being longitudinal and transverse non-dimensional variables in the cylindrical coordinate system along the optical axis, can be analytically expressed based on the Huygens Fresnel Principle. In consequence, the PSF is expressed with convergent series of Bessel function and complex exponential terms as [15] Z 2pia2 A if =a2 u 1 1=2iur2 hu; v e e J 0 vrr dr; (1) lf 2 0

ARTICLE IN PRESS
212 J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223

Fig. 3. Illustration of an image formation near the focal point.

where r r=a (r is the radial co-ordinate and a is the radius of the exit pupil),       2p a 2 2p a 2p a p z and v u r x2 y2 . l f l f l f Note that the Bessel function term is a function of only transverse v variable, and the exponential term is a function of only longitudinal u variable. This means that the lateral and axial diffraction patterns of a PSF are produced in terms of the exponential function and the Bessel function, respectively [16]. For the case of conventional WF microscopy, the captured image intensity is described as a square of the PSF, i.e., I u; v jhu; vj2 . 2.2. PSF of confocal microscopy The PSF of CLSM is derived from the coherent optical imaging formation in scanning microscopes using a detector, the area of which is limited by a pinhole [17]. The light illuminated from a laser source is focused by an objective into a specimen, and the reected light is refocused onto the detector on the image plane mostly by the same objective lens. The pinhole allows only the focused light rays pass through but the off-focused lights to be blocked. Since the pinhole may be regarded as an additional point source from the detectors point-of-view, in addition to the original illuminating point source, the resulting PSF of the CLSM system can be expressed as a multiplication of the PSF, Eq. (1). The PSF of the excitation light, hex(u,v), is given by the Hankel transform of the pupil function Z 1 2 Pex re1=2iur J 0 vrr dr: (2) hex u; v 2
0

The PSF of the emission light, hem(u,v), is given by Z 1 2 Pem re1=2iur J 0 vrr dr. hem u; v 2
0

(3)

ARTICLE IN PRESS
J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223 213

In the above equation, r r=a, where r is the actual radial co-ordinate and a is the radius of the pupil, v krsin a1 , where k is the wave number ( 2p=l), l is the wavelength, and a1 is the angular aperture of the objective, u 4kzsin2 a1 =2, and Pr is the pupil function of the objective. The image formation depends critically on the form of the pupil function, Pr [17]. In general, the PSF of the conventional microscopy having an in-coherent optical system can be expressed by only one of both PSFs hconventional u; v jhu; vj. (4) However, the CLSM has the form of an effective PSF, which is given by the product of the excitation PSF and the emission PSF, and is expressed by hCLSM u; v jhem u; vhex u; vj. (5) In CLSM, the coherent lights of excitation and emission are employed simultaneously to image the object and the image is built up by scanning the specimen and the detector in synchronism. The characteristic of CLSM makes two PSFs of lights equal, and the PSF of CLSM can be expressed as a product of two PSFs of conventional microscopy. hCLSM u; v jhu; vj2 hconventional u; v2 . (6)

In consequence, the intensity distribution of CLSM can be simply expressed as the product of two intensity distributions of conventional microscopy. I CLSM u; v I conventional u; v2 jhu; vj4 . (7)

Figs. 4a and b show the calculated PSF proles for WFM and CLSM, respectively. The maximum image intensity at Dz 0 (the focal plane) of the CLSM case is substantially higher than that of the WFM case. This ensures that the CLSM provides more distinctive and brighter images than the WFM. In addition, since the CLSM intensity is a square of the WFM intensity as seen in Eq. (7), the modulation intensity of the CLSM is far more quickly diminishing with increasing defocusing distance (Dz) in comparison with the WFM case. This describes the key advantage of CLSM imaging in that more distinctive optical slicing can be achieved with less ambiguity occurring from the off-focused images. Fig. 5a shows measured PSF proles for the WFM imaging in comparison with the corresponding CLSM imaging shown in Fig. 5b. For a spatially xed 500-nm uorescence particle,1 the focal plane of the objective lens (40x, 0.75NA) was controlled mechanically with minimum adjustment resolution of one micron. Note that Dz indicates the defocus distance between the corresponding point source (P) and the back focal point (O) in an image plane as illustrated in Fig. 3. The measured diffraction intensity patterns show qualitatively similar intensity variations as the calculated PSFs shown in Fig. 4, and also shown is the quicker diminishing of the CLSM diffraction intensity with increasing the defocusing manifests the
1 While 200-nm seeding particles are actually used for the ow vector measurements, the bigger particles are used to generate more distinctive diffraction images for clear comparison between the CLSM and WFM images.

ARTICLE IN PRESS
214 J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223

Fig. 4. Calculated PSF proles: (a) wide-eld microscopy (WFM) and (b) confocal laser scanning microscopy (CLSM).

aforementioned optical slicing advantage in using the CLSM. Note that no such denition for optical slicing is possible for conventional microscopy. More prolonged diffraction images for the WFM case contributes to raise the background

ARTICLE IN PRESS
J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223 215

Fig. 5. Measured PSF proles: (a) WFM and (b) CLSM.

ARTICLE IN PRESS
216 J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223

noise and can result in serious uncertainty problems particularly for m-PIV applications with the case of high magnication imaging.

3. Experiment The experimental facility (Fig. 6) consists of a CCD camera (UP-1830, UNIQ), a confocal module (CSU-10, Yokogawa Inc.), an inverted microscope (IX-50, Olympus Corp.), an argon-ion laser (488 nm50 mW, Laser Physics Inc.), a computer with a frame grabber (QED Imaging), and a test section. The confocal module is attached to the side port of the microscope, and the CCD camera is connected to the other side of the confocal module. The magnication and numerical aperture of the objective lens are M 10X and NA 0:3 for the present experiment, and its working distance is 10 mm. The illuminating laser light passes through a single-mode ber optic cable to enter the confocal module. PIV images are consecutively captured with a frame rate of 30 fps and a shutter speed of 33-ms. The images from the CCD camera are transferred through a 16 bit digital cable to the frame grabber installed in PC. The commercially available Davis PIV software from LaVision Inc. is used to process the recorded images to calculate the ow velocity vector distributions based on the cross correlation schemes. Fig. 7 shows the experimental setup to provide a micro Couette ow developed between the submerged rotating disk and the stationary bottom plate. The gab

Fig. 6. Experimental setup of a micro rotating Couette ow measurement.

ARTICLE IN PRESS
J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223 217

Fig. 7. Test section of a micro rotating Couette ow experiment.

between the disk and the bottom is measured as 180 mm and a number of measurement planes are dened at different heights to show optically sliced rotating ow elds. A stable DC power supply (6033A, Hewlett Packard Corp.) provides a constant disk rotating speed at 17.0 rpm. 200-nm yellowgreen uorescent microspheres (SG 1.05, 505 nm absorption/515 nm emission, Molecular Probes Inc.) are used as tracers mixed with de-ionized water at 0.002% volume concentration.

4. Results and discussion Fig. 8 shows the particle images captured by the CLSM system in the left column and by the WFM system in the right column, at three different heights of h 142, 88, and 35 mm, respectively, measured from the bottom plane. The distinctive advantage of the CLSM system in achieving optically sliced imaging is clearly seen while the conventional WFM shows blur and obscured images from the out-of-focus foreground and background images. The resulting blended images will obscure the particle displacements to create underestimated, overestimated, biased and/or incorrect ow eld calculations. The data will also suffer from excessive spatial uncertainties of the measurement plane locations. The distinction of the CLSM measured data is also quantitatively shown by the ow contour maps presented in Fig. 9. The ow contours should be concentric for the case of an ideal rotating Couette ow. The ow velocity contours by the WFM shown in the right column (Figs. 9b-1 to 6) do not show the concentricity and accuracy of the Couette ow characteristics. The concentric feature has been substantially improved by the CLSMs optically sliced images as shown in the left column (Figs. 9a-1 to 6). The deviation of the WFM data from the CLSM data becomes more severe with increasing height, i.e., in the vicinity of the rotating disk. This is attributed to the more complicated ow proles and the initiation of the 3-D secondary ows as the ow rotation increases.

ARTICLE IN PRESS
218 J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223

Fig. 8. Particle images captured by the CLSM (a-1, 2 and 3) and by the WFM (b-1, 2 and 3) at three different heights measured from the stationary bottom.

ARTICLE IN PRESS
J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223 219

Fig. 9. Flow contours calculated from the PIV images taken by the CLSM (a-1 to 6) and by the WFM (b-1 to 6) at six different heights.

In order to quantitatively examine the measurement accuracy and uncertainties, the tangential (rotational) component of the ow velocity vectors are presented along the height h ranging from 35 to 170 mm at four different radial locations of R 50, 100, 150 and 200 mm in Figs. 10ad, respectively. Each symbol represents the

ARTICLE IN PRESS
220 J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223

Fig. 9. (Continued)

ensemble average of total 464 velocity vectors calculated at 16 different angular locations per image and for total 30 PIV images. The solid extended bars represent the uctuation ranges of the measured velocity vector magnitudes with 99% condence range for the CLSM, and the dashed bars for the WFM. When the radial location is closer to the axis, at R 50 and 100 mm, the average data of both the

ARTICLE IN PRESS
J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223 221

(a)

(b)

(c)

(d)

Fig. 10. Comparison of tangential velocity distributions along the height at several different radial (R) locations: the symbols represent the ensemble average data from 464 realizations; the solid extended bars indicate the data scattering range measured by the CLSM; and the dashed extended bars indicate the data scattering range measured by the WFM.

CLSM and the WFM measurements show fairly good agreement with the predicted linear Couette ow proles as shown by the straight dashed lines. However, at relatively larger radii of 150 and 200 mm, particularly as the upper rotating disk is approached at high h, the average data start deviating from the predictions showing slight underestimation for the CLSM measurements (the triangle symbols), and excessive underestimation for the WFM measurements (the square symbols). The gradually progressing underestimation of the CLSM data reects the 3-D secondary ow initiation with increasing radius in the vicinity of the rotating disk. As a result, the local ow starts deviating from the solid rotation of the ideal and two-dimensional rotational Couette ow. The rapidly progressive deviation

ARTICLE IN PRESS
222 J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223

of the WFM data is attributed to the obscurity created by the out-of-focus-plane imaging of the complicated 3-D secondary ows. The data uctuations of the WFM measurements are substantially larger than the CLSM data uctuations for all the regions and dramatically exemplied when the rotating disk is approached because of the more pronounced obscured imaging of the out-of-focus planes. The CLSM data uctuations are gradually increasing with increasing line-of-sight distance, or equivalently height h, and this is believed due to the inevitable reduction of the signal-to-noise level for deeper imaging planes inside the test uid.

5. Conclusions A dual-disk confocal laser scanning microscopy (CLSM) was successfully applied to a PIV measurement of a micro rotating Couette ows and the measurement results were compared with the corresponding wide-eld microscopy (WFM) measurements. Because of the distinctive optical slicing capability, the CLSM predicts the exact solution of an ideal two-dimensional Couette ow far more accurately than the WFM data. More crucially, the WFM data uctuations are excessively large, particularly when the secondary ow is initiated near the rotating disk. The use of the CLSM for m-PIV applications can alleviate such a critical problem occurring from the spatial uncertainties of a depth-of-eld.

Acknowledgments The authors acknowledge that the current research has been partially sponsored by the NASA-Fluid Physics Research Program Grant No. NAG 3-2712, and partially by a subcontract from the R4D Program at the National Center for Microgravity Research (NCMR). The presented technical contents are not necessarily the representative views of NASA or NCMR.

References
[1] Minsky M. Memoir on inventing the confocal scanning microscope. Scanning 1998;10:12838. [2] Webb RH. Confocal optical microscopy. Rep Prog Phys 1996;59:42771. [3] Wilhelm S, Grobler B, Gluch M, Heinz H. Confocal laser scanning microscopy: principles. Carl Zeiss; 2003. [4] Sandison DR, Webb WW. Background rejection and signal-to-noise optimization in confocal and alternative uorescence microscopes. Appl Opt 1994;33:60315. [5] Conchello JA, Lichtman JW. Theoretical analysis of a rotating-disk partially confocal scanning microscope. Appl Opt 1994;33:58596. [6] Tiziani HJ, Uhde HM. Three-dimensional analysis by a microlens-array confocal arrangement. Appl Opt 1994;33:56772. [7] Homepage origin: http://www.solameretech.com; 2003. S, Spring KR. Video microscopy: the fundamentals. New York: Plenum Press; 1997. [8] Inoue

ARTICLE IN PRESS
J.S. Park, K.D. Kihm / Optics and Lasers in Engineering 44 (2006) 208223 223

[9] Ichihara A, Tanaami T, Isozaki K, Sugiyama Y, Kosugi Y, Mikuriya K, et al. High-speed confocal uorescence microscopy using a Nipkow scanner with microlenses for 3-D imaging of single uorescent molecule in real time. Bioimages 1996;4:5762. [10] Park JS, Choi CK, Kihm KD. Optically sliced micro-PIV using confocal laser scanning microscopy (CLSM). Exp Fluids 2004;37:10519. [11] Park JS, Kihm KD, Allen JS. Three dimensional microuidic measurements using optical sectioning by confocal microscopy: ow around a moving bubble in a micro-channel. Proceedings of 2002 ASME IMECE, Paper No. IMECE2002-32790, November 2002. [12] Park JS, Choi CK, Kihm KD, Allen JS. Optically sectioned micro PIV measurements using CLSM. ASME J Heat TransferPhotogallery 2003;125:542. [13] Meinhart CD, Wereley ST, Gray MHB. Volume illumination for two-dimensional particle image velocimetry. Meas Sci Technol 2000;11:80914. [14] Olsen MG, Adrian RJ. Out-of-focus effects on particle image visibility and correlation in microscopic particle image velocimetry. Exp Fluids 2000(Suppl.):S16674. [15] Born M, Wolf E. Principles of optics. New York: Pergamon; 1965. p. 437. [16] Gibson FS, Lanni F. Experimental test of an analytical model of aberration in an oil-immersion objective lens used in three-dimensional light microscopy. J Opt Soc Am A 1991;8:160113. [17] Wilson T. Confocal microscopy. London: Academic Press; 1990.

You might also like