Open Access
Add to BibSonomyAbstract
Abstract
A digital light modulation microscope (DLMM) that utilizes a digital micromirror device (DMD) on an epifluorescence microscope has been developed to modulate excitation light in spatial and temporal domains for phosphorescence lifetime detection. Local O2 concentration can be inferred through the detected lifetime around an O2-quenching phosphorescent porphyrin microsensor. Combined with microsensor arrays, the DLMM can sequentially address light to each microsensor element to construct a discrete lifetime image or O2 distribution. In contrast to conventional phosphorescence lifetime imaging, the new method eliminates the need for a pulsed light source and a time-gated camera. To demonstrate O2 sensing with lab-on-a-chip devices, an array of 150-µm-diameter micro-wells coated with phosphorescent porphyrin were observed. The locations of the sensor elements were automatically identified though image analysis. The goal of this platform is to measure the O2 consumption of individual cells trapped in the microwells.
©2007 Optical Society of America
1. Introduction
Cellular oxygen uptake rate is a critical parameter to achieve comprehensive understanding of metabolic responses. At an individual-cell level, conventional electrochemical oxygen measurement methods have the disadvantages of low signal-to-noise ratios and inherent oxygen consumption of the sensor. The porphyrin-based optical detection method does not have these disadvantages, and it is compatible with widely accepted fluorescence microscopy methods for genomics studies [1, 2]. Here, we report on a digital light modulation microscope (DLMM) that can modulate excitation light in both space and time domains using a digital micro-mirror device (DMD, Texas Instruments, Dallas, TX) [3]. A DMD chip is the key component of commercial digital projectors utilizing Digital Light Processing (DLP) technology. DMD chips have arrays of approximately one million microfabricated mirrors (exact numbers depending on the models). Each mirror is a tiny light switch controlled by an electronically addressable CMOS circuit. The mirrors are actuated by the electrostatic force generated by the electrodes beneath individual mirrors.
The DMD has been introduced to a variety of microscopy techniques. The early motivation for using DMDs on microscopes was to replace the Nipkow disk of common spinning disk confocal microscopy for flexible, programmable operation [4–6]. The Jovin group adopted DMD techniques for epifluorescence confocal microscopy for biological applications with extensive theoretical and experimental studies [7–9]. Bansal et al. implemented a DMD-based fluorescence microscope to observe specimens. Unlike Jovin’s approach, the DMD only control the emission patterns to achieve optical sectioning [10]. Fukano et al. combined DMD and fringe-projection techniques to reconstruct optically sectioned images [11]. MacAulay et al. used a DMD to control aperture iris and field stop to compensate for illumination uniformity [12], and integrated a DMD with a fiber-optic bundle as a confocal endoscope [13].
Although the previous DMD-based microscope designs exploited the spatial light modulation ability of DMD techniques, none of them utilized the fast switching feature of the DMD for time domain measurements. The design of the proposed DLMM uses a DMD to generate light pulses to excite phosphorescence emission of the oxygen-sensitive porphyrin, allowing for real-time O2 concentration monitoring through phosphorescence lifetime measurement [14]. The DLMM is especially suited to lab-on-a-chip devices and large-scale living cell assays for high content screening, such as the microfabricated living cell array (LCA) that is designed for single-cell oxygen uptake analysis [15, 16].
2. Optical oxygen concentration measurement
Porphyrins are large tetrapyrrolic macrocyclic compounds that are able to coordinate metal atoms. They are phosphorescent materials that absorb higher-energy excitation photons to reach triplet states, and then emit lower-energy photons as the porphyrin molecules return to low-energy states. A global phosphorescence process has an exponential decay response from excitation to emission. The time exponent of such a process is the phosphorescence lifetime. The emission decay I em(t) as a function of time t can be expressed as
where Ī em is the emission level when excitation is on and τ is the phosphorescence lifetime. Excited triplet states (resulting in phosphorescence) of metalloporphyrins are quenched effectively with dynamic quenchers such as oxygen molecules [17]. Phosphorescence intensity and lifetime are both functions of O2 concentration (denoted as [O2]), so measuring either emission intensity or phosphorescence lifetime can be used to estimate [O2]. Although measuring intensity is much easier than measuring lifetime, phosphorescence lifetime measurement is more robust and is immune to common problems such as photobleaching, variation of lamp illumination, and the amount of phosphorescent material present.
The relation between phosphorescence lifetime τ and [O2] is described by the Stern-Volmer equation:
where Ksv is the Stern-Volmer quenching constant and τ 0 is the phosphorescence lifetime in the absence of oxygen. As indicated by Eq. (2), a high [O2] results in a short phosphorescence lifetime. In addition, the quenching process does not consume oxygen so the measurement is less invasive as compared to electrochemical oxygen probes.
Platinum-porphyrin was selected for oxygen measurements because of its good stability during chemical and temperature variation, its bright emission, and its commercial availability as a compound embedded in polystyrene microspheres. The microscope filter configuration is simple because the porphyrin excitation and emission spectra are well separated (e.g., 390-nm excitation peak, 650-nm emission peak). The phosphorescence lifetime is on the order of 25 to 100 µs, much longer than ~10 ns decay time of conventional fluorophores, and long enough for the DLMM to detect. We deposited 1-µm-diameter carboxylate-modified Pt(II) meso-tetra(pentafluorophenyl)porphyrine (Pt-MTP) microspheres (F-20888, Invitrogen, Carlsbad, CA) to the outer edges of the microwells on the LCA (Fig. 1), which has an array of 150-µm-diameter, 50-µm-deep microwells etched in borosilicate glass [15]. For high-content living cell screening, individual living cells will be seeded into single wells and then the wells will be sealed to monitor the O2 consumption of single cells. The utility of the LCA for this purpose has been successfully demonstrated by using a pulse-laser and a time-gated CCD camera [18].
Fig. 1. The microfabricated living cell array (LCA) with microwells etched on a glass substrate, where each well works as a cell trap with ring-shaped O2-sensitive phosphorescent material.
3. Microscope and DMD configuration
A schematic of the DLMM is shown in Fig. 2. The construction is similar to common fluorescence microscopes except that a DMD is located on the conjugated plane of the image and object planes. The 1027×768 mirrors on the DMD perform as electronically addressable optical switches, each having a ~15 µs switching time [19]. We purchase the DMD as a part of DMD Discovery kit that included control electronics (Productivity Systems Inc. Richardson, TX). Although the switching time is shorter than the phosphorescence lifetimes of the O2-quenching porphyrin (~25–100 µs), the detectable emission decay (exponent denote as ) is a convolution of DMD switching and phosphorescence emission decay with the exponent: τ in Eq. (2). For our experiments, the frequency and duty cycle of the DMD modulation are 1kHz and 25%, respectively.
The structured light pattern of the DMD and the switching rate of mirrors are controlled by a central computer (Precision 450 with one Xeon processor, Dell Computer Corp., Round Rock, Texas). A Xenon arc lamp (X-Cite 120, EXFO Life Sciences, Mississauga, ON, Canada) illuminates the DMD uniformly and then projects to the specimen through a filter cube (XF101-2, Omega Optical, Inc., Brattleboro, VT) for the selected phosphorescent sensor compound. The emitted light is detected by a CCD camera (Infinity 2, Lumenera Corp., Ottawa, ON, Canada) and a photo multiplier tube (PMT, H6780-01, Hamamatsu Photonics K.K., Shizuoka, Japan) simultaneously with a 50–50 beam splitter. We selected a broad-bandwidth PMT (0.78 ns nominal rise time) so the phosphorescence lifetime detection is not affected by the speed of the PMT electronics. An oscilloscope is used to record the emission signal waveform from the PMT. In contrast to other phosphorescence lifetime imaging techniques in the literature [18, 20–22], the new method eliminates the need for a pulsed light source and a time-gated camera.
Although the field of view of the DLMM covers multiple microwells, the DMD can modulate illumination spatially so light only excites the microwell of interest (Fig. 3). The inserts on the lower-left of all images indicate the pattern on the DMD (white areas represent switched-on micro-mirrors). Figure 3(a) was acquired by the CCD camera when all mirrors were switched on; Figure 3(b) were acquired when the DMD projected two rings to the sensor rings. By comparing Fig. 3(a) to Fig. 3(b), projecting light only to the region of interest provides better contrast and sharpness. Figure 3(c-d) show lights addressed to individual sensors to measure local parameters independently. The phosphorescence lifetime in an individual microwell can be detected by projecting light only to the ring area as in Fig. 3(c) and 3(d), and quickly switching it on and off to generate an excitation pulse train. Therefore, the PMT only detects the phosphorescence emission of the projected ring.
Fig. 3. Images of two 150-µm diameter, ring-shaped phosphorescent patches imaged by the DLMM. The inserts on the lower-left of all images indicate the light pattern in the DMD (white areas represent switched-on micro-mirrors).
4. Image alignment
The DLMM is designed to automatically recognize the locations of the O2 microsensors through the CCD camera on the microscope. The key to achieving this goal is the registration of CCD pixels to the DMD micromirrors [23]. The 2D mirror/pixel indices of the DMD and CCD camera can be seen as two orthogonal coordinate systems. By projecting the DMD coordinates to the CCD camera, the relationship between the two coordinate systems is a combination of rotation and translation, as illustrated in Fig. 4(a) (assuming that the optical distortion is negligible). This relationship can be expressed as
where [x,y]T is the coordinate vector in the DMD domain, [x̄, ȳ]T is the coordinate vector in the CCD domain, [x̄ 0,ȳ 0]T is the translation vector between two original points on the CCD domain, and θ and M are the rotational angle and magnification between the two coordinate systems, respectively.
Fig. 4. (a) The relationship of the coordinate system of the CCD camera to that of the DMD. (b-c) Calibration of the coordinate systems using a user interface that processes images from the CCD camera: (b) a DMD generated a horizontal line for calculating the rotation angle between two coordinate systems. (c) DMD generated squares for calculating the magnification between the two coordinate systems.
The values of θ, M, and [x̄ 0,ȳ 0]T are determined empirically. The angle θ is determined by generating a straight line on the DMD to project to the specimen. Fig. 4(b) shows the line captured by the CCD camera. Note that rough alignment of θ can be easily adjusted by changing the mounting angle of the camera, and can be minimized physically.
We used another vertical straight line to find the translation vector [x̄ 0,ȳ 0]T, where the intersection of the horizontal and vertical lines is defined as an origin. The magnification M is found by projecting bright areas with known dimensions on the DMD domain and calculating the corresponding dimensions on the CCD image. Figure 4(c) shows the use of two 64x64 pixels on the DMD to calculate the scaling factor M. With the values of θ, M, and [x̄ 0,ȳ 0]T, a CCD pixel can precisely register to the DMD locations, and an interactive graphical user interface can be built to control the pattern of the DMD.
5. Sensor element identification
This section describes the procedure to identify the regions of interest (ROI, i.e., the regions of individual wells) and use the DMD to match the interrogation areas with the ROI. For our particular LCA, we take advantage of the phosphorescent sensor rings to identify the outer diameters and centroids of individual rings, which can be inferred as the diameters and centers of the wells.
The raw fluorescence image from the CCD camera is shown in Fig. 5(a). A threshold operation is performed and the images are converted into binary as shown in Fig. 5(b). The white regions in the binary images are recognized as identified objects (i.e., possible sensor rings). Next, this image is processed using morphological open and close operations to smooth the edges of the objects but without significantly changing the sizes and locations of either the objects or their associated centroids, as shown in Fig. 5(c). This step is advantageous because it removes background noise and shortens the computing time required to estimate feature dimensions. The outer edges of the objects in Fig. 5(c) are fit as circles using the method described in Appendix 3 in [24]. Figure 5(d) shows the outlines, diameters, and center locations of the fit circles. Objects of erroneous dimensions, usually due to the speckle noise in the image, will be removed.
Fig. 5. The procedure and demonstrated results of micro-well identification: a) raw fluorescence image of five micro-wells, b) thresholding the raw image into a binary image, c) morphological processing to remove undesired artifacts, and d) estimating the geometrical parameters (unit: pixel).
6. Results and discussion
Figure 6 demonstrates the phosphorescent lifetime detection of a single well on the LCA under two different gaseous state O2 concentrations. The excitation was switched off at 0 µs in the time axis. The emission waveforms were averaged with eight pulses. Fig. 6(a) shows the phosphorescent response in ambient conditions (~21% O2) where a decay time ~52.6 µs was measured. Fig. 6(b) shows the decay signal when the sensors were purged with nitrogen (~0% O2); a decay time ~102.3 µs was measured. The phosphorescent lifetimes of Pt-MTP microspheres as a response to [O2] in gaseous states have not been found in the literature. The range of phosphorescent lifetimes of Pt-MTP is usually longer than that of well-studied platinium(II)-octaethyl-porphyrin (Pt-OEP), which varied from 20 µs (in 21% O2 solution) to 70 µs (in 0% O2 solution) [25].
Actual lifetimes τ can be estimated by removing the DMD switching response through deconvolution. However, we select to present raw measurements because deconvolution introduces extra variations to the inferred lifetimes. For the purpose of [O2] measurement, a direct calibration between and [O2] can be established.
Fig. 6. The phosphorescence decay curves of the detected O2 sensor patch (solid lines) and fit curves (dashed lines). (a) ambient conditions (~21% O2); (b) depleted O2 conditions (~0% O2).
Measurement of the phosphorescence lifetime of multiple microwells is enabled by both the spatial and temporal modulation capabilities of the DLMM. Measured phosphorescence lifetimes of five individual wells of the LCA are shown in Fig. 7(a). During these sequential measurements, the DMD pulses light onto individual wells by rapidly switching the mirrors mapped inside the squares. As the mirrors turn off, a photosensor detects the decay of each well. Figure 7(b) shows the phosphorescence decays of the five wells in ambient conditions due to single light pulses without averaging. The vertical axis is normalized light intensity. The numbers in the bottom are the estimated decay times . Although the phosphorescence emission of some wells is very dim (i.e., the lower-right well), the result of the exponential fit is in good agreement with all other wells having identical conditions. The scan rate for an LCA using the DLMM is a function of the phosphorescence lifetime of the porphyrin sensor compound, the number of microwells to be observed, and the desired signal-to-noise ratio (i.e., the number of samples to be averaged). The high detection speed and detection parameter reconfigurability offered by the DLMM match well with the requirements of high content living cell screening.
Fig. 7. (a) Images of five wells with O2 sensing phosphorescent rings. (b) The measured phosphorescence decay curves of the O2 sensors and the inferred decay time .
7. Conclusions
We are developing a new digital light modulation microscope (DLMM) using a digital micromirror device (DMD) to control illumination patterns in spatial and temporal domains. The rapid switching capability of the DMD is utilized to generate high-speed light pulses to excite phosphorescence in sensor regions having arbitrary geometries. Oxygen concentration sensing using a microfabricated living cell array (LCA) has been demonstrated using this detection scheme. The rapid operation of DLMM is especially beneficial to high-content living cell viability screening where throughput is important. In addition to phosphorescence lifetime detection, temporal illumination modulation can also be used to implement a homodyne detection configuration using a lock-in amplifier to increase signal-to-noise ratio, to remove background interference, and to reduce photobleaching using lower dose of excitation light. The operation of the DLMM is fully programmable, making scanner configurations easily modified as LCA designs evolve. As such, a complete DLMM will provide a powerful and evolving tool for multiparameter scanning of large-scale living cell microwell arrays during long duration stimulus/response experiments that address questions central to the mission of the Microscale Life Sciences Center.
Acknowledgments
We gratefully acknowledge Dr. Joe Dragavon for the helpful discussions and the support of this research by the National Institutes of Health, National Human Genome Research Institute, Centers of Excellence in Genomic Science, grant 1 P50 HG002360 CEGS: Microscale Life Sciences Center.
References
1. W. Trettnak, W. Gruber, F. Reininger, and I. Klimant, “Recent progress in optical oxygen sensor instrumentation,” Sens. Actuators B , 1–3 (1995).
2. F. Reininger, C. Kolle, W. Gruber, and W. Trettnak, “Current progress in O2 measurement by luminescence quenching,” Med. Biol. Eng. Comput . 34, 113–114 (1996).
3. L. J. Hornbeck, “The DMD (TM) projection display chip: A MEMS-based technology,” Mrs. Bull . 26, 325–327 (2001). [CrossRef]
4. M. Liang, R. L. Stehr, and A. W. Krause, “Confocal microscope system that uses a binary spatial light modulator,” in Proceedings of the 1997 Conference on Lasers and Electro-Optics, CLEO, May 18–23 1997 (Optical Society of America, Washington DC, 1997), p. 154.
5. S. Cha, P. C. Lin, L. Zhu, E. L. Botvinick, P.-C. Sun, and Y. Fainman, “3D profilometry using a dynamically configurable confocal microscope,” Proceedings of SPIE of the Three-Dimensional Image Capture and Applications II, Jan 25-Jan 26 1999 3640, 246–253 (1999).
6. C. Sungdo, P. C. Lin, Z. Lijun, E. L. Botvinick, and S. Pang Chen, “3D profilometry using a dynamically configurable confocal microscope,” Proceedings of the SPIE The International Society for Optical, for-Optical (1999).
7. P. J. Verveer, Q. S. Hanley, P. W. Verbeek, L. J. Van Vliet, and T. M. Jovin, “Theory of confocal fluorescence imaging in the programmable array microscope (PAM),” J. Microsc. 189, 192–198 (1998). [CrossRef]
8. Q. S. Hanley, P. J. Verveer, M. J. Gemkow, D. Arndt Jovin, and T. M. Jovin, “An optical sectioning programmable array microscope implemented with a digital micromirror device,” J. Microsc . 196, 317–331 (1999). [CrossRef] [PubMed]
9. Q. S. Hanley, P. J. Verveer, and T. M. Jovin, “Optical sectioning fluorescence spectroscopy in a programmable array microscope,” Appl. Spectrosc . 52, 783–789 (1998). [CrossRef]
10. V. Bansal, S. Patel, and P. Saggau, “High-speed addressable confocal microscopy for functional imaging of cellular activity,” J. Biomed. Opt . 11, 34003–34011 (2006). [CrossRef] [PubMed]
11. T. Fukano and A. Miyawaki, “Whole-field fluorescence microscope with digital micromirror device: imaging of biological samples,” Appl. Opt . 42, 4119–4124 (2003). [CrossRef] [PubMed]
12. C. MacAulay and A. Dlugan, “Use of digital micro mirror devices in quantitative microscopy,” in Optical Investigations of Cells In Vitro and In Vivo, Jan 25–28 1998 (The International Society for Optical Engineering, San Jose, CA), 201–206.
13. P. M. Lane, A. L. P. Dlugan, R. Richards-Kortum, and C. E. MacAulay, “Fiber-optic confocal microscopy using a spatial light modulator,” Opt. Lett . 25, 1780–1782 (2000). [CrossRef]
14. S.-H. Chao, M. R. Holl, S. C. McQuaide, and D. R. Meldrum, “Oxygen concentration measurement with a phosphorescence lifetime based micro-sensor array using a digital light modulation microscope,” Proceedings of SPIE of Biomedical Optics and Imaging, 2006 (San Jose, CA), 6088, 60880S.
15. S. C. McQuaide, M. R. Holl, L. Burgess, T. Molter, J. Dragavon, A. C. Young, T. Strovas, J. Anderson, A. Jen, B. Karlsgodt, M. Lidstrom, and R. D. Meldrum, “A Living Cell Array (LCA) for Multiparameter Cell Metabolism Studies,” in IEEE/RAS-EMBS International Conference on Biomedical Robotics and Biomechatronics (Pisa, Italy, 2006), 971–976.
16. M. E. Lidstrom and D. R. Meldrum, “Life-on-a-chip,” Nat. Rev. Microbiol . 1, 158–164 (2003). [CrossRef]
17. D. B. Papkovsky, “Luminescent Porphyrins as Probes for Optical (Bio)Sensors,” Sens. Actuators, B-Chemical 11, 293–300 (1993). [CrossRef]
18. T. W. Molter, M. R. Holl, J. M. Dragavon, S. C. McQuaide, J. B. Anderson, A. C. Young, L. W. Burgess, M. E. Lidstrom, and D. R. Meldrum, “A new approach for measuring single cell oxygen consumption rates,” IEEE T. Automat. Sci. Eng. (to appear).
19. “DMD discovery kit brochure,” in Productivity systems Inc.(2002).
20. R. D. Shonat and A. C. Kight, “Oxygen tension imaging in the mouse retina,” Ann. Biomed. Eng . 31, 1084–1096 (2003). [PubMed]
21. G. Liebsch, I. Klimant, B. Frank, G. Holst, and O. S. Wolfbeis, “Luminescence lifetime imaging of oxygen, pH, and carbon dioxide,” Appl. Spectrosc . 54, 548–559 (2000). [CrossRef]
22. M. Shahidi, A. Shakoor, N. P. Blair, M. Mori, and R. D. Shonat, “A method for chorioretinal oxygen tension measurement,” Curr. Eye. Res . 31, 357–366 (2006). [CrossRef] [PubMed]
23. S.-H. Chao, T. T. H. Ren, S. A. Gales, M. R. Holl, S. C. McQuaide, and D. R. Meldrum, “Automated digital light modulation microscope (DLMM) for living cell array analysis: Pattern recognition and spatial alignment,” in IEEE/RAS-EMBS International Conference on Biomedical Robotics and Biomechatronics (Pisa, Italy, 2006), 977–981.
24. R. M. Haralick and L. G. Shapiro, “Computer and robot vision”. Reading, Mass.: Addison-Wesley Pub. Co, (1992).
25. G. Liebsch, I. Klimant, B. Frank, G. Holst, and O. S. Wolfbeis, “Luminescence lifetime imaging of oxygen, pH, and carbon dioxide,” Appl. Spectrosc . 54, 548–559 (2000). [CrossRef]
