Anal. Chem. 2007, 79, 2658-2673
Raman Chemical Imaging Spectroscopy Reagentless Detection and Identification of Pathogens: Signature Development and Evaluation Kathryn S. Kalasinsky,† Ted Hadfield,†,‡ April A. Shea,† Victor F. Kalasinsky,† Matthew P. Nelson,§ Jason Neiss,§ Amy J. Drauch,§ G. Steven Vanni,§ and Patrick J. Treado*,§
Division of Microbiology, Department of Environmental & Infectious Disease Sciences, Armed Forces Institute of Pathology, 6825 16th Street NW, Washington, D.C. 20306-6000, and ChemImage Corporation, 7301 Penn Avenue, Pittsburgh, Pennsylvania 15208
An optical detection method, Raman chemical imaging spectroscopy (RCIS), is reported, which combines Raman spectroscopy, fluorescence spectroscopy, and digital imaging. Using this method, trace levels of biothreat organisms are detected in the presence of complex environmental backgrounds without the use of amplification or enhancement techniques. RCIS is reliant upon the use of Raman signatures and automated recognition algorithms to perform species-level identification. The rationale and steps for constructing a pathogen Raman signature library are described, as well as the first reported Raman spectra from live, priority pathogens, including Bacillus anthracis, Yersinia pestis, Burkholderia mallei, Francisella tularensis, Brucella abortus, and ricin. Results from a government-managed blind trial evaluation of the signature library demonstrated excellent specificity under controlled laboratory conditions.
Detection and identification of biological threat organisms in 10 min or less within complex environmental backgrounds without requiring the use of costly reagents is a clear, but currently unmet, military and societal need. This paper presents the progress of ongoing studies that are evaluating Raman chemical imaging spectroscopy (RCIS)-based detection and identification of biological threat agents. RCIS has the potential to satisfy the need for rapid detection of threat agents, within complex matrixes and without the use of costly consumables or liquid reagents. In this paper, the rationale for RCIS as a phenomenology for pathogen detection and identification is presented. In addition, Raman signature development efforts are described, as well as requirements for exploitation in threat sensing applications and a detailed treatment of ongoing efforts to evaluate the performance and limitations of RCIS. Currently, there are a variety of microbiology, immunoassay, genetic, and molecular-based approaches for the identification of * To whom correspondence should be addressed. E-mail: treado@ chemimage.com. † Armed Forces Institute of Pathology. ‡ Current address: Midwest Research Institue, 1470 Treeland Blvd., Palm Bay, FL, 32909. § ChemImage Corp.
2658 Analytical Chemistry, Vol. 79, No. 7, April 1, 2007
biological threat organisms.1-6 However, traditional methods are limited by long analysis times associated with culture sample collection and preparation, DNA extraction, and reliance on reagents with limited shelf lives. These limitations complicate autonomous operation and require traditional methods to have costly logistics trails, which increases overall cost of ownership. Reagentless optical detection strategies including measurement of particle size and intrinsic fluorescence7-10 and laser-induced breakdown spectroscopy (LIBS)11-13 are potential trigger sensors capable of differentiating biological from nonbiological particles, but they have not shown capability as robust, identification methods. In contrast, mass spectrometry-based sensing strategies are showing potential for rapid detection and classification of biological threat organisms.14-19 Unfortunately, mass spectrometrybased detection methods are susceptible to false alarms due to (1) Higgins, J. A.; Ibrahim, M. S.; Knauert, F. K.; Ludwig, G. V.; Kijek, T. M.; Ezzell, J. W.; Courtney, B. C.; Henchal, E. A. Ann. N.Y. Acad. Sci. 1999, 894, 130-148. (2) Fatah, A. A.; Barrett, J. A.; Arcilesi, R. D., Jr.; Ewing, K. J.; Lattin, C. H.; Moshier, T. F. An Introduction to Biological Agent Detection Equipment for Emergency First Responders; NIJ Guide 101-00, 2001; pp 29-32. (3) Vitko, J., Jr.; Franz, D. R.; Alper, M.; Biggins, P. D.E.; Brandt, L. D.; Bruckner-Lea, C.; Burge, H. A.; Ediger, R.; Hollis, M. A.; Laughlin, L. L.; Mariella, R. P., Jr.; McFarland, A. R.; Schaudies, R. P. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases; The National Academies Press: Washington, DC, 2005. (4) Hindson, B. J.; McBride, M. T.; Makarewicz, A. J.; Henderer, B. D.; Setlur, U. S.; Smith, S. M.; Gutierrez, D. M.; Metz, T. R.; Nasarabadi, S. L.; Venkateswaran, K. S.; Farrow, S. W.; Colston, B. W., Jr.; Dzenitis, J. M. Anal. Chem. 2005, 77, 284-289. (5) Stratis-Cullum, D. N.; Griffin, G. D.; Mobley, J.; Vass, A. A.; Vo-Dinh, T. Anal. Chem. 2003, 75, 275-280. (6) Karasinski, J.; Andreescu, S.; Sadik, O. A. Anal. Chem. 2005, 77, 79417949. (7) Ho, J. Anal. Chim. Acta 2002, 457, 125-148. (8) Brosseau, L. M.; Vesley, D.; Rice, N.; Goodell, K.; Nellis, M.; Hairston, P. Aerosol Sci. Technol. 2000, 32, 545-558. (9) Pan, Y. L.; Hartings, J.; Pinnick, R. G.; Hill, S. C.; Halverson, J.; Chang, R. K. Aerosol Sci. Technol. 2003, 37, 628-639. (10) Sivaprakasam, V.; Huston, A.; Scotto, C.; Eversole, J. Opt. Express 2004, 12, 4457-4466. (11) Hybl, J. D.; Lithgow, G. A.; Buckley, S. G. Appl. Spectrosc. 2003, 57, 12071215. (12) Dixon, P. B.; Hahn, D. W. Anal. Chem. 2005, 77, 631-638. (13) Munson, C. A.; DeLucia, F. C., Jr.; Piehler, T.; McNesby, K. L.; Miziolek, A. W. Spectrochim. Acta, B 2005, 60, 1217-1224. (14) Snyder, A. P.; Maswadeh, W. M.; Parsons, J. A.; Tripathi, A.; Meuzelaar, H. L.C.; Dworzanski, J. P.; Kim, M. G. Field Anal. Chem., Technol. 1999, 3, 315-326. 10.1021/ac0700575 CCC: $37.00
© 2007 American Chemical Society Published on Web 03/06/2007
background clutter and lack mass spectrum signature robustness in the presence of biological variability, primarily due to mixed signature complexities. These limitations exist, in large part, due to the extremely high sensitivity of mass-based detection. Raman spectroscopy-based detection can address a number of limitations of more conventional biological threat organism sensing strategies and may become an optimal threat detection and identification system technology. An ideal sensor would be capable of detecting and identifying multiple pathogens in real or near-real time. The sensor would operate in an autonomous, unattended fashion for an extended period of time, would not require frequent maintenance, and would have a low overall life cycle cost. The sensor would have sufficient sensitivity to detect exposure to life-threatening pathogens and have low false alarm rates. Raman spectroscopy has an extensive history as a materials identification tool for the evaluation of a wide variety of analytes, including foods, semiconductors, industrial chemicals, pharmaceuticals, drugs of abuse, radiological hazards, explosives, chemical warfare agents, and aerosolized particulate matter.20-26 In addition, a growing body of evidence shows that infrared27,28 and Raman spectroscopy have utility as tools for biological organism detection, classification, and identification. A variety of Raman approaches for pathogen detection have been reported, including use of normal Raman,29-39 Raman imaging,40-43 UV resonance Raman,44-49 surface-enhanced Raman,50-55 and nonlinear Raman.56 (15) Victor, R.; Hathout, Y.; Fenselau, C. Appl. Environ. Microbiol. 2000, 66, 3828-3834. (16) Beverly, M. B.; Voorhees, K. J.; Hadfield, T. L.; Cody, R. B. Anal. Chem. 2000, 72, 2428-2432. (17) Fergenson, D. P.; Piteskey, M. E.; Tobias, H. J.; Steele, P. T.; Czerwieniec, G. A.; Russell, S. C.; Lebrilla, C. B.; Horn, J. M.; Coffee, K. R.; Srivatava, A.; Pillai, S. P.; Shih, M. P.; Hall, H. L.; Ramponi, A. J.; Chang, J. T.; Langlois, R. G.; Estacio, P. L.; Hadley, R. T.; Frank, M.; Gard, E. E. Anal. Chem. 2004, 76, 373-378. (18) Steele, P. T.; Srivastava, A.; Pitesky, M. E.; Fergenson, D. P.; Tobias, H. J.; Gard, E. E.; Frank, M. Anal. Chem. 2005, 77, 7448-7454. (19) Demirev, P. A.; Feldman, A. B.; Kowalski, P.; Lin, J. S. Anal. Chem. 2005, 77, 7455-7461. (20) Chalmers, J. M., Griffiths, P. R., Eds. Handbook of Vibrational Spectroscopy; Wiley: New York, 2002. (21) Schaeberle, M. D.; Tuschel, D. D.; Treado, P. J. Appl. Spectrosc. 2000, 54, 257-266. (22) Schoonover, J. R.; Saab, A.; Bridgewater, J. S.; Havrilla, G. J.; Zugates, C. T.; Treado, P. J. Appl. Spectrosc. 2000, 54, 1362-1371. (23) Doub, W. H.; Adams, W. P.; Spencer, J. A.; Buhse, L. F.; Nelson, M. P.; Treado, P. J. Pharm. Res., in press. (24) Nelson, M. P.; Zugates, C. T.; Treado, P. J.; Casuccio, G. S.; Exline, D. L.; Schlaegle, S. F. Aerosol Sci. Technol. 2001, 34, 108-117. (25) Christesen, S. D. Appl. Spectrosc. 1988, 42, 318-321. (26) Fell, N. F., Jr.; Vanderhoff, J. A.; Pesce-Rodriguez, R. A.; McNesby, K. L. J.Raman Spectrosc. 1998, 29, 165-172. (27) Naumann, D. In Encyclopedia of Analytical Chemistry; Meyers, R. A., Ed.; Wiley: Chichester, 2000; pp 102-131. (28) Naumann, D.; Keller, S.; Helm, D.; Schultz, C.; Schrader, B. J. Mol. Struct. 1995, 347, 399-405. (29) Goodacre, R.; Timmins, E. M.; Burton, R.; Kaderbhai, N.; Woodward, A. M.; Kell, D. B.; Rooney, P. J. Microbiology 1998, 144, 1157-1170. (30) Carmona, P. Spectrochim. Acta, Ser. A 1980, 36, 705-712. (31) Carey, P. R.; Fast, P.; Kaplan, H.; Pozsgay, M. Biochim. Biophys. Acta 1986, 872, 169-176. (32) Choo-Smith, L. P.; Maquelin, K.; Endtz, H. P.; Bruining, H. A.; Puppels, G. J. Spectrosc. Biol. Mol. New Dir. 1999, 8, 537-540. (33) Choo-Smith, L. P., Maquelin, K.; Van Vreeswijk, T.; Bruining, H. A.; Puppels, G. J.; Thi, N. A. N.; Kirschner, C.; Naumann, D.; Ami, D.; Villa, A. M.; Orsini, F.; Doglia, S. M.; Lamfarraj, H.; Sockalingum, G. D.; Manfait, M.; Allouch, P.; Endtz, H. P. Appl. Environ. Microbiol. 2001, 67, 1461-1469.
The basis for Raman spectroscopy being capable of discriminating closely related species of bacteria is twofold: (1) Raman is sensitive to the molecular components within cells; and (2) variations in genome organization between organisms, as well as extrachromosomally encoded phenotypes, such as the Bacillus anthracis toxins, polyglutamic acid capsules, and Bacillus thuringiensis toxins, lead to variations in patterns of gene expression, protein synthesis, and metabolite accumulation that are anticipated to contribute to species-specific compositional signatures.57 In operational settings, species-level detection and identification has the potential to be automated by comparing a measured Raman spectrum (target) against a Raman signature library. Therefore, a comprehensive, reproducible, and robust Raman spectral library is a key enabling technology for a successful Raman-based detection and identification system. At present, there is an incomplete understanding of the taxonomic resolution achievable by Raman-based identification methods, especially in (34) Kirschner, C., Maquelin, K.; Pina, P.; Thi, N. A. N.; Choo-Smith, L. P.; Sockalingum, G. D.; Sandt, C.; Ami, D.; Orsini, F.; Doglia, S. M.; Allouch, P.; Mainfait, M.; Puppels, G. J.; Naumann, D. J. Clin. Microbiol. 2001, 39, 1763-1770. (35) Maquelin, K.; Kirschner, C.; Choo-Smith, L. P.; van den Braak, N.; Endtz, H. P.; Naumann, D.; Puppels, G. J. J. Microbiol. Methods 2002, 51, 255271. (36) Esposito, A. P.; Talley, C. E.; Huser, T.; Hollars, C. W.; Schaldach, C. M.; Lane, S. M. Appl. Spectrosc. 2003, 57, 868-871. (37) Huang, W. E.; Griffiths, R. I.; Thompson, I. P.; Bailey, M. J.; Whiteley, A. S. Anal. Chem. 2004, 76, 4452-4458. (38) Schuster, K. C.; Reese, I.; Urlaub, E.; Gapes, J. R.; Lendl, B. Anal. Chem. 2000, 72, 5529-5534. (39) Gardner, C. W., Jr.; Maier, J. S.; Nelson, M. P.; Schweitzer, R. C.; Treado, P. J.; Vanni, G. S.; Wolfe, J. U.S. Patent 6,765,668, 2004. (40) Tripathi, A.; Jabbour, R. E.; Treado, P. J.; Neiss, J. H.; Nelson, M. P.; Jensen, J. L.; Snyder, A. P. Appl. Spectrosc., submitted. (41) Wang, X.; Voigt, T. C.; Bos, P. J.; Nelson, M. P.; Treado, P. J. Appl. Opt., submitted. (42) Escoriza, M. F.; VanBriesen, J. M.; Stewart, S.; Maier, J. S.; Treado, P. J. J. Microbiol. Methods 2006, 66, 63-72. (43) Rosch, P.; Harz, M.; Klaus-Dieter, P.; Ronneberger, O.; Burkhardt, H.; Schule, A.; Schmauz, G.; Lankers, M.; Hofer, S.; Thiele, H.; Motzkus, H.; Popp, J. Anal. Chem. 2006, 78, 2163-2170. (44) Chadha, S.; Manoharan, R.; Moenne-Loccoz, P.; Nelson, W. H.; Peticolas, W. L.; Sperry, J. F. Appl. Spectrosc. 1993, 47, 38-43. (45) Dalterio, R. A.; Nelson, W. H.; Britt, D.; Sperry, J. F. Appl. Spectrosc. 1987, 41, 417-422. (46) Ghiamati, E.; Manoharan, R.; Nelson, W. H.; Sperry, J. F. Appl. Spectrosc. 1992, 46, 357-364. (47) Lopez-Diez, E. C.; Goodacre, R. Anal. Chem. 2004, 76, 585-591. (48) Manoharan, R.; Ghiamati, E.; Chadha, S.; Nelson, W. H.; Sperry, J. F. Appl. Spectrosc. 1993, 47, 2145-2150. (49) Nelson, W. H.; Dasari, R.; Feld, M.; Sperry, J. F. Appl. Spectrosc. 2004, 58, 1408-12. (50) Grow, A. E.; Wood, L. L.; Claycomb, J. L.; Thompson, P. A. J. Microbiol. Methods 2003, 53, 221-233. (51) Alexander, T. A.; Pellegrino, P. M.; Gillespie, J. B. Proc. SPIE 2003, 5085, 91-100. (52) Jarvis, R. M.; Goodacre, R. Anal. Chem. 2004, 76, 40-47. (53) Farquharson, S.1; Gift, A. D.; Maksymiuk, P.; Inscore, F. E. Appl. Spectrosc. 2004, 58, 351-354. (54) Zhang, X.; Young, M. A.; Lyandres, O.; Van Duyne, R. P. J. Am. Chem. Soc. 2005, 127, 4484-4489. (55) Cao, Y. C.; Jin, R.; Mirkin, C. A. Science 2002, 297, 1536-1540. (56) Scully, M. O.; Kattawar, G. W.; Lucht, R. P.; Opatrny, T.; Pilloff, H.; Rebane, A.; Sokolov, A. V.; Zubairy, M. S. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 10994-11001. (57) Hoffmaster, A. R.; Ravel, J.; Rasko, D. A.; Chapman, G. D.; Chute, M. D.; Marston, C. K.; De, B. K.; Sacchi, C. T.; Fitzgerald, C.; Mayer, L. W.; Maiden, M. C. J.; Priest, F. G.; Barker, M.; Jiang, L.; Cer, R. Z.; Rilstone, J.; Peterson, S. N.; Weyant, R. S.; Galloway, D. R.; Read, T. D.; Popovic, T.; Fraser, C. M. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 8449-8454.
Analytical Chemistry, Vol. 79, No. 7, April 1, 2007
2659
the presence of signature variability arising from variable growth and sample handling conditions in the presence of clutter. Applications where Raman detection of pathogens may be appropriate include the following: surface contamination; bioaerosols; waterborne pathogens; foodborne pathogens, and in clinical samples. In each of these application examples, background interferents, which can be highly variable in composition and concentration, are present in sufficiently high abundance to complicate the detection of pathogens. In the event of an intentional release of pathogens, threat organisms will tend to predominate, but susceptibility to ambient, variable background is a significant challenge to application of reagentless sensing strategies, including Raman-based sensors. For detection of pathogens in complex, environmental samples, RCIS as described in this paper has the following advantages: (1) detectivity at the single organism level through use of high numerical aperture illumination and light gathering optics; (2) reliance on normal Raman scattering (i.e., without resonance or surface enhancement), which provides a linear regime of operation, more reproducible signatures, and more robust bioagent recognition; (3) monolayer sample deposition on a Raman-inactive immobilizing surface, typically an optically flat metal substrate, which dissipates laser-induced heat and minimizes sample photodegradation; (4) use of widefield illumination, which provides an efficient means to photobleach a large sample population simultaneously without having to scan the laser beam, even in the presence of clutter; (5) visible laser excitation, while not intuitive due to ν4 considerations (UV would be preferable) or due to fluorescence considerations (near-IR would be preferable), provides the best demonstrated compromise between maximizing scattering efficiency (tradeoff relative to the NIR), minimizing laser-induced photodegradation (tradeoff relative to the UV) at an acceptable level of intrinsic fluorescence background. Despite the use of visible laser excitation, fluorescence can be minimized through use of photobleaching and imaging as background clutter is typically the source of interfering fluorescence; (6) Raman imaging provides efficient means to generate “pure pixel” Raman spectra in the presence of background clutter (a means to increase signal to background ratio), which enables more consistent correlation with end member spectra resident in the signature library; (7) imaging provides a method to increase sample throughput and manage relative insensitivity of Raman method; and (8) sampling (“imaging”) from multiple bioagents provides improved overall sampling statistics, which enables use of probabilistic decision-making algorithms and exponential improvement in false alarm performance. Due to these underlying advantages, to the best of the authors’ knowledge, the authors were the first to successfully collect Raman spectra and Raman chemical images from trace levels of live pathogens using visible wavelength laser excitation. This article provides representative results collected over the past several years in a series of Raman studies that have been performed without the use of enhancement (resonance, SERS, nonlinear) methods. MATERIALS AND METHODS Raman Signature Development at the Armed Forces Institute of Pathology. Autonomous Raman-based pathogen identification is reliant upon the use of a Raman signature library 2660 Analytical Chemistry, Vol. 79, No. 7, April 1, 2007
Table 1. Signature Library threats CDC Category A and B biothreats anthrax (B. anthracis) botulism (Clostridium botulinum toxin) plague (Y. Pestis) tularemia (F. tularensis) salmonella Escherichia coli O157:H7 cholera (Vibrio cholerae) Cryptosporidium parvum ricin toxin from Ricinus commmunis brucellosis (B. abortus, B. melitensis, B. suis) glanders (Burkholderia mallei) melioidosis (Burkholderia pseudomallei) staphylococcal enterotoxin B chemical threat chemical warfare agents and simulants TICs TIMs explosives radiological hazards forensics environmental interferents food pharmaceutical clinical/medical polymer consumer products semiconductor and electronics laboratory products
A
B
library entries 281
85 4 5 38 5 2 1 1 6 17 5 9 2 146 13 75 5 8 13 32 310 53 133 64 32 44 16 25
measured under controlled conditions from pathogen reference materials. The Raman signature library must contain priority pathogens, near neighbors to the pathogens, and clutter specific to defined detection applications (i.e., surface contamination, bioaerosols, waterborne pathogens, foodborne pathogens, or clinical diagnostics). The signatures must be reproducible and exhibit minimal variability induced by sensor uncertainty, biological growth conditions, and sample preparation or dissemination conditions. A comprehensive pathogen signature library has been constructed using RCIS technology, and as of December 2006, the Raman library has over 1000 end member signatures. Table 1 provides an overview of the pathogen Raman signature library entries. Like mass spectrometry and LIBS, Raman is a pathogen detection/identification modality that can also detect and identify chemical, radiological, and explosive hazards. Therefore, Raman signatures of a variety of hazardous agents have been captured, but the focus of this paper is on biological agents. Keys to the development of a valid library include the following: access to extensive, pathogen collections that are microbiologically well characterized, free of contaminating strains, and prepared with well-defined growth materials and conditions; access to live agents, near neighbors, and real-world environmental threat samples; performing periodic review and enhancements of the signature library to address gaps; archiving reference copies of the organisms and materials used to create the library in order to allow future independent and internal validation; and conducting blind challenges of the robustness of the signature library, where the blind trials are managed independently of the developers to minimize bias.
Library End-Member Selection. (a) Pathogens and Near Neighbors. Due to the heightened interest in detection of B. anthracis (Ba) at the time this research effort was initiated, biological organisms relevant to Ba detection, including evaluation of 12 Ba strains, as well as genetic near neighbors to Ba (other Bacillus species, including Bacillus cereus (Bc), Bacillus thurengiensis (Bt), Bacillus globigii (Bg)), and other Gram-positive organisms (such as Streptococcus agalactiae and Streptococcus pyogenes), have been a historical focus area. Another focus area driving the selection of pathogen Raman signatures was evaluation of CDC Category A and Category B pathogens,58 along with their respective genetic and morphological near neighbors. While not all Category A pathogens were assessed, including viral pathogens requiring access to a BSL 4 laboratory, a majority of high-priority pathogens were analyzed and have entries in the library. To date, Category C agents have not been a major focus area. (b) Interferent Materials. In addition to pathogen and nearneighbor biological signatures, a variety of biological and nonbiological background interferent materials have been included in the Raman Signature Library. These interferent materials include common white powders, pharmaceutical drugs, excipients, explosives, drugs of abuse, indoor and outdoor environmental contaminants, industrial solvents, toxic industrial chemicals (TICs), and toxic industrial materials (TIMs). Selection of interferent materials was driven by the operational experience of personnel at AFIP, who serve in a forensic pathology role as a member of the Laboratory Rapid Response Network. Since 2001, AFIP has processed hundreds of suspected pathogen samples. The majority of these samples have been hoax samples. However, the materials employed in these cases provide valuable insight into candidate interferent materials. Sample Preparation. Most strains were obtained from the American Type Culture Collection (ATCC, http://www.atcc.org) while others came from established government collections. Most samples were stored at -70 °C; however, storage conditions are species/strain specific and differ according to standard operating procedure. Additional details on specific growth conditions are provided as Supporting Information. Samples were transferred to aluminum-coated microscope slides for analysis. The Al-coated slides consist of a standard 1 in. by 3 in. glass microscope slide with vacuum-deposited Al coating on one side. The slides exhibit good optical properties enabling high-quality optical image detection. The Al-coated slides exhibit low visible fluorescence background when excited with UV light suitable for intrinsic fluorescence imaging. In addition, the slides exhibit low Raman background when excited with laser light suitable for Raman detection. Raman Chemical Imaging Spectroscopy. RCIS technology has been described in detail.59,60 Briefly, RCIS combines digital imaging with one or more modes of molecular spectroscopy to provide molecular images detailing material morphology, composition, and structure. A diagram of a multimodal RCIS instru(58) http://www.bt.cdc.gov/agent/agentlist-category.asp. (59) Morris, H. R.; Hoyt, C. C.; Miller, P.; Treado, P. J. Appl. Spectrosc. 1996, 50, 805-811. (60) Treado, P. J.; Nelson, M. P. In Handbook of Vibrational Spectroscopy; Chalmers, J. M., Griffiths, P. R. Eds.; Wiley: New York, 2002; pp 14291459.
ment is shown in Figure 1. The sensor architecture combines multiple independent detection strategies, including digital optical imaging, intrinsic fluorescence chemical (i.e., hyperspectral) imaging, dispersive Raman spectroscopy, and Raman chemical imaging spectroscopy. Devices with this architecture have been demonstrated in laboratory (ChemImage Corp., Falcon) and field transportable (ChemImage Corp., Eagle) configurations. When applied to pathogen detection, each of the RCIS detection modalities has critical functions. Specifically, optical imaging enables rapid, sensitive, low-specificity discrimination of pathogens from background particulate matter. Optical imaging detection modalities include bright-field reflectance imaging (appropriate for particulate on opaque, Al-coated substrates), polarized light microscopy (PLM), and differential interference contrast (DIC) microscopy. Optical imaging is an effective strategy for identifying the boundaries of target particles. This becomes critical information in subsequent fluorescence and Raman imaging analysis. If a signature arises from a region where particulate matter does not exhibit characteristic pathogen morphology, it can be rejected, thus reducing the probability of false alarm (Pfa). Next, fluorescence chemical imaging enables rapid screening of large surface areas for biological versus nonbiological particulate discrimination based on the tendency of biological materials to exhibit autofluorescence under UV excitation. Nonbiological material is not evaluated in subsequent Raman interrogation. Fluorescence imaging is performed using the UV-filtered Hg arc lamp emitting at 365 nm to induce pathogen autofluorescence emission in the visible spectral range (400-720 nm). Pathogen fluorescence is primarily attributed to flavin chemistry.9 A liquid crystal tunable filter (LCTF) and a thermoelectric (TE)-cooled CCD detector were used for fluorescence imaging. Alternatively, total integrated fluorescence may be imaged using the Hg arc lamp for excitation, an excitation filter, dichroic beam splitter and barrier filter combination for discrimination of sample fluorescence, and a video camera for detection. While the total integrated fluorescence detection mode provides less specificity than hyperspectral fluorescence imaging, it can operate in real time. Presumptive pathogen particles exhibiting properties consistent with pathogen morphology and fluorescence signatures are analyzed using RCIS. A solid-state, frequency-doubled, diodepumped, 532-nm Nd:YVO4 laser is employed as the laser source for Raman measurements. Typical laser power and laser power density at the sample were 13 mW and 2.96 × 103 W/cm2, respectively. Laser illumination conditions were selected to prevent laser-induced sample degradation. In RCIS, two independent Raman detection channels are employed to provide Raman imaging and spectroscopy. Namely, a LCTF imaging spectrometer is employed for high-definition Raman imaging in which Raman spectra are collected at each image pixel. In addition, a dispersive Raman spectrometer generates a single, integrated Raman spectrum from the laser focal volume interacting with the sample. Collecting multiple Raman detection channels provides the following: (1) the Raman imaging channel is efficient in collecting high-definition spatial information, but one wavelength at a time; (2) the dispersive Raman spectroscopy channel is efficient in collecting spectral information, but one pixel at a time. Through the combination of both techniques, high spatial, high spectral Raman image collection can be generated at diffraction-limited Analytical Chemistry, Vol. 79, No. 7, April 1, 2007
2661
Figure 1. RCIS instrument schematic. Mercury arc lamp illumination for visible bright-field reflectance (Vis) and ultraviolet excitation fluorescence (UV Fl.) microscopy. In Vis microscopy mode, a 50/50 beam splitter (BS1) is employed, and the resulting reflectance image can be viewed through crossed polarizers and a DIC attachment if desired. In UV Fl. mode, an excitation (Exc.) bandpass filter in combination with a dichroic beam splitter (BS2) and emission (Em.) bandpass filter is employed. For bright-field and fluorescence microscopy, mirror 1 (M1) is removed from the optical path and the image can be reflected by mirror 2 (M2) onto a video charge-coupled device (CCD) imaging detector to perform video microscopy or directed to a fluorescence liquid crystal tunable filter (Fl. LCTF) to perform hyperspectral fluorescence imaging. For Raman analysis, laser light is directed by mirror 3 (M3) to a dichroic beam splitter (BS3) prior to impinging on M1. Raman scatter is transmitted through BS3 to a laser rejection filter (LRF) before impinging on a polarization beam splitter (PBS), which directs a linear polarization to a Raman LCTF and the orthogonal linear polarization to a fiber array spectral translator (FAST), which couples the spatially resolved Raman scatter to a dispersive imaging spectrometer.
spatial resolution (
