Table of Contents Author Guidelines Submit a Manuscript
Journal of Spectroscopy
Volume 2016, Article ID 5436821, 9 pages
Research Article

Reagentless Bacterial Identification Using a Combination of Multiwavelength Transmission and Angular Scattering Spectroscopy

1Claro Scientific LLC, 10100 MLK Street North, St. Petersburg, FL 33716, USA
2University of South Florida, 4202 E. Fowler Avenue, Tampa, FL 33620, USA
3OneBlood Inc., 10100 MLK Street North, St. Petersburg, FL 33716, USA

Received 27 September 2015; Revised 21 January 2016; Accepted 9 February 2016

Academic Editor: Elizabeth A. Carter

Copyright © 2016 Debra E. Huffman et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


Optics based technologies are being advanced by many diagnostic companies around the globe. This resurgence is being driven by several factors including novel materials, enhanced computer power, nonlinear optics, and advances in algorithmic and statistical analysis. This study expands on a previous paper that evaluated the capability of a reagent-free optical profiling platform technology that used multiwavelength transmission spectroscopy to identify bacterial pathogens from pure culture. This study combines multiwavelength angular scattering with transmission based analysis into a single algorithm that will identify bacterial pathogens. Six predominant organisms, S. aureus, E. coli, K. pneumoniae and P. aeruginosa, E. faecalis, and coagulase negative Staphylococcus, were analyzed from a total of 753 clinical isolates received from three large community hospital systems. The bacterial identification method used for comparison in this study was the Vitek-2 (bioMerieux) which utilizes a biochemically based identification system. All of the clinical isolates received were blinded as to their identification until completion of the optical analysis. Sensitivities ranged from 87.7 to 94.6% with specificities ranging from 97.2 to 99.9% indicating that optical profiling is a powerful and exciting new technology that could be developed for the rapid identification of pathogens without the use of chemical reagents.

1. Introduction

Accurate and rapid bacterial identification is essential for correct disease diagnosis, treatment of infection, and tracing back of disease outbreaks associated with microbial infections. Traditionally, clinical methods for bacterial identification have relied on phenotypic identification of the causative organism using Gram staining, cultural growth characteristics, and biochemical reaction based techniques. These methods of bacterial identification, while in broad use today, suffer from a major drawback as certain strains exhibit nontraditional and unique biochemical characteristics that do not fit into the classical patterns that have been established to characterize a particular genus and species. Further, phenotypic properties can be unstable at times and expression can be dependent upon changes in environmental conditions, for example, growth substrate, temperature, and pH. Despite these limitations, early advances designed to produce more rapid microbial detection focused on the automation of these phenotypic identification methods [1].

Alternatives to the phenotypic methods for bacterial identification available today include a variety of spectroscopy and spectrometry based methodologies in various stages of investigation and development [214]. Many of these systems require less than 6 hours for specimen analysis, can analyze whole bacterial cells [6], are reproducible over a broad mass range, and are often specific enough to identify antibiotic-resistant bacteria [12, 13]. Despite this renewed interest in spectroscopic analysis, there exists a persistent focus on the use of either the scattering properties of the cells [9, 15, 16] or the absorption properties of the cellular constituents [3, 14]. Neither of these approaches in isolation has been successfully brought to market for bacterial identification. The combination of multiwavelength transmission and angular scattering measurements shown in this study is unique to the SpectraWave technology.

SpectraWave is being developed as a low-cost, optical profiling platform for the characterization of particulates which in this case are the bacterial pathogens of interest. At current stage of development, it integrates multiwavelength transmission and angular scattering measurements to ascertain the physical, structural, and compositional characteristics of bacterial cells. In this study, the performance of SpectraWave for bacterial identification was assessed using Vitek-2 (bioMerieux), a biochemically based bacterial identification system, as a reference system. The characteristics of a generic matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry system is also presented for comparison in Table 1. The techniques differ significantly in three main areas, namely, the assay format, a priori testing requirements, and the time to results (Table 1).

Table 1: Comparison of SpectraWave (Claro Scientific) and Vitek-2 (bioMerieux) bacterial identification techniques.

With regard to the assay format, SpectraWave is an optically based method which uses no chemical reagent for either sample preparation or analysis. All biochemically based systems currently on the market require the laboratory personnel to perform a Gram stain prior to initiating the bacterial identification procedure. This information is used to select the correct reagent card (Gram positive or Gram negative) that will be used for analysis in the Vitek-2. This step is not necessary for optical profiling as no a priori information is needed in the identification algorithm. Lastly, optical profiling can provide identification within minutes while results from the biochemically based systems are available in approximately 4–10 hours. Although manual sample preparation was used during this study, bacterial identification using SpectraWave was completed in less than 15 minutes. When fully automated, the time to identification will be less than 5 minutes. This compares very favorably to methods employed in the clinical laboratory today.

Optical profiling in its most basic sense encompasses the analysis of the interactions of light with matter. Previous studies have demonstrated that a rich body of quantitative information can be extracted from spectra when the combined contribution of the absorption and scattering properties are considered jointly as a function of wavelength [1720]. This technique enables the detection, identification, and characterization of a variety of cells, pathogens, and disease markers using their optical signatures [19, 20]. Individual modes of interaction (absorption, scattering, angular scattering, etc.) have their respective advantages, yet they are restricted by the amount of information that can be extracted from the measurements. When combined, they supply a unique set of independent, complimentary, and confirmatory data that can be used to identify, characterize, and quantify a sample with high levels of sensitivity and specificity. Combining the simplicity of the spectroscopic measurement and reagentless nature of the technique, optical profiling has a great potential to be developed for rapid, low-cost, and easy to operate diagnostic devices.

In this study SpectraWave was evaluated for the rapid and accurate identification of bacterial pathogens from pure culture. For this purpose, 753 clinical isolates of six bacterial pathogens (S. aureus, E. coli, K. pneumoniae and P. aeruginosa, E. faecalis, and coagulase negative Staphylococcus spp.) were received from three large local community hospitals. These six target species accounted for over 85% of the total number of organisms isolated from positive blood cultures in participating hospitals during the study period. All samples were analyzed in a blinded fashion.

2. Materials and Methods

2.1. Sample Preparation

Bacterial isolates from positive blood cultures that had been subcultured onto either MacConkey and/or blood agar were numbered sequentially and packaged at the participating community hospitals and shipped via courier to Claro Scientific on a twice weekly basis. Upon receipt, the samples were stored at room temperature until assayed. Each shipment contained a sealed envelope with the identification of each isolate as determined by the Vitek-2 (bioMerieux) by hospital laboratory personnel.

All culture plates were visually inspected prior to use for obvious contamination events. All plates that were deemed contaminated were excluded from the study. A single colony was selected from each acceptable culture plate and put in Erlenmeyer flasks containing 50 mL of sterile tryptic soy broth (BD Franklin Lakes, NJ). All samples were grown to stationary phase (19–24 hrs at 37°  ±  2°C) using standard techniques. Samples were prepared for spectroscopic analysis by collection of 1 mL aliquots from the stationary phase culture transferred to 1.5 mL sterile Eppendorf tubes. Samples were centrifuged for three minutes at 13,000 RPM (ThermoFisher Accuspin Microfuge 12). Tubes were removed from the Microfuge and the supernatant was slowly drawn off and discarded. The remaining pellets were resuspended in deionized water and vortexed until homogeneous distribution of cells in the suspension. The washing process (centrifugation, aspiration, and resuspension) was repeated three times until the suspensions were clear of the growth media.

2.2. Transmission Measurement

Appropriate aliquots of the final cell suspensions were diluted into 3 mL of deionized water in a 1 cm path length quartz cuvette. The degree of dilution for each sample was selected to yield maximal optical density values between 0.4 and 0.8 absorbance units (AU) in the wavelength range of 190–220 nm. The transmission measurements were collected using diode array spectrometers. The following diode array spectrometers were tested and found to be equivalent for pathogen identification: Agilent 8453 (Santa Clara, California), Ocean Optics HR4000, and Ocean Optics HR2000 (Dunedin, Florida). The data reported herein were collected with the Agilent 8453. Other commercial spectrophotometers can be used provided that the spectrophotometers have a minimum of 1 nm wavelength resolution, a high signal-to noise ratio (>99.9%), and an acceptance angle smaller than 2°. All transmission spectra were visually inspected and removed from the data set prior to any analysis if they exhibited aberrant absorption and/or scattering signal in the visible-IR wavelength region due to manual sample preparation errors, diffraction artifacts, and so forth.

2.3. Angular Scattering Measurements

For multiwavelength angular scattering measurements the sample concentrations were approximately twice what were used for the transmission measurements. Multiwavelength angular scattering measurements were collected at 80- and 90-degree observation angles over a wavelength range of 400–800 nm. The relative scattering intensity () was computed by scaling the difference between measured scattering intensities () of a sample () and background () at 80 degrees to those at 90 degrees at each wavelength as follows:The multiwavelength angular scattering measurements were collected using Ocean Optics HR4000 (Dunedin, Florida) spectrometer.

A typical sinusoid curve with two peaks in the wavelength region of interest (400–800 nm) is shown in Figure 1. The differences in the amplitudes () of the peaks () and the troughs () were used for the analysis. The following ratios of the amplitude differences were also used:

Figure 1: A typical AS spectrum has two major peaks and two major troughs.
2.4. Bacterial Identification Models

The approach for bacterial identification included transmission based identification models generated through statistical analysis based on principal component regressions combined with models based on the analysis of multiwavelength angular scattering data. The generation of transmission based bacterial identification models is described in detail in Smith et al. [20]. Since six organisms were investigated in this study, two new transmission based identification models, namely, E. faecalis and coagulase negative Staphylococcus spp. (CNS) models, were generated herein in addition to preexisting models of E. coli, K. pneumoniae, P. aeruginosa, and S. aureus described in [20]. For this purpose, training sets of spectra that maximally represented the spectral variations that can be exhibited by each organism due to natural variability related to its size, shape, and chemical composition were collected using repositories of known isolates. Of these training sets, 10 spectra were selected to create the identification models through solution of the generalized eigenvalue problem as described in [20]. The resulting sets of 10 eigenvectors gave the optimal representations of the underlying spectral features for the corresponding microorganisms and constituted their identification models. Table 2 summarizes the performance of the identification models with respect to the corresponding training sets. The outcome sensitivity values were >97% and specificity values were >99% for the first three target organisms (E. coli, K. pneumoniae, and P. aeruginosa) and therefore, their identification assignment relied only on transmission based models in this study.

Table 2: Results of the training set for the sensitivity and specificity for each of the target organisms based on transmission data only.

These results confirm the validity of the model for each organism. The outcome values for other three target organisms were somewhat lower. Table 2 shows that there were some incorrect assignments between S. aureus, E. faecalis, and the CNS models. Therefore, the multiwavelength angular scattering spectra were analyzed to aid in the differentiation between E. faecalis, S. aureus, and CNS.

The features of a multiwavelength angular scattering spectrum of a blind sample compared to the characteristics of known species are summarized in Table 3. If the spectral characteristics of a blind sample fit those of the species in Table 3, it was further evaluated with the scatter diagrams (Figures 2 and 3). If a blind sample fell into E. faecalis, S. aureus, or CNS regions on both scatter plots, it was identified as E. faecalis, S. aureus, or CNS with high confidence.

Table 3: The mean (2 SD) and range values of the characteristics of the multiwavelength angular scattering spectra of S. aureus, CNS, and E. faecalis training sets.
Figure 2: Scatter plot of Ratio 2 versus Ratio 1 characteristics of multiwavelength angular scattering spectra of target species training sets.
Figure 3: Scatter plot of the difference versus the characteristics of multiwavelength angular scattering spectra of target species training sets.

If a blind sample fell into overlap region on one of the scatter plots, it would be given low confidence identification result. Transmission identification assignment was equally weighted in the decision in this case. If a blind sample fell into overlap region on both scatter plots, the sample will be regrown and reanalyzed. However, if the repeats were still inconclusive, only transmission based identification assignment was used.

In cases of some CNS species, samples were frequently dominated by large cellular aggregates of variable sizes that resulted in diminished amplitudes of the major peaks and other additional features in the angular scattering spectra (Figure 4). Low confidence identification assignment was given to samples exhibiting such angular scattering spectra and the final identification assignment for such samples was given with transmission based models.

Figure 4: Representative multiwavelength angular scattering spectra of CNS (grey), S. haemolyticus (red), S. hominis (green), and S. warneri (blue).

3. Results and Discussion

The results of the analysis of the transmission spectra of 753 blind isolates with six identification transmission based models are presented in Table 4. Good sensitivity values (>90%) were achieved for E. coli, P. aeruginosa, and E. faecalis. Fair sensitivity values (>85%) were achieved for the other two species and CNS. Very good specificity values exceeding 97% were achieved for all target species. Yet, these values were somewhat lower than those obtained in the previous 204-isolate study of only four target organisms, E. coli, K. pneumoniae, P. aeruginosa, and S. aureus [20]. Obviously, the greater number of isolates was analyzed herein and, therefore, the greater variability in their spectral features could have been expected. For instance, only 16 isolates of P. aeruginosa were observed in the previous study [20] and all showed very narrow spectral variability and exceptionally distinct spectral features when compared to other target species. It was, in general, confirmed in this study; yet, of 62 isolates of P. aeruginosa, 6 were misassigned as E. coli or K. pneumoniae and, indeed, carried somewhat different spectral features than other P. aeruginosa isolates (Figure 5(c)).

Table 4: Sensitivity and specificity results for the identification assignment of clinical isolates using transmission models for six common bacterial pathogens.
Figure 5: Comparison of the transmission spectra of clinical isolates of target organisms: E. coli (a), K. pneumoniae (b), P. aeruginosa (c), S. aureus (d), CNS (e), and E. faecalis (f).

Further, cross-identification between E. coli and K. pneumoniae was noted previously [20] and was also observed in the results of this study: 14 isolates of E. coli were misassigned as K. pneumoniae and 9 isolates of K. pneumoniae were misassigned as E. coli (Table 4). Lower total number of isolates resulted in the lower sensitivity result for K. pneumoniae compared to E. coli. These two microorganisms are closely related and, therefore, have substantial similarities in their spectral features (Figures 5(a) and 5(b)). Nevertheless, the results of this study indicate that the transmission based models are quite successful in correctly assigning the vast majority of the clinical isolates of these closely related species.

The transmission based sensitivity and specificity values for S. aureus were also lower in this study due to cross-identification with other CNS and E. faecalis (Table 4). These two organisms were not included in the previous study. CNS was largely incorrectly assigned as S. aureus. Notably, E. faecalis had only few incorrectly assigned isolates: 1 as S. aureus and 3 as CNS (Table 4).

Combination of transmission results with angular scattering results improved the sensitivity values of the S. aureus assignment (Table 5). The number of incorrectly assigned S. aureus isolates was reduced by more than 50%. Consequently, the sensitivity value for S. aureus increased to 94.6%. However, the addition of the results of the angular scattering analysis did not affect the results for E. faecalis and made only a minor improvement for the CNS results (Table 5).

Table 5: Sensitivity and specificity results for the identification assignment of clinical isolates using combination of transmission and angular scattering identification models.

The far greater overlap in the ranges of the angular scattering parameters used for organism differentiation and identification for clinical isolates of E. faecalis, S. aureus, and CNS was noted compared to those of target species in the training sets (compare Tables 6 and 3). The most likely contributing factor to the misidentification of organism with multiwavelength angular scattering was greater sensitivity of the angular spectra to the larger sizes in the size distribution of the particles in a suspension. S. aureus and CNS species are known to form aggregates of various sizes and even though these aggregates might constitute only a few percent fraction of the total number of cells in suspension, they make significant contribution to the angular scattering spectral features. These effects can be seen as additional side peaks and troughs in the multiwavelength angular scattering spectra as many of those, for example, illustrated in Figure 4. The solutions to this problem are envisioned to be automated sample preparation that includes particle size separation step and/or a theoretical description of the angular scattering spectra that would allow for mathematically removing the aggregate effect.

Table 6: The mean (2 SD) and range values of the characteristics of the multiwavelength angular scattering spectra of clinical isolates of target organisms.

4. Conclusions

This paper presents a small focused study of a novel combination of multiwavelength scattering combined with the multiwavelength transmission measurements. The presentation of the data is complete and well described in comparison with the current method being used for bacterial identification in two large metropolitan hospitals in our area. The paper does not overextend the value of this technology or does it overstate the level of accuracy that has been shown thus far. This paper raises new investigative areas for exploration of the understanding of the combination of analysis that can be brought forth to increase its usefulness for bacterial characterization. This study validated the capability of SpectraWave, an optical profiling technology, to identify a menu of six clinical bacterial pathogens without the use of chemical reagents. The good to excellent sensitivity and specificity values obtained in this study support the relevance of the continued development and automation of the technology for rapid differentiation and identification of the pathogens causing the most frequent bloodstream infections in patients from local hospitals during the study period.

The organism with the highest sensitivity of detection within this study set was S. aureus with a sensitivity of 94.6%. This was followed in descending order by E. faecalis, E. coli, P. aeruginosa, K. pneumoniae, and CNS. The most numerous organism classified was E. coli (253 samples) representing 36% of the isolates considered for identification. The species with the fewest number of isolates examined was E. faecalis with 54 isolates (7% of the isolates). The errors in identification were seen between the most closely related organisms such as CNS and S. aureus as well as between E. coli and K. pneumoniae. Complementation of transmission measurements with angular scattering improved the performance of the system, in particular, for identification of S. aureus.

The demonstrated ability to identify most common bacterial isolates supports the fact that as the menu of pathogens analyzed through SpectraWave expands, this system can replace the need for several subjective manual laboratory tests currently in use including the Gram stain, catalase, oxidase, and coagulase tests as well as the use of specialized media for biochemical analysis.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.


Financial support for this study was provided by Claro Scientific LLC, St. Petersburg, Florida. Bacterial isolates were provided by BayCare Health Systems and Tampa General Hospital. Facilities and clinical sample transportation support was provided by OneBlood Inc.


  1. P.-E. Fournier, M. Drancourt, P. Colson, J.-M. Rolain, B. L. Scola, and D. Raoult, “Modern clinical microbiology: new challenges and solutions,” Nature Reviews Microbiology, vol. 11, no. 8, pp. 574–585, 2013. View at Publisher · View at Google Scholar · View at Scopus
  2. J. O. Lay, “MALDI-TOF mass spectrometry of bacteria,” Mass Spectrometry Reviews, vol. 20, no. 4, pp. 172–194, 2001. View at Publisher · View at Google Scholar · View at Scopus
  3. K. Maquelin, C. Kirschner, L.-P. Choo-Smith et al., “Identification of medically relevant microorganisms by vibrational spectroscopy,” Journal of Microbiological Methods, vol. 51, no. 3, pp. 255–271, 2002. View at Publisher · View at Google Scholar · View at Scopus
  4. D. J. Ecker, R. Sampath, C. Massire et al., “Innovation—Ibis T5000: a universal biosensor approach for microbiology,” Nature Reviews Microbiology, vol. 6, no. 7, pp. 553–558, 2008. View at Publisher · View at Google Scholar
  5. M. Nayak, A. Kotian, S. Marathe, and D. Chakravortty, “Detection of microorganisms using biosensors-A smarter way towards detection techniques,” Biosensors & Bioelectronics, vol. 25, no. 4, pp. 661–667, 2009. View at Publisher · View at Google Scholar · View at Scopus
  6. A. Ahmed, J. V. Rushworth, N. A. Hirst, and P. A. Millner, “Biosensors for whole-cell bacterial detection,” Clinical Microbiology Reviews, vol. 27, no. 3, pp. 631–646, 2014. View at Publisher · View at Google Scholar · View at Scopus
  7. C. J. Keys, D. J. Dare, H. Sutton et al., “Compilation of a MALDI-TOF mass spectral database for the rapid screening and characterisation of bacteria implicated in human infectious diseases,” Infection, Genetics and Evolution, vol. 4, no. 3, pp. 221–242, 2004. View at Publisher · View at Google Scholar · View at Scopus
  8. M. Venkatapathi, B. Rajwa, K. Ragheb et al., “High speed classification of individual bacterial cells using a model-based light scatter system and multivariate statistics,” Applied Optics, vol. 47, no. 5, pp. 678–686, 2008. View at Publisher · View at Google Scholar · View at Scopus
  9. A. Dunn and R. Richards-Kortum, “Three-dimensional computation of light scattering from cells,” IEEE Journal on Selected Topics in Quantum Electronics, vol. 2, no. 4, pp. 898–905, 1996. View at Publisher · View at Google Scholar · View at Scopus
  10. P. P. Banada, S. Guo, B. Bayraktar et al., “Optical forward-scattering for detection of Listeria monocytogenes and other Listeria species,” Biosensors and Bioelectronics, vol. 22, no. 8, pp. 1664–1671, 2007. View at Publisher · View at Google Scholar · View at Scopus
  11. M. Pignone, K. M. Greth, J. Cooper, D. Emerson, and J. Tang, “Identification of mycobacteria by matrix-assisted laser desorption ionization-time-of-flight mass spectrometry,” Journal of Clinical Microbiology, vol. 44, no. 6, pp. 1963–1970, 2006. View at Publisher · View at Google Scholar · View at Scopus
  12. S.-Y. Hsieh, C.-L. Tseng, Y.-S. Lee et al., “Highly efficient classification and identification of human pathogenic bacteria by MALDI-TOF MS,” Molecular and Cellular Proteomics, vol. 7, no. 2, pp. 448–456, 2008. View at Publisher · View at Google Scholar · View at Scopus
  13. V. Edwards-Jones, M. A. Claydon, D. J. Evason, J. Walker, A. J. Fox, and D. B. Gordon, “Rapid discrimination between methicillin-sensitive and methicillin-resistant Staphylococcus aureus by intact cell mass spectrometry,” Journal of Medical Microbiology, vol. 49, no. 3, pp. 295–300, 2000. View at Publisher · View at Google Scholar · View at Scopus
  14. U. Neugebauer, U. Schmid, K. Baumann et al., “Towards a detailed understanding of bacterial metabolism—spectroscopic characterization of Staphylococcus epidermidis,” Chemphyschem, vol. 8, no. 1, pp. 124–137, 2007. View at Google Scholar
  15. P. J. Wyatt, “Identification of bacteria by differential light scattering,” Nature, vol. 221, no. 5187, pp. 1257–1258, 1969. View at Publisher · View at Google Scholar · View at Scopus
  16. P. J. Wyatt and D. T. Phillips, “Structure of single bacteria from light scattering,” Journal of Theoretical Biology, vol. 37, no. 3, pp. 493–501, 1972. View at Publisher · View at Google Scholar · View at Scopus
  17. C. E. Alupoaei and L. H. García-Rubio, “Growth behavior of microorganisms using UV-Vis spectroscopy: Escherichia coli,” Biotechnology and Bioengineering, vol. 86, no. 2, pp. 163–167, 2004. View at Publisher · View at Google Scholar · View at Scopus
  18. C. E. Alupoaei and L. H. García-Rubio, “An interpretation model for the UV-VIS spectra of microorganisms,” Chemical Engineering Communications, vol. 192, no. 1-3, pp. 198–218, 2005. View at Publisher · View at Google Scholar · View at Scopus
  19. Y. M. Serebrennikova and L. H. Garcia-Rubio, “Modeling and interpretation of extinction spectra of oriented nonspherical composite particles: application to biological cells,” Applied Optics, vol. 49, no. 23, pp. 4460–4471, 2010. View at Publisher · View at Google Scholar · View at Scopus
  20. J. M. Smith, A. Roth, D. E. Huffman, Y. M. Serebrennikova, J. Lindon, and L. H. García-Rubio, “Multi-wavelength transmission spectroscopy revisited for micron and submicron particle characterization,” Applied Spectroscopy, vol. 66, no. 10, pp. 1186–1196, 2012. View at Publisher · View at Google Scholar · View at Scopus