Applications of Raman Spectroscopy to Virology and Microbial Analysis - Emerging Raman Applications and Techniques in Biomedical and Pharmaceutical Fields - page 445

Table 19.1. Average recognition rate for single bacterial cell identification by means
of a support vector machine

Recognition rate

Number of strains Number of spectra for species (%)

B. pumilus 7 594 93.2

B. megaterium 1 94 87.3

B. sphaericus 3 275 77.0

B. subtilis 7 924 91.6

E. coli 7 641 98.4

M. luteus

7 1259 93.0

M. lylae

2 186 94.1

Staph. cohnii 6 665 96.9

Staph. epidermidis 8 1885 89.0

Staph. warneri 2 245 99.2

50 6768

92.0

B. cereus is an ubiquitous soil bacterium and an opportunistic pathogen which
commonly causes food poisoning [65].
Calcium dipicolinate (CaDPA), which is the calcium chelate of dipicolinic
acid (2,6-pyridinedicarboxylic acid), is a major chemical component in the en-
dospores, accounting for 5-15% of its molecular weight. The Raman scattering
of this mineral, found in the core and cortex of endospores, dominates their
Raman spectra (s. endospores of B. pumilus , Fig 19.2). Nevertheless, the
Raman spectra of endospores exhibit sucient species-related spectral char-
acteristics to allow for an investigation of spore-to-spore variability and dif-
ferences in spores attributable to environmental variation. For example De
Gelder et al. monitored the sporulation of single B. licheniformis endospores
and correlated it to the CaDPA content [66]. Also a classification of endospores
of five different species has been achieved according to inherent strain differ-
ences and culturing conditions.
Beside the identification of single bacterial cell or spores by means of
Raman spectroscopy, the localization of these cells inside partly complex
matrices has to be performed. One approach is the combination of Raman
spectroscopy, fluorescence spectroscopy, and digital imaging techniques. This
method was applied to detect traces of endospores and other biothreat organ-
isms even in the presence of complex environmental matrices like bioaeroso-
lic background, nasal mucin [67], or tap water [68]. Another fully auto-
mated device was built to analyze bioaerosols in clean room environments,
where prior to the Raman identification method a particle preselection took
place [69].
The differentiation in abiotic and biotic particles before the actual Raman
measurement allows a higher sample throughput. Here, especially morpho-
logical features can be evaluated by bright light or glancing light illumina-