FTIR spectra of poultry meat specific bacteria viz.
Salmonella enteritidis, Pseudomonas ludensis, Listeria
monocytogenes and Escherichia coli were collected and investigated
for identification of spectral windows capable of
bacterial classification and quantification. Two separate
datasets obtained at different times were used in the study to
check reproducibility of results. Multivariate data analysis
techniques viz. principal component analysis (PCA), partial
least-squares discriminant analysis (PLSDA) and soft independent
modelling of class analogy (SIMCA) were used in the
analysis. Using full cross-validation and separate calibration
and prediction datasets, the highest correct classification results
for SIMCA and PLSDA were achieved in spectral window
(1800-1200 cm-1) for both datasets. The window was
also tested then for quantification of different bacteria and it
had been observed that PLS models had better R values for
classification (R=0.984) than predicting various concentration
levels (R=0.939) of all four poultry specific bacteria inoculated
in distilled water. The identified spectral window 1800–
1200 cm-1 also demonstrated potential for 100% correct
classification of chicken salami samples contaminated with
S. enteritidis and P. ludensis from control using SIMCA.
However, this wavenumber range yielded few misclassifications
using PLS-DA approach. Thus FTIR spectroscopy in
combination with chemometrics is a powerful technique that
can be developed further to differentiate bacteria directly on
poultry meat surface.