Identification of <i>Listeria</i> Species Using a Low-Cost Surface-Enhanced Raman Scattering System With Wavelet-Based Signal Processing

Abstract

We investigated the ability to distinguish between six species within the Listeria genus (including the human pathogen Listeria monocytogenes ) based on a bacteria sample's surface-enhanced Raman scattering (SERS) spectrum. Our measurement system consists of a portable low-cost Raman spectral acquisition unit and associated signal processing and classification modules. First, <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Listeria</i> was cultured and then adsorbed onto silver colloidal nanoparticles for SERS measurements. A total of 483 SERS spectra were collected and preprocessed (using a stationary wavelet transform decomposition) to remove noise and baseline artifact. Distinguishing features were extracted by retaining detail wavelet coefficients of significant value across multiple scales. Using a linear classifier in association with ldquoleave one outrdquo cross-validation, the system achieved maximum classification accuracies of 96.1% (six-category) and 97.9% (two-category, <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">L.</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">monocytogenes</i> versus all others). Dimensionality reduction was used to decrease the number of features from 74 to 5 while maintaining similar classification accuracy. In the future, it is envisioned that a measurement system such as this, which is a combination of low-cost hardware with sophisticated signal processing, could play a complementary role with existing methods in realizing a rapid inexpensive means of identifying food-borne bacterial pathogens.