Classification of Pathogenic Bacteria Using Near-Infrared Diffuse Reflectance Spectroscopy

Abstract

Near-infrared diffuse reflectance spectroscopy is proposed for the classification of pathogenic bacteria using optical properties. The spectrally resolved data are analyzed using a diffuse reflectance model to extract the local optical properties, including the reduced scattering coefficient and absorption coefficient. The optical properties at different wavelengths form the feature set. A particle swarm optimization-based support vector machine is used to classify seven categories of bacteria. The experimental results demonstrate the feasibility of the method for the rapid and noninvasive classification of pathogenic bacteria using optical properties.