Hyperspectral application in bacterial rapid detection using optical scattering technology via integration of a supercontinuum laser

Abstract

We developed a hyperspectral elastic light scatter (ELS) phenotyping instrument to explore the relationship between the wavelength of the incident beam and the elastic light-scatter patterns of a bacterial colony, and, ultimately, to enhance the classification efficiency of non-invasive ELS-based systems employed in microbiology. The new instrument consists of a supercontinuum (SC) laser and acousto-optic tunable filter (AOTF), which enables the selection of the wavelength of interest allowing multiple spectral patterns in a single measurement. An initial experiment with microflora found on green leafy vegetables derived an encouraging result, showing over 90% of average classification accuracy when classifying colonies from two different bacterial species utilizing 70 spectral bands from SC-laser. This observation suggested the notion that colonies of varied species may form distinguishable scatter features at separate different spectral bands. The increase in the number of employed bands consequently led to the rise in the number of scatter pattern features, potentially resulting in the classifier's overfitting and decrease in real-life accuracy. Therefore, the presented hyperspectral ELS system employs feature reduction and selection procedures to enhance the robustness and ultimately lessen the complexity of data collection.