This Is AuburnElectronic Theses and Dissertations

Characterization and differentiation of bacteria by hyperspectral analysis

Date

2019-07-19

Author

Riggs, Rebecca

Type of Degree

PhD Dissertation

Department

General Veterinary Medicine

Restriction Status

EMBARGOED

Restriction Type

Full

Date Available

07-17-2024

Abstract

Characterization and differentiation of microorganisms by hyperspectral analysis is a promising alternative to conventional techniques in clinical, food, and environmental microbiology to be used as a rapid, non-destructive tool. In this work we used the Cytoviva hyperspectral imaging system to elicit the individual hyperspectral profiles produced by three food-borne pathogens, Salmonella enterica serovar Typhimurium, Escherichia coli O157:H7, and Listeria monocytogenes at 18, 21, and 24 hours of growth. Additionally, the individual hyperspectral profiles produced by four Salmonella enterica serovar Typhimurium lipopolysaccharide mutants at 18 hours post inoculation were also analyzed. The distinctive feature of the hyperspectral system is that it employs a super-resolution condenser that allows for the capture of high-resolution spectral data of a single microorganism. The microscopic system with hyperspectral capability produces uncomplicated single peak optical spectra of individual cells that can be analyzed by simple computerized methods. In contrast, to the conventional multi-cell hyperspectral methods that require spectral convolution and principal component analysis. Data were collected by the pushbroom method through which an image was made that has both spectral and spatial data by moving the sample pixel line by pixel line to produce distinct profiles for each cell. This super-resolution hyperspectral imaging created sharp spectral profiles based upon the unique surface property of each organism. All organisms were cultured, pelleted, washed, transferred to a poly-l-lysine coated slide, and cover-slipped to obtain the spectral profiles of live, metabolically active bacterial cells. The single-peak bacterial spectra were analyzed by the non-statistical Local Maximum method and by the non-linear fitting of spectra to a particular statistical model. Gaussian, Lorentzian, Logistic, Inverse polynomial, Gumbel, and Gram-Charlier peak functions were applied to bacterial spectra in the range of 400 – 1000 nm. A one-way ANOVA was conducted to compare fitting parameters. With this technique, we were able to characterize and differentiate Salmonella, E. coli, Listeria, and four Salmonella mutant bacteria with a high level of confidence. This novel use of hyperspectral imaging has the potential to be used as a point-of-care testing and safety scanning at all points of food production, including growing, harvesting, preparation, transportation, distribution, and storage.