Location: Quality and Safety Assessment Research UnitTitle: New Application of Hyperspectral Imaging for Bacterial Cell Classification
Submitted to: NIR news (Near Infrared Reflectance News)
Publication Type: Research Notes
Publication Acceptance Date: 10/28/2016
Publication Date: 11/23/2016
Citation: Park, B., Eady, M.B. 2016. New Application of Hyperspectral Imaging for Bacterial Cell Classification. NIR news (Near Infrared Reflectance News). Volume 27 No.8:4-6.
Interpretive Summary: Salmonella and Campylobacter are common foodborne bacteria associated with poultry. Detection of the organism can take days or weeks, requiring specialized process including enumerating the bacteria for measurement. Hyperspectral microscopy has shown potential as a means of rapidly detecting foodborne pathogenic bacteria. This research demonstrated hyperspectral microscope imaging technique as a method for discriminating between species of Campylobacter as well as serotypes of Salmonella. Five Salmonella serotypes and three Campylobacter species were imaged and through the use of statistical data analysis method correctly identified bacteria cells with high classification accuracy, showing potential for future validation methods leading towards the use of hyperspectral microscopy as a rapid and sensitive bacterial detection tool for food safety
Technical Abstract: Hyperspectral microscopy has shown potential as a method for rapid detection of foodborne pathogenic bacteria with spectral characteristics from bacterial cells. Hyperspectral microscope images (HMIs) are collected from broiler chicken isolates of Salmonella serotypes Enteritidis, Typhimurium, Infantis, Heidelberg, Kentucky and Campylobacter species coli, fetus, and jejuni for classification. The process requires a sample preparation time of 20 minutes after bacteria enumeration, and has the sensitivity capabilities to detect single cells of bacteria. Individual cells from a trial data set were extracted from the hypercube of 89 spectral images and classified through multivariate data analysis with classification accuracy of 99.8% with support vector machine (SVM) method for Salmonella and 98.9% for Campylobacter with principal component linear discriminant analysis (PC-LDA) method, respectively. Thus, the results showed hyperspectral microscope imaging technique as a rapid and sensitive detection method for foodborne pathogenic bacteria.