Submitted to: Applied Sciences
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/15/2021
Publication Date: 1/15/2021
Citation: Eady, M.B., Park, B. 2021. Unsupervised prediction model for Salmonella detection with hyperspectral microscopy: A Multi-year validation. Applied Sciences. https://doi.org/10.3390/app11030895.
Interpretive Summary: Salmonellosis is one of the leading causes of food poisoning around the world. It can be time consuming and expensive to confirm the presence of Salmonella in a food product along with the technical training requirement. There is a need for rapid and non-destructive testing methods for foodborne bacteria in an effort to reduce worldwide illness. Hyperspectral microscopy combines image collection and processing with across the visible light range. Previously, hyperspectral microscopy has shown to differentiate between bacteria species and sub-species as well. Here, a method is proposed and validated for a non-destructive and rapid (less than 1 hr) sampling method to detect Salmonella and validated over five years of data collection. When testing Salmonella against 14 non-Salmonella microorganisms the method had an overall classification accuracy of above 95%. This work suggests that similar models could also be constructed for additional foodborne or waterborne bacteria of interest.
Technical Abstract: Hyperspectral microscope images (HMI) have been previously explored as a tool for early and rapid detection of common foodborne pathogenic bacteria. A robust unsupervised classification approach to differentiate bacterial species with potential for single cell sensitivity benefits for a real-world application confirming the identification of pathogenic bacteria isolated from a food product. Here, a one-class soft independent model classification analogy (SIMCA) was used to determine if individual cells are Salmonella positive or negative. The model was constructed and validated with a spectral library built over five years, containing 13 Salmonella serotypes and 14 non-Salmonella foodborne pathogens. An image processing method designed to take less than one minute paired with the one-class Salmonella prediction algorithm resulted in an overall classification accuracy of 95.4%, with a Salmonella sensitivity of 0.97, and specificity of 0.92. SIMCA’s prediction accuracy was only achieved after a robust model incorporating multiple serotypes was established. These results demonstrate the potential for HMI as a sensitive and unsupervised presumptive screening method, moving towards early (< 8 h) and rapid (< 1 h) identification of Salmonella from food matrices.