Classification and detection of Salmonella, Escherichia coli O157:H7, and Listeria monocytogenes using near infrared spectroscopy coupled with machine learning

Authors: OZTURK, SAMET - Oak Ridge Institute For Science And Education (ORISE); Huang, Lihan; Hwang, Cheng An; Sheen, Shiowshuh

Submitted to: Food Research International
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/19/2026
Publication Date: 1/20/2026
Citation: Ozturk, S., Huang, L., Hwang, C., Sheen, S. 2026. Classification and detection of Salmonella, Escherichia coli O157:H7, and Listeria monocytogenes using near infrared spectroscopy coupled with machine learning. Food Research International. https://doi.org/10.1016/j.foodres.2026.118485.
DOI: https://doi.org/10.1016/j.foodres.2026.118485

Interpretive Summary: Salmonella, Escherichia coli O157:H7, and Listeria monocytogenes are three major pathogens compromising the safety of the US food supply. Rapid detection and classification may prevent foodborne outbreaks caused by these pathogens. This study was conducted to combine near infrared (NIR) spectroscopy and machine learning for rapid detection and classification of these pathogens. The results showed that NIR spectroscopy and machine learning can detect and classify these pathogens with greater than 95% accuracy. The results from this study may be used by the food industry, regulatory agencies, and food retailers and distributors to detect these pathogens and prevent foodborne outbreaks.

Technical Abstract: This study was conducted to investigate the potential use of near-infrared (NIR) spectroscopy combined with machine learning (ML) algorithms to accurately identify foodborne pathogens, including Salmonella spp., Listeria monocytogenes, and Escherichia coli O157:H7 strains. Six different strains (two per bacterium) were individually cultured and purified through sequential washing steps using ethanol–deionized (DI) water solutions. Following purification, 0.1 ml of the bacterial strain suspended in 0.4 ml of 75% ethanol-25% DI water solution was transferred into a 96-well cell culture plate covered with filter paper and then vacuum dried at 50°C under 20'kPa for 1 hour. The absorbance spectra of dehydrated bacterial cells were then acquired across the range of 1000-2400 nm using a diffuse reflectance probe connected to a NIR-process analyzer. To identify the optimal classification pipeline, the acquired spectra were then analyzed using ten different pre-processing methods, three feature selection methods, and supervised algorithms including partial least squares discriminant analysis (PLS-DA), support vector machine (SVM), random forest (RF), artificial neural network (ANN), and convolutional neural network (CNN). Furthermore, the accuracy of the developed ML pipelines was further enhanced through under-sampling and boosting steps. The results demonstrated that SVM, RF, ANN and CNN outperformed PLS-DA, where the classification accuracies were above 90%. The findings demonstrate that using Savitzky-Golay first derivative (SG1) filtering to pre-process the full spectra followed by SVM classification yielded the highest accuracy, achieving 95.3% overall accuracy in ML pipeline. This study highlights the powerful capability of NIR spectroscopy coupled with ML algorithms to detect and identify foodborne pathogens on dehydrated surfaces in process environments.