Skip to main content
ARS Home » Research » Publications at this Location » Publication #209247


item Chao, Kuanglin - Kevin Chao
item Nou, Xiangwu
item Kim, Moon
item Chan, Diane
item Patel, Jitu
item Sharma, Manan

Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/13/2007
Publication Date: 2/1/2008
Citation: Chao, K., Nou, X., Liu, Y., Kim, M.S., Chan, D.E., Yang, C., Patel, J.R., Sharma, M. 2008. Detection of fecal/ingesta contaminants on poultry processing equipment surfaces by visible and near-infrared reflectance spectroscopy. Applied Engineering in Agriculture. 24(11):49-55.

Interpretive Summary: The Agricultural Research Service of U.S. Department of Agriculture has developed a method based on visible (Vis) or near-infrared (NIR) spectra to differentiate fecal contaminants, ingesta contaminants, and bare poultry processing equipment surfaces at a commercial poultry processing plant. Because poultry feces are the most likely source of pathogenic contamination in a poultry plant, FSIS inspectors use the established guidelines to identify fecal residues on hard surfaces in the processing environment, including equipment and tools. Evaluation and inspection of sanitation effectiveness is usually performed through one or more of the following methods: human visual inspection, microbiological culture analysis, bioluminescent ATP-based assays, and antibody-based microbiological tests. However, these labor-intensive and time-consuming procedures do not meet the needs of the poultry processing industry for an accurate, high speed, and non-invasive method that can provide near immediate results that are useful for monitoring the processing line in real-time. Thus, development of low-cost, reliable, and portable sensor systems is being pursued, such as personal goggle and binocular devices. One key factor in successful applications is the use of a few essential spectral bands, which should not only reflect the chemical / physical information in the samples, but also maintain successive discrimination and classification efficiency. This study found that the visible band ratio using the 518 nm and 576 nm wavelength pair and the NIR band ratio using the 1565 nm and 1645 nm wavelength pair were both able to identify 100% of fecal and ingesta contaminants. The bare stainless steel surfaces were easily differentiated from the contaminant samples, but a small percentage of bare rubber belt surfaces were misclassified by both the visible and NIR band ratios. The NIR ratio performed slightly better, achieving 95.0% correct identification for bare equipment surfaces, while the visible ratio achieved 92.5%. Microbiological analysis of contaminant and equipment surface samples showed significant EBC values for fecal contaminant samples, compared to the other three sample types. Compared to fecal contaminants, ingesta contaminants showed significantly lower EBC values, but were not easily differentiated spectrally. Consequently, for the development of a device implementing these Vis/NIR waveband ratios for rapid and accurate surface sanitation verification purposes, fecal and ingesta contaminants should be included together for target detection. This information is useful to the Food Safety and Inspection Service (FSIS), and poultry processing plants.

Technical Abstract: Visible and near-infrared (NIR) spectra and samples for laboratory microbial analysis were acquired of fecal contaminants, ingesta contaminants, and bare processing equipment surfaces (rubber and stainless steel) in a commercial poultry processing plant. Spectra were analyzed in the visible region of 450 to 748 nm and the NIR region of 920 to 1680 nm and microbial sampling for Enterobacteriaceae counts (EBC) was conducted for 89 fecal contaminant samples, 52 ingesta contaminant samples, 40 bare rubber belt areas, and 40 bare stainless steel areas. Two-wavelength band ratios in the visible and NIR regions were selected for separating contaminants from equipment areas. Principal component analysis (PCA) was performed to analyze the spectral data set and 2-class soft independent modeling of class analogy (SIMCA) models were developed for comparison with band ratio classification. Fecal and ingesta contaminants were difficult to separate from each other but both were easily differentiated from the equipment areas. The visible ratio using 518 nm and 576 nm correctly classified 100% of contaminant samples and 92.5% of equipment area samples. The NIR ratio using 1565 nm and 1645 nm correctly classified 100% of the contaminant samples and 95% of the equipment area samples. Microbiological analysis found the highest EBC levels for fecal contaminants; mean EBC for ingesta contaminants was significantly lower than that for fecal contaminants. The high EBC levels for fecal contaminants indicate that these contaminants should be targeted for spectral-based detection methods for sanitation monitoring and verification purposes; although their EBC levels are significantly lower, ingesta contaminants should also be included due to difficulty of separation from fecal contaminants.