Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Environmental Microbial & Food Safety Laboratory » Research » Publications at this Location » Publication #342599

Research Project: Sensing Technologies for the Detection and Characterization of Microbial, Chemical, and Biological Contaminants in Foods

Location: Environmental Microbial & Food Safety Laboratory

Title: Hyperspectral determination of fluorescence wavebands for multispectral imaging detection of multiple animal fecal species contamination on romaine lettuce

item CHO, HYUNGJEONG - Us Forest Service (FS)
item Kim, Moon
item KIM, SUNGYOUN - Korean Rural Development Administration
item LEE, HOONSOO - Us Forest Service (FS)
item OH, MIRAE - Us Forest Service (FS)
item CHUNG, SOOHYUN - Korea University

Submitted to: Food and Bioprocess Technology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/14/2017
Publication Date: 1/9/2018
Citation: Cho, H., Kim, M.S., Kim, S., Lee, H., Oh, M., Chung, S. 2018. Hyperspectral determination of fluorescence wavebands for multispectral imaging detection of multiple animal fecal species contamination on romaine lettuce. Food and Bioprocess Technology. 11:774-784.

Interpretive Summary: A large portion of past foodborne illness outbreaks involving E.coli and Salmonella have been found to be associated with greater consumption of raw produce as part of healthful lifestyles and increased popularity of ready-to-eat products such as packaged fresh salad greens. Since the primary sources of those bacteria are animal feces, which may contaminate vegetable fields directly via animal intrusion, or indirectly via contaminated irrigation water or natural rain/groundwater flow through regions with livestock or wild animal activity, this research used hyperspectral fluorescence imaging to evaluate and select fluorescence imaging wavebands to detect fecal contamination on the surfaces of green vegetable leaves. Single-waveband and two-waveband-ratio classification methods were evaluated for detecting feces from dairy cattle, pigs, chickens, and sheep. To simulate contamination scenarios, spots of both undiluted and diluted feces were applied to romaine lettuce leaves. The simple classification methods were selected and evaluated for detection of each fecal species as well as for "common" use to detect traces of all four fecal species undifferentiated as one group. The results show that single-waveband and two-waveband-ratio imaging are effective methods for fecal detection in the lab, although further research is needed to evaluate their effectiveness when implemented in portable imaging devices for real-world field use. This research will benefit the fresh produce industry and food safety regulators by providing science-based tools to help ensure the safety of fresh vegetables consumed by the public.

Technical Abstract: Consumption of fresh produce has been linked to multiple outbreaks of serious foodborne illnesses over the past two decades, and the popularity growth of ready-to-eat fruit and vegetable products may be related to the increased incidence of produce-related outbreaks. Because the sources of the pathogenic microorganisms most frequently involved in these outbreaks, E.coli O157: H7 and Salmonella, have been attributed primarily to animal fecal matter, research examining routes of fecal contamination and developing methods to prevent them have been areas of recent emphasis. This investigation used non-destructive hyperspectral fluorescence imaging to evaluate relatively simple spectral classification algorithms—using single wavebands or two-band ratios—for detecting animal feces contamination on leafy greens. In particular, this study sought to detect and discriminate between feces contamination from four animal species on romaine lettuce leaves. Single fluorescence wavebands found to be effective for discriminating feces from dairy cattle, pig, chicken, and sheep were F641 nm, F505 nm, F633 nm, and F645 nm, respectively. A two-band ratio in the range of F664±4 nm/F694±2 nm was shown to have a detection accuracy of over 93% for undiluted feces, and 1:20 and 1:50 fecal dilutions for three of the four fecal species. The spectral bands identified in this hyperspectral imaging study can be implemented for use in a relatively simple portable hand-held imaging device for on-site safety evaluation of produce in the field.