Hyperspectral, time-resolved, fluorescence imaging system for large sample sizes: Part II. Detection of fecal contamination on spinach

Authors: TEWEY, KEVIN - University Of Maryland; Lefcourt, Alan; TASCH, URI - University Of Maryland; SHILTS, PATRICK - University Of Maryland; Kim, Moon

Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/12/2017
Publication Date: 4/1/2018
Citation: Tewey, K., Lefcourt, A.M., Tasch, U., Shilts, P., Kim, M.S. 2018. Hyperspectral, time-resolved, fluorescence imaging system for large sample sizes: Part II. Detection of fecal contamination on spinach. Transactions of the ASABE. 61(2):391-398. https://doi.org/10.13031/trans.11571.
DOI: https://doi.org/10.13031/trans.11571

Interpretive Summary: Contamination of produce with pathogens prior to harvest is a recognized food safety concern. In an attempt to reduce the potential for contaminated produce from entering the food supply, produce fields are visually inspected for possible sources of fecal contamination and identified problem sites are not harvested. Our laboratory developed an imaging system with the goal of using the system in produce fields to increase the effectiveness of current inspections. The system was tested by applying four different dilutions of dairy feces to spinach leaves and evaluating algorithms to detect the contaminated sites. The imaging system consists of a gated, intensified, camera; a spectral adapter; and a lens with a pulsed UV laser to provide illumination. The UV illumination elicited fluorescence responses from both the spinach and the feces. The camera captured the fluorescence responses of both in a nanosecond time frame. Using the 450-500 nm waveband (blue region of the visible spectrum) and proper exposure settings of the camera, algorithms were developed that essentially allowed for 100% identification of all but the most dilute fecal sites with no errors. These findings demonstrate that the imaging system works and has the potential to be able to reliably detect feces in produce fields prior to harvest. Given that produce accounts for the majority of outbreaks of foodborne illnesses and that field inspections for fecal material occur at the beginning of the produce supply chain, even a small increase in the effectiveness of inspections could have a material impact on risks of foodborne illness. This information should be useful to other scientists, the produce industry and regulatory agencies.

Technical Abstract: To reduce the risk of foodborne illnesses resulting from fecal contamination events in produce fields, a novel hyperspectral, line-scan, laser-induced fluorescence, imaging system was developed with the goal of eventually incorporating the imaging system in a pre-harvest detection apparatus for fecal contamination. The imaging system includes an intensified, gated, camera, a 400-800 nm spectral adapter, a 355-nm pulsed laser, and optical expansion optics that produce a line-illumination profile. To validate and test the system, spinach leaves acquired from a local grocery store and inoculated with 1:2, 1:10, 1:100, and 1:200 dilutions of bovine fecal material using a 50 µl pipette were imaged repeatedly using a pre-define set of imaging parameters. Detection methods investigated included ratio detection, edge detection, threshold detection, and slope detection. Differences in the magnitude of averaged intensities for the spectral range of 450-500 nm for regions within fecal contamination sites and in nearby uncontaminated surface areas suggested that the 450-500 nm waveband would be a good region for use in detection tests. Validation tests using threshold or slope detection, the 450-500 nm waveband, and that took advantage of the slower fluorescence decay rates of fecal contamination sites relative to uncontaminated surface areas showed almost 100% detection of 1:2, 1:10, and 1:100 dilution sites and over 70% detection of 1:200 dilution sites with essentially zero false positives. These results suggest the imaging system has potential for the development of a commercially viable system for the pre-harvest detection of fecal contamination in produce fields, and for the detection of fecal contamination of leafy green vegetables in general.