Location: Environmental Microbial & Food Safety Laboratory
Project Number: 8042-42000-021-021-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Sep 1, 2026
End Date: Aug 31, 2027
Objective:
Meatpacking and food-processing facilities rely heavily on visual, naked-eye inspection of carcasses, products, and equipment surfaces — a process that is subjective and prone to error. Rapid, user-friendly, and objective methods are needed to quantify contamination severity for Contamination and Sanitation Inspection (CSI) in agricultural and food-safety environments. This project will investigate and further develop a fluorescence-based contamination assessment metric, a novel image-based metric intended to quantify contamination severity from fluorescence imaging data. The specific objectives are to: (1) establish the scientific foundation necessary to evaluate this fluorescence-based contamination assessment metric as a quantitative hygiene-assessment framework; (2) develop advanced image-processing and machine-learning methodologies to compute and refine the index across diverse surfaces and operational conditions; and (3) validate the index against trained human evaluators and against accepted contamination-assessment techniques, including ATP bioluminescence, microbial culturing, and environmental residue testing.
Approach:
The work will collect large-scale, well-characterized fluorescence image datasets spanning multiple surface types and contamination scenarios. A machine-learning framework will be developed to annotate these fluorescence images and to compute the contamination assessment metric, with image augmentation (e.g., rotation, scaling, contrast adjustment) used to expand the training corpus for artificial-intelligence model development. The metric outputs will be benchmarked against trained human evaluators and correlated with established contamination-assessment methods (ATP bioluminescence, microbial culturing, and environmental residue testing) to characterize the robustness, generalizability, and practical utility of the index. The project will additionally explore next-generation AI approaches for contamination characterization and risk assessment. The resulting datasets, algorithms, and validation studies will provide the evidence base needed to refine the index, identify its limitations, establish performance benchmarks, and guide future development toward a scientifically grounded contamination-assessment methodology — ultimately informing the feasibility of transitioning fluorescence imaging from a qualitative inspection aid into a quantitative decision-support tool for hygiene verification and contamination monitoring.