Related Experiment Video
Updated: Apr 14, 2026
06:27
Automated Analysis of C. elegans Fluorescence Images using SegElegans
Published on: October 10, 2025
796
Explainable deep learning framework for fecal contamination detection on chicken eggshells via portable fluorescence
Insuck Baek1, Lalit Mohan Kandpal1, Chansong Hwang2
1Environmental Microbial and Food Safety Laboratory, Agricultural Research Service, United States, Department of Agriculture, Beltsville, MD 20705, USA.
Poultry Science
|April 12, 2026
Summary
Portable fluorescence imaging effectively detects fecal contamination on chicken eggs. Optimized deep learning models achieve high accuracy, enhancing food safety without darkroom conditions.
Area of Science:
- Food Science
- Biophotonics
- Artificial Intelligence
Background:
- Fecal contamination on eggshells poses a food safety risk.
- Current detection methods may lack sensitivity or require controlled environments.
- Fluorescence properties of fecal matter offer a potential detection avenue.
Purpose of the Study:
- To evaluate optimized deep learning models for in-situ fecal contamination detection on chicken eggshells using portable fluorescence imaging.
- To leverage spectral characteristics of fecal matter for accurate identification.
- To enhance food safety standards in the poultry industry.
Main Methods:
- Utilized a Contamination and Sanitization Inspection device to capture fluorescence images (600-720 nm emission).
- Developed and evaluated nine deep learning architectures (MobileNet, ViT Base 384) with 365 nm and 405 nm excitation.
- Employed Explainable AI frameworks to validate detection reliability.
Main Results:
- Fluorescence signals remained stable under ambient lighting.
- Structural Similarity Index Measure (SSIM) consistently exceeded 0.9200.
- MobileNet (365 nm excitation) achieved 0.9000 accuracy on brown eggs; ViT Base 384 (405 nm excitation) reached 0.9333 accuracy on white eggs.
Conclusions:
- Portable fluorescence imaging with optimized deep learning provides a robust method for detecting fecal contamination on eggshells.
- The technology enhances food safety by enabling objective, in-situ detection.
- This approach eliminates the need for strict darkroom environments, offering practical application in the poultry industry.
