Deep learning enhanced multiplex detection of viable foodborne pathogens in digital microfluidic chip

Area of Science:

• Microfluidics
• Deep Learning
• Foodborne Pathogen Detection

Background:

• Culture plating is the standard for quantifying foodborne pathogens but is slow and labor-intensive.
• There is a need for rapid, accurate, and accessible methods for pathogen detection.

Purpose of the Study:

• To develop a deep learning-enhanced digital microfluidic platform for multiplex detection of viable foodborne pathogens.
• To improve the speed and accuracy of bacterial typing and quantification.

Main Methods:

• Utilized a Time-Lapse images driven EfficientNet-Transformer Network (TLENTNet) for bacterial analysis.
• Employed a microfluidic chip with 800 microwells for single-bacterium encapsulation.
• Captured time-lapse images of bacterial growth every 15 minutes for spatiotemporal feature analysis.

Main Results:

• Achieved an average accuracy of 97.72% for detecting four bacterial species (S. typhimurium, E. coli O157:H7, S. aureus, B. cereus).
• Demonstrated a detection limit of 63 CFU/mL.
• Completed detection and quantification within 7 hours, significantly faster than traditional methods.

Conclusions:

• The developed deep learning-based digital microfluidic platform offers a rapid and accurate alternative to culture plating for foodborne pathogen detection.
• This technology has the potential to enhance food safety monitoring and diagnostics.