Development of a smartphone-based lateral-flow imaging system using machine-learning classifiers for detection of...

Area of Science:

• Food Microbiology
• Biosensing Technology
• Machine Learning Applications

Background:

• Salmonella spp. is a significant foodborne pathogen causing public health concerns, often found in raw foods.
• Rapid and reliable detection methods for Salmonella are crucial for food safety.
• Existing methods can struggle with low bacterial concentrations, leading to inaccurate results.

Purpose of the Study:

• To develop a smartphone-based lateral-flow assay (SLFA) for sensitive and accurate detection of Salmonella spp.
• To enhance the detection of ambiguous low concentrations of Salmonella using machine learning.
• To improve the confidence and reduce errors in microbial detection compared to visual inspection.

Main Methods:

• Designed a smartphone cradle with optimized lighting for lateral-flow assay image capture.
• Applied machine learning algorithms (SVM, KNN) combined with L*a*b and RGB color spaces for Salmonella detection.
• Utilized image analysis of test lines from varying Salmonella concentrations.

Main Results:

• Achieved a high accuracy of 95.56% in detecting Salmonella spp. concentrations.
• Demonstrated superior performance over visual inspection in a blind test, with fewer errors.
• Established a detection limit of 5 x 10^4 CFU/mL for the developed SLFA.

Conclusions:

• The smartphone-based lateral-flow assay (SLFA) offers a reliable and accurate method for detecting Salmonella.
• Machine learning integration significantly improves the interpretation of ambiguous results in microbial assays.
• This technology presents a promising advancement for rapid, on-site foodborne pathogen detection.