Quantitative image analysis of microplastics in bottled water using artificial intelligence.

Area of Science:

• Environmental Science
• Analytical Chemistry
• Food Safety

Background:

• Microplastics (MPs) are widespread in the environment and food chain.
• Current analytical methods lack validation for reliable MP quantification and risk assessment.

Purpose of the Study:

• To develop and validate a novel, automated method for microplastic quantification in bottled water.
• To utilize artificial intelligence for enhanced speed and accuracy in MP analysis.

Main Methods:

• Nile Red staining and fluorescent microscopy coupled with AI-driven image processing (Random Forest classifier).
• Automated workflow for particle count, size, and size distribution analysis.
• Method validation for sensitivity, linearity, precision, and accuracy.

Main Results:

• Highly sensitive detection of MPs down to 10 μm.
• Quantification limits: 28 items/500 mL (LOD), 85 items/500 mL (LOQ).
• Excellent linearity (R²=0.99), precision (9-16% RSD), and accuracy (92-98% recovery).
• Successful application to commercial bottled water samples, with MP levels up to 7237 items/500 mL.

Conclusions:

• The developed AI-assisted method is validated and suitable for regulatory compliance.
• Provides a reliable tool for assessing MP contamination in bottled water.
• Enables accurate estimation of consumer exposure and food safety risks.