Air quality monitoring using mobile microscopy and machine learning

Area of Science:

• Environmental Science
• Biomedical Engineering
• Computer Science

Background:

• Particulate matter (PM) air pollution poses significant health risks, with PM2.5 classified as carcinogenic by the WHO.
• Accurate, high-throughput monitoring of airborne particles is essential for effective air quality management.
• Existing monitoring methods can be costly and lack portability for widespread deployment.

Purpose of the Study:

• To develop and validate a cost-effective, field-portable platform for high-throughput quantification of particulate matter.
• To integrate computational lens-free microscopy and machine learning for real-time aerosol analysis.
• To create a smartphone-compatible system for accessible air quality data.

Main Methods:

• Development of the c-Air platform, a portable device utilizing computational lens-free microscopy.
• Integration of a machine-learning algorithm for image analysis and particle quantification.
• Field testing in diverse indoor/outdoor environments and comparison with EPA-approved beta-attenuation monitoring.

Main Results:

• The c-Air platform demonstrated high-throughput screening of 6.5 L of air in 30 seconds with ~93% sizing accuracy.
• Strong correlation was observed between c-Air measurements and a reference EPA-approved device.
• Air quality mapping around LAX revealed elevated PM concentrations over 7 km away, particularly along flight paths.

Conclusions:

• The c-Air system offers a cost-effective, mobile solution for accurate and distributed air quality sensing.
• Its machine-learning interface allows adaptive detection of specific airborne particles like pollen and mold.
• This technology can significantly enhance real-time air quality monitoring and public health initiatives.