We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Classification of microplastics in field-collected stream water using a submersible single-shot lensless polarimetric holographic system
Summary
Scientists built a compact, waterproof camera-like device that can be dropped directly into streams to detect and identify different types of microplastic particles in real time, with about 99% accuracy in tests. This matters because current methods for finding microplastics usually require collecting water samples and analyzing them later in a lab, which is slow and can miss important details — a fast, on-site tool like this could help researchers track plastic pollution in our water supplies more efficiently, an early step toward better understanding how these particles move through the environment and potentially into our bodies.
Microplastic pollution in aquatic environments demands field-ready tools that can rapidly detect and discriminate particles in complex aquatic environments. Here, we present a single-shot polarimetric off-axis lensless holographic imaging system integrated into a compact submersible prototype. The system combines a simple off-axis interferometric layout, requiring only one mirror and one beam splitter, with a division-of-focal-plane (DoFP) polarimetric sensor to acquire four polarization-resolved holograms per exposure, enabling Fourier-domain order isolation and complex-field recovery while preserving polarization-dependent contrast. We validate joint phase and polarimetric reconstruction using a birefringent USAF resolution target and demonstrate polymer discrimination directly from raw four-channel hologram patches using a lightweight detection–classification pipeline (no phase reconstruction or handcrafted features), achieving approximately 99% test accuracy across four different polymers. Finally, submerged experiments in unfiltered, field-collected stream water with controlled microplastic spiking confirm accurate detection and time-resolved classification under realistic conditions.