We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
From microplastics to pixels: Testing the robustness of two machine learning approaches for automated, Nile red-based marine microplastic identification.
Summary
Researchers tested the robustness of decision tree and random forest machine learning classifiers combined with Nile red fluorescent staining for automated detection and identification of microplastic polymers weathered under semi-controlled surface water and deep-sea conditions for up to one year. They found both models achieved comparable accuracy above 90% for pristine plastics, but assessed how environmental weathering affected classification reliability, also evaluating analysis time, model complexity, size detection limits, and interoperability.
Despite the pressing need for accurate and robust observations of microplastics in the marine environment to assess current and future environmental risks, existing procedures remain labour-intensive. Additionally, microplastic analysis is challenged by environmental weathering, affecting the reliability of studies relying on pristine plastics. This study addresses these knowledge gaps by testing the robustness of two automated analysis techniques which combine machine learning algorithms with fluorescent colouration of Nile red-stained particles. The study evaluated and compared the accuracy of models based on decision tree and random forest classifiers in detecting and identifying microplastic polymers weathered under semi-controlled surface water and deep-sea conditions, for a duration of one year. Furthermore, the analysis time, model complexity, lower size limit, and interoperability of the machine learning based approach were assessed. The decision tree and random forest models demonstrated a comparable accuracy in detecting and identifying pristine plastic polymers (both ¿ 90 Also see: https://micro2024.sciencesconf.org/559377/document