Defining Irregular Microplastics: A Machine Learning Approach for Morphometric Characterization

AbstractIntroduction: It is accepted that nano- and micro-plastic (NMP) pollutants threaten ecosystems and human health by their bioaccumulation but, interestingly, their toxicity is shape-dependent. However, a clear definition of irregular NMPs, as the dominant shape in environmental and biological samples, is currently lacking when compared to spherical and fibrous NMPs. Consequently, this issue has become an obstacle in terms of health risk assessments relating to NMPs.Objectives: This study quantifies morphometric descriptors in order to develop a standardized definition for irregular NMPs.Methods: Hyperspectral images of 34 spherical, 50 fibrous, and 45 irregular NMPs were collected from the literature. All shape-related features reported previously were analyzed using a machine learning model. Prominent morphometric descriptors and their optimal thresholds were determined using a decision tree (DT) classifier to accurately distinguish irregular NMPs from spherical and fibrous NMPs.Results: Five morphometric descriptors, including circularity, roundness, solidity, long diameter, and perimeter-area ratio were identified with the highest accuracy, for all spherical, fibrous, and irregular NMPs. Optimal thresholds for defining irregular NMPs were as follows: circularity (0.39–0.77), roundness (0.25–0.78), solidity (0.80–0.96), long diameter (65.62–212.87 µm), and perimeter-area ratio (>10.97). This definition generated a 96.0% macro-averaged accuracy across spherical, fibrous, and irregular NMPs, with 100% precision and 89.0% recall for irregular NMPs.Conclusions: Our results show that irregular NMPs may be characterized as five morphometric descriptors, such as circularity, roundness, solidity, long diameter, and perimeter-area ratio, with high accurate discrimination from spherical and fibrous NMPs.