Laundry to Laboratory:\nAutomated Image Analysis for\nthe Characterization of Fibrous Microplastics

Microfibers from clothes laundering\nare the predominant\nsource\nof ocean microplastics. Despite this, existing research relies on\noutdated methods and is inconsistent. This study addresses the critical\nneed for improved characterization of the prevalence and morphology\nof fibrous microplastics by introducing an automated image analysis\napproach. Utilizing digital or smartphone cameras, significant improvements\nare achieved in the characterization rate without losing accuracy,\nthus reducing variability and uncertainty. A novel Gaussian-offset\nthreshold methodology for automatic microplastic segmentation demonstrates\na 95% binary classification accuracy and a Matthews correlation coefficient\nof 0.87. Our methods were primarily evaluated using polyester microplastics\nfrom clothes washing produced at a concentration of 6280 fibers/g\nwith a median length of 435 μm. Additional testing was conducted\nwith fibers of varied makeup, width, and topological complexity. The\npractical application of this research was exemplified through a froth\nflotation study. Bubble flux was optimized for microplastic removal,\nand microplastic concentrations and length distributions were tracked\nover time. This work can be easily integrated into existing practices,\nsignificantly improves the labor-intensive nature of characterization,\nand ultimately contributes to a more standardized and reliable approach\nto understanding and mitigating fibrous microplastic pollution.