Advances in Machine Learning for the Detection and Characterization of Microplastics in the Environment

Mst. Maksuda Khanam¹M Khabir Uddin^1*Julhash U Kazi^2*

• ¹Department of Environmental Sciences, Jahangirnagar University, Dhaka, Bangladesh
• ²Division of Translational Cancer Research, Department of Laboratory Medicine, Faculty of Medicine, Lund University, Lund, Sweden

Microplastics are increasingly recognized as a pervasive pollutant in both aquatic and terrestrial environments, raising pressing concerns about their ecological impacts and implications for human health. Traditional detection and quantification methods—including manual microscopy and standalone spectroscopic techniques—offer reliable accuracy but are limited by labor-intensive procedures and low throughput. Recent advances in machine learning (ML) have revolutionized the field of microplastic research by automating and enhancing detection processes. In particular, algorithms such as support vector machines, random forests, and convolutional neural networks have demonstrated considerable success in classifying microplastics based on chemical signatures and visual characteristics. This review offers a comprehensive overview of ML approaches utilized for monitoring microplastic contamination across diverse aquatic settings. Spectral techniques, including infrared and Raman spectroscopy, leverage molecular vibrations to facilitate highly specific identification of polymer types, even within heterogeneous matrices. Image-based methods make use of sophisticated computer vision techniques to classify microplastics by shape, size, and color, reducing the subjectivity inherent in manual counting. Extending these capabilities further, hyperspectral imaging combines spatial and spectral data to generate comprehensive chemical maps, enabling the simultaneous assessment of polymer composition and distribution. Integrating ML algorithms into these various approaches has improved sensitivity, speed, and scalability, thereby addressing critical challenges in high-throughput and real-time monitoring. Despite these advances, key obstacles remain, including the need for larger, higher-quality datasets and the development of robust models capable of handling complex environmental conditions. Nevertheless, ongoing improvements in imaging hardware and ML methodologies hold significant promise for establishing more effective, automated, and accurate strategies for microplastic detection. By providing a comprehensive overview of current technologies and future opportunities, this review aims to guide researchers and stakeholders in developing science-based solutions for mitigating the global threat of microplastic pollution.