We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Microplastic particles in the Arctic marine environment: database of IR spectra and its analysis by machine learning methods
Summary
Researchers compiled a database of infrared spectra from microplastic particles collected in the Arctic marine environment and applied machine learning methods to automate polymer identification, addressing the labor-intensive nature of manual spectral analysis. They developed and evaluated ML classification models using real environmental polymer spectra to improve the speed and scalability of microplastic chemical characterization in polar research.
Understanding microplastic marine pollution is gradually becoming the most important problem for the modern environmental science. A decisive step towards its solution requires the detailed analysis of the microplastic samples with respect to their chemical composition in various marine regions. Typically, such analysis is performed by various spectral methods and takes a lot of time. In recent years, machine learning (ML) has emerged as a powerful tool for overcoming this limitation, thus paving the way for less labor-intensive investigations. However, the development of the ML models also relies on the manual collection of the polymer samples database. Up to the data, there are only a limited number of freely available polymer datasets containing spectra of the real environmental polymer. Given the fact that all of these datasets are typically used at the stage of model development and testing, the question of the final verification of the proposed models is typically left for further research. In the present work, we examine this question by focusing on the analysis of the novel database of more than 1500 microplastic particles collected from the seas in the East Siberian Arctic during expeditions in 2019-2023. All of the microplastic particles were selected from the sea surface layers using a neuston net ranging from 5 to 0.5 mm, and subsequently analyzed using a Perkin Elmer FT-IR ATR Spectrum Two in the four-time scanning mode without any preliminary mineral/organic purification of the particles. The collected database is utilized for estimating the accuracy of the microplastic classification performed by the ML models already reported in the literature. Also, an analysis of the database diversity is performed with respect to other open-source microplastic IR spectra datasets. This study was supported by the Ministry of Science and Higher Education of the Russian Federation (state contract no. 075-15-2024-629, MegaGrant). Also see: https://micro2024.sciencesconf.org/559724/document
Sign in to start a discussion.
More Papers Like This
Microplastic particles in the Arctic marine environment: database of IR spectra and its analysis by machine learning methods
Researchers built a database of IR spectra from microplastic particles collected across Arctic marine environments and applied machine learning methods to enable faster and less labor-intensive chemical composition analysis, identifying polymer types from spectral signatures at broad regional scales.
Classifying polymers with mid-IR spectra and machine learning: From monitoring to detection
Researchers applied machine learning to mid-infrared spectra to automatically classify different types of plastic polymers found in the environment. Accurate polymer identification is essential for microplastic research, and this automated approach could improve monitoring efficiency and data consistency across studies.
A comparison of machine learning techniques for the detection of microplastics
This German-language study compared machine learning algorithms for classifying microplastics based on their infrared spectra, finding that several methods could reliably distinguish polymer types. Automating microplastic identification through machine learning could greatly increase the speed and throughput of environmental monitoring.
A machine learning algorithm for high throughput identification of FTIR spectra: Application on microplastics collected in the Mediterranean Sea
Researchers developed a machine learning method to automatically identify the chemical composition of microplastics from FTIR spectroscopy data collected during the Tara Mediterranean expedition. The algorithm performed well for common polymers like polyethylene and was applied to classify over 4,000 unidentified microplastic spectra. The study demonstrates that automated identification tools can significantly speed up large-scale microplastic pollution surveys while maintaining acceptable accuracy.
Training and evaluating machine learning algorithms for ocean microplastics classification through vibrational spectroscopy
Researchers evaluated multiple machine learning algorithms for automatically classifying ocean microplastics using infrared spectroscopy data across 13 polymer types. The study found that Support Vector Machine classifiers provided the best balance of simplicity and accuracy, offering a practical tool for faster and more reliable identification of microplastic contaminants.