Development of automated microplastic identification workflow for Raman micro-imaging and evaluation of the uncertainties during micro-imaging

Artificial intelligence in microplastic detection and pollution control

The rising prevalence of microplastics (MPs) in various ecosystems has increased the demand for advanced detection and mitigation strategies. This review examines the integration of artificial intelligence (AI) with environmental science to improve microplastic detection. Focusing on image processing, Fourier transform infrared spectroscopy (FTIR), Raman spectroscopy, and hyperspectral imaging (HSI), the review highlights how AI enhances the efficiency and accuracy of these techniques. AI-driven image processing automates the identification and quantification of MPs, significantly reducing the need for manual analysis. FTIR and Raman spectroscopy accurately distinguish MP types by analyzing their unique spectral features, while HSI captures extensive spatial and spectral data, facilitating detection in complex environmental matrices. Furthermore, AI algorithms integrate data from these methods, enabling real-time monitoring, traceability prediction, and pollution hotspot identification. The synergy between AI and spectral imaging technologies represents a transformative approach to environmental monitoring and emphasizes the need to adopt innovative tools for protecting ecosystem health.

Spectral analysis of environmental microplastic polyethylene (PE) using average spectra

In this study, microplastic samples from surface seawater of Tokyo Bay were collected, polyethylene (PE) microplastics were used to calculate carbonyl index (CI), and average spectra of PE were analyzed and compared with a previous study applying agitation during chemical treatment. It was found that PE and polypropylene (PP) were the predominant polymer type in the samples. Among PE samples, fragments were the most commonly observed shape, with white being the dominant color. Deviations were found in the average spectra among different shapes and colors when compared to the standard PE spectrum. A comparison of the average spectra between the two datasets suggests that pronounced peaks related to oxidation are most likely resulted from agitation during the chemical treatment. Additionally, it was found a closer spectral resemblance between the sample spectra and the spectrum of standard sample of oxidized PE (PEOx) than with the standard PE spectrum, suggesting that using the oxidized PE as a reference spectrum might be more effective for identification. These findings highlight the complex factors affecting the spectral properties of microplastics and highlight the importance of understanding these variations to enhance the accuracy of microplastic identification workflows and understanding of environmental fate of microplastics.

Analysis of micro(nano)plastics based on automated data interpretation and modeling: A review

The widespread presence of micro(nano)plastics (MNPs) in the environment threatens ecosystem integrity, and thus, it is necessary to determine and assess the occurrence, characteristics, and transport of MNPs between ecological components. However, most analytical approaches are cost- and time-inefficient in providing quantitative information with sufficient detail, and interpreting results can be difficult. Alternative analyses integrating novel measurements by imaging or proximal sensing with signal processing and machine learning may supplement these approaches. In this review, we examined published research on methods used for the automated data interpretation of MNPs found in the environment or those artificially prepared by fragmenting bulk plastics. We critically reviewed the primary areas of the integrated analytical process, which include sampling, data acquisition, processing, and modeling, applied in identifying, classifying, and quantifying MNPs in soil, sediment, water, and biological samples. We also provide a comprehensive discussion regarding model uncertainties related to estimating MNPs in the environment. In the future, the development of routinely applicable and efficient methods is expected to significantly contribute to the successful establishment of automated MNP monitoring systems.

Accessing the intrinsic factors of carbonyl index of microplastics: Physical and spectral properties, baseline correction, calculation methods, and their interdependence

Carbonyl index (CI) is a measure of the degradation status of microplastics. While many studies address environmental factors of microplastic degradation, intrinsic factors like physical properties, spectral properties, baseline correction, and CI calculation methods are less explored. This research focused on these aspects using surface seawater samples. We found that color and shape have limited dependence on particle size or signal-to-noise ratio (SNR). Baseline correction can significantly alter CI values, with the direction of the shift depending on the methods used. Additionally, most CI values before and after baseline correction and those calculated using different methods tend to be strongly correlated. Using the selected CI calculation methods, we found that CI values varied significantly by shape and color. CI's relation to the similarity between the sample and its pristine form suggests an alternative degradation measure. Our findings emphasize the need for standardized CI calculation methods.

Challenges of Raman spectra to estimate carbonyl index of microplastics: A case study with environmental samples from sea surface

This study investigates the feasibility of using the carbonyl index (CI) derived from Raman spectra as an indicator of plastic degradation and its relationship with the CI calculated from Fourier transform infrared (FTIR) spectra, using microplastic samples of polyethylene (PE) from surface seawater. Multiple methods were used to calculate the CI values of FTIR spectra, while proposed methods were used to calculate the corresponding CI values of Raman spectra. Some significant relations between FTIR CI and Raman CI were observed. However, small R2 values suggest weak functional relationships, which can be attributed to the low signal-to-noise ratio (SNR) of Raman spectra. These results highlight the challenges of establishing a functional relationship between FTIR CI and Raman CI, including challenges such as the uniformity of Raman spectra, determining optimal Raman measurement parameters, selecting appropriate peaks for Raman CI calculation, deciding on spectral processing methods, and addressing the interdependence of these issues.