Comparison of two rapid automated analysis tools for large FTIR microplastic datasets.

One of the biggest issues in microplastic (MP, plastic items <5 mm) research is the lack of standardisation and harmonisation in all fields, reaching from sampling methodology to sample purification, analytical methods and data analysis. This hampers comparability as well as reproducibility among studies. Concerning chemical analysis of MPs, Fourier-transform infrared (FTIR) spectroscocopy is one of the most powerful tools. Here, focal plane array (FPA) based micro-FTIR (µFTIR) imaging allows for rapid measurement and identification without manual preselection of putative MP and therefore enables large sample throughputs with high spatial resolution. The resulting huge datasets necessitate automated algorithms for data analysis in a reasonable time frame. Although solutions are available, little is known about the comparability or the level of reliability of their output. For the first time, within our study, we compare two well-established and frequently applied data analysis algorithms in regard to results in abundance, polymer composition and size distributions of MP (11-500 µm) derived from selected environmental water samples: (a) the siMPle analysis tool (systematic identification of MicroPlastics in the environment) in combination with MPAPP (MicroPlastic Automated Particle/fibre analysis Pipeline) and (b) the BPF (Bayreuth Particle Finder). The results of our comparison show an overall good accordance but also indicate discrepancies concerning certain polymer types/clusters as well as the smallest MP size classes. Our study further demonstrates that a detailed comparison of MP algorithms is an essential prerequisite for a better comparability of MP data.

Classification of target tissues of Eisenia fetida using sequential multimodal chemical analysis and machine learning.

Acquiring comprehensive knowledge about the uptake of pollutants, impact on tissue integrity and the effects at the molecular level in organisms is of increasing interest due to the environmental exposure to numerous contaminants. The analysis of tissues can be performed by histological examination, which is still time-consuming and restricted to target-specific staining methods. The histological approaches can be complemented with chemical imaging analysis. Chemical imaging of tissue sections is typically performed using a single imaging approach. However, for toxicological testing of environmental pollutants, a multimodal approach combined with improved data acquisition and evaluation is desirable, since it may allow for more rapid tissue characterization and give further information on ecotoxicological effects at the tissue level. Therefore, using the soil model organism Eisenia fetida as a model, we developed a sequential workflow combining Fourier transform infrared spectroscopy (FTIR) and matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) for chemical analysis of the same tissue sections. Data analysis of the FTIR spectra via random decision forest (RDF) classification enabled the rapid identification of target tissues (e.g., digestive tissue), which are relevant from an ecotoxicological point of view. MALDI imaging analysis provided specific lipid species which are sensitive to metabolic changes and environmental stressors. Taken together, our approach provides a fast and reproducible workflow for label-free histochemical tissue analyses in E. fetida, which can be applied to other model organisms as well.

Computer-Assisted Analysis of Microplastics in Environmental Samples Based on µFTIR Imaging in Combination with Machine Learning.

The problem of automating the data analysis of microplastics following a spectroscopic measurement such as focal plane array (FPA)-based micro-Fourier transform infrared (FTIR), Raman, or QCL is gaining ever more attention. Ease of use of the analysis software, reduction of expert time, analysis speed, and accuracy of the result are key for making the overall process scalable and thus allowing nonresearch laboratories to offer microplastics analysis as a service. Over the recent years, the prevailing approach has been to use spectral library search to automatically identify spectra of the sample. Recent studies, however, showed that this approach is rather limited in certain contexts, which led to developments for making library searches more robust but on the other hand also paved the way for introducing more advanced machine learning approaches. This study describes a model-based machine learning approach based on random decision forests for the analysis of large FPA-µFTIR data sets of environmental samples. The model can distinguish between more than 20 different polymer types and is applicable to complex matrices. The performance of the model under these demanding circumstances is shown based on eight different data sets. Further, a Monte Carlo cross validation has been performed to compute error rates such as sensitivity, specificity, and precision.

A graph-based clustering method with special focus on hyperspectral imaging.

A common trait of the more established clustering algorithms such as K-Means and HCA is their tendency to focus mainly on the bulk features of the data which causes minor features to be attributed to larger clusters. For hyperspectral imaging this has the consequence that substances which are covered by only a few pixels tend to be overlooked and thus cannot be separated. If small lateral features such as particles are the research objective this might be the reason why cluster analysis fails. Therefore we propose a novel graph-based clustering algorithm dubbed GBCC which is sensitive to small variations in data density and scales its clusters according to the underlying structures. The analysis of the proposed method covers a comparison to K-Means, DBSCAN and KNSC using a 2D artificial dataset. Further the method is evaluated on a multisensor image of atmospheric particulate matter composed of Raman and EDX data as well as an FTIR image of microplastics.