Quantitative and qualitative evaluation of microplastic contamination of shrimp using Vis-NIR multispectral imaging technology combined with a modified self-organizing map

Microplastics (MPs) severely impact ecosystems and human health. Therefore, detecting and identifying them is essential for assessing environmental problems. Visible-Near Infrared Multispectral Imaging (Vis-NIR MSI) technology is effective for polymer detection because it captures spatial and spectral data and provides fast and non-destructive measurements. However, this technique faces challenges, such as the high number of spectra per image, resulting in a high number of spectra as well as complex spectra per sample. To address this, chemometrics is needed to extract crucial information. Self-organizing maps (SOMs) are a remarkable area of chemometrics, revealing complex correlations in polymer detection. Herein, we propose a novel approach using the Vis-NIR MSI technique combined with modified SOMs to detect and characterize 1-4 mm MPs, including polyethylene (PE) and polypropylene (PP) in a biological sample (shrimp in this case) both quantitatively and qualitatively based on the number of pixels present in the object image. The modified SOMs were applied to the Vis-NIR MSI image to generate a color index, which was projected onto the object image, with different classes represented by different color shades. In qualitative analysis, the results allow for the visual identification and differentiation of PE and PP, providing insights into the types and distribution of MPs in the samples. In quantitative analysis, individual PE and PP were mixed with minced shrimp samples at concentrations ranging from the limit of quantification (LOQ) of 0.04% to 1% w/w. The results show that the modified model achieved a high R² over 0.99 for PE and PP. This suggests a strong correlation between the predicted and actual concentrations, indicating that the model can accurately predict concentrations of MPs in shrimp samples. The research shows that advanced imaging technologies and machine learning could be used to detect microplastic contamination in seafood products. Future efforts might focus on different types of seafood and MPs to improve food safety and safeguard consumer health.