Qualitative and quantitative analysis of microplastics in chicken meat using near-infrared spectroscopy

The aim of this study was to present a rapid qualitative and quantitative analysis approach of microplastics (MPs) in chicken meat based on near infrared (NIR) diffuse reflectance spectroscopy. First, the NIR reflectance spectra of all samples were measured, and the spectral properties of non- and MPs-contaminated chicken meat were investigated. Then, the data driven soft independent modeling of class analogy (DD-SIMCA) was used to build models for the three target classes. The results showed that it was possible to identify the non-contaminated and MPs-contaminated chicken meat samples with 88.46-100% of sensitivity and 82-98.11% of specificity. Moreover, the competitive adaptive reweighted sampling (CARS) was extracted the characteristic variables. Finally, PLS, BPNN and linear SVM regression models were built for the determination of concentration of polypropylene (PP) and polyvinyl chloride (PVC) in chicken meat using the characteristic variables and the full-spectrum, respectively. PLS and linear SVM achieve similar results with high coefficient of determination of prediction (Rp2) and low root mean square error of prediction (RMSEP). CARS-PLS models obtained the best performance, with the Rp2 of 0.97 and 0.99 for PP and PVC, the RMSEP of 0.04% and 0.02% for PP and PVC, respectively. This is the first study to investigate the feasibility of detection of MPs in chicken meat using NIR spectroscopy coupled with DD-SIMCA and machine learning algorithms. The results showed that NIR spectroscopy for the rapid detection of MPs in chicken were feasible and achieved excellent analysis results.