The performance of three near-infrared spectroscopy (NIRS) instruments was compared through the investigation of coriander seed authenticity. The Thermo Fisher iS50 NIRS benchtop instrument, the portable Ocean Insights Flame-NIR and the Consumer Physics handheld SCiO device were assessed in conjunction with chemometric modelling in order to determine their predictive capabilities and use as quantitative tools through regression analysis. Two hundred authentic coriander seed samples and ninety adulterated samples were analysed on each device. Prediction models were developed and validated using SIMCA 15 chemometric software. All instruments correctly predicted 100% of the adulterated samples. The best models resulted in correct predictions of 100%, 98.5% and 95.6% for authentic coriander samples using spectra from the iS50, Flame-NIR and SCiO, respectively. The development of regression models highlighted the limitations of the Flame-NIR and SCiO for quantitative analysis, compared to the iS50. However, the results indicate their use as screening tools for on-site analysis of food, at various stages of the food supply chain.

Background: Given the recent detection of tetrodotoxin (TTX) in bivalve molluscs but the absence of a full collaborative validation study for TTX determination in a large number of shellfish samples, interlaboratory assessment of method performance was required to better understand current capabilities for accurate and reproducible TTX quantitation using chemical and immunoassay methods.Objective: The aim was to conduct an interlaboratory study with multiple laboratories, using results to assess method performance and acceptability of different TTX testing methods.Methods: Homogenous and stable mussel and oyster materials were assessed by participants using a range of published and in-house detection methods to determine mean TTX concentrations. Data were used to calculate recoveries, repeatability, and reproducibility, together with participant acceptability z-scores.Results: Method performance characteristics were good, showing excellent sensitivity, recovery, and repeatability. Acceptable reproducibility was evidenced by HorRat values for all LC–MS/MS and ELISA methods being less than the 2.0 limit of acceptability. Method differences between the LC–MS/MS participants did not result in statistically different results. Method performance characteristics compared well with previously published single-laboratory validated methods and no statistical difference was found in results returned by ELISA in comparison with LC–MS/MS.Conclusion: The results from this study demonstrate that current LC–MS/MS methods and ELISA are on the whole capable of sensitive, accurate, and reproducible TTX quantitation in shellfish. Further work is recommended to expand the number of laboratories testing ELISA and to standardize an LC–MS/MS protocol to further improve interlaboratory precision.Highlights: Multiple mass spectrometric methods and a commercial ELISA have been successfully assessed through an interlaboratory study, demonstrating excellent performance.