Efficiency of NIR spectroscopy and chemometrics to assess mislabeling of fish fillets

Mislabeling and species substitution are widely affecting the fishery sector, especially considering fish fillets. It is difficult for both consumers and official control authorities to ascertain the identity information provided in the label. For this reason, there is a constant need of innovative, fast, and reliable instruments to improve the detection of such frauds. The present study explores how near-infrared (NIR) spectroscopy coupled with chemometric tools can identify fillets from five different fish species, namely: Atlantic cod (Gadus morhua), Goldblotch grouper (Epinephelus costae), Tub gurnard (Chelidonichthys lucerna), Guinean sole (Synaptura cadenati), and European plaice (Pleuronectes platessa). Fifty fresh samples of each species were collected and analyzed using three NIR devices: the portable MicroNIR by VIAVI, the handheld SCiO by Consumer Physics, and the benchtop MPA by Bruker. The acquired spectra were processed and analyzed to build classification models. Results showed high accuracy: SCiO performed with over 93 % accuracy in both cross-validation and test set prediction; MPA followed with accuracy over 83 %, while MicroNIR showed accuracy slightly lower than 81 %. These results suggest great reliability of the approach combining NIR spectroscopy and chemometric methods, in the context of food authenticity and fight against seafood frauds.