Evaluation of spectral handheld devices for freshness assessment of carp and trout fillets in relation to standard methods including non-targeted metabolomics

The potential of three different portable near infrared (NIR) spectrometers (TellSpec, SCiO, MicroNIR) to monitor freshness of organic carp and lake trout fillets was assessed. The intercomparison was supported by data obtained with standard methods (for biogenic amine content, microbiological parameters, total volatile basic nitrogen content, fatty acid profiles, drip loss, color and texture change) including non-targeted metabolomics by dynamic headspace - gas chromatography - time-of-flight mass spectrometry (DHS-GC-TOFMS). Fillets were measured after 1, 3, 6 and 9 days of storage at +4 degrees C. Discriminant and class modelling approaches (orthogonal partial least square discriminant analysis (OPLS-DA) and data-driven soft independent modelling of class analogy (DDSIMCA)) were applied in order to evaluate the performance of the NIR handhelds to distinguish fresh from spoiled fish. NIR data were correlated with data from standard methods in order to identify a set of characteristic wavelengths describing the spoilage process. Correlated wavelengths were evaluated regarding the improvement of the chemometric models. Handheld NIR spectrometers were able to discriminate carp fillets according to freshness with high sensitivity (88-100% for OPLS-DA, 100% for DD-SIMCA) and specificity (88-100% for OPLS-DA, 60-100% for DD-SIMCA). However, for trout fillets, both classification methods led to lower sensitivity and specificity compared to carp. Reduction of wavelengths to only those correlating with standard methods including DHS-GC-TOFMS data, improved the models' performance for trout samples measured with TellSpec and MicroNIR but did not improve the performance of carp models generated using MicroNIR and SCiO data. The results showed that the studied NIR handheld devices yielded different classification results depending on the investigated fish species. It can be concluded that NIR handhelds are a useful tool for on-site fish freshness assessment, however, primarily as a screening tool. For confirmation of results obtained in the field, lab-based analysis for freshness assessment by common spoilage indicators is required.

Automated summary

The potential of three different portable near infrared (NIR) spectrometers (TellSpec, SCiO, MicroNIR) to monitor the freshness of organic carp and lake trout fillets was evaluated. Discriminant and class modelling approaches were applied to assess the performance of the NIR handhelds in distinguishing fresh from spoiled fish. The results showed that handheld NIR spectrometers were able to discriminate carp fillets according to freshness with high sensitivity and specificity, while for trout fillets, both classification methods led to lower sensitivity and specificity compared to carp.