Non-destructive Visible/NIR Spectroscopy for Differentiation of Fresh and Frozen-thawed Fish

Nondestructive visible/near‐infrared (NIR) spectroscopy was evaluated to investigate whether fish has been frozen‐thawed. Fresh or frozen‐thawed red sea bream Pagrus major (n= 108) were scanned using a NIRSystems 6500 spectrophotometer equipped with a surface interactance fiber‐optic accessory then discriminated by soft independent modeling of class analogy (SIMCA) and linear discriminant analysis (LDA) based on principal component analysis (PCA) scores. The major effect of freeze‐thawing treatment involves a gross change in total reflectance after freezing and thawing; this arises from changes in light scatter presumably arising from alterations in the physical structure of at least the surface layer of fish. Untreated original absorbance spectra achieved much better (100%) classification accuracy for the prediction samples while the same figures for multiplicative scatter correction (MSC) treated spectra are considerably worse, indicating that scattering is the major information that makes classification work. No incorrect type of classification at all and also there are no samples classified to both groups either. This faster technique has the potential to differentiate fresh and frozen‐thawed fish and could be applied for online or at‐line processing control.