Quantification of fat content in the liver of different aquaculture fish species using hyperspectral image analysis

Abstract

Lipid accumulation in the liver compromises the health and robustness of farmed fish, yet there is a lack of high-throughput methods for accurately measuring liver fat. Hyperspectral imaging holds promise as a rapid and cost-effective method for estimating tissues fat content in aquaculture species. In this study, we evaluated the efficacy of hyperspectral imaging in estimating liver fat content in three commercially relevant aquaculture species: Atlantic salmon, European Seabass, and Atlantic Cod. Two hyperspectral cameras were used to cover different spectral ranges, including the Visible and Near-Infrared (VNIR) and Shortwave Infrared (SWIR) regions. Partial least squares regression (PLSR) models based on VNIR and SWIR spectra show medium to high accuracy for predicting liver fat (R2 = 0.62–0.89), depending on the species and wavelength region. Spectral differences between liver tissues from the three species were distinct, as were regression coefficients for PLSR models to predict fat content. These results demonstrate the utility of hyperspectral imaging as a high-throughput method to assess liver fat in aquaculture finfish.