Non-destructive Through-Package Assessment of Salmon Fat Content Via Electrical Impedance Spectroscopy: A Preliminary Study
摘要
This study establishes the feasibility of using electrical impedance spectroscopy (EIS) as a non-destructive technique for assessing crude fat content in packaged salmon. Whole specimens of Coho salmon and Sockeye salmon were analyzed in various anatomical positions using a parallel circular plate electrode system. The electrical impedance |Z| and parallel capacitance Cp of the samples were measured at five representative frequencies between 100 Hz and 100 kHz in four progressive sample-preparation states: whole fish, half-section, fillet, and homogenized. Soxhlet extraction was applied to correlate the measured values with crude fat content, and the results were assessed using leave-one-out cross-validation (LOOCV). The resulting root mean square error of cross-validation (RMSECV), R2cv, and relative predictive deviation (RPD) were used as the primary prediction metrics. We found that sample thickness and tissue heterogeneity were important factors in determining prediction accuracy. Whole fish and half-section samples showed negligible correlations (R2cv < 0.53, RPD values < 1.5), whereas fillet preparations showed a strong correlation for Sockeye salmon (R2cv = 0.86, RPD = 2.23) and limited correlation for Coho salmon (R2cv = 0.44, RPD ≈ 1.2). The homogenized samples showed the highest correlation for both species (Sockeye: R2cv = 0.87, RPD = 2.43; Coho: R2cv = 0.71, RPD = 1.72). Notably, measuring through 0.075-mm polyethylene packaging yielded promising predictive correlations, which highlights the potential of through-package assessment. Thus, this study provides preliminary evidence of the technical feasibility of rapid fat assessment for commercial seafood quality control.