Hyperspectral Imaging and Chemometrics for Nondestructive Quantification of Total Volatile Basic Nitrogen in Pacific Oysters (Crassostrea gigas)

Abstract

Total volatile basic nitrogen (TVB-N) content is used to evaluate Pacific oyster (Crassostrea gigas) freshness. In this work, hyperspectral imaging (HSI; 400–1000 nm) was used to measure the TVB-N content in Pacific oysters. Accordingly, Pacific oyster samples stored in 15 °C water were assessed at intervals after 1, 3, 5, 7, or 9 days. Minimum noise separation processing of the hyperspectral images was performed before determining the region of interest for data dimension reduction. The effects of multiplicative scatter correction (MSC) on the obtained data were then investigated. To simplify the calibration model, 12 characteristic wavelengths were selected from the Pacific oyster hypercube using the correlation coefficient method. Finally, multiple linear regression (MLR) and back-propagation artificial neural network (BP-ANN) models were built from the selected wavelengths. The experimental results showed that the correlation coefficients between the corrected, predicted, and cross-validated datasets were lower in the MLR model than in the BP-ANN. However, the MLR model outperformed the BP-ANN in terms of the root-mean-square errors of correction, prediction, and interaction verification. Overall, both the MLR and BP-ANN models demonstrated that the combination of HSI with chemometric methods can be used to detect and accurately predict Pacific Oyster freshness during storage.