Salmon, riched in various nutrients such as amino acids, minerals, and other micronutrient elements is popular and accepted worldwide. (Jääskeläinen et al., 2019). Due to the limit in production areas, the storage method for salmon has become a vital factor affecting salmon quality. Still, salmon has a short shelf life which is prone to deterioration. Chilled storage shows the outstanding feature in preserving the quality of salmon to the utmost extent. With freezing below − 30℃, frozen salmon has a longer shelf life owing to the inhibition of enzyme activity and microorganisms growth. However, repeated freezing and thawing will destroy tissue structures of salmon, resulting in the loss of abundant flavoring substances like nucleotides and amino acids, as well as coloring substances like astaxanthin and carotenoids, which leads to a decrease in tenderness, blur color, and off-taste of the salmon (Fernández-Segovia et al., 2012).
Umami is considered one of the most important characteristics of aquatic products, which mainly refers to the taste of sodium glutamate (Wang, Zhou, & Liu, 2020). Besides, previous research showed that the 5'-nucleotide also provides umami taste as the forms of GMP (disodium 5'-guanosine monophosphate), AMP (disodium 5'-adenosine monophosphate), and IMP (disodium 5'-inosine monophosphate)(Bruno Siewe, Kudre, & Narayan, 2021). Noteworthy, a synergistic effect exists between GMP, AMP, IMP, and MSG, which generates a strong umami taste in a certain proportion (Kong et al., 2017). Therefore, a single nucleotide or amino acid cannot fully characterize meat's umami intensity. Equivalent umami concentrations (EUC), proposed by a Japanese scholar Yamaguchi to characterize the umami intensity of food, represents the MSG (monosodium glutamate) concentration equal to the umami intensity generated by the mixtures of MSG-like amino acids and the flavor 5'-nucleotides (Yamaguchi et al., 1971). After that, some researchers also utilized EUC to analyze and evaluate the umami intensity of Chinese mitten crab, Ganoderma, canned mushrooms, aquacultured pufferfish, and shiitake. (Chen & Zhang, 2007; Tseng, Lee, Li, & Mau, 2005; Chiang, Yen, & Mau, 2006; Zhang, Wang, Li, & Liu, 2019; Harada-Padermo et al., 2020). Conventional procedures used for the determination of the EUC include the HPLC method, which measures the 5′-nucleotides and free amino acids, and then obtains the value through the formula. Although this method has high precision, it is time-consuming and requires a lot of toxic and harmful organic reagents. Therefore, it is necessary to establish a rapid and non-destructive measurement to characterize the umami intensity of salmon.
Hyperspectral imaging is a prominent detection technique that integrates advantages of both spectroscopy technology and computer vision technology, which can obtain the hyperspectral image in the form of a three-dimensional data block with spectral and spatial information (Yu et al.,2021;Jiang et al., 2019; Kamruzzaman, Makino, & Oshita, 2016; Lorente et al., 2012). With this characteristic, hyperspectral imaging technology has attracted many researchers for safety and quality evaluation of agricultural products such as vegetables (Furlanetto et al., 2020), grains (Nie, Zhang, Feng, Yu, & He, 2019; Zhang, Sun, Rao, & Ji, 2020), fruits(Li et al., 2018; Tian, Fan, Huang, Wang, & Li, 2020), meat (Anderssen, Stormo, Skåra, Skjelvareid, & Heia, 2020) and others. Furthermore, HSI has also been largely applied to predict the quality attributes like tenderness (Jiang et al., 2018; Reis et al., 2018), freshness (Guo, Huang, Zhu, Guo, & Qin, 2018), color (Kamruzzaman et al., 2016), moisture (Zhang et al., 2020), fat (Zhao, Esquerre, Downey, & O’Donnell, 2017) and protein content (Ma, Cheng, Sun, & Liu, 2019) and microorganismof the meat (He & Sun, 2015). However, no report has been found on the detection and characterization of the umami intensity of meat (especially in salmon) using this technology.
This study aims to establish a rapid and non-destractive method for umami intensity (EUC) detection and characterization of salmon based on the visible and near-infrared hyperspectral imaging system with multivariate feature variable screening methods. The specific goals of this study focused on the following five key points: (1) investigate the EUC changes in salmon with different freeze-thaw times; (2) select the optimal pretreatment algorithm; (3) construct the quantitative correlation between spectral data and EUC by PLS; (4) select the optimal variable screening algorithm to extract characteristic wavelengths related to nucleotides and free amino acids; (5) develop pseudo-color images to visualize the value and distribution of EUC. The complete procedures of this study are shown in Fig. 1
[Fig. 1]