ABSTRACT Aerobic plate counts (APCs) and color are key quality indicators of salmon slices during low‐temperature storage. This study aims to develop a rapid and nondestructive method using multispectral imaging (MSI) for the detection of APC and color in 4°C stored salmon slices, which presents less time‐taking, high accuracy, and a safer tool as compared with traditional biological strategies. Machine learning (ML) models, including partial least squares (PLS), support vector machine (SVM), back propagation neural network (BPNN), and genetic algorithms (GAs) integrated (GA‐PLS, GA‐SVM, and GA‐BPNN), are used for precise prediction of APC and color in salmon slices. Results demonstrated high prediction efficiencies for APC, with R 2 p > 97% in all models and the highest RPD of 7.59 and 7.25 achieved by GA‐BPNN and PLS, respectively. For color, SVM achieved excellent accuracy with R 2 p of 94% and 84% for L * and a *, and the lowest error of 0.62 and 0.83, respectively. Finally, our study suggests that MSI combined with chemometrics is a promising and multipurpose detection method for bacterial contamination and color values in salmon slices.
Jin et al. (Mon,) studied this question.