Hyperspectral Imaging Analysis for SSC Prediction in Actinidia arguta: Impact of Short-Term Anaerobic Treatment

In this study, hyperspectral imaging technology was utilized to monitor the alterations and spatial distribution of soluble solid content (SSC) in Actinidia arguta during postharvest storage. The fruit were exposed to a 24 h anaerobic treatment in a pure N₂ atmosphere and then stored at ambient temperature for 10 days. These findings collectively affirm that N₂ treatment can effectively decelerate the softening process of Actinidia arguta by impeding firmness loss and SSC progression. After preprocessing, feature band extraction was conducted using competitive adaptive reweighted sampling (CARS), interval variable iterative space shrinkage approach (iVISSA), and a synergistic iVISSA-CARS algorithm. Partial least squares regression (PLSR) and particle swarm optimization extreme learning machine (PSO-ELM) models were developed for SSC prediction, with the PSO-ELM model yielding the most accurate predictions. In the test set, the CARS-PSO-ELM model for the control group achieved an R_p² of 0.877, an RMSEP of 0.611, and an RPD of 1.953, while the iVISSA-CARS-PSO-ELM model for the N₂ treatment group achieved an R_p² of 0.904, an RMSEP of 0.554, and an RPD of 2.236. Finally, SSC visualization maps of Actinidia arguta were generated for both the control and treatment groups based on their respective optimal models, providing valuable references for comprehensive quality assessment during subsequent processing, transportation, and commercialization stages.