Comparative Analysis of Chilling Injury in Banana Fruit During Storage: Physicochemical and Microstructural Changes, and Early Optical-Based Nondestructive Identification

Abstract

Chilling injury (CI) during postharvest storage seriously impairs bananas’ quality and marketability. This study systematically investigated CI mechanisms through physicochemical, microstructural, and optical analyses and innovatively developed a hyperspectral imaging (HSI)-based approach for early CI detection. Bananas stored at suboptimal (7 °C) and optimal (13 °C) conditions exhibited distinct physicochemical changes. CI progression was related to increased browning symptoms, an abnormal moisture redistribution (reduced pulp moisture content), and delayed softening. Microstructural analysis revealed membrane destabilization, cellular lysis, intercellular cavity formation, and inhibited starch hydrolysis under chilling stress. Hyperspectral microscope imaging (HMI) captured chilling-induced spectral variations (400–1000 nm), enabling the t-SNE-based clustering of CI-affected tissues. Machine learning models using first derivative (1-st)-processed spectra achieved a high accuracy. Both PLS-DA and RF had a 99% calibration accuracy and 98.5% prediction accuracy for CI classification. Notably, HSI detected spectral signatures of early CI (2 days post-chilling treatment) before visible symptoms, achieving a 100% identification accuracy with an optimized PLS-DA combined with 1-st processing. This study provides a theoretical basis for studying fruit CI mechanisms and a novel nondestructive optical method for early CI monitoring in postharvest supply chains.