Pectinolytic enterobacteria of the genus Pectobacterium (e.g., Pectobacterium carotovorum subsp. carotovorum, PCC) are the most common cause of major diseases, such as soft rot disease, in healthy vegetables, causing serious deterioration in quality of various healthy produce1. Soft rot is an important attribute that significantly influences cabbage texture through enzymatic maceration2. In addition, sticky patches, a common symptom of spoilage, are the root cause of unpleasant odors and may adversely affect adjacent products. Therefore, research has been performed to identify plant pathogens associated with soft rot disease and to develop protocols for detection and classification of infected vegetables3,4. In addition, PCC has long-lasting biological properties in environmental matrices such as surface water, groundwater, soil, and infected agricultural products such as seeds and vegetables5. Recently, Kang et al. reported that PCC contamination has seriously affected quality of kimchi due to the deterioration of commercially cut healthy napa cabbage (Brassica rapa L. subsp. pekinensis Hanelt)6.
Napa cabbage is an essential ingredient of kimchi, a traditional Korean fermented food standardized by the Codex Alimentarius Commission (CXS 223–2001) in 20017. It accounts for 70–80% of kimchi ingredients, thus directly affecting the quality of kimchi8. To improve kimchi quality, identifying infected napa cabbage before its processing is crucial. Generally, during the kimchi manufacturing process, soft rot disease of cabbage can be clearly determined while halving the cabbage; however, identification of soft rot disease symptoms in napa cabbage by the naked eye is challenging because of the large amount of cabbage that passes through the cutting machine. Moreover, even if an experienced employee visually inspects the softness of the cabbage tissue, efficiency issues may arise because of the unclear criteria of the operator. Therefore, developing rapid, nondestructive, and unbiased detection methods based on imaging analysis techniques for maintaining kimchi quality is necessary.
Research related to monitoring technologies for preventing soft rot symptoms has been conducted. Numerous studies have reported that several plant pathogens (e.g., Botrytis cinerea, Fusarium sambucinum, Pectobacterium carotovorum ssp. atrosepticum, P. carotovorum ssp. carotovorum, Phytophthora infestans, and Pythium ultimum) produce volatile markers that can be used to identify fruit and vegetable infections. Morath et al. reported that volatile compound (VC) emanation by fungi has biotechnological potential in the control of postharvest decay9. Strobel demonstrated the application of VCs for controlling postharvest fruit diseases using Muscodor albus10. Extensive evidence suggests that the VCs generated by antagonist bacteria could be effective in controlling postharvest decay caused by plant pathogens; however, those studies often remain unreported11,12. Several studies have demonstrated that plant pathogenic bacteria can alter the profiles of VCs emitted from healthy vegetable tissues13,14. Interestingly, these biological characteristics can be considered specific disease markers. Several monitoring studies on VC emission using nondestructive methods have been conducted; however, few studies have monitored and classified the pathogenic symptoms of PCC-infected postharvest cabbage using imaging analysis techniques based on an active sensing system.
Hyperspectral imaging (HSI) is an innovative platform technique that can integrate spectroscopy15 and computer vision16 and simultaneously provide information on the spatial and spectral properties of samples. It has been widely employed in the evaluation of food safety and quality17,18, to study defects or bruise classification19,20, firmness, and soluble solid content21, and to monitor pear quality22. As a nondestructive technology, HSI has also been applied for early detection of fruit diseases in apples and peaches16,20. Although several lab-scale studies based on HSI have been performed to classify the spectral characteristics of bean, potato, tomato, and lettuce diseases23,24, no study has focused on soft rot in napa cabbage.
In this study, we investigated the potential of HSI processing techniques at different spectral ranges based on near-infrared (NIR) region for classifying napa cabbage quality using nondestructive measurement. The specific objectives were: determining the physicochemical and microbiological quality properties of napa cabbage for intercomparison of HSI information, screening the HSI characteristic of 900 to 1,700 nm spectral ranges for evaluating napa cabbage freshness, and suggesting a novel approach for the classification of healthy and rotten napa cabbage. Consequently, the spectral preprocessing algorithms were verified, and different classification algorithms, such as partial least square discriminant analysis (PLS-DA), support vector machine (SVM), and random forests (RF), were used for rapid classification of soft rot symptoms from random napa cabbage spectra.