The potential use of non destructive optical-based techniques for early detection of chilling injury and freshness in horticultural commodities

The increasing concern and awareness of the modern consumer regarding food including fruits and vegetables, has been oriented the research in the food industry to develop rapid, reliable and cost effective methods for the evaluation of food products including the traceability of the product history in terms of storage conditions. Since the conventional destructive analysis methods are time consuming, expensive, targeted and labor intensive, non-destructive methods are gaining significant popularity. These methods are being utilized by the food industry for the early detection of fruits defects, for the classification of fruits and vegetables on the basis of variety, maturity stage, storage history and origin and for the prediction of main internal constituents. Since chilling injury (CI) occurrence is a major problem for chilling sensitive products, as tropical and sub-tropical fruit and vegetables, prompt detection of CI is still a challenge to be addressed. The incorrect management of the temperature during storage and distribution causes significant losses and wastes in the horticultural food chain, which can be prevented if the product is promptly reported to the correct temperature, before that damages become irreversible. For this reason, rapid and fast methods for early detection of CI are needed. In the first work of this thesis, non-destructive optical techniques were applied for the early detection of chilling injury in eggplants. Eggplant fruit is a chilling sensitive vegetable that should be stored at temperatures above 12°C. For the estimation of CI, fruit were stored at 2°C (chilling temperature) and at 12°C (safe storage temperature) for a time span of 10 days. CIE L*a*b* measurements, reflectance data in the wavelength range 360–740 nm, Fourier Transformed (FT)-NIR spectra (800–2777 nm) and hyperspectral images in the visible (400–1000 nm) and near infrared (900–1700 nm) spectral range were acquired for each fruit. Partial least square discriminant analysis (PLSDA), supervised vector machine (SVM) and k-nearest neighbor (kNN) were applied to classify fruit according to the storage temperature. According to the results, although CI symptoms started being evident only after the 4th day of storage at 2°C, it was possible to discriminate fruit earlier using FT-NIR spectral data with the SVM classifier (100 and 92% non-error-rate (NER) in calibration and cross validation, respectively, in the whole data set. Color data and PLSDA classification possessed relatively lower accuracy as compared to SVM. These results depicted a good potential of for the non-destructive techniques for the early detection of CI in eggplants. Similarly, in the second experimental part of the thesis, hyperspectral imaging in Vis-NIR and SWIR regions combined with chemometric techniques were used for the early estimation of chilling injury in bell peppers. PLSDA models accompanied by wavelength selection algorithms were used for this purpose, with accuracies ranging from 81% and 87% non-error-rate (NER) based on the wavelength ranges used and variables selected. PLSR models were developed for the prediction of days of cold storage resulting in R²CV = 0.92 for full range and R²CV = 0.79 using selected variables. Based on the results, it was concluded, that Vis-NIR hyperspectral imaging is a reliable option for on-line classification of fresh versus refrigerated fruit and for identifying early incidence of CI. Inspired by the results obtained from previous studies a third study regarded the use of nondestructive techniques for the estimation of freshness of eggplants using color, spectral and hyperspectral measurements. To this aim, fruit were stored at 12°C for 10 days. Fruit were left at room temperature (20°C) for 1 day after sampling which was done with a 2-day interval, simulating one-day of shelf life in the market. PLSR models were developed using the spectral and hyperspectral data and the storage days, allowing safe assessment of the freshness of the fruits along with the utilization of SPA for variable reduction. The results depicted strong correlation between storage days, FT-NIR spectra and the hyperspectral data in the Vis-NIR range with accuracies as high as RC> 0.98, RCV> 0.94, RMSEC < 0.4 and RMSECV< 0.8, followed by lower accuracies using color data. The results of this study may set the basis to develop a protocol allowing a rapid screening and sorting of eggplants according to their postharvest freshness either upon handling in a distribution center or even upon the reception in the retail market. In the last work, as a deeper investigation, the effect of temperature and storage time on the FTNIR spectra was statistically investigated using ANOVA-simultaneous component analysis (ASCA) on eggplant fruit as a crop model. Also in this case, fruit were stored at 2 and 12 °C, for 10 days. Sensorial analysis, electrolyte leakage (EL), weight loss and firmness were used, as the reference measurements for CI. ASCA model proved that both temperature, duration of storage, and their interaction had a significant effect on the spectral changes over time of eggplant fruit. Followed by ASCA, PLSDA was conducted on the data to discriminate fruit based on the storage temperature. In this case, only the WL significant in the ASCA approach for temperature were considered, allowing to reach 87.4±2.7% as estimated by a repeated double-cross-validation procedure. The outcomes of all these studied manifested a promising, non-invasive, and fast tool for the control of CI and the prevention of food losses due to the incorrect management of the temperature in the horticultural food chain.