Detection of chilling injury in pickling cucumbers using dual-band chlorophyll fluorescence imaging

Authors: LU, YUZHEN - Mississippi State University; Lu, Renfu

Submitted to: Foods
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/13/2021
Publication Date: 7/14/2021
Publication URL: https://handle.nal.usda.gov/10113/7387360
Citation: Lu, Y., Lu, R. 2021. Detection of chilling injury in pickling cucumbers using dual-band chlorophyll fluorescence imaging. Foods. 10(5). Article 1094. https://doi.org/10.3390/foods10051094.
DOI: https://doi.org/10.3390/foods10051094

Interpretive Summary: Pickling cucumbers are susceptible to chilling injury during postharvest refrigerated storage at temperatures below10 degrees Celsius, which could result in quality degradation and economic loss. It is thus desirable to remove these defective fruit before they are marketed as fresh or processed into pickled products. When green plant products like cucumber fruit are illuminated with short-wave ultraviolet or blue light, they would emit longer-wavelength light in the visible range due to the presence of chlorophylls, which is called chlorophyll fluorescence. Chilling-injured tissues would emit lower chlorophyll fluorescence compared to normal tissues. This study evaluated the feasibility of using a dual-band chlorophyll fluorescence imaging (CFI) technique for detecting chilling-injured pickling cucumbers. Three hundred normal pickling cucumbers were first subjected to different chilling injury treatments at 5 degrees Celsius and 80% relative humidity for 3 to 12 days. Chlorophyll fluorescence images at the two visible wavebands of 675 nm and 750 nm were then acquired from the treated pickling cucumbers, followed with the visual grading of them into three classes (normal, mild injury and severe injury). Characteristic features were extracted from the processed fluorescence images. Finally, machine learning algorithms (i.e., support vector machine) were used to develop classification models for automatic classification of the pickling cucumbers into two (normal and injured) or three classes. Results showed that combination of the images at 675 nm and 750 nm resulted in overall accuracies of 96.9% and 91.2% for two-class and three-class classification, respectively. Further, by using a fewer image features that were selected by using a technique called the neighborhood component feature selection technique, we were able to achieve 97.4% accuracy for the two-class classification. This study demonstrated that dual-band CFI is an effective modality for chilling injury detection of pickling cucumbers. Compared to other imaging techniques, this technique is faster and lower in instrumentation cost. With further research, the dual-band CFI technique is promising for real-time inspection of pickling cucumbers for chilling injury.

Technical Abstract: Pickling cucumbers are susceptible to chilling injury (CI) during postharvest refrigerated storage, which would result in quality degradation and economic loss. It is thus desirable to remove the defective fruit before they are marketed as fresh or processed into pickled products. Chlorophyll fluorescence is sensitive to CI in green fruits, because exposure to chilling temperatures can induce alterations in chlorophylls of tissues. This study evaluated the feasibility of using a dual-band chlorophyll fluorescence imaging (CFI) technique for detecting CI-affected pickling cucumbers. Chlorophyll fluorescence images at 675 nm and 750 nm were acquired from pickling cucumber samples under the excitation of ultraviolet-blue light. The raw images were processed for vignetting corrections through bi-dimensional empirical mode decomposition and subsequent image reconstruction. The fluorescence images were found effective for ascertaining CI-affected tissues, which appeared as dark areas in the images. Support vector machine models were developed for classifying pickling cucumbers into two or three classes using the features extracted from the fluorescence images. Concatenating the features at 675 nm and 750 nm resulted in overall accuracies of 96.9% and 91.2% for two-class (normal and injured) and three-class (normal, mildly and severely injured) classification, respectively, which are significantly better than those obtained using the features at a single wavelength. Further, a subset of features, selected based on the neighborhood component feature selection technique, achieved a significantly improved accuracy of 97.4% for the two-class classification. This study demonstrated that dual-band CFI is an effective modality for CI detection of pickling cucumbers.