Hyperspectral imaging coupled with machine learning for classification of anthracnose infection on mango fruit.

Anthracnose is the major plant disease causing an economic loss of mango fruit. Anthracnose symptom is not visible at a quiescent stage and the infected fruit often enters the food chain before the infection is known. Detection of a pre-symptomatic anthracnose infection is thus, crucial to prevent the infected fruit from entering the food chain. This research applied hyperspectral imaging (HSI) spectroscopy integrated with machine learning (ML) including principal component analysis (PCA) and support vector machine (SVM) for rapid identification of quiescent infection of anthracnose in mango fruit. Mango fruit (Nam Dok Mai Si Thong) was artificially infected with Colletotrichum gloeosporioides and stored at 20 °C and 90 % RH. The HSI was used to collect the spectral and spatial data of the samples. PCA and SVM were respectively performed to explore the hyperspectral data and to classify different symptom severities. The obtained spectral data can be recognized as fingerprints ascribing to the metabolites produced by C. gloeosporioides and the decomposed fruit tissues caused by the fungal infection. The HSI integrated with ML was able to not only detect the anthracnose infection at a latent stage before the onset of disease symptoms but also correctly classify different symptom severities. The symptom maps were also constructed using false-color image processing to simplify the data visualization of different symptom severities. The capability of detecting a pre-symptomatic anthracnose infection is a key advantage of the developed ML-assisted HSI.

Effect of variable selection algorithms on model performance for predicting moisture content in biological materials using spectral data.

Variable selection is a critical step for designing a dedicated multispectral real-time system from multicollinearity spectral data. It improves the prediction ability of the calibration model and provides faster prediction by reducing the curse of dimensionality. The main objective of this study was to compare the effect of variables selection algorithms on model performance for predicting moisture content in red meat using visible and near-infrared (VNIR) hyperspectral imaging in the spectral range of 400-1000 nm and corn using near-infrared (NIR) spectroscopy in the spectral range of 1100-2498 nm. Six variable selection algorithms including the size of the regression coefficient (RC), variable importance in projection (VIP), genetic algorithm (GA), competitive adaptive reweighted sampling (CARS), successive projection algorithm (SPA), and stepwise regression (SWR) were tested and compared to realize their effects on the model performance for predicting moisture content in red meat and corn. The model based on competitive adaptive reweighted sampling-partial least squares regression (CARS-PLSR) was the best model to predict moisture content in red meat and corn. The results indicated the effectiveness of variable selection for providing the feature wavelengths to design a low-cost, real-time multispectral system.

Efficient preservation of sprouting vegetables under simulated microgravity conditions.

The effectiveness of a simulated microgravity environment as a novel method for preserving the freshness of vegetables was investigated. Three types of vegetables were selected: vegetable soybean, mung bean sprouts, and white radish sprouts. These selected vegetables were fixed on a three-dimensional rotary gravity controller, rotated slowly. The selected vegetables were stored at 25°C and 66% of relative humidity for 9, 6, or 5 d while undergoing this process. The simulated microgravity was controlled utilizing a gravity controller around 0 m s-2. The mung bean sprouts stored for 6 d under simulated microgravity conditions maintained higher thickness levels than the vegetable samples stored under normal gravity conditions (9.8 m s-2) for the same duration. The mass of all three items decreased with time without regard to the gravity environment, though the samples stored within the simulated microgravity environment displayed significant mass retention on and after 3 d for mung bean sprout samples and 1 d for white radish sprout samples. In contrast, the mass retention effect was not observed in the vegetable soybean samples. Hence, it was confirmed that the mass retention effect of microgravity was limited to sprout vegetables. As a result of analysis harnessing a mathematical model, assuming that the majority of the mass loss is due to moisture loss, a significant difference in mass reduction coefficient occurs among mung bean sprouts and white radish sprouts due to the microgravity environment, and the mass retention effect of simulated microgravity is quantitatively evaluated utilizing mathematical models. Simulated microgravity, which varies significantly from conventional refrigeration, ethylene control, and modified atmosphere, was demonstrated effective as a novel method for preserving and maintaining the freshness of sprout vegetables. This founding will support long-term space flight missions by prolonging shelf life of sprout vegetables.

Digitization of Broccoli Freshness Integrating External Color and Mass Loss.

Yellowing of green vegetables due to chlorophyll decomposition is a phenomenon indicating serious deterioration of freshness, and it is evaluated by measuring color space values. In contrast, mass reduction due to water loss is a deterioration of freshness observed in all horticultural crops. Therefore, in this study, we propose a novel freshness evaluation index for green vegetables that combines the degree of greenness and mass loss. The green color retention rate was measured using a computer vision system, and the mass retention rate was measured by weighing. Linear discriminant analysis (LDA) was performed using both variables (greenness and mass) as covariates to obtain a single freshness evaluation value (first canonical variable). The correct classification of storage period length by LDA was 96%. Green color retention alone allowed for classification of storage durations between 0 day and 10 days, whereas LDA could classify storage durations between 0 day and 12 days. The novel freshness evaluation index proposed by this research, which integrates greenness and mass, has been shown to be more accurate than the conventional evaluation index that uses only greenness.

Hyperspectral Imaging Coupled with Multivariate Analysis and Image Processing for Detection and Visualisation of Colour in Cooked Sausages Stuffed in Different Modified Casings.

A hyperspectral imaging system was for the first time exploited to estimate the core colour of sausages stuffed in natural hog casings or in two hog casings treated with solutions containing surfactants and lactic acid in slush salt. Yellowness of sausages stuffed in natural hog casings (control group, 20.26 ± 4.81) was significantly higher than that of sausages stuffed in casings modified by submersion for 90 min in a solution containing 1:30 (w/w) soy lecithin:distilled water, 2.5% wt. soy oil, and 21 mL lactic acid per kg NaCl (17.66 ± 2.89) (p < 0.05). When predicting the lightness and redness of the sausage core, a partial least squares regression model developed from spectra pre-treated with a second derivative showed calibration coefficients of determination (Rc2) of 0.73 and 0.76, respectively. Ten, ten, and seven wavelengths were selected as the important optimal wavelengths for lightness, redness, and yellowness, respectively. Those wavelengths provide meaningful information for developing a simple, cost-effective multispectral system to rapidly differentiate sausages based on their core colour. According to the canonical discriminant analysis, lightness possessed the highest discriminant power with which to differentiate sausages stuffed in different casings.

Prediction of Degreening Velocity of Broccoli Buds Using Hyperspectral Camera Combined with Artificial Neural Networks.

Developing a noninvasive technique to estimate the degreening (loss of green color) velocity of harvested broccoli was attempted. Loss of green color on a harvested broccoli head occurs heterogeneously. Therefore, hyperspectral imaging technique that stores spectral reflectance with spatial information was used in the present research. Using artificial neural networks (ANNs), we demonstrated that the reduction velocity of chlorophyll at a site on a broccoli head was related to the second derivative of spectral reflectance data at 15 wavelengths from 405 to 960 nm. The reduction velocity was predicted using the ANNs model with a correlative coefficient of 0.995 and a standard error of prediction of 5.37 × 10-5 mg·g-1·d-1. The estimated reduction velocity was effective for predicting the chlorophyll concentration of broccoli buds until 7 d of storage, which was established as the maximum time for maintaining marketability. This technique may be useful for nondestructive prediction of the shelf life of broccoli heads.

Induction of Terminal Oxidases of Electron Transport Chain in Broccoli Heads under Controlled Atmosphere Storage.

Controlled atmosphere (CA) storage, that is, at low O2 and high CO2 concentrations, effectively extends the shelf life of horticultural products. The influence of CA storage (O2/CO2: 2.5%/6.0% or 2.5%/0.0%) and in normal air (both at 1 °C for 21 d) on the physicochemical (O2 uptake, mass loss and L-ascorbate) and biological properties of broccoli (Brassica oleracea var. italica, Plenck, 1794) via amounts and activities of terminal oxidases of the electron transport chain was investigated. Mass loss, a sensitive index of freshness for broccoli heads under CA, was significantly lower under CA than under normoxia (p < 0.05). Mass loss was depressed 7 d earlier under CA, including 6.0% CO2 than under CA without CO2. High CO2 effectively depressed the degradation of L-ascorbate. During storage, the activity of the alternative oxidase (AOX) was lower under CA than in normal air (p < 0.05), while the amount of cytochrome c oxidase (COX), and the AOX/COX activity ratio (based on oxygen isotope discrimination), were not affected during storage. Our results indicate that CA storage effectively retained the freshness of broccoli heads by depressing the induction of AOX. However, depression of AOX amount was not associated with CO2 around broccoli heads.

Influence of low O2 and high CO2 environment on changes in metabolite concentrations in harvested vegetable soybeans.

Influence of atmosphere and storage period on the physicochemical and biological properties of harvested vegetable soybeans stored for 10 d at 25 °C was investigated. Storing vegetable soybeans under modified atmosphere (low O2 and high CO2), was more effective in maintaining its green color and mass than storing them under normoxia. Principal component 1 (PC1; contribution rate: 25%) was related to the atmospheres, whereas PC2 (contribution rate: 19%) was related to storage period. Cluster analysis showed that some types of sugars decreased, whereas some types of organic and amino acids increased with deterioration. Alanine, an indicator of low O2 stress, was maintained for 3 d under modified atmospheres, whereas alanine significantly decreased under normoxia. The concentrations of inositol and niacinamide (functional ingredients) under the modified atmospheres were significantly higher than those under normoxia. Thus, storage under modified atmospheres was effective in maintaining freshness and increasing the nutritional content of vegetable soybeans.

Meat quality evaluation based on computer vision technique: A review.

Nowadays people tend to include more meat in their diet thanks to the improvement in standards of living as well as an increase in awareness of meat nutritive values. To ensure public health, therefore, there is a need for a rise in worldwide meat production and consumption. Further attention is also required as to how the safety and the quality of meat production process should be assessed. Classical methods of meat quality assessment, however, have some disadvantages; expensive and time-consuming. This study intends to introduce an alternative method known as Computer Vision (CV) for the assessment of various quality parameters of muscle foods. CV has several advantages over the traditional methods. It is non-destructive, easy, and quick, hence, more efficient in meat quality assessments. This study aims to investigate different quality characteristics of some muscle foods using CV. It closes with a discussion on the future challenges and expected opportunities of the practical application of CV in the meat industry.

Effect of the storage atmosphere on metabolomics of harvested tomatoes (Solanum lycopersicum L.).

Harvested tomatoes were stored under atmospheres that were normoxic, anoxic, or modified (altered O2 and CO2 concentrations). Each atmosphere was created by storing the tomatoes at 25°C for up to 8 days in different kinds of pouches. During storage, metabolites of the tomatoes were measured using metabolomics. We obtained score plots of the metabolites on eighth day of storage by principal component analysis. There was a tendency for groups to be divided on the basis of score plot according to the composition of each gas. PC1 and PC2 seemed to correspond to the influence of O2 and CO2 concentrations, respectively, and the total contribution rate of the two axes was 72%, so that we concluded that the metabolites were affected mainly by O2 and CO2 concentrations. The results indicate that metabolomics may be an effective tool to reveal the relationship between metabolic state of harvested fruits and the atmosphere.

Simultaneous assessment of various quality attributes and shelf life of packaged bratwurst using hyperspectral imaging.

A simultaneous evaluation of various quality attributes of packaged bratwurst using hyperspectral imaging (HSI) was developed. Changes in physicochemical (L*, a*, b* color values, pH and thiobarbituric acid (TBA)), microbiological (total viable counts (TVC) and lactic acid bacteria (LAB)) and sensory (color, odor and overall acceptability) characteristics of the packaged sausages were monitored during storage at 4 ±â€¯1 °C. Reflectance spectra covering a wavelength range of 400-1000 nm of the samples were acquired using HSI. The relationships between the quality attributes and the spectroscopic reflectance were investigated using canonical correlation analysis. Among all quality attributes, L* color value, TBA, TVC, LAB, odor and overall acceptability appeared to be highly associated with the reflectance. To facilitate the HSI for rapid image acquisition and data processing, partial least squares regression (PLSR) analysis was employed for selection of optimal wavelengths. The selected wavelengths were then assembled into multispectral data and used as input variables to optimize the PLSR and artificial neural network models for the prediction of quality attributes of the sausage samples. The HSI technique can be used for rapid and nondestructive evaluation of the product's quality and shelf life.

Estimation of adenosine triphosphate content in ready-to-eat sausages with different storage days, using hyperspectral imaging coupled with R statistics.

A hyperspectral imaging (HSI) system (380-1000 nm) was investigated for non-invasively estimating adenosine triphosphate (ATP) content in ready-to-eat sausages during 5 days storage at 35 °C. A set of pretreated combinations were carried out on preprocessing the spectra to improve the performance of partial least squares regression (PLSR). According to the regression coefficient values, ten important wavelengths (385, 390, 395, 505, 580, 670, 745, 780, 855, and 955 nm) were selected in this study. PLSR models developed using full wavelengths and optimal wavelengths showed the prediction coefficient of determination (rp2) up to 0.8324 and 0.8606, respectively. The concentration and location of the ATP content in sausages were for the first time displayed via chemical imaging developed by R statistics. Combining HSI and multivariate analysis can quantify and visualize ATP dynamic changes during storage and a great potential in the processed meat industry for real-time inspection.

Storage in high-barrier pouches increases the sulforaphane concentration in broccoli florets.

Sulforaphane is a phytochemical that is usually found in cruciferous vegetables and is known to have a depressive effect on gastric cancer. Preliminary investigations showed that the sulforaphane concentration in broccoli (Brassica oleracea var. italica) florets increased under anoxia. Therefore, in the present study, we examined the effect of different atmospheric conditions on the sulforaphane concentration in broccoli and also tested whether there are concurrent effects on the concentration of ethanol, which is an unfavorable byproduct of fermentation. The sulforaphane concentration in broccoli florets was significantly elevated by 1.9- to 2.8-fold after 2 d of storage under hypoxia at ca. 0% O2 and ca. 24% CO2 at 20°C, whereas no such increase was observed following storage under normoxia at ca. 0% O2 without CO2 at 20°C. Furthermore, after 2 d, the sulforaphane concentration under hypoxia was 1.6- to 2.3-fold higher than that under normoxia. These results suggest that storage under hypoxia with high CO2 levels can elevate the sulforaphane concentration in broccoli florets. However, the elevated sulforaphane concentration could not be maintained beyond 2 d. There was no significant difference in the concentration of ethanol between florets that were stored under hypoxia with/without CO2 or normoxia at 2 d. However, the ethanol concentrations inside the pouches significantly increased between 2 d and 7 d. These findings indicate that the quality of broccoli florets can be improved through storage under hypoxia with high CO2 levels at 20°C for 2 d.

Real-time prediction of pre-cooked Japanese sausage color with different storage days using hyperspectral imaging.

BACKGROUND: Redness can greatly influence the freshness of sausages. A precise, rapid and noncontact analytical method or tool is needed to quantify the color. Hyperspectral imaging (HSI) is an emerging technique that integrates spectroscopy and imaging to obtain the spectral and spatial information simultaneously. In the present study, the redness of cooked sausages stored up to 57 days was predicted using HSI in tandem with multivariate data analysis. The mean spectra of the sausages were extracted from the hyperspectral images. Partial least squares regression (PLSR) and forward stepwise multiple regression (FSMR) models were used to develop the relavent spectral profiles with the redness of the cooked sausages. RESULTS: Ten important wavelengths were selected based on the regression coefficient values from the PLSR model. The PLSR model established using the full wavelengths presented a good performance, with Rc of 0.934 and a root mean square error of calibration of 0.642 (redness ranged between 14.99 and 21.48). The prediction maps for demonstrating evolution of redness in sausages were developed for the first time using R statistics (R Foundation for Statistical Computing) and Matlab (MathWorks Inc., Natick, MA, USA). CONCLUSION: HSI combined with PLSR and FSMR can be used to quantify and visualize evolution of sausage redness under different storage days. © 2017 Society of Chemical Industry.

Hyperspectral Imaging in Tandem with R Statistics and Image Processing for Detection and Visualization of pH in Japanese Big Sausages Under Different Storage Conditions.

The potential of hyperspectral imaging with wavelengths of 380 to 1000 nm was used to determine the pH of cooked sausages after different storage conditions (4 °C for 1 d, 35 °C for 1, 3, and 5 d). The mean spectra of the sausages were extracted from the hyperspectral images and partial least squares regression (PLSR) model was developed to relate spectral profiles with the pH of the cooked sausages. Eleven important wavelengths were selected based on the regression coefficient values. The PLSR model established using the optimal wavelengths showed good precision being the prediction coefficient of determination (Rp2 ) 0.909 and the root mean square error of prediction 0.035. The prediction map for illustrating pH indices in sausages was for the first time developed by R statistics. The overall results suggested that hyperspectral imaging combined with PLSR and R statistics are capable to quantify and visualize the sausages pH evolution under different storage conditions. PRACTICAL APPLICATION: In this paper, hyperspectral imaging is for the first time used to detect pH in cooked sausages using R statistics, which provides another useful information for the researchers who do not have the access to Matlab. Eleven optimal wavelengths were successfully selected, which were used for simplifying the PLSR model established based on the full wavelengths. This simplified model achieved a high Rp2 (0.909) and a low root mean square error of prediction (0.035), which can be useful for the design of multispectral imaging systems.

Influence of a Modified Atmosphere on the Induction and Activity of Respiratory Enzymes in Broccoli Florets during the Early Stage of Postharvest Storage.

Modified atmosphere packaging and controlled atmosphere storage (hypoxia conditions) extend shelf lives of horticultural products by depressing the O2 uptake rate. We investigated the relationship between atmospheres and alternative oxidase (AOX) to cytochrome c oxidase (COX) activities (on the basis of oxygen isotope discrimination) and the relative amounts of two respiratory enzymes, AOX and COX, during the early stage of storage. Broccoli florets, with high O2 uptake rates, were stored under hypoxia and normoxia at 25 °C. O2 uptake rates, weight loss, and yellowing of broccoli florets were significantly lower when stored under hypoxia than when stored under normoxia. Significantly more AOX proteins were produced during storage under normoxia, but COX proteins were more consistent than those of AOX proteins. Hypoxia may depress the expression of AOX and prolong the shelf life. Oxygen isotope discrimination was elevated under hypoxia after 50.5 h. AOX production in broccoli was controlled more by changing atmospheres than by COX.

Influence of Sampling Component on Determination of Soluble Solids Content of Fuji Apple Using Near-Infrared Spectroscopy.

Fuji apples from two production areas were separated into six batches by different experimenters. After applying light (500-1010 nm) on the surface of intact ones for their visible and near-infrared (NIR) spectra, destructive samples of three apple components were taken to determine the soluble solids content (SSC). Correlation and regression coefficients between the second Savitzky-Golay derivative of the spectra and SSC were analyzed to reveal that SSC values derived from the different apple components showed significantly different responses in the visible region. However, similar responses, particularly in the NIR section (730-932 nm), remained, including two sugar bands at 890 and 906 nm. On the basis of applying above characteristic bands to remove the interference signals, partial least square (PLS) and multiple linear regression (MLR) showed similar effective performances. According to the analysis of variance (ANOVA) method, sampling methods had significant effect on quantitative accuracy, and the model, using SSC values detected from the outer flesh cuboid (2.5 × 2.5 × 1.5 cm3), provided the best performance with lower root mean square error of prediction and higher correlation coefficient.

Identification of ROS Produced by Nanobubbles and Their Positive and Negative Effects on Vegetable Seed Germination.

Exogenous reactive oxygen species (ROS) produced by nanobubble (NB) water offer a reasonable explanation for NBs' physiological promotion and oxidation effects. To develop and exploit the NB technology, we have performed further research to identify the specific ROS produced by NBs. Using a fluorescent reagent APF, a Fenton reaction, a dismutation reaction of superoxide dismutase and DMSO, we distinguished four types of ROS (superoxide anion radical (O2·-), hydrogen peroxide (H2O2), hydroxyl radical (·OH), and singlet oxygen (1O2)). ·OH was confirmed to be the specific ROS produced by NB water. The role of ·OH produced by NB water in physiological processes depends on its concentration. The amount of exogenous ·OH has a positive correlation with the NB number density in the water. Here, spinach and carrot seed germination tests were repeatedly performed with three seed groups submerged in distilled water, high-number density NB water, and low-number density NB water under similar dissolved oxygen concentrations. The final germination rates of spinach seeds in distilled water, low-number density NB water, and high-number density NB water were 54%, 65%, and 69%, respectively. NBs can also promote sprout growth. The sprout lengths of spinach seeds dipped in NB water were longer than those in the distilled water. For carrot seeds, the amount of exogenous ·OH in high-number density NB water was beyond their toxic threshold, and negative effects were shown on hypocotyl elongation and chlorophyll formation. The presented results allow us to obtain a deeper understanding of the physiological promotion effects of NBs.

Online monitoring of red meat color using hyperspectral imaging.

A hyperspectral imaging system in the spectral range of 400-1000 nm was tested to develop an online monitoring system for red meat (beef, lamb, and pork) color in the meat industry. Instead of selecting different sets of important wavelengths for beef, lamb, and pork, a set of feature wavelengths were selected using the successive projection algorithm for red meat colors (L*, a*, b) for convenient industrial application. Only six wavelengths (450, 460, 600, 620, 820, and 980 nm) were further chosen as predictive feature wavelengths for predicting L*, a*, and b* in red meat. Multiple linear regression models were then developed and predicted L*, a*, and b* with coefficients of determination (R(2)p) of 0.97, 0.84, and 0.82, and root mean square error of prediction of 1.72, 1.73, and 1.35, respectively. Finally, distribution maps of meat surface color were generated. The results indicated that hyperspectral imaging has the potential to be used for rapid assessment of meat color.

Parsimonious model development for real-time monitoring of moisture in red meat using hyperspectral imaging.

A hyperspectral imaging system in the spectral range of 400-1000 nm was investigated to develop a multispectral real-time imaging system allowing the meat industry to determine moisture content in red meat including beef, lamb, and pork. Multivariate calibration models were developed using partial least-squares regression (PLSR) and least-squares support vector machines (LS-SVM) in the full spectral range. Instead of selection of different sets of feature wavelengths for beef, lamb, and pork, a set of 10 feature wavelengths was selected for convenient industrial application for the determination of moisture content in red meat. A quantitative linear function was then established using MLR based on these key feature wavelengths for predicting moisture content of red meat in an online system and creating moisture distribution maps. The results reveal that the combination of hyperspectral imaging and multivariate has great potential in the meat industry for real-time determination of moisture content.