Rapid classification of commercial Cheddar cheeses from different brands using PLSDA, LDA and SPA–LDA models built by hyperspectral data

Predicting wheat gluten concentrations in potato starch using GPR and SVM models built by terahertz time-domain spectroscopy

The purpose of this study was for the first time to explore the feasibility of terahertz (THz) spectral imaging for the detection of gluten contents in food samples. Based on the obtained 80 THz spectrum data, Gaussian process regression (GPR) and support vector machine (SVM) models were established to predict wheat gluten concentrations in 40 potato starch mixture samples. The prediction performances of GPR and SVM obtained were R2 = 0.859 and RMSE = 0.070, and R2 = 0.715 and RMSE = 0.101 in the gluten concentration range of 1.3%-100%, respectively, showing that the linear SVM algorithm had better prediction performance. The results indicated that THz spectral imaging combined with GPR could be used to predict the gluten content in food samples. It is thus hoped that this research should provide a novel technique for gluten content detection to ensure gluten-free food samples.

Identifying Chemical Differences in Cheddar Cheese Based on Maturity Level and Manufacturer Using Vibrational Spectroscopy and Chemometrics

Cheese is a nutritious dairy product and a valuable commodity. Internationally, cheddar cheese is produced and consumed in large quantities, and it is the main cheese variety that is exported from Australia. Despite its importance, the analytical methods to that are used to determine cheese quality rely on traditional approaches that require time, are invasive, and which involve potentially hazardous chemicals. In contrast, spectroscopic techniques can rapidly provide molecular information and are non-destructive, fast, and chemical-free methods. Combined with partner recognition methods (chemometrics), they can identify small changes in the composition or condition of cheeses. In this work, we combined FTIR and Raman spectroscopies with principal component analysis (PCA) to investigate the effects of aging in commercial cheddar cheeses. Changes in the amide I and II bands were the main spectral characteristics responsible for classifying commercial cheddar cheeses based on the ripening time and manufacturer using FTIR, and bands from lipids, including β'-polymorph of fat crystals, were more clearly determined through changes in the Raman spectra.

Analyzing the effects of nonthermal pretreatments on the quality of microwave vacuum dehydrated beef using terahertz time-domain spectroscopy and near-infrared hyperspectral imaging

Both nonthermal pretreatment and nondestructive analysis are effective technologies in improving drying processes. This study evaluated the effects of different pretreatment methods on the quality of beef dehydrated by microwave vacuum drying (MVD) and compared the MVD process performance comprising real-time moisture content (MC), MC loss, colour content, and shrinkage rate using different optical sensing methods including terahertz time-domain spectroscopy (THz-TDS) and near-infrared hyperspectral imaging (NIR-HSI). Results indicated that osmotic pretreatment improved the drying rate of MVD beef with lower changes in colour and shrinkage rate. Both THz-TDS-based and NIR-HSI-based on-site direct scanning and in-situ in-direct sensing showed accurate prediction results, with best R²_p of 0.9646 and 0.9463 for MC and R²_p of 0.9817 and 0.9563 for MC loss prediction, respectively. NIR-HSI visualisation of MC results showed that ultrasound pretreatment curbed but osmotic pretreatment promoted nonuniform distribution during MVD. This research should guide improving the industrial MVD drying process.

Non-Destructive Classification of Organic and Conventional Hens' Eggs Using Near-Infrared Hyperspectral Imaging

Eggs that are produced using organic methods retail at higher prices than those produced using conventional methods, but they cannot be differentiated reliably using visual methods. Eggs can therefore be fraudulently mislabeled in order to increase their wholesale and retail prices. The objective of this research was therefore to test near-infrared hyperspectral imaging (NIR-HSI) to identify whether an egg has been produced using organic or conventional methods. A total of 210 organic and 210 conventional fresh eggs were individually scanned using NIR-HSI to obtain absorbance spectra for discrimination analysis. The physical properties of each egg were also measured non-destructively in order to analyze the performance of discrimination compared with those of the NIR-HSI spectral data. Principal component analysis (PCA) showed variation for PC1 and PC2 of 57% and 23% and 94% and 4% based on physical properties and the spectral data, respectively. The best results of the classification using NIR-HSI spectral data obtained an accuracy of 96.03% and an error rate of 3.97% via partial least squares-discriminant analysis (PLS-DA), indicating the possibility that NIR-HSI could be successfully used to rapidly, reliably, and non-destructively differentiate between eggs that had been produced using organic methods from eggs that had been produced using conventional methods.

Research on highly sensitive quantitative detection of aflatoxin B2 solution based on THz metamaterial enhancement

Food such as cereal crops, oil crops and dairy products are very easy to produce highly toxic and carcinogenic aflatoxins during inappropriate storage. Therefore, it is of great significance to achieve rapid, non-destructive and highly sensitive detection of aflatoxin. A terahertz metamaterial sensor with "X" compound double-peak structure is designed based on electromagnetic theory to realize highly sensitive detection of aflatoxin B2 solution. It is found that the amplitude of the transmission peak of the terahertz transmission spectrum of aflatoxin B2 (AFB2) solution around 1.2 THz and 2.0 THz gradually decreased with the increase of the concentration of aflatoxin B2 solution, and the frequency of the transmission peak gradually shifted to high frequency with the increase of the concentration of aflatoxin B2 solution, hence a full concentration model was established. And a strategy of first classifying concentration intervals and then building a grouped quantitative model was proposed. The Limit of Detection (LOD) of the interval sub-model of low and medium concentration of aflatoxin B2 solution has been greatly improved with the LOD of the optimal grouping model was 7.28 × 10^-11 mg/ml, 4.19 × 10^-9 mg/ml and 1.22 × 10^-7 mg/ml, respectively. This research verifies the feasibility of terahertz metamaterial sensor based on "X" composite double-peak structure combined with THz-TDS technology for highly sensitive detection of aflatoxin B2 solution. And it provides a new rapid, non-destructive and highly sensitive detection of aflatoxin in food.

Differentiation of Transylvanian fruit distillates using supervised statistical tools based on isotopic and elemental fingerprint

BACKGROUND

The spirit drinks industry is one of the largest in the world. Fruit distillates require adequate analysis methods combined with statistical tools to build differentiation models, according to distinct criteria (geographical and botanical origin, producer's fingerprint, respectively). Over time a database of alcoholic beverage fingerprints can be generated, being very important for product safety and authenticity control.

RESULTS

To control the distillates' geographical origin, linear discriminant analysis (LDA) revealed that the cross-validation classification was correct for 88.2% of samples, but partial least squares discriminant analysis (PLS-DA) was slightly better suited for this purpose, with a correct classification rate of 91.2%. LDA effectiveness was proven for the trademark fingerprint differentiation, which was achieved at 93.5%, compared to 89.1% for PLS-DA. The principal predictors obtained by LDA were the same both for geographical origin and producer differentiation: B, δ¹³ C, Na, Cu, Ca and Be; highlighting the fact that in the production process of distillates each producer used fruits coming from the respective specific region. Through PLS-DA, some of the discrimination markers were the same for geographical origin and producer's identification, but others were completely specific: the rare earth elements Eu and Er only for geographical origin differentiation, and Cu solely as predictor for producer's identification. Regarding distillates' fruit variety, the correct discrimination rates of plum spirits from the rest were 84.2% for PLS-DA and 63% for LDA.

CONCLUSION

LDA and PLS-DA were suitable for differentiation models development of fruits spirits according to geographical region, producer and fruit variety based on isotopic and elemental fingerprint. © 2022 Society of Chemical Industry.

Applications of machine learning techniques for enhancing nondestructive food quality and safety detection

In considering the need of people all over the world for high-quality food, there has been a recent increase in interest in the role of nondestructive and rapid detection technologies in the food industry. Moreover, the analysis of data acquired by most nondestructive technologies is complex, time-consuming, and requires highly skilled operators. Meanwhile, the general applicability of various chemometric or statistical methods is affected by noise, sample, variability, and data complexity that vary under various testing conditions. Nowadays, machine learning (ML) techniques have a wide range of applications in the food industry, especially in nondestructive technology and equipment intelligence, due to their powerful ability in handling irrelevant information, extracting feature variables, and building calibration models. The review provides an introduction and comparison of machine learning techniques, and summarizes these algorithms as traditional machine learning (TML), and deep learning (DL). Moreover, several novel nondestructive technologies, namely acoustic analysis, machine vision (MV), electronic nose (E-nose), and spectral imaging, combined with different advanced ML techniques and their applications in food quality assessment such as variety identification and classification, safety inspection and processing control, are presented. In addition to this, the existing challenges and prospects are discussed. The result of this review indicates that nondestructive testing technologies combined with state-of-the-art machine learning techniques show great potential for monitoring the quality and safety of food products and different machine learning algorithms have their characteristics and applicability scenarios. Due to the nature of feature learning, DL is one of the most promising and powerful techniques for real-time applications, which needs further research for full and wide applications in the food industry.

Recent advances in muscle food safety evaluation: Hyperspectral imaging analyses and applications

As there is growing interest in process control for quality and safety in the meat industry, by integrating spectroscopy and imaging technologies into one system, hyperspectral imaging, or chemical or spectroscopic imaging has become an alternative analytical technique that can provide the spatial distribution of spectrum for fast and nondestructive detection of meat safety. This review addresses the configuration of the hyperspectral imaging system and safety indicators of muscle foods involving biological, chemical, and physical attributes and other associated hazards or poisons, which could cause safety problems. The emphasis focuses on applications of hyperspectral imaging techniques in the safety evaluation of muscle foods, including pork, beef, lamb, chicken, fish and other meat products. Although HSI can provide the spatial distribution of spectrum, characterized by overtones and combinations of the C-H, N-H, and O-H groups using different combinations of a light source, imaging spectrograph and camera, there still needs improvement to overcome the disadvantages of HSI technology for further applications at the industrial level.

Near-infrared techniques for fraud detection in dairy products: A review

The dairy products sector is an important part of the food industry, and their consumption is expected to grow in the next 10 years. Therefore, the authentication of these products in a faster and precise way is required for the sake of public health. This review proposes the use of near-infrared techniques for the detection of food fraud in dairy products as they are faster, nondestructive, environmentally friendly, do not require sample preparation, and allow multiconstituent analysis. First, we have described frequent forms of food fraud in dairy products and the application of traditional techniques for their detection, highlighting gaps and counterproductive characteristics for the actual global food chain, as longer sample preparation time and use of reagents. Then, the application of near-infrared spectroscopy and hyperspectral imaging for the detection of food fraud mainly in cheese, butter, and yogurt are described. As these techniques depend on model development, the coverage of different dairy products by the literature will promote the identification of food fraud in a faster and reliable way.

A dual AE-GAN guided THz spectral dehulling model for mapping energy and moisture distribution on sunflower seed kernels

Energy and moisture contents are important food chemical attributes. In the current study, a nondestructive Terahertz (THz) time-domain imaging system was first time used for evaluating the energy and moisture distributions of sunflower seed kernels inside shells. For this task, a dual autoencoders (AE)-generative adversarial nets (GAN) spectral dehulling semi-supervised model was developed. The model could automatically learn the kernel information from the latent representations of the spectra of the intact seeds through adversarial learning to achieve feature disentanglement. Results indicated that the generated kernel images had similar features to the original kernel images and high-quality chemical distribution maps for energy and moisture contents of sunflower seed kernels inside shells were successfully obtained. As the current method took the advantage of the characteristics of THz imaging and selected a suitable deep learning algorithm, it has the potential to generalize for imaging other chemical substances of other dry shelled seeds or biological samples (moisture content and thickness below 15% and 5 mm, respectively).

Effects of dielectric properties and microstructures on microwave-vacuum drying of mushroom (Agaricus bisporus) caps and stipes evaluated by non-destructive techniques

This research work aimed to investigate the effects of microstructures, dielectric property and temperature distributions on drying feature difference between the mushroom cap and stipe during the microwave-vacuum drying (MVD) process. Near-infrared hyperspectral imaging (NIR HSI) was employed to visualize distribution maps for moisture content (MC), dielectric constant ε' and dielectric loss factor ε'' of mushroom slices during the MVD process. Infrared (IR) thermal imaging was used to evaluate the temperature distribution of the mushroom slices. Results demonstrated higher MC, ε' and ε'' values in MVD mushroom stipes. Nevertheless, the centre area of the mushroom slice showed the highest temperature, while the MVD mushroom cap displayed a more porous structure. The effect of microstructure could encounter effects of dielectric properties and temperature to cause higher water evaporation in the MVD cap. This work highlights the novelty to combine different detection techniques to investigate the effects of microstructures, dielectric property and temperature distributions on drying patterns of mushroom slices.

Recent developments in vibrational spectral analyses for dynamically assessing and monitoring food dehydration processes

Dehydration is one of the most widely used food processing techniques, which is sophisticated in nature. Rapid and accurate prediction of dehydration performance and its effects on product quality is still a difficult task. Traditional analytical methods for evaluating food dehydration processes are laborious, time-consuming and destructive, and they are not suitable for online applications. On the other hand, vibrational spectral techniques coupled with chemometrics have emerged as a rapid and noninvasive tool with excellent potential for online evaluation and control of the dehydration process to improve final dried food quality. In the current review, the fundamental of food dehydration and five types of vibrational spectral techniques, and spectral data processing methods are introduced. Critical overtones bands related to dehydration attributes in the near-infrared (NIR) region and the state-of-the-art applications of vibrational spectral analyses in evaluating food quality attributes as affected by dehydration processes are summarized. Research investigations since 2010 on using vibrational spectral technologies combined with chemometrics to continuously monitor food quality attributes during dehydration processes are also covered in this review.

Application and interpretation of deep learning methods for the geographical origin identification of Radix Glycyrrhizae using hyperspectral imaging

Radix Glycyrrhizae is used as a functional food and traditional medicine. The geographical origin of Radix Glycyrrhizae is a determinant factor influencing the chemical and physical properties as well as its medicinal and health effects. The visible/near-infrared (Vis/NIR) (376–1044 nm) and near-infrared (NIR) hyperspectral imaging (915–1699 nm) were used to identify the geographical origin of Radix Glycyrrhizae. Convolutional neural network (CNN) and recurrent neural network (RNN) models in deep learning methods were built using extracted spectra, with logistic regression (LR) and support vector machine (SVM) models as comparisons. For both spectral ranges, the deep learning methods, LR and SVM all exhibited good results. The classification accuracy was over 90% for the calibration, validation, and prediction sets by the LR, CNN, and RNN models. Slight differences in classification performances existed between the two spectral ranges. Further, interpretation of the CNN model was conducted to identify the important wavelengths, and the wavelengths with high contribution rates that affected the discriminant analysis were consistent with the spectral differences. Thus, the overall results illustrate that hyperspectral imaging with deep learning methods can be used to identify the geographical origin of Radix Glycyrrhizae, which provides a new basis for related research.

Hyperspectral imaging provides an effective way to identify the geographical origin of Radix Glycyrrhizae to assess its quality.