Next Article in Journal
Reliability of Different Nanofluids and Different Micro-Channel Configurations on the Heat Transfer Augmentation
Next Article in Special Issue
Analysis of the Unintended Effects of the Bacillus thuringiensis Insecticidal Protein in Genetically Modified Rice Using Untargeted Transcriptomics
Previous Article in Journal
Optimization and Performance Evaluation of Foam Acid Systems for Plugging Removal in Low Pressure Oil and Gas Reservoirs
Previous Article in Special Issue
Classification for GM and Non-GM Maize Kernels Based on NIR Spectra and Deep Learning
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Processes
Volume 11
Issue 3
10.3390/pr11030651
processes-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Identification of Transgenic Agricultural Products and Foods Using NIR Spectroscopy and Hyperspectral Imaging: A Review

by
Jun Zhang
1,
Zihao Liu
2,
Yaoyuan Pu
1,
Jiajun Wang
1,
Binman Tang
1,
Limin Dai
3,
Shuihua Yu
4 and
Ruqing Chen
1,*
1
Jiaxing Nanhu University, 572 Yuexiu South Road, Jiaxing 314001, China
2
School of Information Science and Engineering, Jiaxing University, 118 Jiahang Road, Jiaxing 314033, China
3
School of Agricultural Engineering, Jiangsu University, 301 Xuefu Road, Zhenjiang 212013, China
4
Seed Management Station of Jiuquan City, 1 Baoquan West Road, Jiuquan 735000, China
*
Author to whom correspondence should be addressed.
Processes 2023, 11(3), 651; https://doi.org/10.3390/pr11030651
Submission received: 20 January 2023 / Revised: 16 February 2023 / Accepted: 20 February 2023 / Published: 21 February 2023
(This article belongs to the Special Issue Genetically Modified Organisms Testing: Current Technology and Applications)
Browse Figures
Versions Notes

Abstract

:
Spectroscopy and its imaging techniques are now popular methods for quantitative and qualitative analysis in fields such as agricultural products and foods, and combined with various chemometric methods. In fact, this is the application basis for spectroscopy and spectral imaging techniques in other fields such as genetics and transgenic monitoring. To date, there has been considerable research using spectroscopy and its imaging techniques (especially NIR spectroscopy, hyperspectral imaging) for the effective identification of agricultural products and foods. There have been few comprehensive reviews that cover the use of spectroscopic and imaging methods in the identification of genetically modified organisms. Therefore, this paper focuses on the application of NIR spectroscopy and its imaging techniques (including NIR spectroscopy and hyperspectral imaging techniques) in transgenic agricultural product and food detection and compares them with traditional detection methods. A large number of studies have shown that the application of NIR spectroscopy and imaging techniques in the detection of genetically modified foods is effective when compared to conventional approaches such as polymerase chain reaction and enzyme-linked immunosorbent assay.

1. Introduction

Currently, genetics is widely used in various science fields. Transgenic technology is used to transfer the genes with known functional traits, such as high yield, resistance to disease and insects, and improvement of nutritional quality, into the target organism through modern scientific and technological means, so that new varieties and products are produced by adding new functional characteristics to the recipient organism. Many countries regard transgenic technology as a strategic choice to support development. Transgenics has become a strategic focus for countries to seize the commanding heights of science and technology, and to enhance the international competitiveness of agriculture.
At present, applications of transgenic technology in various fields, including to improve crops, produce vaccines, food, etc., are experiencing a very high growth rate. As a result, genetically modified (GM) production is increasing on the global market. Genetically modified crops are cultivated in 29 countries, with an area under cultivation of 190.4 million hectares [1]. Although GM crops have advantages such as insect resistance, weed resistance, disease resistance, improved nutritional value, and increased yield [2], the use of GM technology may have unintended negative effects on food and environmental safety, and therefore, GM foods have been severely restricted in most parts of the world due to legal pressure from regulatory agencies to control the production of GM products. Thus, it is a very necessary and important task to identify GM products.
Today, several methods for the identification of GM products are available on the market, including the protein-based ELISA assays, the DNA-based qualitative PCR assays, and also some chromatographic techniques such as HPLC and GC. Although DNA-based methods for the identification of GM products are more reliable compared to other methods [1,2]. However, these traditional detection methods have the disadvantages of being destructive, time-consuming, and costly, and cannot satisfy the demands of online applications [1]. These methods require the sample to be damaged, a series of experimental operations to be conducted, and professional personnel to analyze the results, and finally, to identify the gene; the results are relatively convincing.
In addition to these methods described above, some methods such as near infrared (NIR) spectroscopy, terahertz spectroscopy, and hyperspectral imaging have been found to be effective in the identification of GM crops. These methods do not require sample processing, and the time consumed for data collection is short. However, data processing, including pretreatment, wavelength selection, and model establishment (described in Section 2), also requires professional personnel to analyze and process, and the results have a certain dependence on the model. However, if a good model can be established, it is helpful for follow-up detection, promotion, and online detection. The advantages and disadvantages of the methods to detect GM products are shown in Table 1.
Compared with previous transgenic detection methods, the advantages of this technology are low cost, no need for sample preparation, and less time consumption. The data of NIR spectroscopy and hyperspectral imaging can be obtained in less than 5 min, and these two methods have wide application in the quality detecting of agricultural products. This paper will focus on the basic principles, detection methods, and applications of NIR spectroscopy and hyperspectral imaging techniques for the detection of transgenic agricultural products and foods. Figure 1 represents the steps of review of the paper.

2. The Principles and Characteristics of NIRS

NIRS, with a wavelength range between 780 and 2500 nm, can be divided into short-wave NIR (with a range of 780–1100 nm) and long-wave NIR (with a range of 1100–2500 nm) [3], and is sometimes used together with a range of 350–780 nm visible range light to form a Vis-NIR spectrum for relevant detection. The state, composition, and structure of the molecule can be obtained by analyzing the primary overtones and oscillations between the hydrogen-containing groups, such as C-H, N-H, O-H, etc., by NIRS [4]. The common near-infrared spectrometers consist of a light source, a beam splitter system (wavelength selector), a sample detector, and an optical detector, and some are equipped with a data processing/analysis system for simplicity. The use of these parts should be chosen according to their use. NIR spectroscopy has transmission, diffuse reflection, transmission and reflection detection methods, and the choice of different detection methods is also demand-dependent.
After data acquisition with the spectrometer, the general steps for spectral analysis include: (1) spectral data preprocessing [5]; (2) feature wavelength selection [6]; (3) model establishment and evaluation [7]. The main analysis steps are shown in Figure 2.

2.1. The Spectral Preprocessing Methods

The sample spectrum data collected by the spectrometer contains only not the chemical information of the sample itself, but also other irrelevant information and noise, such as electrical noise, sample background, stray light, etc. [5]. Therefore, in the application of chemometric methods for spectral analysis, it is necessary to preprocess the original spectral data to eliminate the irrelevant information and noise in the data, which is a necessary step in the analysis.
Smoothing, derivative, multiple scattering correction (MSC), baseline correction, standard normal transformation (SNV), orthogonal signal correction (OSC), and combinations of these methods are common spectral preprocessing methods [8,9,10].
Smoothing preprocessing is one of the most widely used methods for removing spectral noise. Moving average smoothing (MS) and Savitzk-Golay (SG) smoothing are commonly used smoothing methods. Derivative preprocessing is to eliminate the baseline offset and drift, enhance the spectral band features, and overcome the spectral band overlap [11]. The direct difference method and SG derivative method are the commonly used derivative preprocessing methods [12]. Baseline correction pretreatment successfully eliminates baseline drift and tilt caused by the instrument’s backdrop and the uneven surface of the sample by artificially pulling the baseline of the absorbance spectrum back to 0 baseline [13]. Multiple scattering correction (MSC) preprocessing is mainly used to eliminate the effect of scattering on the spectrum and effectively enhance the spectral information related to the content of sample components; the spectral errors due to factors such as optical path changes or sample dilution can be eliminated [4,14]. Standard normal variate transformation (SNV) is the processing of spectral data with a mean value of 0 and a standard deviation of 1 [15]. Orthogonal signal correction (OSC) is a spectral preprocessing algorithm based on the involvement of physical and chemical values of samples [16,17]. In order to improve the robustness and prediction ability of the model, the information unrelated to the physical and chemical values of spectral data is removed by orthogonal projection and then analyzed by corresponding modeling methods.
In NIRS preprocessing analysis, MSC and SNV are two well-known methods for reducing spectral distortion due to dispersion, and they have been proven to be effective in correcting the problems of inhomogeneous particle distribution and refractive index variation in food applications [18,19]. Although these preprocessing methods were aimed to reduce unmodeled variability in the spectra data in order to improve the features sought in the spectra, which are usually linearly related to the phenomenon of interest. However, if incorrect preprocessing techniques are used, the essential information may at risk of information removal [10].

2.2. The Feature Wavelength Selection Methods

When all wavelength variables are used for modeling, it may be computationally intensive and time consuming, and sometimes the absorption of NIR spectra is not obvious and the overlap is serious, which contains redundant information, so it is normal to eliminate the irrelevant information and filter out the independent variables with high correlation when modeling. When the useless variable is introduced into the model, it will affect model stability and prediction precision. Therefore, it is necessary to extract the feature wavelength variables from the full spectrum before modeling. At present, the commonly used methods for selecting the characteristic wavelengths [6,20,21] include principal component analysis (PCA), competitive adaptive reweighting (CARS), the genetic algorithm (GA), the successive projection algorithm (SPA), and uninformative variable elimination (UVE), etc.
PCA is a popular linear dimensionality reduction approach that is used to map high-dimensional data into a low-dimensional space using some type of linear projection. It is expected that the variance of the projected dimension is the largest, so that fewer data dimensions can be used and more original data points can be retained, which can reduce dimension and eliminate redundant information [4]. CARS is a variable selection method proposed to simulate the “survival of the fittest” principle in Darwin’s evolution theory [22]. The idea of GA is to optimize the PLSR model based on the RMSECV of selected variables by genetic iteration [23,24]. SPA is a method to improve modeling speed and prediction accuracy by reducing the covariance between variables and obtaining the wavelength with the least redundant information [6,25]. UVE is a wavelength selection algorithm based on the PLSR coefficients, which is used to eliminate the full-wavelength variables, the stability of which is less than the noise, thereby improving the predictive power of the model [26]. Sometimes, one feature wavelength selected algorithm is used, and the modeling effect is not very effective, and is therefore often used in combination with other feature wavelength selection methods [26].

2.3. Model Establishment and Evaluation

For NIR spectroscopy, a calibration model of the spectra is finally established in a linear and nonlinear way for qualitative or quantitative analysis after the pretreatment or feature wavelength selection.
With the rapid development of statistics, it is an inevitable trend to use mathematical analysis methods [7] for more scientific classification and quantitative detection, which can be linear, non-linear, or supervised or unsupervised modes. The common qualitative and quantitative methods are k-nearest (KNN) [27], linear discriminant analysis (LDA) [11,28], partial least squares discriminant analysis (PLS-DA) [29,30], extreme learning machine (ELM) [27], Support vector machine (SVM) [12,31,32], back propagation neural network (BPNN) [33], partial least squares regression (PLSR) [34,35], and radial basis function neural network (RBFNN) [27], etc.
After the model is established, the stability and accuracy of the model is evaluated, and the high-quality correction model is selected. Indicators often employed include accuracy, correlation coefficient, standard deviation of calibration and prediction set samples, etc.
NIRS and chemometrics methods are a pair of twin technologies that have been developing in tandem with each other. In recent years, deep learning algorithms, represented by convolutional neural networks (CNN), have been used for quantitative and qualitative modeling of NIR spectra [36,37]. Compared with traditional machine learning methods, the convolutional neural network can extract the features embedded in the spectral data step by step through multiple convolution and pooling layers, and to a certain extent, the preprocessing of spectra and the selection of variables before modeling can be reduced.
Among the most popular deep learning-based models, the DeepSpectra model has outperformed all the other models [38]. The combination of deep learning and spectral detection methods is a promising approach for the quality assessment of food and agricultural products, as well as for genetic modification detection [38,39].

3. The Applications of NIRS for the Detection of Transgenic Agricultural Products and Foods

In the last few decades, NIRS has demonstrated its power in the detection of agricultural products and foods, and there are now a series of applications in meat detection [40], agricultural materials and foods safety control [41,42,43], and fruits and vegetables detection [44,45]. Taking maize testing as an example, NIRS has been used in a range of applications in the identification of variety purity identification [27,46,47], vigor [48], internal components such as moisture and protein [49,50,51], fungal toxins [52,53], and frost damage [54]. Today, there is also equipment that can be used for online monitoring of agricultural products and foods using handheld/portable NIR spectroscopy for industrial applications [55,56].
With the in-depth study of spectroscopy technology, researchers have started to introduce NIRS technology into the identification of transgenic food and agricultural products, as shown in Table 2. In this section, articles in the application of NIRS technology to GM products and foods (including vegetables, cereals, fruits, other plants (soybean, cotton), oil, etc.) are screened and reviewed.
In the detection of vegetables, Soo-In Sohn et al. used Vis-NIR spectroscopy to achieve the discrimination of transgenic Canola (Brassica napus L.) and their hybrids. The results showed that in the identification of transgenic Canola (Brassica napus L.) and their hybrids with B. rapa, the accuracy rate of the combination of SG and SVM reached 100%, and the classification accuracy using CNN combined with Normalization was 98.9%, and in the identification of transgenic Canola (Brassica napus L.) and their hybrids with B. juncea, the accuracy of the combination of SNV and SVM reached 99.4%, while the classification accuracy using CNN combined with SG reached 99.1% [11,28]. Lijuan Xie et al. used NIRS technology to detect the transgenic tomato and its parents’ leaves by Vis/NIR spectroscopy. The effects of different spectral pretreatments (MSC, 1st, and 2nd derivatives) and modeling methods (SIMCA, DPLS, DA, PLSDA, etc.) on the identification results were compared and analyzed. The quantitative models of chlorophyll and ethylene synthesis in tomato leaves were also developed, and the size of the leaves and the content of chlorophyll and ethylene were measured as the wet lab chemistry values; the results showed that NIRS could provide a theoretical basis for the rapid identification of transgenic agricultural products [14,15,35,57,58].
In the detection of cereal food, Wenchao Zhu et al. applied Vis/NIR spectroscopy to achieve rapid identification of transgenic rice leaves and rapid detection of chlorophyll content (SPAD). The spectra were processed using MSC and OSC preprocessing methods, and the SPA algorithm was applied to extract the effective wavelengths; the correct rate of the prediction set of the SPA-LS-SVM model reached 87.27%. In the quantitative analysis, the measured SPAD values were used as the wet lab chemistry values during modeling. Finally, the optimal SPAD value prediction model was SPA-LS-SVM with a correlation coefficient and a root mean square error of prediction (RMSEP) of 0.9022 and 1.3121, respectively [17]. Takefumi Hattori et al. developed a rapid NIRS method to predict lignin and starch contents and the enzymatic saccharification efficiency of transgenic rice straw; the content of lignin and starch contents and enzymatic saccharification showed the difference between the transgenic and non-transgenic rice straw. The results showed a strong correlation between laboratory wet chemistry values and NIR predictions [34]. Long Zhang et al. used NIRS combined with PLS-DA to distinguish transgenic rice. SNV pretreatment and the PLS-DA modeling algorithm were used, and the results showed that transgenic rice and wild-type rice, transgenic rice TCTP, and mi166 could be distinguished from each other; the correct classification rate was 100.0% [59]. Hao Yong et al. used near-infrared diffuse reflectance spectroscopy to identify transgenic (BT63)/non-transgenic rice varieties. Spectral preprocessing methods such as Norris-Williams smoothing, SNV, MSC, and SG 1st-der were used to effectively reduce spectral noise and enhance spectral information. Multivariate correction methods such as PCA, PLS-DA, and SVM were used, the results showed that the combination of SG-1st-Der pre-processing and SVM provided the optimal model for distinguishing different rice varieties. The SNV-SVM model, MSC-SVM model, and SG 1st-Der-PLS-DA model all achieved good analysis results with 100% accuracy [60]. Mayara Macedo da Mata et al. proposed a novel approach to identify genetically modified organisms using PLS-DA based on NIR and Raman data to differentiate between conventional and transgenic cottonseed genotypes. The results showed that the classification errors of the prediction sets were 2.23% and 0.0%, respectively [29]. Jin Hwan Lee et al. used NIR reflectance spectroscopy for the nondestructive detection of herbicide-resistant transgenic soybean seeds, and the PLSDA model with second-derivative preprocessing of the raw spectra had the best calibration and prediction ability, with 97% accuracy [30]. In addition, Jiang Wu et al. investigated the feasibility of the nondestructive identification of transgenic soybean using NIR spectroscopy. SNV pretreatment and PCA combined with the BPNN method were applied for analysis and identification. The results showed that the model had a 100% correct identification rate for transgenic soybean [33]. Xuping Feng et al. used NIR spectra for the screening of transgenic maize. The SG smoothing preprocessing method, combined with three variable selection algorithms of weighted regression coefficients, principal component analysis-loading, and second-order derivatives, were used to extract the feature wavelengths. Five classification methods, KNN, SIMCA, NBC, ELM, and RBFNN, were used to establish the identification models. The results showed that the full-spectrum classification rate was 100% and the feature wavelength classification rate was 90.83% in the ELM model [27]. Cheng Peng et al. used near-infrared spectroscopy to investigate the trans-bivalent gene (cry1Ab/cry2Aj-G10evo) maize kernels and maize flour. The SG smoothing algorithm was used to remove noise from the extracted spectral data. The results showed that in the model based on PCA, the accuracy of the SVM model set and the prediction set was 100%. The accuracy of the SVM model based on full-band spectra of transgenic maize powder was 90.625% [12]. Haosong Guo et al. applied the SG method and the moving window wavelength selection method to the combined model of PCA and LDA. A spectral pattern recognition method based on SG preprocessing (MW-PCA-LDA) was proposed and successfully applied to the Vis-NIR identification of transgenic sugarcane leaves. The corresponding validation recognition rates of transgenic and non-transgenic samples achieved 99.1% and 98.0%, respectively [61]. At the same time, Guisong Liu et al. applied SG smoothing, PCA combined with supervised LDAm and unsupervised systematic cluster analysis (HCA) to the Vis-NIR nondestructive detection of transgenic sugarcane breeding. The results showed that the recognition rates of the optimal SG-PCA-LDA model for positive and negative samples were 94.3% and 96.0%, respectively, and those of the optimal SG-PCA-HCA model for positive and negative samples were 92.5% and 98.0%, respectively [62]. Yafeng Zhai et al. proposed a method for the rapid identification of transgenic wheat seeds using NIRS. The raw data were firstly pre-processed with normalization, and then a model for identification of wheat varieties was developed using PCA combined with the biomimetic pattern recognition method [63].
Some researchers have applied NIRS to detect oil quality. Aderval S. Luna et al. applied NIR spectroscopy and multivariate classification techniques to distinguish non-GMO and GMO soybean oil samples. Pre-processing methods such as MC, MSC, OSC, SG 1st, and 2nd derivatives were used to develop SVM-DA and PLS-DA models. The results showed that the classification rate of SVM-DA was 100% in the training group and 100% and 90% in the validation group for non-GMO and GMO soybean oil samples. In the PLS-DA model, the classification rates were 95% and 100% for the training group and 100% and 80% for the validation group of non-GMO and GM soybean oil samples, respectively [64]. Jianguo Zhu et al. used NIRS to study the mixed solution of transgenic oil and non-transgenic oil. The original spectra were preprocessed by MSC, FD, MWS, SG1, etc. The feature wavelength was extracted by SPA, and the SVM model was established. The result showed that the model preprocessed by MSC had the best prediction performance, with an accuracy rate of 91.6%. The prediction accuracy of SVM was improved to 98.3% by using the SPA algorithm [31].
From the introduction of the above research, it can be seen that NIRS-based transgenic identification does not require sample processing, and it is simple and fast compared with DNA-and protein-based transgenic identification. In addition, from the above review, research on vegetables and cereals accounts for the majority. Studies on transgenic vegetables have mainly focused on tomatoes, including tomatoes, tomato leaves, and the chlorophyll content. In research on cereal, the detection of corn and rice has been a relatively large proportion. The selection of these as research objects is mainly related to the fact that they are the main food crops at present and the popularization of transgenic technology.

4. The Principles and Characteristics of Hyperspectral Imaging Technique

The hyperspectral imaging technique (HSI) is an emerging technology that combines traditional two-dimensional imaging with spectral technique. Compared to NIRS, HSI is characterized by its multiband, high resolution, wide spectral range, and the unity of spectrum and image [65]. HSI combines image and spectral information, which can collect the spectral information and spatial information of the object at the same time, and can obtain the internal and external information of the sample in a larger range [66], reflecting not only the apparent traits such as color, shape, size, but also the internal traits such as chemical components and physiological structure of the object. After acquiring the hyperspectral image, the image data is analyzed and processed as Figure 3. It can be seen that, in general, the final processing of this method is still back to the processing of spectral data, only one difference from the NIRS is that it extracts the spectra of the region of interest.

5. The Applications of HSI for the Detection of Transgenic Agricultural Products and Foods

Hyperspectral imaging technology, which combines image and spectral information, is being increasingly used in the identification of agricultural products/foods [42,65,66,67,68] and in nondestructive testing of the quality of meat [69], fruits [45], etc. Taking maize as an example, this method has a series of analysis and application in purity variety identification [70,71], vigor detection [72,73], internal components such as moisture [74,75], oil [76], etc., hardness [77,78], mycotoxin [79,80], and freeze injury of maize seeds [4,33].
At the same time, hyperspectral imaging has also been used in the identification of genetic modification in food and agricultural products, as shown in Table 3.
Priscilla Dantas Rocha et al. proposed a NIR-based hyperspectral imaging technique to differentiate between conventional and transgenic cotton seeds. A cotton seed pixel-based and two individual cotton chemometric models were developed using SNV, the SG smoothing preprocessing method, and modeling using PLS-DA. The results showed that the specificity and sensitivity values of the different methods ranged from 0.78–0.92 and 0.62–0.93, respectively [81]. Xuping Feng et al. used NIR hyperspectral imaging and multivariate data analysis to differentiate transgenic maize kernels by using WT, MSC, SNV preprocessing methods and constructing discriminant models by applying PCA, SVM, and PLS-DA to classify transgenic maize. The results showed that there were significant differences between transgenic and non-transgenic maize, and the SVM and PLS-DA models could obtain good performance with almost 100% accuracy [3], and also implemented the detection of shikimic acid concentrations in transgenic maize using hyperspectral imaging with the pretreatment method of SNV, MSC, WT, SG smoothing and chemometric methods (PLS-DA, PLSR, SPA, RF). The leaf shikimic acid concentration and the content of chlorophyll a and b were measured as the wet lab chemistry values during modeling. The results showed that a PLSR model based on the optimal wavelength was effective in predicting shikimic acid concentrations in transgenic maize with a coefficient of determination value of 0.79 for the calibration set, and a coefficient of determination of 0.82 for the prediction set [82]. Hailong Wang et al. used NIR hyperspectral imaging combined with chemometric methods to explore three different non-transgenic parents (HC6, JACK, TL1) and their transgenic soybean. The spectral data were analyzed using moving average (MA) smoothing preprocessing. A soybean detection model was developed using PLS-DA, and the results showed that hyperspectral imaging techniques could be used for the identification of non-GM soybeans [83]. Xuping Feng et al. proposed a rapid and effective method to identify HD wild-type and CRISPR/Cas9 mutant rice seeds by building RBFNN, ELM, and KNN models based on the full spectrum and feature wavelengths, respectively. The results showed that the RBFNN models based on the 24 feature bands extracted from the 2nd Derivative correctly achieved 92.25% and 89.50% of the modeling set and prediction set, respectively [84].
From the above review, the number of research papers on the application of hyperspectral imaging technology in transgenic agricultural products and food is less than that of NIRS, which may be related to the characteristics of hyperspectral imaging technology. Although the data collected by this technology includes image information and spectral information, the image information has not been used much in the research of transgenic products. Thus, researchers prefer to focus on the NIRS technology, which is relatively simpler to operate and requires less computation. However, overall, the technology has certain application in the identification of corn, soybeans, and cotton and other mainstream genetically modified products.
In fact, in addition to the above two techniques, terahertz spectroscopy and mid-infrared spectroscopy, are also used for the detection of transgenic products; e.g., Feiyu Lian et al. used terahertz spectroscopy and PCA-SVM to identify transgenic components in maize, and the results showed that the accuracy of sample identification was nearly 92.08% [85]. Jianjun Liu et al. realized the identification of transgenic cotton based on terahertz spectroscopy. The results showed that the recognition rate of the proposed method was 98.3% [86]. Xiaochen Shen et al. identified three transgenic and non-transgenic cotton seeds based on terahertz spectroscopy [87], and Hui Fang et al. identified transgenic soybeans based on mid-infrared spectroscopy, and used PLS-DA to identify and analyze the spectroscopy. The results showed that the correct discriminations of the modeling and prediction sets of the three soybean species were more than 80% and 75% [88]. Xiaodan Liu et al. applied mid-infrared spectroscopy to identify genetically modified maize; WT was used to preprocess the data, the SPA algorithm was used to select optimal wavelengths for the chemometrics analysis of PLS-DA, KNN, and ELM. The overall results indicated that a 100% recognition rate in the calibration set and a 98.75% recognition rate in the prediction set were obtained using the ELM model. The use of these non-destructive techniques has laid the foundation for the production and testing of transgenic products [89].

6. Conclusions and Future Perspectives

The development of transgenic technology has attracted the attention of many countries, representing a strategic choice to support current and future. However, the use of GM technology may have unintended negative effects on food and environmental safety, therefore it is a very necessary and important task to identify GM products. The identification of products that have been genetically modified is now an area of study that merits consideration on a global scale. However, conventional methods of detection based on DNA or proteins have the disadvantages of being time-consuming and destructive, while NIR spectroscopy and its imaging techniques offer a variety of uses in the agricultural and food sectors.
These two methods have the advantages of being environmental-friendly, easy to use, quick testing, etc.; thus, more scholars have applied them in the field of agriculture and food. However, NIRS and its imaging techniques are model-reliable, and the collected data is affected by environmental factors. Thus, some processing methods need to be introduced to eliminate the influence of the environment; otherwise, the established model may be unstable. With the development of data processing methods, deep learning methods may be more widely used in spectral processing.
According to the studies described above, NIRS and its imaging techniques combined with chemometrics tools have been applied to the detection of transgenic products, and reliable results have been obtained. Therefore, this technology is expected to become an alternative to the traditional lossy method, used in the application of transgenic agricultural and food products in future life production. In order to better serve life and make these two technologies available to farmers or manufacturers, portable and online detection may become possible, making transgenic detection technology more advanced.

Author Contributions

Conceptualization: J.Z. and R.C.; Writing–original draft preparation: J.Z., Y.P., J.W. and B.T. Writing–review and editing: J.Z., Z.L. and L.D.; Supervision: Z.L., S.Y. and R.C.; Project Administration and Funding Acquisition: J.Z. All authors have read, revised and approved the final manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the project of Jiaxing Nanhu University (No. 70500000/050) and the student research training project (No. 8517223023).

Data Availability Statement

No data available.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Sohn, S.; Pandian, S.; Oh, Y.; Zaukuu, J.Z.; Kang, H.; Ryu, T.; Cho, W.; Cho, Y.; Shin, E.; Cho, B. An overview of near infrared spectroscopy and its applications in the detection of genetically modified organisms. Int. J. Mol. Sci. 2021, 22, 9940. [Google Scholar] [CrossRef]
  2. Alishahi, A.; Farahmand, H.; Prieto, N.; Cozzolino, D. Identification of transgenic foods using nir spectroscopy: A review. Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 2010, 75, 1–7. [Google Scholar] [CrossRef]
  3. Feng, X.; Zhao, Y.; Zhang, C.; Cheng, P.; He, Y. Discrimination of transgenic maize kernel using nir hyperspectral imaging and multivariate data analysis. Sensors 2017, 17, 1894. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  4. Zhang, J.; Dai, L.; Cheng, F. Classification of frozen corn seeds using hyperspectral vis/nir reflectance imaging. Molecules 2019, 24, 149. [Google Scholar] [CrossRef] [Green Version]
  5. Candolfi, A.; De Maesschalck, R.; Jouan-Rimbaud, D.; Hailey, P.A.; Massart, D.L. The influence of data pre-processing in the pattern recognition of excipients near-infrared spectra. J. Pharm. Biomed. Anal. 1999, 21, 115–132. [Google Scholar] [CrossRef] [PubMed]
  6. Liu, D.; Sun, D.; Zeng, X. Recent advances in wavelength selection techniques for hyperspectral image processing in the food industry. Food Bioprocess Technol. 2014, 7, 307–323. [Google Scholar] [CrossRef]
  7. Ramirez, C.A.M.; Greenop, M.; Ashton, L.; Rehman, I.U. Applications of machine learning in spectroscopy. Appl. Spectrosc. Rev. 2021, 8–10, 733–763. [Google Scholar] [CrossRef]
  8. Lee, L.C.; Liong, C.; Jemain, A.A. A contemporary review on data preprocessing (dp) practice strategy in atr-ftir spectrum. Chemom. Intell. Lab. Syst. 2017, 163, 64–75. [Google Scholar] [CrossRef]
  9. Jiao, Y.; Li, Z.; Chen, X.; Fei, S. Preprocessing methods for near-infrared spectrum calibration. J. Chemom. 2020, 34, e3306. [Google Scholar] [CrossRef]
  10. Rinnan, A.; Berg, F.V.D.; Engelsen, S.B. Review of the most common pre-processing techniques for near-infrared spectra. TrAC Trends Anal. Chem. 2009, 28, 1201–1222. [Google Scholar] [CrossRef]
  11. Sohn, S.; Pandian, S.; Oh, Y.; Zaukuu, J.Z.; Na, C.; Lee, Y.; Shin, E.; Kang, H.; Ryu, T.; Cho, W.; et al. Vis-NIR Spectroscopy and Machine Learning Methods for the Discrimination of Transgenic Brassica napus L. and Their Hybrids with B. juncea. Processes 2022, 10, 240. [Google Scholar] [CrossRef]
  12. Cheng, P.; Xuping, F.; Yong, H.; Chu, Z.; Yiying, Z.; Junfeng, X. Discrimination of transgenic maize containing the cry1ab/cry2aj and g10evo genes using near infrared spectroscopy (nir). Spectrosc. Spectr. Anal. 2018, 38, 1095–1100. [Google Scholar]
  13. Workman, J.J. Review of Process and Non-invasive Near-Infrared and Infrared Spectroscopy: 1993–1999. Appl. Spectrosc. Rev. 1999, 34, 1–89. [Google Scholar] [CrossRef]
  14. Xie, L.; Ying, Y.; Ying, T. Combination and comparison of chemometrics methods for identification of transgenic tomatoes using visible and near-infrared diffuse transmittance technique. J. Food Eng. 2007, 82, 395–401. [Google Scholar] [CrossRef]
  15. Xie, L.; Ying, Y.; Ying, T.; Yu, H.; Fu, X. Discrimination of transgenic tomatoes based on visible/near-infrared spectra. Anal. Chim. Acta 2007, 584, 379–384. [Google Scholar] [CrossRef]
  16. Wold, S.; Antti, H.; Lindgren, F.; öhman, J. Orthogonal signal correction of near-infrared spectra. Chemom. Intell. Lab. Syst. 1998, 44, 175–185. [Google Scholar] [CrossRef]
  17. Wenchao, Z.; Fang, C. Analysis of transgenic and non-transgenic rice leaves using visible/near-infrared spectroscopy. Spectrosc. Spectr. Anal. 2012, 32, 370–373. [Google Scholar]
  18. Hashimoto, A.; Kameoka, T. Applications of infrared spectroscopy to biochemical, food, and agricultural processes. Appl. Spectrosc. Rev. 2008, 43, 416–451. [Google Scholar] [CrossRef]
  19. Zinia Zaukuu, J.; Aouadi, B.; Lukács, M.; Bodor, Z.; Vitális, F.; Gillay, B.; Gillay, Z.; Friedrich, L.; Kovacs, Z. Detecting low concentrations of nitrogen-based adulterants in whey protein powder using benchtop and handheld nir spectrometers and the feasibility of scanning through plastic bag. Molecules 2020, 25, 2522. [Google Scholar] [CrossRef]
  20. Dai, Q.; Cheng, J.H.; Sun, D.W.; Zeng, X.A. Advances in feature selection methods for hyperspectral image processing in food industry applications: A review. Crit. Rev. Food Sci. Nutr. 2015, 55, 1368–1382. [Google Scholar] [CrossRef]
  21. Li, H.; Liang, Y.; Xu, Q.; Cao, D. Model population analysis for variable selection. J. Chemom. 2010, 24, 418–423. [Google Scholar] [CrossRef]
  22. Li, H.; Liang, Y.; Xu, Q.; Cao, D. Key wavelengths screening using competitive adaptive reweighted sampling method for multivariate calibration. Anal. Chim. Acta 2009, 648, 77–84. [Google Scholar] [CrossRef] [PubMed]
  23. Niazi, A.; Leardi, R. Genetic algorithm in chemometrics. J. Chemom. 2012, 26, 345–351. [Google Scholar] [CrossRef]
  24. Depczynski, U.; Frost, V.J.; Molt, K. Genetic algorithms applied to the selection of factors in principal component regression. Anal. Chim. Acta 2000, 420, 217–227. [Google Scholar] [CrossRef]
  25. Araújo, M.C.U.; Saldanha, T.C.B.; Galvão, R.K.H.; Yoneyama, T.; Chame, H.C.; Visani, V. The successive projections algorithm for variable selection in spectroscopic multicomponent analysis. Chemom. Intell. Lab. Syst. 2001, 57, 65–73. [Google Scholar] [CrossRef]
  26. Centner, V.; Massart, D.L.; de Noord, O.E.; de Jong, S.; Vandeginste, B.M.; Sterna, C. Elimination of uninformative variables for multivariate calibration. Anal. Chem. 1996, 68, 3851–3858. [Google Scholar] [CrossRef]
  27. Feng, X.; Yin, H.; Zhang, C.; Peng, C.; He, Y. Screening of transgenic maize using near infrared spectroscopy and chemometric techniques. Span. J. Agric. Res. 2018, 16, e203. [Google Scholar] [CrossRef]
  28. Sohn, S.; Pandian, S.; Zaukuu, J.Z.; Oh, Y.; Park, S.; Na, C.; Shin, E.; Kang, H.; Ryu, T.; Cho, W.; et al. Discrimination of transgenic canola (brassica napus L.) And their hybrids with b. Rapa using vis-nir spectroscopy and machine learning methods. Int. J. Mol. Sci. 2022, 23, 220. [Google Scholar] [CrossRef]
  29. Mata, M.M.D.; Rocha, P.D.; Farias, I.K.T.D.; Silva, J.L.B.D.; Medeiros, E.P.; Silva, C.S.; Simões, S.D.S. Distinguishing cotton seed genotypes by means of vibrational spectroscopic methods (nir and raman) and chemometrics. Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 2022, 266, 120399. [Google Scholar] [CrossRef]
  30. Lee, J.H.; Choung, M. Nondestructive determination of herbicide-resistant genetically modified soybean seeds using near-infrared reflectance spectroscopy. Food Chem. 2011, 126, 368–373. [Google Scholar] [CrossRef]
  31. Jianguo, Z.; Yajing, W.; Zhiqin, Y.; Leiying, X.; Na, W.; Duo, C. Research on fast identification of transgenic oil based on near infrared spectroscopy. Opt. Instrum. 2020, 42, 61–66. [Google Scholar]
  32. Zhang, J.; Dai, L.; Cheng, F. Identification of corn seeds with different freezing damage degree based on hyperspectral reflectance imaging and deep learning method. Food Anal. Meth. 2021, 14, 389–400. [Google Scholar] [CrossRef]
  33. Jiang, W.; Furong, H.; Caihuan, H.; Jun, Z.; Xingdan, C. Study on near infrared spectroscopy of transgenic soybean identification based on principal component analysis and neural network. Spectrosc. Spectr. Anal. 2013, 33, 1537–1541. [Google Scholar]
  34. Hattori, T.; Murakami, S.; Mukai, M.; Yamada, T.; Hirochika, H.; Ike, M.; Tokuyasu, K.; Suzuki, S.; Sakamoto, M.; Umezawa, T. Rapid analysis of transgenic rice straw using near-infrared spectroscopy. Plant Biotechnol. 2012, 29, 359–366. [Google Scholar] [CrossRef] [Green Version]
  35. Xie, L.; Ying, Y.; Ying, T. Rapid determination of ethylene content in tomatoes using visible and short-wave near-infrared spectroscopy and wavelength selection. Chemom. Intell. Lab. Syst. 2009, 97, 141–145. [Google Scholar] [CrossRef]
  36. Lecun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436–444. [Google Scholar] [CrossRef]
  37. Yang, J.; Xu, J.; Zhang, X.; Wu, C.; Lin, T.; Ying, Y. Deep learning for vibrational spectral analysis: Recent progress and a practical guide. Anal. Chim. Acta 2019, 1081, 6–17. [Google Scholar] [CrossRef]
  38. Zhang, X.; Lin, T.; Xu, J.; Luo, X.; Ying, Y. Deepspectra: An end-to-end deep learning approach for quantitative spectral analysis. Anal. Chim. Acta 2019, 1058, 48–57. [Google Scholar] [CrossRef]
  39. Zhang, X.; Yang, J.; Lin, T.; Ying, Y. Food and agro-product quality evaluation based on spectroscopy and deep learning: A review. Trends Food Sci. Technol. 2021, 112, 431–441. [Google Scholar] [CrossRef]
  40. Prieto, N.; Roehe, R.; Lavín, P.; Batten, G.; Andrés, S. Application of near infrared reflectance spectroscopy to predict meat and meat products quality: A review. Meat Sci. 2009, 83, 175–186. [Google Scholar] [CrossRef]
  41. Qu, J.H.; Liu, D.; Cheng, J.H.; Sun, D.W.; Ma, J.; Pu, H.; Zeng, X.A. Applications of near-infrared spectroscopy in food safety evaluation and control: A review of recent research advances. Crit. Rev. Food Sci. Nutr. 2015, 55, 1939–1954. [Google Scholar] [CrossRef]
  42. Fu, X.; Ying, Y. Food safety evaluation based on near infrared spectroscopy and imaging: A review. Crit. Rev. Food Sci. Nutr. 2016, 56, 1913–1924. [Google Scholar] [CrossRef]
  43. Huang, H.; Yu, H.; Xu, H.; Ying, Y. Near infrared spectroscopy for on/in-line monitoring of quality in foods and beverages: A review. J. Food Eng. 2008, 87, 303–313. [Google Scholar] [CrossRef]
  44. Nicolaï, B.M.; Beullens, K.; Bobelyn, E.; Peirs, A.; Saeys, W.; Theron, K.I.; Lammertyn, J. Nondestructive measurement of fruit and vegetable quality by means of nir spectroscopy: A review. Postharvest Biol. Technol. 2007, 46, 99–118. [Google Scholar] [CrossRef]
  45. Chandrasekaran, I.; Panigrahi, S.S.; Ravikanth, L.; Singh, C.B. Potential of near-infrared (nir) spectroscopy and hyperspectral imaging for quality and safety assessment of fruits: An overview. Food Anal. Methods 2019, 12, 2438–2458. [Google Scholar] [CrossRef]
  46. Ge, W.; Zhang, L.; Li, X.; Zhang, C.; Sun, M.; An, D.; Wu, J. Applying multimodal data fusion based on manifold learning with nuclear magnetic resonance (nmr) and near infrared spectroscopy (nirs) to maize haploid identification. Biosyst. Eng. 2021, 210, 299–309. [Google Scholar] [CrossRef]
  47. Cui, Y.; Xu, L.; An, D.; Zhang, X.; Zhu, D. Identification of maize seed varieties based on near infrared reflectance spectroscopy and chemometrics. Int. J. Agric. Biol. Eng. 2018, 11, 177–183. [Google Scholar] [CrossRef]
  48. Qiu, G.; Lü, E.; Lu, H.; Xu, S.; Zeng, F.; Shui, Q. Single-kernel ft-nir spectroscopy for detecting supersweet corn (Zea mays L. Saccharata sturt) seed viability with multivariate data analysis. Sensors 2018, 18, 1010. [Google Scholar] [CrossRef] [Green Version]
  49. Egesel, C.Ö.; Kahrıman, F.; Ekinci, N.; Kavdır, O.; Büyükcan, M.B. Analysis of fatty acids in kernel, flour, and oil samples of maize by nir spectroscopy using conventional regression methods. Cereal Chem. J. 2016, 93, 487–492. [Google Scholar] [CrossRef]
  50. Rosales, A.; Galicia, L.; Oviedo, E.; Islas, C.; Palacios-Rojas, N. Near-infrared reflectance spectroscopy (nirs) for protein, tryptophan, and lysine evaluation in quality protein maize (qpm) breeding programs. J. Agric. Food Chem. 2011, 59, 10781–10786. [Google Scholar] [CrossRef] [PubMed]
  51. Zhong, J.; Qin, X. Rapid quantitative analysis of corn starch adulteration in konjac glucomannan by chemometrics-assisted ft-nir spectroscopy. Food Anal. Methods 2016, 9, 61–67. [Google Scholar] [CrossRef]
  52. Levasseur-Garcia, C.; Bailly, S.; Kleiber, D.; Bailly, J.; Pan, L. Assessing risk of fumonisin contamination in maize using near-infrared spectroscopy. J. Chem. 2015, 2015, 485864. [Google Scholar] [CrossRef] [Green Version]
  53. Darnell, R.E.; Harvey, J.J.; Fox, G.P.; Fletcher, M.T.; Wainaina, J.; Wanjuki, I.; Turner, W.J. Nirs calibration of aflatoxin in maize. Aust. J. Chem. 2018, 71, 868. [Google Scholar] [CrossRef]
  54. Jia, S.; Yang, L.; An, D.; Liu, Z.; Yan, Y.; Li, S.; Zhang, X.; Zhu, D.; Gu, J. Feasibility of analyzing frost-damaged and non-viable maize kernels based on near infrared spectroscopy and chemometrics. J. Cereal Sci. 2016, 69, 145–150. [Google Scholar] [CrossRef]
  55. Rady, A.; Adedeji, A. Assessing different processed meats for adulterants using visible-near-infrared spectroscopy. Meat Sci. 2018, 136, 59–67. [Google Scholar] [CrossRef]
  56. Pullanagari, R.R.; Yule, I.J.; Agnew, M. On-line prediction of lamb fatty acid composition by visible near infrared spectroscopy. Meat Sci. 2015, 100, 156–163. [Google Scholar] [CrossRef]
  57. Xie, L.; Ying, Y.; Ying, T. Quantification of chlorophyll content and classification of nontransgenic and transgenic tomato leaves using visible/near-infrared diffuse reflectance spectroscopy. J. Agric. Food Chem. 2007, 55, 4645–4650. [Google Scholar] [CrossRef]
  58. Xie, L.; Ying, Y.; Ying, T.; Tian, H.; Niu, X.; Fu, X. Application of vis/nir diffuse reflectance spectroscopy to the detection and identification of transgenic tomato leaf. Spectrosc. Spectr. Anal. 2008, 28, 1062–1066. [Google Scholar]
  59. Long, Z.; Shan-Shan, W.; Yan-Fei, D.; Jia-Rong, P.; Cheng, Z. Discrimination of transgenic rice based on near infrared reflectance spectroscopy and partial least squares regression discriminant analysis. Rice Sci. 2015, 22, 245–249. [Google Scholar] [CrossRef] [Green Version]
  60. Hao, Y.; Geng, P.; Wu, W.; Wen, Q.; Rao, M. Identification of rice varieties and transgenic characteristics based on near-infrared diffuse reflectance spectroscopy and chemometrics. Molecules 2019, 24, 4568. [Google Scholar] [CrossRef] [Green Version]
  61. Guo, H.; Chen, J.; Pan, T.; Wang, J.; Cao, G. Vis-nir wavelength selection for non-destructive discriminant analysis of breed screening of transgenic sugarcane. Anal. Methods 2014, 6, 8810–8816. [Google Scholar] [CrossRef]
  62. Guisong, L.; Haosong, G.; Tao, P.; Wang, J.; Cao, G. Vis-nir spectroscopic pattern recognition combined with sg smoothing applied to breed screening of transgenic sugarcane. Spectrosc. Spectr. Anal. 2014, 34, 2701–2706. [Google Scholar]
  63. Yafeng, Z.; Qian, S.; Wenjin, W.; Zhentian, H.; Zongying, Z.; Jiashuang, A.; Jin, D.; Xin, D.; Chenggui, H.; Jialin, Y.; et al. Fast discrimination of varieties of transgene wheat based on bionimetic pattern recognition and near infrared spectra. Spectrosc. Spectr. Anal. 2010, 30, 924–928. [Google Scholar]
  64. Luna, A.S.; Da Silva, A.P.; Pinho, J.S.A.; Ferré, J.; Boqué, R. Rapid characterization of transgenic and non-transgenic soybean oils by chemometric methods using nir spectroscopy. Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 2013, 100, 115–119. [Google Scholar] [CrossRef]
  65. Gowen, A.; Odonnell, C.; Cullen, P.; Downey, G.; Frias, J. Hyperspectral imaging–an emerging process analytical tool for food quality and safety control. Trends Food Sci. Technol. 2007, 18, 590–598. [Google Scholar] [CrossRef]
  66. Mahesh, S.; Jayas, D.S.; Paliwal, J.; White, N.D.G. Hyperspectral imaging to classify and monitor quality of agricultural materials. J. Stored Prod. Res. 2015, 61, 17–26. [Google Scholar] [CrossRef]
  67. Qin, J.; Chao, K.; Kim, M.S.; Lu, R.; Burks, T.F. Hyperspectral and multispectral imaging for evaluating food safety and quality. J. Food Eng. 2013, 118, 157–171. [Google Scholar] [CrossRef]
  68. Dale, L.M.; Thewis, A.; Boudry, C.; Rotar, I.; Dardenne, P.; Baeten, V.; Pierna, J.A.F. Hyperspectral imaging applications in agriculture and agro-food product quality and safety control: A review. Appl. Spectrosc. Rev. 2013, 48, 142–159. [Google Scholar] [CrossRef]
  69. Kamruzzaman, M.; Elmasry, G.; Sun, D.; Allen, P. Prediction of some quality attributes of lamb meat using near-infrared hyperspectral imaging and multivariate analysis. Anal. Chim. Acta 2012, 714, 57–67. [Google Scholar] [CrossRef]
  70. Zhang, J.; Dai, L.; Cheng, F. Corn seed variety classification based on hyperspectral reflectance imaging and deep convolutional neural network. J. Food Meas. Charact. 2021, 15, 484–494. [Google Scholar] [CrossRef]
  71. Wang, L.; Sun, D.; Pu, H.; Zhu, Z. Application of hyperspectral imaging to discriminate the variety of maize seeds. Food Anal. Methods 2016, 9, 225–234. [Google Scholar] [CrossRef]
  72. Wakholi, C.; Kandpal, L.M.; Lee, H.; Bae, H.; Park, E.; Kim, M.S.; Mo, C.; Lee, W.; Cho, B. Rapid assessment of corn seed viability using short wave infrared line-scan hyperspectral imaging and chemometrics. Sens. Actuators B Chem. 2018, 255, 498–507. [Google Scholar] [CrossRef]
  73. Ambrose, A.; Kandpal, L.M.; Kim, M.S.; Lee, W.; Cho, B. High speed measurement of corn seed viability using hyperspectral imaging. Infrared Phys. Technol. 2016, 75, 173–179. [Google Scholar] [CrossRef]
  74. Wang, Z.; Fan, S.; Wu, J.; Zhang, C.; Xu, F.; Yang, X.; Li, J. Application of long-wave near infrared hyperspectral imaging for determination of moisture content of single maize seed. Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 2021, 254, 119666. [Google Scholar] [CrossRef] [PubMed]
  75. Huang, M.; Zhao, W.; Wang, Q.; Zhang, M.; Zhu, Q. Prediction of moisture content uniformity using hyperspectral imaging technology during the drying of maize kernel. Int. Agrophys. 2015, 29, 39–46. [Google Scholar] [CrossRef] [Green Version]
  76. Weinstock, B.A.; Janni, J.; Hagen, L.; Wright, S. Prediction of oil and oleic acid concentrations in individual corn (Zea mays L.) Kernels using near-infrared reflectance hyperspectral imaging and multivariate analysis. Appl. Spectrosc. 2006, 60, 9–16. [Google Scholar] [CrossRef]
  77. Qiao, M.; Xu, Y.; Xia, G.; Su, Y.; Lu, B.; Gao, X.; Fan, H. Determination of hardness for maize kernels based on hyperspectral imaging. Food Chem. 2022, 366, 130559. [Google Scholar] [CrossRef]
  78. Williams, P.J.; Kucheryavskiy, S. Classification of maize kernels using nir hyperspectral imaging. Food Chem. 2016, 209, 131–138. [Google Scholar] [CrossRef]
  79. Zhang, J.; Xu, B.; Wang, Z.; Cheng, F. Application of hyperspectral imaging in the detection of aflatoxin b1 on corn seed. J. Food Meas. Charact. 2022, 16, 448–460. [Google Scholar] [CrossRef]
  80. Tao, F.; Yao, H.; Hruska, Z.; Kincaid, R.; Rajasekaran, K.; Bhatnagar, D. A novel hyperspectral-based approach for identification of maize kernels infected with diverse aspergillus flavus fungi. Biosyst. Eng. 2020, 200, 415–430. [Google Scholar] [CrossRef]
  81. Rocha, P.D.; Medeiros, E.P.; Silva, C.S.; Da Silva Simões, S. Chemometric strategies for near infrared hyperspectral imaging analysis: Classification of cotton seed genotypes. Anal. Methods 2021, 13, 5065–5074. [Google Scholar] [CrossRef]
  82. Feng, X.; Yu, C.; Chen, Y.; Peng, J.; Ye, L.; Shen, T.; Wen, H.; He, Y. Non-destructive determination of shikimic acid concentration in transgenic maize exhibiting glyphosate tolerance using chlorophyll fluorescence and hyperspectral imaging. Front. Plant Sci. 2018, 9, 468. [Google Scholar] [CrossRef]
  83. Wang, H.L.; Yang, X.D.; Zhang, C.; Guo, D.Q.; Bao, Y.D.; He, Y.; Liu, F. Fast identification of transgenic soybean varieties based near infrared hyperspectral imaging technology. Spectrosc. Spectr. Anal. 2016, 36, 1843–1847. [Google Scholar]
  84. Xuping, F.; Cheng, P.; Chu, Z.; Xiaodan, L.; Tingting, S.; Yong, H.; Junfeng, X. A simple and efficient method for crispr/cas9-induced rice mutant screening. Spectrosc. Spectr. Anal. 2018, 38, 570–574. [Google Scholar]
  85. Lian, F.; Xu, D.; Fu, M.; Ge, H.; Jiang, Y.; Zhang, Y. Identification of transgenic ingredients in maize using terahertz spectra. IEEE Trans. Terahertz Sci. Technol. 2017, 7, 378–384. [Google Scholar] [CrossRef]
  86. Liu, J.; Li, Z.; Hu, F.; Chen, T.; Zhu, A.; Du, Y.; Xin, H. Method for identifying transgenic cottons based on terahertz spectra and wlda. Opt. Int. J. Light Electron Opt. 2015, 126, 1872–1877. [Google Scholar] [CrossRef]
  87. Shen, X.; Li, B.; Li, X.; Long, Y. Identification of transgenic and non-transgenic cotton seed based on terahertz range spectroscopy. Trans. Chin. Soc. Agric. Eng. 2017, 33, 288–292. [Google Scholar]
  88. Fang, H.; Zhang, Z.; Wang, H.; Yang, X.; He, Y.; Bao, Y. Identification of transgenic soybean varieties using mid-infrared spectroscopy. Spectrosc. Spectr. Anal. 2017, 37, 760–765. [Google Scholar]
  89. Liu, X.; Yu, Y.; Bai, X.; Li, X.; Zhang, J.; Wang, D. Rapid identification of insecticide- and herbicide-tolerant genetically modified maize using mid-infrared spectroscopy. Processes 2023, 11, 90. [Google Scholar] [CrossRef]
Processes 11 00651 g001 550
Figure 1. The steps of review of the paper.
Figure 1. The steps of review of the paper.
Processes 11 00651 g001
Processes 11 00651 g002 550
Figure 2. The general analyzing steps of NIRS data.
Figure 2. The general analyzing steps of NIRS data.
Processes 11 00651 g002
Processes 11 00651 g003 550
Figure 3. The data analysis procedures of hyperspectral imaging data.
Figure 3. The data analysis procedures of hyperspectral imaging data.
Processes 11 00651 g003
Table
Table 1. The methods of detecting transgenic products.
Table 1. The methods of detecting transgenic products.
MethodsAdvantagesDisadvantages
Protein-based methodsWestern blotReliable resultsDifficult to use, destructive, time-consuming (2 days)
ELISAReliable resultsModerate to use, destructive, time-consuming (30–90 min)
Lateral flow stripSimple to use, quick testing (10 min)destructive
DNA-based methodsSouthern blotReliable resultsDifficult to use, destructive, time-consuming (6 h)
Qualitative PCRReliable resultsDifficult to use, destructive, time-consuming (1.5 h)
Real time PCRReliable resultsDifficult to use, destructive, time-consuming (1 day)
MicroscopyClassical microscopyResults visualizationDifficult to use, destructive, time-consuming (1 day)
ChromatographyHPLS, GC-MSReliable resultsDifficult to use, destructive, time-consuming (1–2 days)
Spectroscopy-based methodsNIRSNon-destructive, Quick testing (Less than 1 min), Easy to useModel-reliable
Hyperspectral ImagingNon-destructive, Quick testing (Less than 5 min), Moderate to useModel-reliable
MIRSNon-destructive Quick testing (Less than 5 min), Easy to useModel-reliable, Not widely used
Terahertz SpectroscopyNon-destructive, Quick testing (Less than 15 min), Moderate to useModel-reliable, Not widely used
ELISA: Enzyme linked immunosorbent assay; DNA: Deoxyribonucleic acid; PCR: Polymerase chain reaction: HPLC: High pressure liquid chromatography; GC-MS: Gas chromatography mass spectroscopy; NIR: Near infrared spectroscopy; MIRS: Mid infrared spectroscopy.
Table
Table 2. Studies on the detection of transgenic agricultural products and foods using near-infrared spectroscopy.
Table 2. Studies on the detection of transgenic agricultural products and foods using near-infrared spectroscopy.
AuthorObjectPreprocessing MethodsModelsResultsReference
Soo-In Sohn et al.Transgenic Brassica napus L.SG, smoothing filter, SNV, NormalizationLDA, CNN, GBT, SVM, RFThe highest accuracy of the combination of SG and SVM was 100%.[28]
Soo-In Sohn et al.Transgenic Brassica napus L.Normalization, SNV, SGLDA, Deep Learning, SVM,
GLM, DT, NB, FLM, RF		99.4% classification accuracy for SNV and SVM, 99.1% classification accuracy for SG and deep learning[11]
Lijuan Xie et al.Transgenic TomatoesMSC, 1st and 2nd derivativesDA, PLS-DAPLS-DA with the classification accuracy of 100%[15]
Lijuan Xie et al.Transgenic TomatoesMSC, SG 1st, 2ndSIMCA, DPLSDPLS with the classification accuracy of 100%[14]
Lijuan Xie et al.Chlorophyll Content of Transgenic Tomato LeavesMSC, 1st and 2nd derivativesPLS-DAPLS-DA with the classification accuracy of 100%[57]
Lijuan Xie et al.Transgenic tomato leafMSC, 1st and 2nd derivativesDA, PLSWith the classification accuracy of 89.7%[58]
Lijuan Xie et al.ethylene content in tomatoesSNV, MSC, 1st and 2nd derivativesPLSR, SMLRPLSR and SMLR can determine the ethylene content in tomato.[35]
Wenchao Zhu et al.Leaves of transgenic rice, SPAD in leafMSC, OSCLS-SVMSPA-LS-SVM method can quickly identify transgenic rice leaves and accurately predict the SPAD value.[17]
Takefumi Hattori et al.Transgenic rice straw1st and 2nd derivatives, SNVPLSRSNV-PLSR obtained a strong correlation between laboratory wet chemistry values and NIR predicted values.[34]
Long Zhang et al.Transgenic RiceSNV, PCAPLS-DAThe correct classification rate of the validation test was 100.0%.[59]
Yong Hao et al.Transgenic RiceNWS, SNV, MSC, SG 1st-DerivativePLS-DA, SVMModel achieved good analytical results with 100% accuracy rate.[60]
Mayara Macedo da Mata et al.Transgenic cottonSNV, 1st derivativePLS-DANIR and Raman prediction sets had classification errors of 2.23% and 0.0%, respectively[29]
Jin Hwan Lee et al.Transgenic soybean1st, 2nd derivativesPLS-DA2nd derivatives and PLSDA had results with 97% accuracy[30]
Jiang Wu et al.Transgenic soybeanSNVBPNNBPNN had 100% identification rate[33]
Xuping Feng et al.Transgenic maizeSG smoothingKNN, SIMCA, NBC, ELM, RBFNNThe classification rates of full-spectrum and the feature wavelength were 100% and 90.83% in ELM model.[27]
Cheng Peng et al.Transgenic maizeSG smoothingPLS, SVMThe accuracy of the SVM model based on full-band spectra of transgenic maize powder was 90.625%.[12]
Haosong Guo et al.Transgenic sugarcaneSG, MWLDAThe corresponding validation recognition rates of transgenic and non-transgenic samples achieved 99.1% and 98.0%, respectively.[61]
Guisong Liu et al.Transgenic sugarcaneSGPCA, LDA, HCAThe optimal SG-PCA-LDA model for positive and negative samples were 94.3% and 96.0%, respectively, and that of the optimal SG-PCA-HCA model for positive and negative samples were 92.5% and 98.0%, respectively.[62]
Yafeng Zhai et al.Transgenic wheatNormalizationBPRA model for identification of wheat varieties was developed using PCA combined with biomimetic pattern recognition method.[63]
Aderval S. Lunaet al.Transgenic soybean oilsMC, MSC, OSC, SG 1st, 2nd derivativesSVM-DA, PLS-DAThe classification rate of SVM-DA was 100% in the training group and 100% and 90% in the validation group for non-GMO and GMO soybean oil samples. In PLS-DA model, the classification rates were 95% and 100% for the training group and 100% and 80% for the validation group of non-GMO and GM soybean oil samples, respectively.[64]
Jianguo Zhu et al.Transgenic oilsMSC, first derivative (FD), MWS, SG1 preprocessingSVMMSC had the best prediction performance with the accuracy rate of 91.6%. The prediction accuracy of SVM was improved to 98.3% by using the SPA algorithm.[31]
Table
Table 3. The studies on the detection of transgenic agricultural products and foods using hyperspectral imaging techniques.
Table 3. The studies on the detection of transgenic agricultural products and foods using hyperspectral imaging techniques.
AuthorObjectPreprocessing MethodsModelsResultsReference
Priscilla Dantas Rocha et al.Transgenic cotton seedSNV, SG smoothingPLS-DAThe specificity and sensitivity values of the different methods ranged from 0.78–0.92 and 0.62–0.93, respectively.[81]
Xuping Feng et al.Transgenic maizeWT, MSC, SNVSVM, PLS-DASVM and PLS-DA models could obtain good performance with almost 100% accuracy.[3]
Xuping Feng et al.Shikimic acid concentration in transgenic maize plantSNV, MSC, WT, SG smoothingPLS-DA, PLSR, RFA coefficient of determination value of 0.79 for the calibration set and a coefficient of determination of 0.82 for the prediction set.[82]
Hailong Wang et al.Transgenic soybeansMAPLS-DAThe results showed that hyperspectral imaging techniques could be used for the identification of non-GM soybeans.[83]
Xuping Feng et al.Transgenic riceWTRBFNN, KNN, ELMThe RBFNN models based on the 24 feature bands extracted from the 2nd derivative achieved the accuracy of 92.25% and 89.50% of the modeling set and prediction set, respectively.[84]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Zhang, J.; Liu, Z.; Pu, Y.; Wang, J.; Tang, B.; Dai, L.; Yu, S.; Chen, R. Identification of Transgenic Agricultural Products and Foods Using NIR Spectroscopy and Hyperspectral Imaging: A Review. Processes 2023, 11, 651. https://doi.org/10.3390/pr11030651

AMA Style

Zhang J, Liu Z, Pu Y, Wang J, Tang B, Dai L, Yu S, Chen R. Identification of Transgenic Agricultural Products and Foods Using NIR Spectroscopy and Hyperspectral Imaging: A Review. Processes. 2023; 11(3):651. https://doi.org/10.3390/pr11030651

Chicago/Turabian Style

Zhang, Jun, Zihao Liu, Yaoyuan Pu, Jiajun Wang, Binman Tang, Limin Dai, Shuihua Yu, and Ruqing Chen. 2023. "Identification of Transgenic Agricultural Products and Foods Using NIR Spectroscopy and Hyperspectral Imaging: A Review" Processes 11, no. 3: 651. https://doi.org/10.3390/pr11030651

APA Style

Zhang, J., Liu, Z., Pu, Y., Wang, J., Tang, B., Dai, L., Yu, S., & Chen, R. (2023). Identification of Transgenic Agricultural Products and Foods Using NIR Spectroscopy and Hyperspectral Imaging: A Review. Processes, 11(3), 651. https://doi.org/10.3390/pr11030651

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop