A high-throughput, rapid, non-destructive quantitative analysis method for Fusarium toxins in single grains of cereals based on near-infrared imaging
Through near-infrared image technology and spectral analysis model, the problem that traditional detection methods cannot quickly detect Fusarium toxins in single grain grains is solved, and fast, non-destructive and high-throughput analysis is achieved, reducing the uneven distribution of Fusarium toxins. food waste and health hazards.
Patent Information
- Application Number
- CN202011492721.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-12-17
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2040-12-17
AI Technical Summary
Traditional Fusarium toxin detection methods cannot meet the needs of rapid, non-destructive and high-throughput during wheat storage and processing, especially the detection of individual grain grains is difficult to achieve, resulting in food waste and health hazards caused by local pollution.
Using near-infrared image technology, grain grains are scanned through near-infrared hyperspectral imaging equipment, near-infrared spectrum patterns are obtained, background deduction and spectral extraction are carried out, and quantitative analysis model is constructed to achieve rapid and non-destructive quantitative analysis of Fusarium toxins in individual grain grains.
The rapid, non-destructive and high-throughput detection of Fusarium toxin content in a single grain grain is achieved, effectively avoiding misjudgment and food waste caused by uneven distribution of Fusarium toxins, and ensuring the safety of grains and human health.
Smart Images

Figure CN114965351B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of analysis and detection, and relates to a high-throughput, rapid, non-destructive quantitative analysis method for Fusarium toxins in single grains of cereals based on near-infrared images. Background Art
[0002] Deoxynivalenol (DON), also known as vomitoxin, is a toxic secondary metabolite produced by Fusarium graminearum complex infecting wheat and other grains, causing fusarium head blight. Vomitoxin not only destroys the cell tissue structure of wheat and reduces flour yield, but also causes vomiting, food refusal, diarrhea, bleeding and even death in livestock that accidentally eat DON-contaminated feed. It is also highly toxic to humans.
[0003] In recent years, affected by extreme climate and other environmental factors, the risk of wheat DON contamination has continued to increase, and has become the main risk factor restricting the quality and safety of wheat products in my country and the world. Timely detection and discovery of DON-contaminated foods can effectively prevent DON-contaminated foods from entering the food chain, thereby posing a threat to human and animal health. Studies have shown that Fusarium toxin contamination is unevenly distributed in grains, and batch sampling and testing cannot reflect the true contamination level of vomitoxin. For example, if most of the grains in the sample are not contaminated or the toxin content is low, but only a few grains have a high toxin content, the toxin test of the entire batch of samples may fail. Traditional DON detection methods include chromatography and immunology. Chromatography methods include high performance liquid chromatography, gas chromatography, and liquid chromatography-tandem mass spectrometry, which have high sensitivity and repeatability and can simultaneously perform quantitative analysis of multiple Fusarium toxins. Immunology methods such as fluorescence immunoassay, enzyme-linked immunosorbent assay, and immune biosensor have high sensitivity, which to some extent make up for the deficiency that chromatography analysis methods cannot be used for rapid on-site detection. However, the above two analysis methods require complex pre-treatment and are destructive tests with poor timeliness. They cannot meet the needs of rapid, non-destructive, and high-throughput rapid screening of DON during wheat storage and processing.
[0004] In recent years, near-infrared spectroscopy analysis technology has been applied to the detection of Fusarium toxins in grains, and has shown great application potential. However, the current reports are all batch sampling analysis of grains, or identification of diseased and shriveled grains, and do not involve toxin detection of individual grains. This study proposes a detection method that can achieve rapid and non-destructive detection of DON content in single grains, which is conducive to avoiding food waste caused by excessive DON in local samples and ensuring grain safety. Summary of the invention
[0005] The purpose of the present invention is to provide a high-throughput, rapid and non-destructive quantitative analysis method for Fusarium toxins in single grains of cereals based on near-infrared images.
[0006] The present invention provides a method for rapid non-destructive quantitative analysis of Fusarium toxins in single grains, comprising the following steps:
[0007] 1) Taking cereal grains with different disease levels caused by fungal infection and asymptomatic grains as research objects, using near-infrared hyperspectral imaging equipment to scan and obtain near-infrared images of the research objects, and obtaining near-infrared spectra of the research objects;
[0008] 2) removing the information-free area of the near infrared spectrum obtained in step 1), performing background subtraction and spectrum extraction, and obtaining near infrared spectrum data of a single grain;
[0009] 3) extracting and determining the content of Fusarium toxin from the research object to obtain the content of Fusarium toxin in a single grain;
[0010] 4) constructing a quantitative analysis model based on the near infrared spectroscopy data of the single cereal grain obtained in step 2) and the Fusarium toxin content of the single cereal grain obtained in step 3);
[0011] 5) subjecting the single cereal grain to be tested to the same testing conditions as in step 1) to obtain near infrared spectrum data of the single cereal grain to be tested;
[0012] 6) Substituting the near infrared spectrum data of the single cereal grain to be tested obtained in step 5) into the quantitative analysis model obtained in step 4) to obtain the content of Fusarium toxin in the single cereal grain to be tested.
[0013] In step 1) of the above method, the different levels include at least: severe and moderate levels;
[0014] Specifically, the severity levels are defined as follows: highly shrunken, dull white, with abundant white mycelium on the seed surface;
[0015] The medium grade is defined as follows: the kernels are relatively full, with pink mycelium in some places;
[0016] The number of grains of each grade is at least 35, and specifically 40.
[0017] The Fusarium toxin is vomitoxin.
[0018] In the step of obtaining the near-infrared image of the research object, there must be a certain interval between the grains, the pixel points of a single grain must not be less than 75, and the near-infrared spectral image information of the back and abdomen of the grain must be obtained respectively;
[0019] Specifically, the cereal grains are spaced 0.5 cm apart, the spectrum scanning range is 900-1700 nm, and the minimum number of pixels for a single grain is 80.
[0020] The step 2) of spectrum extraction comprises: extracting the spectrum data of all pixels of each of the cereal grains respectively, and calculating the average value as the near infrared spectrum of the grain;
[0021] The spectral data is specifically the spectral average value of the pixels occupied by a single wheat grain.
[0022] In step 3), the extraction and determination of Fusarium toxins in individual cereal grains requires the steps of weighing, crushing, extraction, filtering, purification and analysis.
[0023] Specifically, the cereal grains are weighed individually using a 1 / 10,000 balance before crushing, and the results are recorded accurately to the ten-thousandth place; when crushing, the cereal grains must be initially squeezed and broken, and then all transferred to a 2mL centrifuge tube, and a steel ball with a particle size of 0.5cm is placed in a high-throughput tissue grinder (Geno 2010, SPEX) for crushing at a speed of 1600 times / min for 90 seconds; the extract is 3mL of 50% acetonitrile water containing 1% formic acid.
[0024] The method of constructing a quantitative analysis model in step 4) comprises:
[0025] The K-stone algorithm was first used to divide the spectral data and vomitoxin content of each of the cereal kernels into a calibration set and a validation set at a ratio of 3:1, and then the partial least squares algorithm (PLS) was used to construct a quantitative analysis model.
[0026] In actual operation, in order to improve the signal-to-noise ratio of the spectrum and remove irrelevant information and noise, it is sometimes necessary to preprocess the spectrum. Specifically, the wheat grain near-infrared spectrum is processed by standard normal variable transformation (SNV), detrend and Savitzky-Golay first-order derivative to eliminate background interference, reduce spectral noise, and improve spectral quality and signal difference.
[0027] The method of the present invention can quickly analyze the Fusarium toxin content in a single grain without destroying the grain grain. It has the characteristics of being rapid, non-destructive and high-throughput. It can effectively deal with the detection difficulties caused by the uneven distribution of Fusarium toxin contamination in grains and the lack of representativeness of sampling detection. It is of great significance for timely discovering Fusarium toxin contamination, reducing food waste caused by local pollution, ensuring grain safety, and protecting people's health, and is suitable for popularization and application. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 These are images of wheat grains with three different disease levels;
[0029] Figure 2It is the weight and DON content of a single wheat kernel;
[0030] Figure 3 Scatter plot of actual and predicted DON values in the calibration and validation sets. DETAILED DESCRIPTION
[0031] The present invention is further described below in conjunction with specific examples, but the present invention is not limited to the following examples. The methods described are conventional methods unless otherwise specified. The raw materials described can be obtained from public commercial channels unless otherwise specified.
[0032] Example 1
[0033] 1) Sample collection. 40 grains with severe, moderate and asymptomatic fusarium fusarium were collected from multiple ears of wheat infected with fusarium fusarium as research objects. Figure 1 As shown, the samples used in this study can all be obtained from wheat ears with fusarium fusarium;
[0034] 2) Acquisition of near-infrared image data. The equipment used is the GaiaField-N17E imaging spectrometer of Sichuan Hepu Co., Ltd., with a lens-sample distance of 30 cm, a spectral range of 900-1700 nm, a spectral resolution of 5 nm, a sample stage movement speed of 0.6 cm / s, and an exposure time of 20.9 ms. First, place the wheat grains on the sample stage with their backs facing up and 0.5 cm between each grain. Use a digital camera to acquire RGB images, and then transfer them to the hyperspectral mode for scanning; then place the wheat grains with their abdomens facing up and perform the same operation.
[0035] 3) Data processing.
[0036] First, the original hyperspectral image is corrected for black and white according to the formula to eliminate the noise caused by dark current and uneven distribution of light source intensity:
[0037]
[0038] Among them, I is the black and white corrected image, I0 is the original hyperspectral image, B is the black calibration image with the lens cap on, and W is the standard polytetrafluoroethylene white plate calibration image.
[0039] Then, the hyperspectral image was cropped using the resize function in ENVI 5.1 software to remove unnecessary areas and bands with large noise and compress the data;
[0040] Secondly, the maximum inter-class variance method (otsu) and PCA combined with threshold segmentation and other methods were used to achieve image segmentation and background removal, and the image containing only wheat grain information was extracted;
[0041] Finally, the kernels in the image were numbered based on the bwlabel function in MATLAB to automatically extract the fingerprint of a single kernel, and the average spectra of the back and abdomen of the kernel were calculated as the fingerprint of the kernel.
[0042] 4) Extraction and determination of wheat kernel toxins. After obtaining near-infrared hyperspectral images, the wheat kernels were weighed, crushed, extracted, filtered, purified and analyzed to obtain the toxin content of each wheat kernel.
[0043] Before crushing, the wheat kernels are weighed individually using a 1 / 10,000 balance and the result is recorded to the nearest ten-thousandth place, recorded as m;
[0044] When crushing, the wheat grains were first squeezed and crushed, and then all transferred to a 2 mL centrifuge tube, and a steel ball with a particle size of 0.5 cm was placed in it, and placed in a high-throughput tissue grinder (Geno 2010, SPEX) for crushing at a speed of 1600 times / min for 90 seconds; the extract was 3 mL of 50% acetonitrile water containing 1% formic acid, which was used to dissolve the crushed sample three times and transferred all to a 4 mL centrifuge tube, vortexed for 5 minutes, and then centrifuged (3500 rpm) for 5 minutes; 2 mL of the supernatant was injected into a dispersed solid phase extraction purification filler (200 mg magnesium sulfate, 100 mg sodium chloride, 200 mg sodium citrate, C18 100mg) in a 4mL centrifuge tube, vortex for 1min and then centrifuge (3500rpm) for 5min; take 0.7mL (V1) of the supernatant, blow dry with nitrogen at room temperature, and reconstitute the residue with 50% acetonitrile water containing 1% formic acid. The volumes of the reconstituted solutions of severe, moderate and asymptomatic wheat kernels are 1.4mL, 0.7mL and 0.2mL, respectively, recorded as V2; filter through a 0.22μm filter membrane, collect the filtrate and test it with LC-MS / MS, and the test result is recorded as Cmeasured. The actual content of vomitoxin in a single wheat kernel, Cactual, is calculated according to the formula:
[0045]
[0046] The results are as follows Figure 2 shown.
[0047] 5) Model construction. First, the single grain spectra and DON content obtained in step 3) and step 4) were divided into a calibration set and a validation set at a ratio of 3:1 using the K-stone algorithm; then, the partial least squares (PLS) algorithm was used to construct a quantitative analysis model, and the leave-one-out method was used for interactive validation.
[0048] 6) Model verification and evaluation: The quantitative analysis model constructed in step 5) is used to predict the external validation set, and the model is evaluated based on the prediction set determination coefficient and prediction root mean square error.
[0049] The above data processing was completed in Matlab 2014a (MathWorks, USA) and PLS-toolbox 8.0 (Eigenvector Research, USA).
[0050] The quantitative analysis model constructed in this study had a correction set determination coefficient of 0.80, and a prediction root mean square error of 51.2 mg / kg; the interaction validation determination coefficient was 0.72, and the interaction validation root mean square error was 60.7 mg / kg; the prediction set determination coefficient was 0.75, and the prediction root mean square error was 54.6 mg / kg. The results showed that rapid and non-destructive analysis of vomitoxin in single wheat kernels based on near-infrared hyperspectral imaging technology is feasible, and can be used for rapid screening of vomitoxin contamination in single wheat kernels, which can help solve the misjudgment and food waste caused by uneven distribution of Fusarium toxins in grains, and is beneficial to protecting people's lives and safety.
Claims
1. A method for non-destructive quantitative analysis of Fusarium toxins in single cereal kernels, comprising: 1) Taking cereal grains with different disease levels caused by fungal infection and asymptomatic grains as research objects, using near-infrared hyperspectral imaging equipment to scan and obtain near-infrared images of the research objects, and obtaining near-infrared spectra of the research objects; In the step 1), in the step of scanning and obtaining the near-infrared image of the research object, there should be a certain interval between the grains, the pixel points of a single grain should not be less than 75, and the near-infrared spectrum image information of the back and abdomen of the grain should be obtained respectively; the spectrum scanning range is 900-1700nm; In the step 1), the different levels include at least: severe and moderate levels; The severity levels are defined as follows: highly shrunken, dull white, with abundant white mycelium on the seed surface; The medium grade is defined as follows: the kernels are relatively full, with pink mycelium in some places; At least 35 grains of each grade of grain; 2) removing the information-free area of the near infrared spectrum obtained in step 1), performing background subtraction and spectrum extraction, and obtaining near infrared spectrum data of a single grain; The step 2) of spectrum extraction comprises: extracting the spectrum data of all pixels of each of the cereal grains respectively, and calculating the average value as the near-infrared spectrum of the grain; the spectrum data is the spectrum average value of the pixels occupied by a single wheat grain; 3) extracting and determining the content of Fusarium toxin from the research object to obtain the content of Fusarium toxin in a single grain; 4) constructing a quantitative analysis model based on the near infrared spectroscopy data of the single cereal grain obtained in step 2) and the Fusarium toxin content of the single cereal grain obtained in step 3); The method of constructing a quantitative analysis model in step 4) comprises: First, the spectral data and the vomitoxin content of each of the cereal kernels were divided into a calibration set and a validation set at a ratio of 3:1 using the K-stone algorithm, and then a partial least squares algorithm (PLS) was used to construct a quantitative analysis model; 5) subjecting the single cereal grain to be tested to the same testing conditions as in step 1) to obtain near infrared spectrum data of the single cereal grain to be tested; 6) Substituting the near infrared spectrum data of the single cereal grain to be tested obtained in step 5) into the quantitative analysis model obtained in step 4) to obtain the content of Fusarium toxin in the single cereal grain to be tested; The Fusarium toxin is vomitoxin.
2. The method according to claim 1, characterized in that: There are 40 grains of each grade.
3. The method according to claim 1 or 2, characterized in that: In the step 1), the interval between two cereal grains is 0.5 cm, and the minimum number of pixels of a single grain is 80.
