A classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy, its establishment method, and detection device

Abstract

The present invention relates to a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy, a method for establishing the same, and a detection device. A method for establishing a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy includes: (1) collecting fresh blood from healthy and diseased sheep, extracting serum to obtain serum samples; (2) detecting the serum samples using a Fourier transform infrared spectroscopy (FT-IR) instrument to obtain FT-IR spectral data; (3) performing processing such as baseline correction and normalization on the FT-IR spectral data to obtain processed spectral data; (4) after reducing the dimensionality of the processed spectral data using the principal component analysis method (PCA), establishing a linear discriminant analysis (LDA) or support vector machine (SVM) classification model to obtain a classification model for echinococcosis granulosus and brucellosis in sheep. The classification model established by the present invention based on machine learning algorithms and FT-IR spectroscopy has the potential for rapid screening of sheep with echinococcosis granulosus and sheep with brucellosis.

Description

Technical Field

[0001] The present invention belongs to the technical field of livestock disease detection, and particularly relates to a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy, a method for establishing the same, and a detection device. Background Art

[0002] Echinococcosis granulosus (also known as hydatid disease) seriously threatens public health safety and animal husbandry, and is widely distributed globally. It is relatively common in the vast rural and pastoral areas of northern and southwestern China. On October 24, 2016, 12 ministries and commissions including the National Health and Family Planning Commission jointly issued the "National Plan for the Prevention and Control of Key Parasitic Diseases such as Echinococcosis (2016-2020)". This plan points out that the areas with severe echinococcosis epidemics include 7 provinces and autonomous regions such as Xinjiang, Qinghai, Gansu, Ningxia, Inner Mongolia, Tibet, and Sichuan, accounting for 44% of the national area. Sheep are the main intermediate hosts of Echinococcus granulosus. Effective prevention and control measures for sheep infected with echinococcosis granulosus are of great significance for blocking the transmission and development of echinococcosis. Currently, the screening techniques for sheep infected with echinococcosis granulosus mainly include imaging diagnosis techniques and serum immunological detection methods. Imaging techniques require expensive B-ultrasound machines, CT scanners, and professional imaging technicians, and the diagnostic process is also relatively cumbersome and requires the interpretation of professional imaging physicians. Before imaging ultrasound detection, it is also necessary to remove the wool and have two staff members fix the body of the sheep so that the imaging ultrasound physician can perform the detection. Although it is relatively easy to collect sheep serum, the serum immunological detection kits for sheep infected with echinococcosis granulosus have disadvantages such as long detection time, difficulty in standardization and cross-reaction between product batches, and low detection specificity. In addition, echinococcosis granulosus is mainly distributed in the poor rural and pastoral areas of western China. It is necessary to develop a new, economical, portable, and easy-to-operate screening technique that can quickly identify sheep infected with echinococcosis granulosus in order to facilitate the regular screening and treatment of echinococcosis.

[0003] Brucellosis (also known as brucellosis) is a zoonotic infectious disease caused by Brucella infection and is highly prevalent in pastoral areas of China. After Brucella infects livestock, it will cause problems such as easy abortion of livestock and reduction in milk and meat production. When humans come into contact with livestock infected with brucellosis, or eat their milk and meat, etc., they will be infected with brucellosis. Currently, the diagnostic methods for sheep brucellosis mainly rely on the rose bengal plate agglutination test and the tube agglutination test, but there are problems such as time-consuming and low accuracy.

[0004] Echinococcosis granulosus and brucellosis are two diseases that sheep are prone to infect, and currently there is no diagnostic technique that can simultaneously detect echinococcosis granulosus and brucellosis in sheep. Therefore, the present invention proposes a classification model for echinococcosis granulosus and brucellosis in sheep based on Fourier transform infrared spectroscopy (FT-IR), a method for establishing the same, and a detection device. Summary of the Invention

[0005] The object of the present invention is to provide a method for establishing a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy. Based on machine learning algorithms and combined with FT-IR spectroscopy, the established classification model has great application potential for screening sheep infected with echinococcosis granulosus and sheep infected with brucellosis at one time.

[0006] In order to achieve the above object, the technical solution adopted is as follows:

[0007] A method for establishing a classification model for echinococcosis granulosus and brucellosis in sheep, comprising the following steps:

[0008] (1) Collect fresh blood from healthy sheep and diseased sheep, let it stand and then centrifuge to extract serum to obtain serum samples;

[0009] (2) Detect the above-mentioned serum samples with a Fourier transform infrared spectrometer to obtain FT-IR spectral data;

[0010] (3) Perform baseline correction and normalization on the above-mentioned FT-IR spectral data to obtain processed spectral data;

[0011] (4) After dimensionality reduction of the processed spectral data, establish a principal component analysis (PCA)-linear discriminant analysis (LDA) or PCA-support vector machine (SVM) classification model to obtain the classification model for echinococcosis granulosus and brucellosis in sheep.

[0012] Further, in the step (1), the diseased sheep are echinococcosis granulosus-infected sheep and / or brucellosis-infected sheep.

[0013] Still further, at least 125 serum samples of healthy sheep are collected;

[0014] The diseased sheep are echinococcosis granulosus-infected sheep, and at least 77 serum samples of diseased sheep are collected;

[0015] The diseased sheep are brucellosis-infected sheep, and at least 101 serum samples of diseased sheep are collected.

[0016] Further, in the step (2), the detection resolution is 4 cm -1 , the number of scans is 32 times, and FT-IR spectral data is obtained within the scanning range of 600 - 4000 cm -1 .

[0017] Further, in the step (4), the principal component analysis method is used to perform dimensionality reduction on the preprocessed data within the modeling spectral range.

[0018] Still further, the modeling spectral range is the full wavelength of 600 - 4000 cm-1 and wavelengths in different bands of 900 - 1200 cm -1 、1200 - 1500 cm -1 、1500 - 1700 cm -1 、2800 - 3080 cm -1 、3090 - 3700 cm -1 。

[0019] Furthermore, in step (4) described above, an SVM classification model is established.

[0020] Another object of the present invention is to provide a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy, obtained by using the above-mentioned establishment method, which can be used to assist in the early screening and prevention of sheep infected with echinococcosis granulosus and brucellosis.

[0021] Another object of the present invention is to provide a detection device for echinococcosis granulosus and brucellosis in sheep, which can be used for screening sheep infected with echinococcosis granulosus and brucellosis, and can achieve the classification screening of sheep infected with echinococcosis granulosus and sheep infected with brucellosis at one time.

[0022] A detection device for echinococcosis granulosus and brucellosis in sheep includes: a laser, an infrared spectrometer, and a calculation and analysis system;

[0023] The Fourier transform infrared spectrometer scans each sample once to obtain serum infrared spectrum data and transmits the obtained serum infrared spectrum data to the calculation and analysis system;

[0024] The calculation and analysis system receives the serum infrared spectrum data collected by the Fourier transform infrared spectrometer and performs preprocessing of baseline correction and normalization on the data;

[0025] After the calculation and analysis system reduces the dimension of the preprocessed serum infrared spectrum data by using the principal component analysis method (PCA), it selects the principal components with the largest cumulative variance contribution rate and statistical differences and inputs them into the support vector machine (SVM) and linear discriminant analysis (LDA) algorithms to establish a classification model, and randomly divides all samples into a training set and a test set, and finally outputs the classification diagnosis results of the diseased group and the healthy group.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] At present, for the screening of sheep infected with Echinococcus granulosus, techniques such as imaging (ultrasound and CT, etc.) and serum immunology are mainly used. Imaging techniques require expensive B-ultrasound machines, CT scanners and professional imaging technicians. The diagnostic process is also relatively cumbersome and requires the determination of professional imaging physicians. Before imaging ultrasound detection, it is also necessary to remove the wool and have two staff members fix the body of the sheep so that the imaging ultrasound physician can perform the detection. Although it is relatively easy to collect sheep serum, the sheep echinococcosis serum immunology detection kit has disadvantages such as a long detection time, difficulty in standardization between product batches and cross-reactions. The main diagnostic techniques for brucellosis are: bacteriological detection, serological detection and molecular biological examination. Bacterial isolation and culture is the gold standard method for diagnosing brucellosis at present, but the disadvantages are that the bacterial culture cycle is long and there is a potential risk of laboratory-acquired brucellosis during the identification of growing bacteria. Although the Rose Bengal plate agglutination test (RBPT) in serological detection has the advantages of fast use and low price, there is a "window period" when antibodies are not produced in the early stage of the disease, and it is also easily affected by cross-reactions of other bacteria, resulting in false positive results. The serum agglutination test (SAT) is mostly used for retrospective diagnosis of several weeks or months in the acute stage of the disease. False negative results will occur when the SAT titer does not change significantly in the early stage of the disease, so the detection efficiency depends on the SAT titer. ELISA has the advantages of high sensitivity and short operation time, but also has disadvantages such as non-standardized reagents and false positive results easily caused by cross-reactions.

[0028] In summary, the current conventional diagnostic techniques for sheep echinococcosis and sheep brucellosis have disadvantages such as high technical thresholds, cumbersome detection processes and low accuracy. In addition, echinococcosis and brucellosis are mainly distributed in remote agricultural and pastoral areas in the western part of China, where the natural conditions are harsh, the economic conditions are poor, and the medical conditions are backward. Therefore, when developing new differential diagnostic techniques for echinococcosis and brucellosis, factors such as the cost of diagnostic instruments, human conditions, portability and ease of operation need to be considered. The technology of this invention has the advantages of simple operation, accuracy, rapidity and non-damage to samples, and can meet the disadvantages of the current conventional diagnostic techniques for sheep echinococcosis and sheep brucellosis.

[0029] For the first time, the serum FT-IR spectroscopy technology is applied in the screening of echinococcosis granulosus and brucellosis in sheep. A Fourier transform infrared spectroscopy (FT-IR) instrument is used to scan each serum sample to obtain the FT-IR data of the sera of sheep with echinococcosis granulosus, sheep with brucellosis, and healthy sheep, and the data is transmitted to the calculation and analysis system. Pretreatments such as baseline correction and normalization are performed on the obtained data. After the calculation and analysis system reduces the dimension of the pretreated spectral data using the principal component analysis method (PCA), the principal components with the largest cumulative variance contribution rate and statistical differences are selected and input into the support vector machine (SVM) and linear discriminant analysis (LDA) algorithms. Therefore, the samples are randomly divided into a training set and a test set. Finally, it has great application potential for realizing the diagnosis of echinococcosis granulosus or brucellosis in sheep respectively and simultaneously realizing the classification diagnosis of echinococcosis granulosus and brucellosis in sheep. Description of the Drawings

[0030] Figure 1 It is the serum Fourier transform infrared spectroscopy detection device for echinococcosis granulosus and brucellosis in sheep used in the experiment;

[0031] Figure 2 It is the process diagram for baseline correction and normalization pretreatment of the infrared spectral data of the collected sheep sera;

[0032] Figure 3 It is the averaged infrared spectrogram of the sera of sheep with echinococcosis granulosus and healthy sheep;

[0033] Figure 4 It is the matrix scatter plot of the meaningful principal component scores obtained from the infrared spectra of the sera of sheep with echinococcosis granulosus and healthy sheep in the 600 - 4000 cm -1 spectral range;

[0034] Figure 5 It is the matrix scatter plot of the meaningful principal component scores obtained from the infrared spectra of the sera of sheep with echinococcosis granulosus and healthy sheep in the 1500 - 1700 cm -1 spectral range;

[0035] Figure 6 It is the averaged infrared spectrogram of the sera of sheep with brucellosis and healthy sheep;

[0036] Figure 7 It is the matrix scatter plot of the meaningful principal component scores obtained from the infrared spectra of the sera of sheep with brucellosis and healthy sheep in the 600 - 4000 cm -1 spectral range;

[0037] Figure 8 It is the matrix scatter plot of the meaningful principal component scores obtained from the infrared spectra of the sera of sheep with brucellosis and healthy sheep in the 1500 - 1700 cm-1 Matrix scatter plot within the spectral range;

[0038] Figure 9 Average infrared spectra of sera from sheep with echinococcosis granulosus, brucellosis, and healthy sheep;

[0039] Figure 10 Meaningful principal component scores obtained from the infrared spectra of sera from sheep with echinococcosis granulosus, brucellosis, and healthy sheep in the range of 600 - 4000 cm -1 Matrix scatter plot within the spectral range;

[0040] Figure 11 Meaningful principal component scores obtained from the infrared spectra of sera from sheep with echinococcosis granulosus, brucellosis, and healthy sheep in the range of 1500 - 1700 cm -1 Matrix scatter plot within the spectral range. Detailed implementation manners

[0041] In order to further elaborate on a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectra, its establishment method, and detection device according to the present invention, and to achieve the expected invention purpose, the following, in combination with preferred embodiments, details the specific implementation manners, structures, features, and effects of a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectra, its establishment method, and detection device according to the present invention. In the following description, different "one embodiment" or "embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0042] The following will further introduce in detail a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectra, its establishment method, and detection device according to the present invention in combination with specific embodiments:

[0043] The present invention uses serum as a biological sample to diagnose echinococcosis granulosus and brucellosis in sheep by FT-IR technology. The experimental results show that the classification model established by the present invention has great application potential for screening echinococcosis granulosus and brucellosis in sheep at one time.

[0044] The technical solution of the present invention is as follows:

[0045] A classification model for serum Fourier transform infrared spectra (FT-IR) of echinococcosis granulosus and brucellosis in sheep based on machine learning algorithms. (Ⅰ) First, detect sera from sheep with echinococcosis granulosus and healthy sheep. The detailed steps are as follows:

[0046] (1) Blood sampling: Select the sheep flock to be tested, conduct classification diagnosis through B-ultrasound examination, divide them into sheep with Echinococcus granulosus disease and healthy sheep, and then collect blood from the left jugular vein of the sheep using a 10 ml sheep blood sampler (without anticoagulant).

[0047] (2) Blood sample processing: Let the collected blood samples stand at room temperature for 2 - 3 hours to allow the blood to clot completely. Then centrifuge the samples at a speed of 4000 revolutions per minute in a centrifuge. Use a pipette to aspirate the separated serum (the upper milky yellow supernatant) and transfer it to a 2.5 ml EP tube, and store it in a -80°C refrigerator for later use.

[0048] (3) Serum spectral data collection: The experiment uses a Fourier transform infrared spectroscopy detection device for the sera of sheep with Echinococcus granulosus disease and Brucella disease. Set the resolution to 4 cm -1 , the number of scans to 32 times, and obtain the FT-IR spectral data of the serum samples obtained in step (2) respectively within the scanning range of 600 - 4000 cm -1 .

[0049] (4) Serum FT-IR spectral data preprocessing: For the spectral data obtained in step (3), perform baseline correction and normalization on all serum FT-IR spectral data, and convert it into the xls format for extraction.

[0050] (5) Data dimensionality reduction processing: Use MATLAB R2021a software, and adopt the principal component analysis method to perform dimensionality reduction on the preprocessed data within the selected modeling spectral ranges (respectively the full wavelength range of 600 - 4000 cm -1 and different band wavelength ranges of 900 - 1200 cm -1 , 1200 - 1500 cm -1 , 1500 - 1700 cm -1 , 2800 - 3080 cm -1 , 3090 - 3700 cm -1 ). Select the principal components with the largest cumulative variance contribution rate and statistical differences.

[0051] (6) Establishment and verification of the diagnostic model: As described in step (5), respectively divide the principal components of the serum data of sheep with Echinococcus granulosus disease and healthy sheep into a training set and a verification set and input them into the LDA and SVM algorithms, and obtain the classification results of sheep with Echinococcus granulosus disease and healthy sheep based on the serum Fourier transform infrared spectroscopy (FT-IR) of the PCA-LDA and PCA-SVM models respectively.

[0052] Classification model of serum Fourier transform infrared spectroscopy (FT-IR) for echinococcosis granulosus and brucellosis in sheep based on machine learning algorithms. (II) Secondly, serum detection of sheep with brucellosis and healthy sheep is carried out, and the detailed steps are as follows:

[0053] (1) Blood sampling: Select the flock to be tested, and then use a 10 ml sheep blood sampler (without anticoagulant) to collect blood from the left jugular vein of the sheep. Then, divide the flock into sheep with brucellosis and healthy sheep through the Rose Bengal plate agglutination test for brucellosis and the tube agglutination test for brucellosis.

[0054] (2) Blood sample processing: Let the collected blood sample stand at room temperature for 2 - 3 hours to completely clot the blood. Place it in a centrifuge and centrifuge at a speed of 4000 revolutions per minute. Use a pipette to suck out the separated serum (the upper milky yellow supernatant) and transfer it to a 2.5 ml EP tube. First, sterilize it at a high temperature (56 °C) for 30 minutes, and then store it in a -80 °C refrigerator for later use.

[0055] (3) Serum spectral data collection: The experiment uses a Fourier transform infrared spectroscopy detection device for the serum of sheep with echinococcosis granulosus and brucellosis. Set the resolution to 4 cm -1 , the number of scans is 32 times, and within the scanning range of 600 - 4000 cm -1 , obtain the FTIR spectral data of the serum samples obtained in step (1) respectively.

[0056] (4) Serum FTIR spectral data preprocessing: For the spectral data obtained in step (3), perform baseline correction and normalization on all serum FTIR spectral data, and convert it into xls format for extraction.

[0057] (5) Data dimensionality reduction processing: Use MATLAB R2021a software, and adopt the principal component analysis method to perform dimensionality reduction on the preprocessed data within the selected modeling spectral ranges (respectively the full wavelength of 600 - 4000 cm -1 and different band wavelengths of 900 - 1200 cm -1 , 1200 - 1500 cm -1 , 1500 - 1700 cm -1 , 2800 - 3080 cm -1 , 3090 - 3700 cm -1 ). Select the principal component with the largest cumulative variance contribution rate and statistical difference.

[0058] (6) Establishment and verification of the diagnostic model: As described in step (5), the principal components of the serum data of Brucella-infected sheep and healthy sheep were randomly divided into a training set and a verification set and input into the LDA and SVM algorithms to obtain the classification results of Brucella-infected sheep and healthy sheep based on the serum Fourier transform infrared spectroscopy (FT-IR) of the PCA-LDA and PCA-SVM models, respectively.

[0059] Serum Fourier transform infrared spectroscopy (FT-IR) classification models for echinococcosis granulosus and Brucellosis in sheep based on machine learning algorithms. (III) Finally, serum tests were performed on sheep infected with echinococcosis granulosus, Brucella-infected sheep, and healthy sheep. The detailed steps are as follows:

[0060] (1) Blood sampling: Select the flock to be tested, collect blood from the left jugular vein of the sheep using a 10 ml sheep blood sampler (without anticoagulant), and then divide the flock into sheep infected with echinococcosis granulosus, Brucella-infected sheep, and healthy sheep through sheep B-ultrasound examination, Brucella melitensis rose bengal plate agglutination test, and Brucella melitensis tube agglutination test.

[0061] (2) Blood sample processing: The collected blood samples were left standing at room temperature for 2 - 3 hours to allow the blood to clot completely, then centrifuged at a speed of 4000 revolutions per minute in a centrifuge. The separated serum (the upper milky yellow supernatant) was aspirated with a pipette and transferred into a 2.5 ml EP tube, and stored in a -80°C refrigerator for later use.

[0062] (3) Serum spectral data collection: The serum Fourier transform infrared spectroscopy detection device for sheep infected with echinococcosis granulosus and Brucella-infected sheep was used in the experiment. The resolution was set to 4 cm -1 , the number of scans was 32 times, and the FTIR spectral data of the serum samples obtained in step (1) were respectively obtained in the scanning range of 600 - 4000 cm -1 .

[0063] (4) Serum FTIR spectral data preprocessing: For the spectral data obtained in step (3), all serum FTIR spectral data were subjected to baseline correction and normalization processing, and then converted into the xls format for extraction.

[0064] (5) Dimensionality reduction processing of data: Using MATLAB R2021a software, the principal component analysis method was used in the selected modeling spectral ranges (respectively the full wavelength of 600 - 4000 cm -1 and different band wavelengths of 900 - 1200 cm -1 , 1200 - 1500 cm -1 , 1500 - 1700 cm -1 , 2800 - 3080 cm -1 3090 - 3700 cm-1 ) For the preprocessed data, dimensionality reduction is performed, and the principal components with the largest cumulative variance contribution rate and significant statistical differences are selected.

[0065] (6) Establishment and verification of the diagnostic model: According to the description in step (5), the principal components of the serum data of echinococcosis granulosus-infected sheep, brucellosis-infected sheep, and healthy sheep are randomly divided into a training set and a verification set and input into the LDA and SVM algorithms to obtain the three-classification results of echinococcosis granulosus-infected sheep, brucellosis-infected sheep, and healthy sheep based on the PCA-LDA and PCA-SVM models, respectively.

[0066] Example 1.

[0067] Establishment of a classification model for serum Fourier transform infrared spectroscopy (FT-IR) of echinococcosis granulosus and brucellosis in sheep based on machine learning algorithms, the specific steps are as follows:

[0068] (1) Collect samples: First, 77 blood samples of echinococcosis granulosus-infected sheep with clear diagnosis and 121 blood samples of healthy sheep are collected by B-ultrasound detection. At the same time, 300 blood samples of sheep in the flock to be tested for brucellosis are collected.

[0069] (2) Obtain serum samples: The collected blood samples are left standing at room temperature for 2 - 3 hours to allow the blood to clot completely, then centrifuged in a centrifuge at a speed of 4000 revolutions per minute, and the separated serum (the upper milky yellow supernatant) is aspirated with a pipette and transferred to a 2.5 ml EP tube; through the rose bengal plate agglutination test and the tube agglutination test for brucellosis, 101 brucellosis-infected sheep and 125 healthy sheep are screened out from the serum collected from the flock to be tested for brucellosis.

[0070] (3) Serum spectral data collection: The experiment uses a Fourier transform infrared spectroscopy detection device for echinococcosis granulosus-infected sheep and brucellosis-infected sheep serum (as shown in the schematic Figure 1 ), the resolution is set to 4 cm -1 , the number of scans is 32 times, and the FT-IR spectral data of the serum samples obtained in step (2) are respectively obtained in the scanning range of 600 - 4000 cm -1 .

[0071] (4) Serum FT-IR spectral data preprocessing: All serum FT-IR spectral data are subjected to baseline correction and normalization and other processes, and are converted into the xls format for extraction, and finally the available FT-IR spectral data are obtained ( Figure 2 as shown).

[0072] (5) Dimensionality reduction processing of data: Using MATLAB R2021a software, the principal component analysis method is used to perform dimensionality reduction in the selected modeling spectral range (respectively the full wavelength 600 - 4000 cm-1 and for the preprocessed data within different wavelength bands of 900 - 1200 cm -1 、1200 - 1500 cm -1 、1500 - 1700 cm -1 、2800 - 3080 cm -1 、3090 - 3700 cm -1 ) perform dimensionality reduction on the data, and select the principal components with the largest cumulative variance contribution rate and statistical differences.

[0073] (6) Establishment and verification of the diagnostic model: As described in step (5), respectively divide the principal components of the serum FT-IR spectral data of echinococcus granulosus-infected sheep, brucellosis-infected sheep, and healthy sheep into training sets and validation sets and input them into the LDA and SVM algorithms to obtain three classification diagnostic models for echinococcus granulosus-infected sheep and healthy sheep, brucellosis-infected sheep and healthy sheep, and echinococcus granulosus-infected sheep, brucellosis-infected sheep, and healthy sheep based on the PCA-LDA and PCA-SVM models of serum Fourier transform infrared spectroscopy (FT-IR). The specific results are as follows:

[0074] Classification results of echinococcus granulosus-infected sheep and healthy sheep: Randomly divide all serum samples into training sets and validation sets. Within the full wavelength range ( Figure 3 as shown), first extract the principal components ( Figure 4 as shown), input the meaningful principal components into the LDA and SVM models, and the results are shown in Table 1. Then perform principal component extraction on the spectral data within the different wavelength bands described in step (5) (taking the 1500 - 1700 cm -1 band as an example, the results Figure 5 are shown), input the meaningful principal components into the LDA and SVM models, and the results are shown in Table 2.

[0075] Table 1

[0076]

[0077] Table 2

[0078]

[0079] Classification results of brucellosis-infected sheep and healthy sheep: Randomly divide all serum samples into training sets and validation sets. Within the full wavelength range ( Figure 6 as shown), first extract the principal components ( Figure 7 as shown), then input the meaningful principal components into the LDA and SVM models, and the results are shown in Table 3. Perform principal component extraction on the spectral data within the different wavelength bands (taking the 1500 - 1700 cm -1 band as an example, the results Figure 8As shown in the figure, the meaningful principal components are input into the LDA and SVM models, and the results are shown in Table 4.

[0080] Table 3

[0081]

[0082]

[0083] Table 4

[0084]

[0085] Classification results of sheep with echinococcosis granulosus, brucellosis and healthy sheep: All serum samples are randomly divided into a training set and a validation set. In the full wavelength range ( Figure 9 as shown), first, the principal components are extracted ( Figure 10 as shown), and then the meaningful principal components are input into the LDA and SVM models. The results are shown in Table 5. The spectral data in different wavelength ranges are subjected to principal component extraction (taking the 1500 - 1700 cm -1 band as an example, the results Figure 11 are shown), and the meaningful principal components are input into the LDA and SVM models. The results are shown in Table 6.

[0086] Table 5

[0087]

[0088] Table 6

[0089]

[0090] The present invention first applies the serum FT-IR spectroscopy technology to the detection of echinococcosis granulosus and brucellosis in sheep. Based on machine learning algorithms and combined with FT-IR spectroscopy, the present invention designs a PCA-LDA and PCA-SVM model for the serum FT-IR spectroscopy of echinococcosis granulosus and brucellosis in sheep based on multivariate analysis. Using the accuracy of the validation set as the evaluation basis, the experimental results show that the LDA and SVM models have good detection results in the screening of sheep with echinococcosis granulosus and brucellosis, and the detection result of the SVM model is better.

[0091] The above are only the preferred embodiments of the embodiments of the present invention, and do not impose any form of limitation on the embodiments of the present invention. Any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the embodiments of the present invention still fall within the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for establishing a classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy, characterized in that, It includes the following steps: (1) Collect fresh blood from healthy sheep and diseased sheep, let it stand and then centrifuge to extract serum, obtaining serum samples; the diseased sheep are echinococcus granulosus diseased sheep and / or brucellosis diseased sheep; (2) Detect the serum samples with a Fourier transform infrared spectrometer to obtain Fourier transform infrared spectral data; (3) Perform processing such as baseline correction and normalization on the Fourier transform infrared spectral data to obtain processed spectral data; (4) After dimensionality reduction of the processed spectral data, establish a principal component analysis - linear discriminant analysis or principal component analysis - support vector machine classification model to obtain the classification models for echinococcus granulosus disease and brucellosis in sheep; The principal component analysis method is used for dimensionality reduction of the preprocessed data within the modeling spectral range; The modeling spectral range is the full wavelength of 600 - 4000 cm -1 and different band wavelengths of 900 - 1200 cm -1 、1200 - 1500 cm -1 、1500 - 1700 cm -1 、2800 - 3080 cm -1 、3090 - 3700 cm -1 。 2. The establishment method according to claim 1, characterized in that At least 125 serum samples of healthy sheep are collected; The diseased sheep are echinococcus granulosus diseased sheep, and at least 77 serum samples of diseased sheep are collected; The diseased sheep are brucellosis diseased sheep, and at least 101 serum samples of diseased sheep are collected.

3. The establishment method according to claim 1, characterized in that In the said step (2), the detection resolution is 4 cm -1 , the number of scans is 32 times, and Fourier transform infrared spectroscopy data is obtained within the scanning range of 600 - 4000 cm -1 .

4. The establishment method according to claim 1, characterized in that In step (4), a support vector machine classification model is established.

5. A classification model for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy, characterized in that, It is obtained by using the establishment method according to any one of claims 1 - 4.

6. A detection device for echinococcosis granulosus and brucellosis in sheep based on infrared spectroscopy, characterized in that, The detection device uses the establishment method according to any one of claims 1 - 4; the detection device includes: a laser, an infrared spectrometer, and a calculation and analysis system; The Fourier transform infrared spectrometer scans each sample once to obtain serum infrared spectral data and transmits the obtained serum infrared spectral data to the calculation and analysis system; The calculation and analysis system receives the serum infrared spectral data collected by the Fourier transform infrared spectrometer and performs preprocessing of baseline correction and normalization on the data; The calculation and analysis system performs dimensionality reduction on the preprocessed serum infrared spectral data by using the principal component analysis method, then inputs the principal components with the largest cumulative variance contribution rate and statistical differences into the support vector machine and linear discriminant analysis algorithms to establish a classification model, and randomly divides all samples into a training set and a test set, and finally outputs the classification diagnosis results of the diseased group and the healthy group.

Images