Axial precise positioning and judgment method of cell micro-Raman detection

Abstract

The present invention relates to the field of cell detection technology, and in particular to a method for precise axial positioning and judgment of cell microscopic Raman detection. It includes: selecting glass slides with different Raman spectra as slides and cover slides; obtaining normalized standard Raman spectra through testing; fixing the cell sample to be tested between two glass slides, placing it under a microscope objective, adjusting the focal length under ordinary white light illumination, and then switching to laser to form a focused light spot that excites the Raman signal on the cell sample, and detecting the Raman spectrum of the sample at this time; constructing a related linear equation group, using a regression algorithm to solve the intensity coefficient of the Raman spectra of the two glass slides, and calculating the cell Raman spectrum; constructing a normalized positioning coefficient of the light spot axial direction according to the intensity coefficient of the Raman spectra of the two glass slides; constructing and training a discrimination model; and using the trained discrimination model to discriminate the cell sample to be tested. The advantage is that no labeling, staining, etc. are required, and the objective positioning and judgment of the cell Raman axial excitation position can be directly achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of cell detection, and in particular to a method for precise axial positioning and judgment of cell microscopic Raman detection. Background Art

[0002] Cells are the basic units of life, mainly composed of molecules such as proteins, lipids, and nucleic acids. There are differences in molecular composition and content between different cells and the same cells at different developmental stages. For example, there are usually metabolic differences in the process of cell carcinogenesis. Raman spectroscopy is a molecular scattering spectrum that can reflect the intrinsic vibration information of molecular covalent bonds, and then obtain the molecular types and content information of cells through cell Raman spectroscopy. Single-cell Raman spectroscopy technology has the advantages of non-labeled and non-destructive detection, and has potential application value in the fields of cell identification, cancer diagnosis, and cancer prognosis. At present, cell Raman spectroscopy detection is combined with a microscope. The analysis sample is placed under a microscope, and the microscope focuses the Raman excitation spot on the cell. After the cell Raman signal is excited, it is collected by the objective lens, and the scattered signal is transmitted to the spectrometer to obtain the cell Raman spectrum for subsequent analysis.

[0003] In the existing Raman microscopy analysis system, a white light microscope is relied on to find and locate cells and find the location of excitation. However, cells are a three-dimensional spatial structure, and the spatial morphology of cells varies, and different structures have significant compositional differences. The spot size of the Raman excitation light focused by the microscope objective is usually smaller than the cell size, and the traditional microscope obtains the planar structure image of a certain layer in the cell. The cell sample is placed at the focus spot position through a galvanometer or a translation stage for horizontal (lateral) positioning. In the horizontal direction, the cell contour has a clear boundary with the external environment, which can assist in cell positioning, but in the direction of the microscope optical axis, the inspector manually adjusts the focal length of the microscope to image and axially (longitudinally) locate different layers of the cell. The inside of the cell is a nearly transparent sample. Staining and adding fluorescent markers will destroy the advantage of Raman spectroscopy without labeling. The signals of dyes and fluorescent markers will interfere with the sample Raman spectrum, and even make Raman spectroscopy detection impossible. In the absence of labeling or staining, the contrast of the internal structure of the cell is low, and the difference between the layers is almost indistinguishable.

[0004] Therefore, conventional Raman detection cannot effectively distinguish the axial focus position of the laser spot on different levels, and the axial resolution is low. It can only rely on the experience of the experimenter to make judgments, resulting in low axial positioning accuracy of the cell Raman excitation spot, and the consistency of the excitation spectrum cannot be guaranteed. Sampling errors are introduced in subsequent analysis, resulting in a decrease in the accuracy of the current overall technical analysis. At present, there is no label-free, non-staining, objective, and quantitative method for axial positioning of cell Raman spectroscopy detection to ensure the consistency of axial sampling of cell Raman spectroscopy. Summary of the invention

[0005] In order to solve the above problems, the present invention provides a method for precise axial positioning and judgment of cell microscopic Raman detection.

[0006] The present invention aims to provide a method for accurate axial positioning and judgment of cell microscopic Raman detection, which specifically comprises the following steps:

[0007] S1. Select two glass slides of different materials with different Raman spectra as slide glass and cover glass respectively; obtain the normalized standard Raman spectra of the slide glass and cover glass respectively through testing;

[0008] S2. Fix the cell sample to be tested between the slide and the cover glass to prepare a test sample; place the test sample under the microscope objective lens, adjust the focus under ordinary white light illumination, and find the position of the target cell;

[0009] S3. After finding the target cell, the light source is switched to laser, a focused light spot is formed on the cell sample to excite the Raman signal, and the Raman spectrum of the sample is detected at this time;

[0010] S4. Use regression algorithm to solve the intensity coefficient of Raman spectra of cover glass and slide, construct relevant linear equations, and calculate the Raman spectrum of cell sample;

[0011] S5. constructing a normalized positioning coefficient of the light spot axis according to the intensity coefficient of the Raman spectra of the cover glass and the slide;

[0012] S6. Construct a discriminant model; use the Raman spectra collected from known types of cell samples as a training set database to train the discriminant model; use the trained discriminant model to discriminate the cell samples to be tested to obtain the discrimination results.

[0013] Preferably, the regression algorithm in step S4 is the least squares method; the expression of the Raman spectrum of the cell sample is as follows:

[0014] ;

[0015] Where S is the total Raman spectrum of the test obtained by detection, S1 and S2 are the normalized standard Raman spectra of the cover glass and the slide, respectively, A1 and A2 are the intensity coefficients of the Raman spectra of the cover glass and the slide during detection, respectively, and R is the Raman spectrum of the cell sample.

[0016] Preferably, the minimum value of the intensity coefficient A1 of the cover glass Raman spectrum or the intensity coefficient A2 of the slide glass Raman spectrum is 0.

[0017] Preferably, the related linear equations in step S4 are expressed as follows:

[0018] ;

[0019] Where S is the total Raman spectrum of the test obtained by detection, S1 and S2 are the normalized standard Raman spectra of the cover glass and the slide, respectively, A1 and A2 are the intensity coefficients of the Raman spectra of the cover glass and the slide during detection, respectively, R is the Raman spectrum of the cell sample, ωi is the Raman frequency shift, and n is the number of spectral channels in the Raman spectrum.

[0020] Preferably, the method for finding the position of the target cell in step S2 is to move the detection sample to the focused light spot on the horizontal imaging plane.

[0021] Preferably, the Raman spectrum of the sample includes the Raman spectrum of the cover glass, the Raman spectrum of the cell sample and the Raman spectrum of the slide glass.

[0022] Preferably, the normalized positioning coefficient is as follows:

[0023] ;

[0024] Wherein, C1 represents the axial distance between the Raman test cover glass and the excitation spot, and C2 represents the distance between the Raman test slide and the excitation spot.

[0025] Preferably, the method for constructing the discriminant model in step S6 specifically includes: selecting data with a normalized positioning coefficient Ci value in the range of 0.4 to 0.6 to establish a training set database; reducing the dimension of the training set data based on principal component analysis, selecting the number of principal components so that the corresponding variance of the selected principal components accounts for more than 80% of the total, and then establishing a discriminant model based on a linear discriminant analysis method.

[0026] Preferably, the discrimination model is trained in step S6 using a machine learning method; the discrimination method specifically includes: collecting the Raman spectrum of the cell sample to be tested, selecting data with a normalized localization coefficient Ci value in the range of 0.4 to 0.6 as test set data, and bringing the test set data into the trained discrimination model to obtain a discrimination result.

[0027] Compared with the prior art, the present invention can achieve the following beneficial effects:

[0028] The present invention proposes a novel method for accurately positioning and judging the axial direction of cell Raman microscopic detection. In the microscopic Raman test, two glass slides of different materials are introduced in the process of making cell samples. After the standard Raman spectrum of the glass slide is measured, the cell sample is placed between the two glass slides. After the overall spectrum of the sample is obtained by testing, the contribution of the two glass slides to the overall test Raman spectrum intensity is calculated by regression algorithms such as the least squares method. Based on this, the normalized positioning coefficient of the cell excitation light spot is innovatively given. Based on this, interference with the cell Raman test can be avoided without staining or marking the sample, and the objective positioning judgment and calibration of the cell Raman axial excitation position in the cell Raman spectrum microscopic detection can be accurately realized. The present invention can realize an objective method for positioning and judging the excitation position of the focused light spot in the cell axial direction (perpendicular to the microscopic imaging surface) in the cell Raman spectrum detection, get rid of the sampling error introduced by human experience observation, realize the normalization of the cell Raman axial excitation position, ensure the accuracy of data acquisition, and lay a solid foundation for subsequent accurate analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of axial precise positioning of cell Raman microscopy detection provided according to an embodiment of the present invention.

[0030] Figure 2 1 is a schematic diagram of the composition of the test Raman spectrum provided according to an embodiment of the present invention; in the figure, (a) is the total Raman spectrum of the test, (b) is the standard Raman spectrum of the cover glass, (c) is the standard Raman spectrum of the slide, and (d) is the Raman spectrum of the cell sample.

[0031] Figure 3 It is an analysis flow chart of the axial precise positioning and judgment method of cell microscopic Raman detection provided according to an embodiment of the present invention.

[0032] Reference numerals:

[0033] 1. Microscope objective lens;

[0034] 2. Cover glass;

[0035] 3. Cell samples;

[0036] 4. Glass slide;

[0037] 5. Focus the light spot. DETAILED DESCRIPTION

[0038] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the following description, the same modules are represented by the same reference numerals. In the case of the same reference numerals, their names and functions are also the same. Therefore, the detailed description thereof will not be repeated.

[0039] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not constitute a limitation of the present invention.

[0040] The cell microscopic Raman detection axial precise positioning and judgment method specifically includes the following steps:

[0041] S1. Select two glass slides of different materials with different Raman spectra as slide glass and cover glass respectively; obtain the normalized standard Raman spectra of the slide glass and cover glass respectively through testing.

[0042] S2. Fix the cell sample to be tested between the slide and the cover glass to prepare a test sample; place the test sample under the microscope objective lens, adjust the focus under ordinary white light illumination, and find the position of the target cell;

[0043] Specifically, finding the location of the target cell is achieved by moving the detection sample to the focused spot on the horizontal (lateral) imaging plane.

[0044] S3. After finding the target cell, the light source is switched to laser, a focused light spot is formed on the cell sample to excite the Raman signal, and the Raman spectrum of the sample is detected at this time;

[0045] The Raman spectrum of the sample includes the Raman spectrum of the cover glass, the Raman spectrum of the cell sample and the Raman spectrum of the slide. In this step, due to the low axial resolution of the microscope system and the large spot size, the Raman spectra of the cover glass, the cell sample and the slide will be excited at the same time.

[0046] S4. Using regression algorithm to solve the intensity coefficients of Raman spectra of coverslip and slide, constructing related linear equations, and calculating the Raman spectrum of cell sample;

[0047] Specifically, the method for constructing the related linear equation group is as follows: constructing the related linear equation group according to the normalized standard Raman spectra of the cover glass and the slide glass and the intensity values ​​at each wave number of the Raman spectrum of the test sample;

[0048] Specifically, the regression algorithm is the least squares method;

[0049] The expression of the Raman spectrum of the cell sample is as follows:

[0050] (1);

[0051] Where R is the Raman spectrum of the cell sample, S is the total Raman spectrum of the test obtained by detection, S1 and S2 are the normalized standard Raman spectra of the cover glass and the slide, respectively, and A1 and A2 are the intensity coefficients of the Raman spectra of the cover glass and the slide during detection, respectively.

[0052] At a specific Raman frequency shift ωi, the following relationship holds:

[0053] (2);

[0054] In the above formula, n is the number of spectral channels in the Raman spectrum, that is, the number of Raman frequency shifts. Combining the above formula, an n-dimensional linear equation group, that is, a related linear equation group, can be constructed.

[0055] Principle: Since the cell Raman spectrum is relatively weak, if the cell Raman spectrum is regarded as the fitting residual, then according to the normalized standard Raman spectra of the cover glass and the slide, n equations about A1 and A2 can be constructed; using regression algorithms such as the least squares method, the values ​​of A1 and A2 corresponding to each Raman detection spectrum S can be solved, that is, the contribution of the Raman spectra of the cover glass and the slide to the overall Raman spectrum during the test can be obtained. At the same time, substituting A1 and A2 into formula (2), the pure Raman spectrum of the cell in this test can be estimated. Since the excitation spot may be overly biased towards a certain slide, the Raman spectrum contribution of another slide in the collected overall spectrum S is almost zero. Under the influence of the Raman spectrum of the cell sample, the coefficient of the other slide appears to be overfitted with a negative value less than zero. At this time, just set the negative coefficient to 0.

[0056] S5. According to the intensity coefficients of the Raman spectra of the cover glass and the slide, the normalized positioning coefficient Ci of the light spot axis is constructed as shown in the following formula:

[0057] (3);

[0058] In the formula, C1 represents the axial distance between the Raman test cover glass and the excitation spot, and C2 represents the distance between the Raman test slide and the excitation spot. The distance is inversely proportional to the coefficient.

[0059] Principle description: Taking C1 as an example, the larger the C1 value, the greater the contribution of the cover glass Raman spectrum to the overall spectral intensity (characterized by A1), which means that the position of the excitation spot is closer to the cover glass. Since the cell sample is in the space between the two slides, the axial position of the excitation spot on the cell sample can be determined based on this, providing an objective criterion for the axial positioning of the excitation. In a special case, when A1 is approximately equal to A2, that is, C1 is approximately equal to 0.5, it means that the distance from the excitation spot to the cover glass and the slide is approximately equal, which means that the position of the excitation spot is located at the axial center of the cell sample.

[0060] S6. Construct a discriminant model; use the Raman spectra collected from known types of cell samples as a training set database to train the discriminant model; use the trained discriminant model to discriminate the cell samples to be tested to obtain the discrimination results.

[0061] Specifically, data with normalized positioning coefficient Ci values ​​ranging from 0.4 to 0.6 are selected to establish a training set database; the training set data is reduced in dimension based on principal component analysis, the number of principal components is selected so that the corresponding variance of the selected principal components accounts for more than 80% of the total, and then a discriminant model is established based on the linear discriminant analysis method;

[0062] The specific training adopts the machine learning method; the Raman spectrum of the cell sample to be tested is collected, and the data with the normalized positioning coefficient Ci in the range of 0.4~0.6 is selected as the test set data. The test set data is brought into the trained discrimination model to obtain the discrimination result, thereby realizing the judgment of the axial position of the excitation spot in the cell.

[0063] Brief description of the principle: In the cell discrimination method based on cell Raman spectroscopy, machine learning and other methods are usually used. The Raman spectra collected from known types of cells are used as the training set database to train the discrimination model. Then, for unknown types of samples, their Raman spectra are collected as the test set, and the obtained Raman spectra are substituted into the trained discrimination model to obtain the discrimination results. In order to ensure the accuracy of discrimination, the experimental conditions for Raman spectrum collection need to be consistent for the Raman spectra of the training set samples and the subsequent test set samples. Therefore, the positioning of the excitation position of the cell is crucial, but there is currently no axial positioning method for cell Raman testing, and the consistency of the test cannot be guaranteed, resulting in a decrease in analysis accuracy. At the same time, when collecting cell Raman spectra, the best axial position is the center of the cell. The molecular components in the center of the cell are rich and the characteristics are obviously different. At the same time, impurities are easily adsorbed near the glass slide to produce impurity Raman spectra, which interferes with the analysis results. Therefore, when establishing a cell training model and testing unknown samples to be tested, effective sample Raman data is selected based on the method provided by the present invention to construct a training set and process the samples to be tested, the test conditions are constructed according to the present invention, different glass slides are selected to make samples, and the standard Raman spectra of each glass slide are tested. The test data are processed by the algorithm of the present invention, and the normalized positioning coefficient C is calculated. It is found that in samples with a large or small normalization coefficient, impurity interference spectral lines will be observed, and when based on the principal component analysis-linear discriminant analysis (PCA-LDA) algorithm, the accuracy of the discriminant model constructed directly using all the data is low. According to the normalized axial positioning coefficient in the present invention, data with C of about 0.5 is selected, and the discrimination accuracy of the constructed discriminant model is significantly improved. The discrimination process is applied as follows Figure 3 shown.

[0064] The normalized axial positioning coefficient is used as a standard to measure the consistency of the sample collection of the test set and the training set. When establishing the training set, the axial positioning coefficient of the cell Raman spectrum test in the training set needs to be between 0.4 and 0.6. After selecting the data to establish the sample library, the sample data is reduced in dimension based on principal component analysis, and the number of principal components is selected so that the corresponding variance of the selected principal components accounts for more than 80% of the total. Then the discriminant model is established based on the linear discriminant analysis method. When this model is actually used for cell identification, the normalized axial positioning coefficient of the Raman spectrum of the test sample must also be between 0.4 and 0.6, otherwise it is unqualified data; the data is substituted into the discriminant model to obtain the discriminant result. Selecting the spectrum at the same excitation position for discriminant model training can ensure the consistency of the data, and in the subsequent discrimination analysis, select the test Raman spectrum with the same normalized positioning coefficient as the qualified test sample Raman spectrum, and substitute it into the model for discriminant analysis, so as to ensure the consistency of the training set spectrum and the test set spectrum sampling in the discriminant model and improve the accuracy of the analysis.

[0065] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the disclosure of the present invention can be performed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in the present invention can be achieved, and this document does not limit this.

[0066] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for precise axial positioning and judgment of cell microscopic Raman detection, characterized in that: The specific steps include: S1. Select two glass slides of different materials with different Raman spectra as slide glass and cover glass respectively; obtain the normalized standard Raman spectra of the slide glass and cover glass respectively through testing; S2. Fix the cell sample to be tested between the slide and the cover glass to prepare a test sample; place the test sample under the microscope objective lens, adjust the focus under ordinary white light illumination, and find the position of the target cell; S3. After finding the target cell, the light source is switched to laser, a focused light spot is formed on the cell sample to excite the Raman signal, and the Raman spectrum of the sample is detected at this time; S4. Use the least square method to solve the intensity coefficients of the Raman spectra of the cover glass and the slide, construct a related linear equation group, and calculate the Raman spectrum of the cell sample; The method for constructing the related linear equation group is as follows: construct the related linear equation group according to the normalized standard Raman spectra of the cover glass and the slide and the intensity values ​​of each wave number of the Raman spectrum of the test sample; the expression of the related linear equation group is as follows: ； Where, S is the total Raman spectrum of the test obtained by detection, S1 and S2 are the normalized standard Raman spectra of the cover glass and the slide, respectively, A1 and A2 are the intensity coefficients of the Raman spectra of the cover glass and the slide during detection, respectively, R is the Raman spectrum of the cell sample, ωi is the Raman frequency shift, and n is the number of spectral channels in the Raman spectrum; The expression of the Raman spectrum of the cell sample is as follows: ； Where, S is the total Raman spectrum of the test obtained by detection, S1 and S2 are the normalized standard Raman spectra of the cover glass and the slide, respectively, A1 and A2 are the intensity coefficients of the Raman spectra of the cover glass and the slide during detection, respectively, and R is the Raman spectrum of the cell sample; S5. constructing a normalized positioning coefficient of the light spot axis according to the intensity coefficient of the Raman spectra of the cover glass and the slide; S6. Construct a discriminant model; use the Raman spectra collected from known types of cell samples as a training set database to train the discriminant model; use the trained discriminant model to discriminate the cell samples to be tested to obtain the discrimination results.

2. The method for accurate axial positioning and judgment of cell microscopic Raman detection according to claim 1, characterized in that: The minimum value of the intensity coefficient A1 of the cover glass Raman spectrum or the intensity coefficient A2 of the slide glass Raman spectrum is 0.

3. The method for accurate axial positioning and judgment of cell microscopic Raman detection according to claim 1, characterized in that: The method for finding the position of the target cell in step S2 is to move the detection sample to the focused light spot on the horizontal imaging plane.

4. The method for accurate axial positioning and judgment of cell microscopic Raman detection according to claim 1, characterized in that: The Raman spectrum of the sample includes the Raman spectrum of the cover glass, the Raman spectrum of the cell sample and the Raman spectrum of the slide glass.

5. The method for accurate axial positioning and judgment of cell microscopic Raman detection according to claim 1, characterized in that: The normalized positioning coefficient is shown in the following formula: ； Wherein, C1 represents the axial distance between the Raman test cover glass and the excitation spot, and C2 represents the distance between the Raman test slide and the excitation spot.

6. The method for accurate axial positioning and judgment of cell microscopic Raman detection according to claim 5, characterized in that: The method for constructing the discriminant model in step S6 specifically includes: selecting data with a normalized positioning coefficient Ci value in the range of 0.4 to 0.6 to establish a training set database; reducing the dimension of the training set data based on principal component analysis, selecting the number of principal components so that the corresponding variance of the selected principal components accounts for more than 80% of the total, and then establishing a discriminant model based on a linear discriminant analysis method.

7. The method for precise axial positioning and judgment of cell microscopic Raman detection according to claim 6, characterized in that: In step S6, the discriminant model is trained using a machine learning method; The discrimination method specifically includes: collecting Raman spectra of the cell samples to be tested, selecting data with normalized localization coefficients Ci in the range of 0.4 to 0.6 as test set data, bringing the test set data into the trained discrimination model to obtain discrimination results.

Images