Automatic removal method of useless scattering in fluorescence spectrum
By fitting and interpolation processing of fluorescence spectral data, the scattered data in the three-dimensional fluorescence spectrum is automatically identified and removed, and the problem of automatic identification and removal of scattering in the prior art is solved, and the efficiency and accuracy of data processing are improved.
Patent Information
- Application Number
- CN202110001826.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-01-04
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2041-01-04
AI Technical Summary
The existing fluorescence spectroscopy data processing methods cannot automatically identify and remove useless scattered data in the three-dimensional fluorescence spectrum, and rely on artificial parameters, resulting in low processing efficiency and large errors.
By preprocessing the fluorescence spectral data, fitting the fluorescence intensity curve, automatically identifying the scattering position and range, and removing the scattering data using the least squares method and interpolation method, data completion and integration are achieved.
Automatic scattered data removal is realized, processing speed and accuracy is improved, human error is reduced, and the data obtained is more accurate and fast.
Smart Images

Figure CN114720433B_ABST
Abstract
Description
Technical Field
[0001] The invention utilizes a numerical calculation method to remove useless scattering data in a three-dimensional fluorescence spectrum, belongs to the field of analytical chemistry, and specifically relates to a calculation method for automatically identifying a scattering area, removing scattering data and filling the removed part. Background Art
[0002] Fluorescence excitation emission matrix measurement has been widely used in food science, analytical chemistry, biochemistry and environmental science. However, due to the generation mechanism of fluorescence spectrum, Rayleigh scattering and Raman scattering exist, and it is often complicated to analyze and remove them. At present, there are several methods to deal with the influence of useless scattering. However, most of the existing methods have corresponding shortcomings and areas for improvement. They often can only process sample data once, and cannot automatically identify the scattering position in the process of processing data, but can only subjectively judge the scattering position.
[0003] The present invention is an improvement based on the existing de-scattering method. The original method relies on manually set parameters and requires the user of the spectrum to subjectively judge the area covered by the useless scattered energy. The present invention is based on the spectral data itself and automatically judges the position and range of the scattering by processing the image data, so as to achieve the purpose of de-scattering and completing the data. Summary of the invention
[0004] The purpose of the present invention is to provide a calculation method for efficiently and automatically removing scattering.
[0005] The technical solution adopted by the present invention to achieve the above-mentioned purpose is:
[0006] The method for automatically removing useless scattering of fluorescence spectrum comprises the following steps:
[0007] 1) Obtaining fluorescence spectrum data in organic matter through a fluorescence spectrometer and preprocessing the fluorescence spectrum data;
[0008] 2) taking out a number of data with fluorescence intensity greater than a threshold value from the preprocessed fluorescence spectrum data, and fitting them to obtain fitting parameters;
[0009] 3) According to the fitting parameters, construct the expression of fluorescence spectrum data: y=kx+b, where y is the excitation wavelength of the fluorescence spectrum, x is the emission wavelength of the fluorescence spectrum, and k and b are fitting parameters;
[0010] 4) When x is a certain value, the maximum value of the fluorescence intensity in the fluorescence spectrum range corresponding to y;
[0011] 5) processing the maximum fluorescence intensity in step 4) to remove the scattering at the emission wavelength;
[0012] 6) Repeat steps 4) to 5) until all fluorescence spectra are processed, and integrate the processed fluorescence spectra at each emission wavelength to obtain a complete fluorescence spectrum;
[0013] 7) Determine whether there is still scattering in the processed fluorescence spectrum. If not, output the fluorescence spectrum to the host computer for visualization; if so, repeat steps 2) to 6) to continue processing it, and output the processed fluorescence spectrum to the host computer for visualization.
[0014] The scattering includes: Rayleigh scattering and Raman scattering, wherein the fitting parameters of Rayleigh scattering and Raman scattering are different. If scattering still exists in step 7), the existing scattering is Raman scattering.
[0015] The least squares method was used to fit the preprocessed fluorescence spectrum data.
[0016] The processing of the maximum fluorescence intensity in step 5) is specifically as follows:
[0017] Smooth the fluorescence intensity curve and find two valleys adjacent to the peak on the left and right sides of the curve;
[0018] Remove the values between two troughs;
[0019] The curve with the data removed is interpolated to obtain the fluorescence intensity curve with scattering removed.
[0020] Fluorescence intensity curves were smoothed using a mean filter.
[0021] The present invention has the following beneficial effects and advantages:
[0022] 1. The present invention can accurately and automatically identify the scattering position and range, avoids the subjectivity of manually inputting the scattering range, increases the universality of the algorithm, and obtains data results faster and in a shorter time.
[0023] 2. The present invention adopts a more accurate method to identify the scattering interval. Compared with the original algorithm that manually inputs the scattering interval, it is more accurate, greatly reduces the error, and is more in line with reality.
[0024] 3. The present invention uses interpolation method to fit the data, reflecting the properties of the non-scattering part of the spectral data. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 It is a flowchart of automatic descattering of three-dimensional spectra;
[0026] Figure 2 is the three-dimensional spectrum of Raman scattering and Rayleigh scattering to be processed;
[0027] Figure 3 It is the flow chart of the highest point of energy retention;
[0028] Figure 4a It is a schematic diagram of the original data of excitation wavelength and fluorescence intensity;
[0029] Figure 4b It is a schematic diagram of the smoothing process of the original data;
[0030] Figure 4c It is a schematic diagram of the first-order derivative of the processed data;
[0031] Figure 4d It is a schematic diagram of the second-order derivative of the processed data;
[0032] Figure 5 is the data before descattering Figure 1 ;
[0033] Figure 6 is the data of the descattered energy value Figure 1 ;
[0034] Figure 7 is the data after de-scattering and interpolation Figure 1 ;
[0035] Figure 8 is the data before descattering Figure 2 ;
[0036] Fig. 9 is the data of the descattered energy value Figure 2 ;
[0037] Fig.10 is the data after de-scattering and interpolation Figure 2 ;
[0038] Fig.11 It is a three-dimensional spectrum without Raman scattering and Rayleigh scattering;
[0039] Fig.12a This is a schematic diagram of the three-dimensional spectral data of Raman scattering and Rayleigh scattering to be processed Figure 1 ;
[0040] Figure 12b This is a schematic diagram of three-dimensional spectral data after removing Raman scattering and Rayleigh scattering. Figure 2 ;
[0041] Fig.13a This is a schematic diagram of the three-dimensional spectral data of Raman scattering and Rayleigh scattering to be processed Figure 1 ;
[0042] Fig.13b This is a schematic diagram of three-dimensional spectral data after removing Raman scattering and Rayleigh scattering. Figure 2 ;
[0043] Fig.14a This is a schematic diagram of the three-dimensional spectral data of Raman scattering and Rayleigh scattering to be processed Figure 1 ;
[0044] Fig.14b This is a schematic diagram of three-dimensional spectral data after removing Raman scattering and Rayleigh scattering. Figure 2 ;
[0045] Fig.15a This is a schematic diagram of the three-dimensional spectral data of Raman scattering and Rayleigh scattering to be processed Figure 1 ;
[0046] Fig.15b This is a schematic diagram of three-dimensional spectral data after removing Raman scattering and Rayleigh scattering. Figure 2 ;
[0047] Fig.16a This is a schematic diagram of the three-dimensional spectral data of Raman scattering and Rayleigh scattering to be processed Figure 1 ;
[0048] Fig.16b This is a schematic diagram of three-dimensional spectral data after removing Raman scattering and Rayleigh scattering. Figure 2 ;
[0049] Fig.17a This is a schematic diagram of the three-dimensional spectral data of Raman scattering and Rayleigh scattering to be processed Figure 1 ;
[0050] Fig.17b This is a schematic diagram of three-dimensional spectral data after removing Raman scattering and Rayleigh scattering. Figure 2 ;
[0051] Fig.18a This is a schematic diagram of the three-dimensional spectral data of Raman scattering and Rayleigh scattering to be processed Figure 1 ; Fig.18b This is a schematic diagram of three-dimensional spectral data after removing Raman scattering and Rayleigh scattering. Figure 2 . DETAILED DESCRIPTION
[0052] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.
[0053] Three-dimensional fluorescence spectroscopy is a fluorescence analysis technique that has been widely used and developed in recent years. It is widely used in the fields of food science, analytical chemistry, biochemistry and environmental science. The three-dimensional fluorescence spectrum is a matrix spectrum characterized by the three-dimensional coordinates of the excitation wavelength (Emission, Em), the emission wavelength (Excitation, Ex), and the fluorescence intensity (the fluorescence intensity value is represented by Z). It is a graph that describes the relationship between the fluorescence intensity and the changes with the excitation wavelength and the emission wavelength at the same time. The three-dimensional fluorescence spectrum can not only obtain the excitation wavelength and the emission wavelength, but also obtain the fluorescence intensity information when it changes.
[0054] There are currently several methods for dealing with scattering effects. However, due to the generation mechanism of the fluorescence spectrum, there is often energy corresponding to Rayleigh scattering and Raman scattering in the fluorescence spectrum. The amplitude of these energies is relatively large, which often has a huge impact on the analysis of the target object. Therefore, before analyzing the energy of the target object, it is necessary to remove it. The present invention is an improvement based on the original three-dimensional fluorescence spectrum descattering algorithm. The three-dimensional fluorescence spectrum is composed of several two-dimensional fluorescence spectra. All the descriptions in the present invention are based on the two-dimensional fluorescence spectrum. Figure 1 As shown, the following steps are included:
[0055] First, for the two-dimensional spectral data, the scattering position is determined by the Raman scattering and Rayleigh scattering mechanisms and their corresponding energies; then, based on the one-dimensional spectral data, the coverage of the scattering energy is determined by the scattering position and scattering intensity, and the useless energy is eliminated; finally, the area where the scattering data is removed is interpolated to achieve the purpose of removing useless scattering and completing the gap.
[0056] Iteratively selecting high energy points, fitting the unwanted scattering position expression, and determining the position of the highest energy point of the unwanted scattering according to the unwanted scattering position straight line expression, including the following steps:
[0057] Import the three-dimensional spectral fluorescence data affected by scattering. Generally, the energy value of Raman scattering or Rayleigh scattering is higher than the scattering energy value of the target object, and it appears as an approximate straight line area on the fluorescence spectrum. Take the highest energy point of the Raman scattering or Rayleigh scattering part, and through iteration, continuously judge whether the fitted scattering position is reasonable. When processing two-dimensional fluorescence data, the position of the highest point of useless scattering energy is judged by the straight line expression.
[0058] Smooth the data, determine the troughs on both sides of the scattering according to the highest energy point of the useless scattering, and remove the useless scattering area, including the following steps:
[0059] The moving mean filter is used to smooth the fluorescence data column vector. The highest energy point determined in the previous step is used to search for the corresponding troughs on both sides, determine the interval position of useless scattering, and remove the scattering data in the useless scattering interval.
[0060] The present invention provides a method for calculating the removal of useless scattering based on three-dimensional fluorescence spectrum. The specific steps of the method are as follows:
[0061] (1) Fitting the linear expression of Rayleigh scattering
[0062] In order to analyze the capability of this algorithm, a set of three-dimensional fluorescence spectrum data was simulated and generated. The data set contains Raman scattering, Rayleigh scattering and fluorescence data of target scattering, which are expressed in the form of contour lines. Figure 2 Take this data set as an example for explanation.
[0063] Instructions for the first step of importing fluorescence spectral data.
[0064] Through the fluorescence spectrometer, select the appropriate excitation wavelength and emission wavelength, adjust the parameters to scan the sample, record the changes in fluorescence intensity, and obtain a three-dimensional spectrum. The three-dimensional fluorescence spectrum of each sample generates a fluorescence data that represents the fluorescence characteristic spectrum of the organic matter in the sample.
[0065] Fluorescence data is a matrix spectrum characterized by the three-dimensional coordinates of excitation wavelength em (corresponding to the y-axis in the figure), emission wavelength ex (corresponding to the x-axis in the figure), and fluorescence intensity (corresponding to the z-axis in the figure). The experimental data set is measured in nanometer intervals, with the excitation wavelength between 200-250 nanometers and the emission wavelength range of 200-250 nanometers.
[0066] The data is stored in the form of dat.xls or dat.txt. The data of the three-dimensional fluorescence spectrum to be descattered is imported in the form of a matrix. The horizontal axis of the matrix is the column number of the matrix (counting starts from 1), and the vertical axis of the matrix is the row number of the matrix. The values corresponding to the specific rows and columns are the fluorescence intensity data of the point (z (l,r) is the specific fluorescence intensity value, l and r are the coordinates of the location). For example, the data matrix is
[0067] With z (50,1) For example, its horizontal coordinate is 50, the vertical coordinate is 1, and the z-axis coordinate is z (50,1) . It means that when the emission wavelength is 50nm and the excitation wavelength is 1nm, the fluorescence intensity is z (50,1) .
[0068] like Figure 2As shown, the three-dimensional fluorescence spectral intensity data is visualized in the form of contour lines. The horizontal axis is the emission wavelength ex; the vertical axis is the excitation wavelength em; the value of the contour line represents the magnitude of the fluorescence intensity, starting from 0.2, with an interval of 0.2 and ending at 1.4. The darker the color of the contour line, the greater the fluorescence intensity. From the contour map of the fluorescence spectrum generated by the data set, it can be observed that there are two contour line data close to a straight line on both sides of the middle circular contour line. The middle circular contour line is the fluorescence data not affected by scattering, and the two contour lines close to a straight line are Raman scattering and Rayleigh scattering data.
[0069] Step 2: Retain the points with high energy values in the data.
[0070] As Figure 3 shown, in the three-dimensional spectral map affected by scattering, generally, the energy value of useless scattering is higher than the fluorescence energy value of the target scattering, presenting as an approximately straight-line region on the three-dimensional fluorescence spectrum. Take the highest energy point of the scattering part and fit the linear expression of the scattering position and energy value. When the slope and intercept of the Rayleigh scattering linear expression are within a certain range, the position of Rayleigh scattering is defined as the excitation wavelength equal to the emission wavelength, that is, the slope (k represents the slope) is around 1, and the intercept (b represents the intercept) is less than 0. To facilitate judging the value range of the slope, and the slope is around 1 with an error not exceeding 0.5, so there is a probability of Rayleigh scattering when the slope is between 0.5 - 1.5. That is, the range of the intercept and slope of the Rayleigh scattering linear expression is b < 0, 0.5 < k < 1.5.
[0071] Judge through the slope and intercept of the fitted linear expression. The expression within the range is the linear expression of the scattering position and energy value (where y corresponds to the value of the excitation wavelength em, and x corresponds to the value of the emission wavelength ex). The specific operation process is as follows:
[0072] First, retain some high-energy points. Retain the data of high-energy points, and set the data points outside high energy to 0. As For the fluorescence matrix that only retains high-energy points, fit the straight line y = kx + b by the least squares method, where n is the number of coordinates. Judge whether the slope k and intercept b of the fitted straight line are within the range of b < 0, 0.5 < k < 1.5.
[0073] If it is within the range of b < 0, 0.5 < k < 1.5, directly output the slope and intercept of the fitted straight line;
[0074] If it is not within the range of b < 0, 0.5 < k < 1.5, that is, b ≥ 0 or k ≥ 1.5 or k ≤ 0.5, increase the number of high-energy points. The process of increasing high-energy points is achieved by increasing the number of points nat% in the spectrum for each iterative fit. Nat is defined as 1 - 5. As If it is still not within the range of b < 0 and 0.5 < k < 1.5 for the slope and intercept, continue to increment the high-energy point by 0.001%, e.g., Until the slope and intercept are within the range of b < 0 and 0.5 < k < 1.5, output the high-energy point, the slope k of the fitted line, and the intercept b.
[0075] Step 3: Fit the expression of the scattered line and find the highest energy position of the scatter.
[0076] According to the slope k and intercept b of the line fitted for the corresponding range of useless scatter, form the line expression of the scatter y = kx + b. Among them, y and x are the two variables of the line expression, which are the positions of the coordinates where the highest energy point of the scatter is located, and are the abscissa ex and ordinate em of the input data matrix respectively.
[0077] To further remove the scatter, it is necessary to extract each group of data affected by the scatter. Set the emission wavelength ex (corresponding to the x-axis in the figure) as a fixed value, and the excitation wavelength em (corresponding to the y-axis in the figure) and the fluorescence intensity value (corresponding to the z-axis in the figure) form a group of data. Then, according to the line expression of the scatter y = kx + b, determine the value of the excitation wavelength em when the emission wavelength ex is a fixed value. The obtained excitation wavelength em is the position of the highest energy of the scatter within this group of data. For example Figure 2 As shown, extract the numbers corresponding to the y-axis coordinate and z-axis coordinate when the x-axis coordinate is fixed at 20; next, according to the line expression of Rayleigh scatter y = kx + b, determine the position y of the highest energy point of the excitation wavelength at x = 20 Figure 2 In which when x = 20, y is approximately around 20.
[0078] (2) Remove Rayleigh scatter according to the scatter position
[0079] The fluorescence data sets affected by scatter generally can appear in two types. Such as Figure 5 , Figure 8 As shown, visualize the excitation wavelength em data in the form of a two-dimensional curve. Among them, the horizontal axis is the excitation wavelength em; the vertical axis is the intensity of the excitation wavelength under the determined emission wavelength ex (for example Figure 2 in which, when the x-axis is equal to 20, the change curves of the y-axis and z-axis); among them, the dashed line is the actual fluorescence data set, the thick line is the data set affected by scatter, and the thin line is the ideal fluorescence data set without scatter influence. The purpose of the present invention is to remove the scatter influence of the thick line according to the actual fluorescence data set of the dashed line, and finally generate a scatter-removed data set approximating the thin line.
[0080] One situation is Figure 5 As shown, the data set affected by scatter (thick line) and the fluorescence data set without scatter (thin line) are relatively close, and there is an overlapping part between the two data. Another situation is Figure 8 As shown, the distance between the data set affected by scattering (thick line) and the fluorescence data set without scattering (thin line) is relatively far, and there is no overlap between the two data types. The two data types are processed separately.
[0081] The fourth step is to smooth the curve.
[0082] In Figure 4, Figure 4a It indicates that the excitation wavelength em data is visualized in the form of a two-dimensional curve, where the horizontal axis is the excitation wavelength em, and the vertical axis is the raw data of the intensity of the excitation wavelength under the determined emission wavelength ex; Figure 4b Then in Figure 4a The data curve is represented by smoothing the two-dimensional curve on the basis of Figure 4c is Figure 4b Take the first-order derivative of the data under the data, and the position marked by the triangle is the point where the derivative is 0; Figure 4d is Figure 4b The second-order derivative of the data is taken under the data, and the circular mark is the point where the second-order derivative is 0.
[0083] Since there are certain fluctuations in the actual spectrum, such as Figure 4a As shown, the calculation results are affected. Each curve affected by scattering is smoothed without affecting the original data, as shown in Figure 4b As shown. A moving mean filter can be used to filter the fluorescence data column vector z n Perform smoothing, the window width is kk = 2m + 1, m is the number of nodes. Return the denoised data column vector Z n When the original data point z n (w) When the sum of the accumulated elements before and after can be equal to the window width kk, When the original data point z n (w) When the accumulated elements before and after are not equal to the window width kk, remove z n (w) The nearest symmetric element outside will include z n (w) is accumulated and then divided by the total number of elements. For example, the window width of the moving mean filter is 5, Z n The elements in are calculated as follows:
[0084] Z n (1) = z n (1)
[0085] Z n (2) = (z n (1)+z n (2)+z n (3)) / 3
[0086] Z n (3)=(z n(1)+z n (2)+z n (3)+z n (4)+z n (5)) / 5
[0087] Z n (4)=(z n (2)+z n (3)+z n (4)+z n (5)+z n (6)) / 5
[0088] Z n (5)=(z n (3)+z n (4)+z n (5)+z n (6)+z n (7)) / 5
[0089] …
[0090] The curve z n Each data point in is denoised and saved as Z n =[Z n (1),Z n (2),Z n (3)…Z(w)].
[0091] Step 5: Find the troughs on both sides of the scatter.
[0092] According to the position of the highest energy scattered by each curve, determine the interval position of the scattering, find the troughs on both sides of the highest energy point, and find out through actual operation Figure 5 and Figure 8 In both cases, the trough position can be found quickly. The specific operations are as follows:
[0093] a) Calculate all the valleys in the curve
[0094] Z n (w) is the set of data after smoothing and noise reduction, Z n (w) is a smooth curve, then Z n (w) is derived to obtain Z′ n (w)(first-order derivative), and find Z n ″(w)=0 (the second-order derivative is equal to zero), judge Z′ n (w) The sign of the first-order derivative on both sides. If Z′ n (w-1)<0Z′ n (w+1)>0 then Z n(w) reaches its minimum value at this root, which is the trough.
[0095] As shown in Figure 4, Figure 4a is the original data set; Figure 4b is the smoothed data set, and its point marks are the locations of all troughs, and the positions of the arrows on both sides are the trough positions on both sides of the peak; Figure 4-3 is the first-order derivative of the data set, and its point marks are the second-order derivative is 0 and Z′ n (w-1)<0Z′ n Points where (w+1)>0; Figure 4d is the second-order derivative, and its point is marked as the point where the second-order derivative is 0. In the second-order derivative graph ( Figure 4d ) to determine Z″ n (w) = 0, such as Figure 4d The point identification in; then determine the first-order derivative graph ( Figure 4c ) in Z′ n (w-1)<0Z′ n The value of m when (w+1)>0, such as Figure 4c Finally, determine the location of the trough in the data set. Figure 4b The boxes in the figure mark the trough locations.
[0096] b) Calculate the two troughs before and after
[0097] After all the troughs are determined, the position of the highest energy point is used to determine the two troughs on the left and right that are closest to the point. Where (x1, y1) is the coordinate of the trough, and the peak is (x2, y2). Determine the minimum R value of the trough on the left and right sides of the peak. Figure 4b The arrows in the figure indicate the positions of the left and right troughs.
[0098] The areas within these two troughs are the scattered areas that need to be removed. For example, if the data set is [Z n (1),Z n (2),Z n (3),Z n (4),Z n (5),Z n (6),Z n (7),Z n (8),Z n (9),Z n (10)], the two valleys are Z n (4) and Z n (9), then Z n (4) and Z n (9) and the range between them is the scattering area to be removed.
[0099] Step 6: Remove the influence of scattering
[0100] After determining the positions of the front and rear troughs, the data within the trough range on both sides are set to empty set values. For example, if the data set is [Z n (1),Z n (2),Z n (3),Z n (4),Z n (5),Z n (6),Z n (7),Z n (8),Z n (9),Z n (10)], the two valleys are Z n (4) and Z n (9), then the data set after removing the scattering effect of this curve is [Z n (1),Z n (2),Z n (3),NAN,NAN,NAN,NAN,NAN,NAN,Z n (10)], NAN stands for non-data type (not a number).
[0101] like Figure 6 , Fig. 9 Shown in Figure 5 , Figure 8 The image after removing the scattered part based on the data, where the horizontal axis is the excitation wavelength em; the vertical axis is the intensity of the excitation wavelength under the determined emission wavelength ex; the dotted line is the data set with an empty set, the thick line is the data set affected by scattering, and the thin line is the ideal fluorescence data set without scattering influence. At this time, the scattered data has been removed.
[0102] The seventh step is to interpolate the removed curve.
[0103] After removing useless scattered data, the changing trend of scattered data is determined based on the remaining data. A complete set of data is obtained. Figure 7 , Fig.10 As shown, Figure 6 , Fig. 9 The image is interpolated after removing useless data based on the data, where the horizontal axis is the excitation wavelength em; the vertical axis is the intensity of the excitation wavelength under the determined emission wavelength ex; the dotted line is the data set after interpolation processing, the thick line is the data set affected by scattering, and the thin line is the ideal fluorescence data set without scattering influence.
[0104] The specific formula is as follows:
[0105] Assume there are n+1 data nodes (y0,z0),(y1,z1),(y2,z2)......(y n ,zn ), y and z correspond to the values of excitation wavelength em and fluorescence intensity z, n is the number of points of emission wavelength, and the calculation step length h i =y i+1 -y i (i=0,1,…,n-1), put the data nodes and the specified first endpoint conditions into the matrix equation. Solve the matrix equation and obtain the quadratic differential value m i . This matrix is tridiagonal. Calculate the coefficients of the spline curve:
[0106] a i =z i
[0107]
[0108]
[0109]
[0110] Where i = 0, 1, ..., n-1. In each subinterval x i ≤x≤x i+1 , create the equation
[0111] g i (y) = a i +b i (yy i )+c i (yy i ) 2 +d i (yy i ) 3 , a i b i c i d i is the coefficient corresponding to the interpolated curve, g i (y) is the curve equation after interpolation.
[0112] In actual calculation practice, it is found that no matter the distance between the scattering peak and the fluorescence data is close or far, the method of processing the scattering is the same, so both cases can be
[0113] To adopt this algorithm.
[0114] The eighth step is to integrate and obtain the three-dimensional spectrum after removing the scattering.
[0115] Each set of data is processed separately to obtain g i (y)(i=0,1,…,n-1), and then reorganize the processed data into a three-dimensional spectrum matrix Draw the image, and the Rayleigh scattering has been completely removed.
[0116] (3) Remove Raman scattering
[0117] Some of the fluorescence spectra have only one type of scattering, while the other part of the fluorescence spectra have both Raman and Rayleigh scattering. Therefore, it is necessary to judge the spectra. Judge whether there are two types of scattering. If there are two types of scattering, they need to be processed separately; if there is only one type of scattering, directly output the matrix of the three-dimensional spectrum in the previous step That's it.
[0118] Step 9 Judge whether there is Raman scattering
[0119] Judge whether there is Raman scattering. By fitting the highest energy point, judge the range of the slope (v represents the slope) and intercept (u represents the intercept) of the straight line expression of the scattering. The position of Raman scattering is at the place where the excitation wavelength is twice the emission wavelength, that is, at the slope of 2. In order to facilitate judging the range of the slope value, and the slope is around 2 with an error not exceeding 0.5, so there is a probability of Raman scattering when the slope is between 1.5 and 3. That is, the range of the slope of the Raman scattering straight line expression is 1.5 < v < 3. If it is not within the range, it means that the spectrum only has Rayleigh scattering, and output the matrix of the three-dimensional fluorescence spectrum If it is within the range, it means that there is Raman scattering, and use the algorithm in the previous step to process and output the matrix of the three-dimensional fluorescence spectrum (i = 0, 1, …, n - 1).
[0120] (4) Output the image after removing scattering
[0121] After processing the scattering, output the data matrix G. The abscissa of this matrix is the number of columns of the matrix representing the emission wavelength ex, and the ordinate is the number of rows of the matrix representing the excitation wavelength em. The specific numerical values corresponding to the rows and columns are the fluorescence intensity data at that point
[0122] By drawing a graph, analyze the three-dimensional fluorescence spectrum diagram after removing scattering, as Fig.11 shown, it is Figure 2 The three-dimensional fluorescence spectrum diagram after removing useless scattering. The horizontal axis is the emission wavelength ex; the vertical axis is the excitation wavelength em; the value of the contour line represents the magnitude of the fluorescence intensity, starting from 0.2, with an interval of 0.2, and ending at 1.4. The darker the color of the contour line, the greater the fluorescence intensity. The Raman scattering and Rayleigh scattering in the figure have been completely removed.
[0123] Figures 12 to 18 visualize the three-dimensional fluorescence spectrum intensity data in the form of contour lines. The horizontal axis is the emission wavelength ex; the vertical axis is the excitation wavelength em; the value of the contour line represents the magnitude of the fluorescence intensity, and the darker the contour line, the greater the fluorescence intensity. The left side a is the original data image, and the right side b is the three-dimensional fluorescence spectrum image after removing the scattering. At this time, the three-dimensional fluorescence spectrum image after removing the useless scattering is observed, and the Raman scattering and Rayleigh scattering that are approximately linear have been removed.
[0124] The present invention can quickly and automatically identify the location of useless scattering, without manually setting the width of scattering energy, and retain data to the maximum extent within an acceptable error range, which greatly improves the accuracy and efficiency of the scattering removal process. This makes it possible to realize batch processing of a large amount of spectral data containing scattering.
Claims
1. A method for automatically removing useless scattering from fluorescence spectrum, characterized in that: The following steps are involved: 1) Obtain fluorescence spectrum data in organic matter through fluorescence spectrometer and preprocess the fluorescence spectrum data; 2) taking out a number of data with fluorescence intensity greater than a threshold value from the preprocessed fluorescence spectrum data, and fitting them to obtain fitting parameters; 3) According to the fitting parameters, construct the expression of fluorescence spectrum data: y=kx+b, where y is the excitation wavelength of the fluorescence spectrum, x is the emission wavelength of the fluorescence spectrum, k and b are fitting parameters, and judge whether the slope intercept is within the specified range. If it is within the range, output the intercept and slope of the straight line. Otherwise, relax the selection criteria of energy points, increase the number of high energy points and refit the straight line equation; 4) When x is a certain value, the maximum value of the fluorescence intensity in the fluorescence spectrum range corresponding to y; 5) Processing the maximum fluorescence intensity in step 4) to remove the scattering at the emission wavelength; 6) Repeat steps 4) to 5) until all fluorescence spectra are processed, and integrate the processed fluorescence spectra at each emission wavelength to obtain a complete fluorescence spectrum; 7) Determine whether there is still scattering in the processed fluorescence spectrum. If not, output the fluorescence spectrum to the host computer for visualization; if so, repeat steps 2) to 6) to continue processing it, and output the processed fluorescence spectrum to the host computer for visualization; The scattering includes: Rayleigh scattering and Raman scattering, wherein the fitting parameters of Rayleigh scattering and Raman scattering are different. If scattering still exists in step 7), the existing scattering is Raman scattering.
2. The method for automatically removing useless scattering in fluorescence spectrum according to claim 1, characterized in that: The least squares method was used to fit the preprocessed fluorescence spectrum data.
3. The method for automatically removing useless scattering in fluorescence spectrum according to claim 1, characterized in that: The processing of the maximum fluorescence intensity in step 5) is specifically as follows: Smooth the fluorescence intensity curve and find two valleys adjacent to the peak on the left and right sides of the curve; Remove the values between two troughs; The curve with the data removed is interpolated to obtain the fluorescence intensity curve with scattering removed.
4. The method for automatically removing useless scattering in fluorescence spectrum according to claim 3, characterized in that: Fluorescence intensity curves were smoothed using a mean filter.
