Chromatographic image recognition method and system based on characteristic peaks

Abstract

The invention relates to the technical field of chromatogram analysis, in particular to a chromatographic image recognition method and system based on characteristic peaks. The method comprises the following steps: acquiring a chromatogram of a target sample, and preprocessing the chromatogram to obtain an initial atlas; obtaining a plurality of target areas in the initial map; dividing all the target areas to obtain a first area and a second area; calculating a first overlapping probability of each first region, and taking the corresponding first region when the first overlapping probability is greater than a first threshold value as an overlapping peak region; calculating a second overlapping probability of each second region, and taking the corresponding second region when the second overlapping probability is greater than a second threshold value as an overlapping peak region; and performing Gaussian fitting on all overlapped peak regions to obtain a curve of a plurality of peaks. According to the scheme provided by the invention, the overlapping peaks can be accurately identified.

Description

Technical Field

[0001] The present invention relates to the technical field of chromatogram analysis, and more specifically, to a chromatogram image recognition method and system based on characteristic peaks. Background Art

[0002] Online chromatographs are particularly in demand in the fields of petrochemicals, environmental monitoring, etc., because they can monitor and analyze chemical components in real time. Among them, the chromatogram obtained by the online chromatograph is a graphical representation with multiple characteristics, which shows the separation of different substances during the chromatographic analysis process and the relationship between their signal intensity and time or other variables (such as mobile phase volume, etc.). The peak shape, peak position, peak area, etc. in the chromatogram contain rich information. For example, the peak area is usually related to the content of the substance, and the peak position may be related to the properties of the substance.

[0003] In the related technology, for example, a patent application document for a chromatogram analysis and processing system based on mass spectrometer software disclosed in publication number CN118130689A discloses that when making a chromatogram data judgment, the data of each sample is first subjected to a Gaussian fitting peak curve, and then the symmetry of the peak shape and the presence of interference peaks are judged to determine whether the chromatogram data is qualified.

[0004] After obtaining the sample data, the above schemes directly perform Gaussian fitting of the sample data to obtain the peak curve. However, since the Gaussian function is usually a single peak, it does not take into account the chromatogram analysis. If there is a high degree of overlap of characteristic peaks and complex peak shapes, if Gaussian fitting is performed directly, it may lead to poor fitting results, and thus it is impossible to accurately identify the position of each peak in the overlapping peaks and the true shape of each peak in the overlapping peaks.

[0005] Therefore, it is also necessary to separate the overlapping peaks in the chromatogram. In the prior art, the following methods are usually used to separate the overlapping peaks in the chromatogram: (1) Geometric method: The geometric method is intuitive and has fast calculation speed. However, the accuracy of the geometric method often decreases with the increase of the overlap between peaks. When the overlap between multiple peaks is high, the error is large and it is often not applicable, especially for the analysis of tailing peaks and front extended peaks.

[0006] (2) Pattern recognition method: Traditional neural network algorithms do not need to consider overlapping peak types and have a wide range of applications. However, the network structure selection is often not single, the amount of calculation is large, and sometimes overfitting may occur.

[0007] (3) Algebraic method: The function fitting method has a wide range of applications. It does not need to consider the degree of overlap and can also be applied to overlapping peaks of the shoulder type. However, when separating multiple peaks, it is heavily dependent on the selected function model, and the selection of the function model often depends on experience. If different function models are selected, the calculation results will also be different. In addition, the unknown parameters in the model need to be constantly adjusted during the calculation, which results in a large amount of calculation.

[0008] Therefore, when faced with situations where peak overlap is high and peak shapes are complex, it is particularly important to accurately, timely and effectively identify chromatograms. Summary of the invention

[0009] The purpose of the present invention is to propose a chromatographic image recognition method and system based on characteristic peaks, so as to solve the problem in the prior art that chromatograms with high peak overlap and complex peak shapes cannot be accurately, timely and effectively identified; to this end, the present invention provides solutions in the following two aspects.

[0010] In a first aspect, the present invention provides a chromatographic image recognition method based on characteristic peaks, comprising: Acquire a chromatogram of a target sample, and preprocess the chromatogram to obtain an initial spectrum; Acquire multiple target regions in the initial map; Divide all target areas into a first area and a second area; wherein the first area includes a peak point, and the second area includes at least two peak points; Calculating the first overlap probability of each first region, and taking the first region corresponding to the first overlap probability being greater than the first threshold as the overlap peak region; the first overlap probability is inversely correlated with the symmetry degree and Gaussian fitting degree of the first region; Calculate the second overlap probability of each second region, and take the second region corresponding to the second overlap probability greater than the second threshold as the overlap peak region; for: ; It is The average of the slope values ​​on both sides of all peak points in the second region, , are the mean of the difference between two adjacent peak points at corresponding moments among all peak points and the mean of the difference in corresponding signal strengths, , Respectively The total number of times and the maximum signal strength in the second area, , , is the weight; Gaussian fitting was performed on all overlapping peak regions to obtain curves of multiple peaks.

[0011] The above scheme divides the target area (the first area and the second area) into smaller and easier to handle parts, that is, for the first area, by analyzing the Gaussian fit degree and left-right symmetry, it is possible to accurately evaluate whether the area is indeed a single, independent peak; and when the left and right are asymmetric or the fit is not ideal, it implies that there are actually overlapping peaks in the area and they need to be separated; for the second area, by observing the characteristics such as the gentleness of the change at the peak point and the horizontal and vertical axis differences between multiple peak points, it is possible to evaluate whether these peaks really overlap, or whether the morphology is sufficiently separated, and then split by the Gaussian function. That is, the present invention finely divides the target area, evaluates the distribution and symmetry of the peaks in each area, and analyzes the fluctuation situation, so as to better judge the degree of overlap and obtain more accurate initial parameters. Finally, the nonlinear least squares method is used to fit multiple Gaussian functions to more accurately decompose the overlapping peaks. This method can improve the accuracy and reliability of the deconvolution process in complex overlapping peak situations.

[0012] Optionally, the first overlap probability is: ; in, For the The Gaussian fit of the first region, For the The degree of symmetry of the first region, For the The first overlap probability that the first region belongs to the overlapping peak region, It is an exponential function with the natural constant e as its base.

[0013] The above scheme provides a method for accurately calculating the first overlap probability.

[0014] Optionally, the degree of symmetry is: ; in, For the The degree of symmetry of the first region, It is the left side of the peak point The signal strength at the moment, , Respectively The start and end time of the first area, It is the right side of the peak point The signal strength at the moment, is the median value, For the The peak point in the first region corresponds to the Signal strength at the moment.

[0015] Optionally, the Gaussian fitting degree for: ; is the signal strength at the jth moment in the ith first region, is the predicted signal strength at the jth moment in the fitted Gaussian curve, Indicates The corresponding signal strength mean at all times in the first area, For the The total number of moments in the first area.

[0016] The above solution can determine the fitting degree of the corresponding first area by performing preliminary Gaussian fitting in advance.

[0017] Optionally, the specific process of obtaining multiple target areas in the initial atlas is: In response to the difference between the signal intensity corresponding to any moment and the baseline signal intensity being greater than or equal to a set threshold, the corresponding moment is the target moment; Continuously adjacent moments among all target moments are divided into the same target area, and multiple target areas are obtained.

[0018] The above scheme reduces the computational complexity of subsequent fitting by screening out multiple target areas.

[0019] Optionally, the process of dividing all target areas into the first area and the second area is: obtaining the maximum points of the signal strength at each time in each target area, and selecting the points at the time corresponding to each maximum point as peak points; taking the target area with one peak point as the first area, and taking the target area with at least two peak points as the second area.

[0020] The above scheme can perform more refined analysis of areas corresponding to different numbers of peaks by dividing the target area.

[0021] Optionally, the peak point acquisition process is: Calculate the probability that the point corresponding to each time in each target area belongs to the peak point; In response to the probability being greater than a threshold, the point at the corresponding moment is a peak point; The probabilities are: ; in, It is The probability that the kth moment in the target area is the peak point, , , Respectively moment, moment, The signal strength corresponding to the moment, n is the total number of settings, a is the number of settings, is the normalization function.

[0022] Optionally, the process of fitting the overlapping peak regions to obtain a curve of multiple peaks for analyzing the target sample is: Obtain the starting position and ending position of each overlapping peak region; Based on the starting position and the end position of each overlapping peak region, and in combination with the Gaussian fitting algorithm, the Gaussian curve of each mid-peak point in the overlapping peak region is obtained; Based on the peak width and peak intensity information of each single peak, a linear regression equation is established to achieve quantitative analysis of the content of each element in the target sample.

[0023] Optionally, the preprocessing of the chromatogram to obtain the initial spectrum includes: performing denoising on the chromatogram using an adaptive filtering algorithm and a baseline correction algorithm.

[0024] In a second aspect, a chromatographic image recognition system based on characteristic peaks comprises: processor; The memory stores computer instructions for chromatographic image recognition based on characteristic peaks. When the computer instructions are executed by the processor, the system executes the above-mentioned chromatographic image recognition method based on characteristic peaks.

[0025] The beneficial effects of the present invention are: The solution of the present invention analyzes the distribution of overlapping peaks in the chromatogram and accurately locates the overlapping peaks according to the symmetry of the peaks and the speed of change of the peak points. Compared with the prior art of directly fitting overlapping peaks by Gaussian functions to obtain separated peaks, the solution of the present invention obtains complex overlapping peaks in the chromatogram by performing a refined analysis of the chromatogram to avoid the problem of directly fitting the entire image when there may be very slight differences between two peaks, resulting in fitting errors or erroneous confusion of different peaks. The overlapping areas screened out are then fitted respectively, thereby improving the efficiency of deconvolution and realizing accurate recognition of the chromatogram image of the characteristic peak. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 The flowchart schematically shows the steps of the chromatographic image recognition method based on characteristic peaks in this embodiment; Figure 2The structural block diagram of the chromatographic image recognition system based on characteristic peaks in this embodiment is schematically shown. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0028] When the present invention analyzes the chromatogram of the target sample with the existing deconvolution method, it is often fitted based on a specific peak shape model (such as a Gaussian peak). However, the actual shape of the chromatographic peak may be affected by many factors and deviate from the ideal peak shape (for example, the peak may show a tailing, front extension or other asymmetric shapes). In this case, the traditional algorithm may not be able to correctly capture the actual characteristics of the peak and it is difficult to locate the exact area of ​​the overlapping peaks.

[0029] Based on the above problems, the present invention provides a chromatographic image recognition method and system based on characteristic peaks.

[0030] Specifically, taking a target sample as an example, the chromatographic image recognition method based on characteristic peaks in this embodiment is introduced. Figure 1 As shown, the following steps are included: Step S1, performing chromatographic analysis on a target sample to obtain a chromatogram of the target sample, and preprocessing the chromatogram to obtain an initial spectrum.

[0031] The target sample in this embodiment may be a petroleum sample in a certain area, or may be dissolved gas in transformer oil in a certain substation, etc.

[0032] Taking the petroleum samples in a certain area as an example, in this embodiment, a gas chromatograph is used to separate and detect the petroleum samples. During the separation process, the chromatogram will show different peak shapes as each component is separated. In the final chromatogram after the chromatographic analysis is completed, the horizontal axis is the x-axis, which represents the retention time in seconds, reflecting the time required for the sample components to pass through the chromatographic column in the gas chromatograph, and the vertical axis is the y-axis, which represents the signal intensity of the detector response in mV (the unit of voltage - millivolt, 1mV=0.001V), reflecting the height of the peak; the area enclosed by the points corresponding to all moments and the horizontal axis is proportional to the concentration of the components in the sample.

[0033] Among them, when performing chromatographic analysis, due to the long acquisition time of the chromatogram, it may be affected by instrument noise and optical noise, resulting in baseline fluctuations. These baseline changes may interfere with the identification of peaks, causing disturbances in the chromatogram, which in turn affects subsequent analysis. Therefore, the chromatogram needs to be preprocessed to eliminate these noises and disturbances (baseline noise and base peak fluctuations will have a significant impact on the results of peak deconvolution; noise interference may cause the deconvolution algorithm to misjudge the number and position of peaks) to ensure the accuracy and reliability of the data.

[0034] Specifically, the process of preprocessing the chromatogram to obtain the initial spectrum is: using an adaptive filtering algorithm to remove the noise components in the final chromatogram, using a baseline correction algorithm to remove the influence of baseline drift in the final chromatogram, making the peak shape in the final chromatogram clearer, and obtaining the initial spectrum.

[0035] Among them, the adaptive filtering algorithm and the baseline correction algorithm are both existing technologies and will not be described in detail here.

[0036] Step S2, analyzing the distribution of the initial spectrum to obtain all overlapping peak regions.

[0037] The specific process of obtaining overlapping peak areas is: Step S21, obtaining multiple target areas in the initial atlas.

[0038] Specifically, the criterion for determining whether each moment in the initial spectrum belongs to the target area is that the difference between the signal intensity corresponding to any moment and the baseline signal intensity is greater than or equal to a set threshold.

[0039] Taking the tth moment as an example, the signal strength at the tth moment is recorded as , then the criterion for determining the target area at time t is: , that is, the signal strength at time t Baseline signal intensity corresponding to the baseline The difference is greater than or equal to the set threshold When the tth moment is considered to belong to the target area, the threshold is set ; Of course, it can also be determined based on actual conditions.

[0040] Then, all the consecutive moments in the target area are divided into the same target area to obtain the total target area.

[0041] The method for obtaining the above-mentioned baseline signal strength is: obtain the signal strength corresponding to each moment in the initial spectrum, and draw a signal strength histogram, with the horizontal axis being the signal strength, arranged from small to large, and the vertical axis being the number of corresponding signal strengths, and the peak value of the signal strength closest to the coordinate origin is taken as the baseline signal strength.

[0042] The above-mentioned baseline signal strength is the minimum signal strength value determined by analyzing all signal strengths. The reason for determining the baseline signal strength is that the baseline signal is small and stable, and since there are fewer components to be detected in petroleum analysis, the baseline area in the chromatogram corresponds to a longer time (the baseline area corresponds to the area without a peak); the area with a peak (the area higher than the baseline) corresponds to a shorter time.

[0043] Step S22, filtering out the peak points in each target area; when the number of the peak point is one, the corresponding target area is the first area, and when the number of the peak point is at least two, the corresponding target area is the second area.

[0044] After determining the target area, the peak points of each target area are screened out, as follows: By obtaining the maximum value points of the signal strength at each time in each target area, the points at the time corresponding to each maximum value point are selected as peak points.

[0045] In one embodiment, the signal strength corresponding to any moment is compared with the signal strength in its vicinity to determine the maximum point.

[0046] In another embodiment, the probability that the signal strength at each moment belongs to a peak point can also be calculated. When the corresponding probability is greater than a set value, the point corresponding to the moment is the peak point. Specifically, when When The kth moment in the target area is the peak point. For example, the set value .

[0047] Specifically, no. The probability that the kth moment in the target area belongs to the peak point is: ; in, It is The probability that the kth moment in the target area is the peak point, , , Respectively moment, moment, The signal strength corresponding to the moment, n is the total number of settings, a is the number of settings, is the normalization function.

[0048] Indicates Time, The maximum value of the signal strength corresponding to the moment is Indicates The signal strength corresponding to the moment The difference between the maximum value and Reflects when traversing to the Time, At time , the difference is compared with the The degree of volatility at a given moment.

[0049] Indicates normalization processing, which can use maximum and minimum value normalization, etc.

[0050] in, To set the number, specifically, its value is greater than or equal to 1, and the specific value can be less than or equal to the first The integer is half of the number of moments between the start moment and the end moment corresponding to the target area, reflecting the distance between the traversed moment and the kth moment.

[0051] If the signal strength corresponding to the kth moment The greater the difference between the maximum value of the signal strength corresponding to each moment traversed, the greater the number of The more hours, the The greater the probability that the kth moment in the target area is the peak point, and vice versa.

[0052] The above n is the total number of settings selected, specifically, , of course, it can also be determined according to actual conditions.

[0053] In the above embodiment, based on comparing the signal strength at the th moment with the signal strength within its vicinity, a setting number is introduced. to avoid The problem of the weakening of the reference of signal strength when the time interval between moments increases is solved, thereby reducing the impact of distant time points on the analysis.

[0054] After determining the peak points of each target area, each target area needs to be classified. Specifically, when the number of peak points is one, the corresponding target area is the first area, and when the number of peak points is at least two, the corresponding target area is the second area. The target area is divided into The first area and second areas, and obtain the signal strength at each moment in each first area and each second area.

[0055] The reason for the above classification is that since the initial spectrum contains areas with peaks and areas without peaks, the target area here is the area with obvious peaks in the initial spectrum, that is, the area where the signal energy is higher than the baseline by a certain range. Since there may be one or more peaks after multiple overlaps, the target area needs to be divided into the first area and the second area.

[0056] Step S23, respectively obtaining the overlapping peak regions of the first region and the second region.

[0057] For the first area with only one peak point, if there is no overlapping peak, the target area should be a shape that is symmetrical left and right with the peak point as the center, and the more the target area conforms to the Gaussian fitting distribution, the clearer the peak shape is; if there are overlapping peaks, the target area is less symmetrical, and the less the target area conforms to the Gaussian fitting distribution, the less clear the peak shape is.

[0058] For the second region with multiple peak points, since the normal peak points are generally steeper, when there is a gentle change in the peak points, it usually means that the signal strength changes less near the peak points, which may indicate that there are overlapping peaks in the signal. At the same time, if the time interval between the peak points and the difference in signal strength are small, it further indicates that these peaks are overlapping.

[0059] Therefore, based on the above analysis, the two target areas (the first area and the second area) with different numbers of peak points need to be analyzed separately, as follows: (1) The first area is analyzed as follows: The first overlap probability of each first region belonging to the overlapping peak region is calculated, specifically: ; For the The Gaussian fit of the first region, For the The degree of symmetry of the first region, For the The first overlap probability that the first region belongs to the overlapping peak region, is an exponential function.

[0060] When The smaller the Gaussian fit of the first region, the smaller the The smaller the symmetry of the first region, the The greater the first overlap probability of the target area belonging to the overlapping peak area, the smaller it is.

[0061] The above Gaussian fitting degree is obtained by performing Gaussian fitting on the i-th first region, obtaining the predicted signal strength after Gaussian fitting at each moment, and obtaining it according to the predicted signal strength and the actual signal strength. Specifically, the Gaussian fitting degree is: ;in, It is The signal strength in the first area at the jth moment, is the predicted signal strength of the fitted Gaussian curve at the jth moment, Indicates The corresponding signal strength mean at all times in the first area, For the The total number of moments in the first area.

[0062] is the residual, reflecting the difference between the actual observed value and the fitted value, and the sum of squares of the residual It measures the total deviation between the Gaussian fitting curve (actual time number corresponds to the predicted signal strength) and the actual curve (actual time number corresponds to the actual signal strength). The smaller this value is, the better the signal strength is. The closer the fitting curve is to the actual curve in the first region. is the total sum of squares, measuring the The total deviation of the signal strength corresponding to each moment in the first region from the mean signal strength. The smaller the value, the smaller the deviation of the fitting. The closer the Gaussian fit of the first region is to 1, the greater the residual square sum is. Total sum of squares The closer the fitting is, the larger the deviation is. The closer the Gaussian fit degree of the first region is to 0, the smaller the overlap is; conversely, the closer the Gaussian fit degree is to 0, the greater the overlap is.

[0063] The degree of symmetry is: ; in, No. The degree of symmetry of the first region, It is the left side of the peak point The signal strength at the moment, , are the starting time and ending time of the i-th first region, It is the right side of the peak point The signal strength at the moment, is the median value, For the In the first area Signal strength at the moment.

[0064] The above symmetry degree is used to measure the symmetry of the peak in the first region. When the number of moments on the left and right sides of the peak point is the same, that is, When, the denominator Used to calculate the peak point corresponding to the The time is the center, the average of the signal strengths on the left and right sides, and the starting time on the left is the Moment. Molecule The larger the The smaller the degree of symmetry of the first area, the greater the symmetry.

[0065] When the number of moments on the left and right sides of the peak point is inconsistent, that is, The signal strength at the intermediate moment and peak signal strength The greater the difference between the intermediate moment and the moment corresponding to the peak point, the greater the The smaller the degree of symmetry of the first region.

[0066] There are two cases when calculating the degree of symmetry, namely, the case where the number of moments on the left and right sides of the peak point is consistent or inconsistent. When inconsistent, it is reflected by the difference between the signal strength corresponding to the middle moment and the signal strength at the peak. It is a time interval used to characterize the interval between the middle moment and the peak point moment. When the interval is larger, it proves that the degree of left-right symmetry is smaller. Conversely, when the interval is 0, the middle moment coincides with the peak point moment, which proves that the curve corresponding to the peak point is left-right symmetric, that is, the degree of symmetry is large.

[0067] In this embodiment, by comparing the calculated first overlap probability with the first threshold, it is determined whether the first region belongs to the overlap peak region. When The first regions belong to the overlapping peak region. Exemplarily, the first threshold .

[0068] (2) Analysis of the second region is as follows: The second overlap probability of each second region belonging to the overlapping peak region is calculated as: ; in, For the The second overlap probability that the second region belongs to the overlapping peak region, It is The average of the slope values ​​on both sides of all peak points in the second region, , are the mean of the difference between two adjacent peak points at corresponding moments among all peak points and the mean of the difference in corresponding signal strengths, , Respectively The total number of times and the maximum signal strength in the second area, , , is the weight.

[0069] For each peak point, find the neighboring points on the left and right sides. Calculate the left slope and the right slope, and further use the average of the left slope and the right slope as the slope corresponding to the peak point. The smaller the average slope value at each peak in the second region, the smaller the average difference in time (horizontal axis) between all peak points, and the smaller the average difference in signal intensity between all peak points, the smaller the second region. The greater the second overlap probability that the second region belongs to the overlapping peak region.

[0070] when When The second region belongs to the overlapping peak region. Exemplarily, the second threshold According to the above operation, the target regions belonging to the overlapping peak regions in all the second regions are obtained.

[0071] The above average Reflects the The overall smoothness of the change at the peak point in the second region; the mean of the difference between all peak points in time (horizontal axis) , reflects the distribution of these peak points in time; the mean of the difference in signal strength between all peak points reflects the distribution of these peak points in signal strength.

[0072] Step S3, fitting each overlapping peak region to obtain a curve of multiple peaks for analyzing the target sample.

[0073] Overlapping peaks are a common problem in chromatographic analysis. Overlapping peaks refer to the situation where the peaks of multiple components overlap each other in the chromatogram, making it impossible to directly distinguish the peak shape and peak parameters (peak width, peak intensity information) of each component. In order to accurately identify each characteristic peak, we need to deconvolve these overlapping peaks, that is, decompose the overlapping peaks into multiple single peaks.

[0074] Specifically, based on the overlapping peak areas in the first area and the second area, the parameters required for deconvolution (the starting position and the end position of the overlapping peak area) are obtained. Assuming that each peak can be described by a Gaussian function, in this embodiment, the overlapping peaks can be decomposed by fitting multiple Gaussian functions by the nonlinear least squares method.

[0075] Specifically, in this embodiment, the overlapping peaks can be decomposed by fitting multiple Gaussian functions by nonlinear least squares method. Specifically, each peak in the overlapping peak region can be represented by a corresponding Gaussian function, wherein the shape of each Gaussian function is determined by its center position, width and amplitude.

[0076] The Gaussian functions are: ; is the signal strength at time t, For the The amplitude of each peak (i.e., signal strength), For the The center position of the peak, For the The width (i.e. duration) of a peak.

[0077] After establishing the signal intensity data of the overlapping peak area, the parameters in the Gaussian function can be optimized by the nonlinear least squares method to minimize the error between the fitting curve and the actual data point, thereby obtaining the Gaussian fitting curve of each peak, and then realizing the deconvolution of the overlapping peaks. After the fitting is completed, each overlapping peak area will be decomposed into multiple independent Gaussian curves, each Gaussian curve corresponding to a separate peak. Through these fitted Gaussian curves, the original overlapping peak area can be decomposed into multiple independent peaks. Based on the peak width, peak intensity and other information of the resolved single peak, a quantitative analysis model (such as a linear regression equation) can be further established to realize the quantitative analysis of the element content in the target sample.

[0078] Among them, the above quantitative analysis is to calculate the content of each element in the target sample based on the peak area and peak height; in the quantitative analysis, it is also necessary to determine the sensitivity factor of the instrument, the amount of sample used, and the surface area of ​​the sample. Through these parameters, the peak area can be converted into the concentration of the element. Since the above quantitative analysis is a prior art, it will not be described in detail here.

[0079] The advantage of the present invention over the existing methods is that, first, by dividing the target area (such as the first area and the second area), the problem is broken down into smaller and easier to handle parts, that is, for the first area (with only one peak point), a good fitting effect can be directly obtained by Gaussian fitting. However, the peak may have certain non-ideal shapes, such as incomplete symmetry or irregular width. The present invention can accurately evaluate whether the area is indeed a single, independent peak by analyzing the degree of Gaussian fitting and left-right symmetry. If the left-right asymmetry or the fit is not ideal, it may imply that there are actually overlapping peaks in the area and separation is required; for the second area (multiple peak points), by observing the characteristics such as the gentleness of the change at the peak point and the horizontal and vertical axis differences between the multiple peak points, it can be evaluated whether these peaks are really overlapping or whether the morphology is sufficiently separated, and then split by Gaussian function. Secondly, the key to deconvolution is to accurately estimate the characteristic parameters of each peak (such as position, width, amplitude, etc.), especially in the case of overlapping peaks. If the target area is not carefully divided and analyzed, direct deconvolution may lead to incorrect peak parameter estimation, affecting the final separation effect. Therefore, the present invention can better judge the degree of overlap and obtain more accurate initial parameters by finely dividing the target area, evaluating the distribution and symmetry of the peaks in each area, and analyzing the fluctuation. Finally, the nonlinear least squares method is used to fit multiple Gaussian functions to more accurately decompose the overlapping peaks. This method can improve the accuracy and reliability of the deconvolution process in the case of complex overlapping peaks.

[0080] The present invention also provides a chromatographic image recognition system based on characteristic peaks. Figure 2 As shown, the system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the chromatographic image recognition method based on characteristic peaks according to the present invention is implemented.

[0081] The identification system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, whose configuration and functions are known in the art and thus will not be described in detail here.

[0082] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0083] In the description of this specification, “plurality” means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0084] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A chromatographic image recognition method based on characteristic peaks, characterized in that: include: Acquire a chromatogram of a target sample, and preprocess the chromatogram to obtain an initial spectrum; Acquire multiple target regions in the initial map; Divide all target areas into a first area and a second area; The first region includes one peak point, and the second region includes at least two peak points; Calculating the first overlap probability of each first region, and taking the first region corresponding to the first overlap probability being greater than the first threshold as the overlap peak region; the first overlap probability is inversely correlated with the symmetry degree and Gaussian fitting degree of the first region; Calculate the second overlap probability of each second region, and take the second region corresponding to the second overlap probability greater than the second threshold as the overlap peak region; for: ; It is The average of the slope values ​​on both sides of all peak points in the second region, , are the mean of the difference between two adjacent peak points at corresponding moments among all peak points and the mean of the difference in corresponding signal strengths, , Respectively The total number of times and the maximum signal strength in the second area, , , is the weight; The overlapping peak regions are fitted to obtain curves of multiple peaks for analysis of target samples.

2. The chromatographic image recognition method based on characteristic peaks according to claim 1, characterized in that: The first overlap probability is: ； in, For the The Gaussian fit of the first region, For the The degree of symmetry of the first region, For the The first overlap probability that the first region belongs to the overlapping peak region, It is an exponential function with the natural constant e as its base.

3. The chromatographic image recognition method based on characteristic peaks according to claim 2, characterized in that: The degree of symmetry is: ； in, For the The degree of symmetry of the first region, It is the left side of the peak point The signal strength at the moment, , Respectively The start and end time of the first area, It is the right side of the peak point The signal strength at the moment, is the median value, For the The peak point in the first region corresponds to the Signal strength at the moment.

4. The chromatographic image recognition method based on characteristic peaks according to claim 2, characterized in that: The Gaussian fit for: ; is the signal strength at the jth moment in the ith first region, is the predicted signal strength at the jth moment in the fitted Gaussian curve, Indicates The corresponding signal strength mean at all times in the first area, For the The total number of moments in the first area.

5. The chromatographic image recognition method based on characteristic peaks according to claim 1, characterized in that: The specific process of obtaining multiple target areas in the initial atlas is as follows: In response to the difference between the signal intensity corresponding to any moment and the baseline signal intensity being greater than or equal to a set threshold, the corresponding moment is the target moment; Continuously adjacent moments among all target moments are divided into the same target area, and multiple target areas are obtained.

6. The chromatographic image recognition method based on characteristic peaks according to claim 1, characterized in that: The process of dividing all target areas into the first area and the second area is as follows: The maximum value points of the signal strength at each time in each target area are obtained, and the points at the time corresponding to each maximum value point are selected as peak points; the target area with one peak point is taken as the first area, and the target area with at least two peak points is taken as the second area.

7. The chromatographic image recognition method based on characteristic peaks according to claim 6, characterized in that: The process of obtaining the peak point is as follows: Calculate the probability that the point corresponding to each time in each target area belongs to the peak point; In response to the probability being greater than a threshold, the point at the corresponding moment is a peak point; The probabilities are: ； in, It is The probability that the kth moment in the target area is the peak point, , , Respectively moment, moment, The signal strength corresponding to the moment, n is the total number of settings, a is the number of settings, is the normalization function.

8. The chromatographic image recognition method based on characteristic peaks according to claim 3, characterized in that: The process of fitting the overlapping peak regions to obtain a curve of multiple peaks for analyzing the target sample is as follows: Obtain the starting position and ending position of each overlapping peak region; Based on the starting position and the end position of each overlapping peak region, and in combination with the Gaussian fitting algorithm, the Gaussian curve of each mid-peak point in the overlapping peak region is obtained; Based on the peak width and peak intensity information of each single peak, a linear regression equation is established to achieve quantitative analysis of the content of each element in the target sample.

9. The chromatographic image recognition method based on characteristic peaks according to claim 1, characterized in that: The preprocessing of the chromatogram to obtain the initial spectrum includes: using an adaptive filtering algorithm and a baseline correction algorithm to perform denoising on the chromatogram.

10. A chromatographic image recognition system based on characteristic peaks, characterized in that: include: processor; A memory storing computer instructions for chromatographic image recognition based on characteristic peaks, wherein when the computer instructions are executed by the processor, the system executes the chromatographic image recognition method based on characteristic peaks according to any one of claims 1-9.

Images