Mass spectrum signal processing method and device, mass spectrometer and storage medium

Abstract

The invention provides a mass spectrum signal processing method and device, a mass spectrometer and a storage medium. The method comprises the following steps: acquiring original mass spectrum data; key parameters in ensemble empirical mode decomposition are optimized based on a ridge regression model, and the ensemble frequency of optimized additive noise is output; decomposing the original mass spectrum data into a plurality of intrinsic mode functions based on the set frequency of the optimized additive noise, and calculating a correlation coefficient of each intrinsic mode function and the original mass spectrum data; the most relevant intrinsic mode component is extracted according to the correlation coefficient, signals are reconstructed based on the most relevant intrinsic mode component, baseline correction is conducted on the reconstructed signals, and optimized mass spectrum data are obtained.According to the method, ensemble empirical mode decomposition and machine learning are combined, and the method has the advantages of being small in influence of human factors and high in applicability; and the accuracy and the stability of the signal can be obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of biomedicine and image processing, and particularly to a mass spectrometry signal processing method, device, mass spectrometer and storage medium. Background Art

[0002] A mass spectrometer is an instrument for analyzing the composition and structure of substances. It obtains detailed information about samples by measuring the mass and relative abundance of sample molecules or ions, and is widely used in fields such as chemistry, life science, environmental science, and drug analysis. Affected by hardware conditions and environmental factors, the data collected by traditional mass spectrometers often has noise interference and baseline drift phenomena, resulting in a decrease in the accuracy of analysis results. Summary of the Invention

[0003] The present invention provides a mass spectrometry signal processing method, device, mass spectrometer and storage medium to solve the defect that the traditional mass spectrometry signal processing method has a baseline drift phenomenon, which affects the accuracy of analysis results.

[0004] The present invention provides a mass spectrometry signal processing method, including: Collecting original mass spectrometry data; Optimizing key parameters in the ensemble empirical mode decomposition based on a ridge regression model, and outputting the number of ensemble times of optimized additive noise; Based on the number of ensemble times of the optimized additive noise, decomposing the original mass spectrometry data into multiple intrinsic mode functions, and calculating the correlation coefficient between each intrinsic mode function and the original mass spectrometry data; Extracting the most relevant intrinsic mode components according to the correlation coefficient, reconstructing a signal based on the most relevant intrinsic mode components, and performing baseline correction on the reconstructed signal to obtain optimized mass spectrometry data.

[0005] According to the mass spectrometry signal processing method provided by the present invention, the original mass spectrometry data includes the corresponding relationship between ion signal intensity and mass-to-charge ratio. After collecting the original mass spectrometry data, it further includes: Preprocessing the corresponding relationship between the ion signal intensity and the mass-to-charge ratio, and the preprocessing includes mapping the data to the [0, 1] interval by using maximum-minimum normalization; And / or, removing high-frequency noise by using moving average or wavelet transform.

[0006] According to the mass spectrometry signal processing method provided by the present invention, the training method of the ridge regression model includes: Dividing the historical mass spectrometry data sample set into a training set and a validation set; Initializing the regularization coefficient and weight; Setting the model objective function, using the gradient descent method to minimize the model objective function, and iteratively updating the weight until the output converges; Evaluate whether the number of times of the additive noise predicted by the ridge regression model can optimize the decomposition effect of the ensemble empirical mode decomposition using the validation set, and adjust the regularization coefficient according to the validation result until the number of times of the additive noise predicted by the ridge regression model can make the decomposition effect of the ensemble empirical mode decomposition meet the preset requirements.

[0007] According to the mass spectrometry signal processing method provided by the present invention, the decomposition of the original mass spectrometry data into a plurality of intrinsic mode functions includes: Adding Gaussian white noise to the original mass spectrometry data to obtain a noisy signal; Performing empirical mode decomposition on the noisy signal once to obtain a set of intrinsic mode functions, where the empirical mode decomposition once includes: finding all local maxima and minima of the noisy signal; connecting the extreme points by interpolation to form upper and lower envelopes; calculating the mean envelope; extracting candidate intrinsic mode functions according to the mean envelope, and outputting the intrinsic mode function when the candidate intrinsic mode function meets the judgment conditions; the judgment conditions include that the difference between the number of extreme points and the number of zero-crossing points does not exceed 1 and the mean of the envelope is zero; The number of times of performing the empirical mode decomposition is the number of times of the optimized additive noise, and the results after decomposing the number of times of the optimized additive noise are averaged to obtain a plurality of intrinsic mode functions.

[0008] According to the mass spectrometry signal processing method provided by the present invention, the plurality of intrinsic mode functions include high-frequency signals, characteristic signals, and trend terms; Among them, the high-frequency signal :

[0009] The characteristic signal :

[0010] The trend term :

[0011] ; Among them, IMF1 represents the noise part in the signal; IMF2 to IMF 13 represents the characteristic signal part, ℎ is the weighting coefficient, and K i is the noise adjustment parameter, and R r and R l are the proportionality factors of the trend term respectively.

[0012] According to the mass spectrometry signal processing method provided by the present invention, the baseline correction of the reconstructed signal includes: Constructing an optimization objective for baseline correction; Dynamically adjust the weight in the baseline correction optimization objective according to the difference between the signal and the baseline, and solve the baseline correction optimization objective after weight adjustment to obtain the optimal baseline.

[0013] According to the mass spectrometry signal processing method provided by the present invention, the dynamically adjusting the weight in the baseline correction optimization objective according to the difference between the signal and the baseline includes: In the signal peak region, reduce the weight to reduce the interference of the peak on the baseline fitting; In the baseline region, maintain the weight to ensure the fitting accuracy.

[0014] The present invention also provides a mass spectrometry signal processing device, including: An acquisition module for acquiring original mass spectrometry data; An optimization module for optimizing key parameters in the ensemble empirical mode decomposition based on a ridge regression model and outputting the optimized number of ensemble times of additive noise; A decomposition module for decomposing the original mass spectrometry data into a plurality of intrinsic mode functions based on the optimized number of ensemble times of additive noise and calculating the correlation coefficient between each intrinsic mode function and the original mass spectrometry data; A correction module for extracting the most relevant intrinsic mode components according to the correlation coefficient, reconstructing a signal based on the most relevant intrinsic mode components, and performing baseline correction on the reconstructed signal to obtain optimized mass spectrometry data.

[0015] The present invention also provides a mass spectrometer, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the mass spectrometry signal processing method described in any one of the above is implemented.

[0016] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the mass spectrometry signal processing method described in any one of the above is implemented.

[0017] The mass spectrometry signal processing method, device, mass spectrometer, and storage medium provided by the present invention collect original mass spectrometry data; optimize key parameters in the ensemble empirical mode decomposition based on a ridge regression model and output the optimized number of ensemble times of additive noise; decompose the original mass spectrometry data into a plurality of intrinsic mode functions based on the optimized number of ensemble times of additive noise and calculate the correlation coefficient between each intrinsic mode function and the original mass spectrometry data; extract the most relevant intrinsic mode components according to the correlation coefficient, reconstruct a signal based on the most relevant intrinsic mode components, and perform baseline correction on the reconstructed signal to obtain optimized mass spectrometry data. This method combines ensemble empirical mode decomposition with machine learning, has the advantages of being less affected by human factors and having strong applicability, and can significantly improve the accuracy and stability of signals. Brief Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 is a schematic flowchart of the mass spectrometry signal processing method provided by an embodiment of the present invention; Figure 2 is a schematic diagram of the result of ensemble empirical mode decomposition (EEMD) processing of the mass spectrometry signal provided by an embodiment of the present invention; Figure 3 is a schematic diagram of the high-frequency component, low-frequency component, and trend term images obtained by processing the mass spectrometry signal through EEMD provided by an embodiment of the present invention; Figure 4 is a schematic diagram of the comparison before and after baseline processing provided by an embodiment of the present invention; Figure 5 is a schematic diagram of the comparison before and after the original mass spectrometry signal and the processed mass spectrometry signal provided by an embodiment of the present invention; Figure 6 is a schematic diagram of the functional structure of the mass spectrometry signal processing device provided by an embodiment of the present invention; Figure 7 is a schematic diagram of the functional structure of the mass spectrometer provided by an embodiment of the present invention. Detailed Embodiments

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0021] Figure 1 is a flowchart of the mass spectrometry signal processing method provided by an embodiment of the present invention. As Figure 1 shown, the mass spectrometry signal processing method provided by an embodiment of the present invention includes: Step 101, collect original mass spectrometry data; Step 102, optimize the key parameters in the ensemble empirical mode decomposition based on the ridge regression model, and output the optimized number of ensemble times of the additive noise; Ensemble Empirical Mode Decomposition (EEMD) is an improved technique based on Empirical Mode Decomposition (EMD). EMD can decompose a signal into a series of Intrinsic Mode Functions (IMFs) by dividing the time scale. However, in practical applications, due to the non-uniform change of the time scale, modal aliasing may occur, that is, a single IMF contains different scale feature components, or similar features appear in different IMFs. To solve this problem, EEMD introduces white noise and regards the observed data as a combination of signal and noise, thus effectively alleviating modal aliasing. This method uses noise interference and decomposes multiple times to take the mean, thereby improving the accuracy and reliability of signal decomposition and finally obtaining a result close to the real signal.

[0022] Step 103: Decompose the original mass spectrometry data into multiple intrinsic mode functions based on the optimized ensemble number of additive noise, and calculate the correlation coefficient between each intrinsic mode function and the original mass spectrometry data; Step 104: Extract the most relevant intrinsic mode components according to the correlation coefficient, reconstruct the signal based on the most relevant intrinsic mode components, and perform baseline correction on the reconstructed signal to obtain the optimized mass spectrometry data.

[0023] Affected by hardware conditions and environmental factors, the data collected by traditional mass spectrometers often have interference, and there is a baseline drift phenomenon in mass spectrometry signal processing, which further affects the accuracy of the analysis results.

[0024] The mass spectrometry signal processing method provided by the embodiments of the present invention includes collecting original mass spectrometry data; optimizing key parameters in ensemble empirical mode decomposition based on a ridge regression model, and outputting the optimized ensemble number of additive noise; decomposing the original mass spectrometry data into multiple intrinsic mode functions based on the optimized ensemble number of additive noise, and calculating the correlation coefficient between each intrinsic mode function and the original mass spectrometry data; extracting the most relevant intrinsic mode components according to the correlation coefficient, reconstructing the signal based on the most relevant intrinsic mode components, and performing baseline correction on the reconstructed signal to obtain the optimized mass spectrometry data. This method combines ensemble empirical mode decomposition with machine learning, has the advantages of less influence by human factors and strong applicability, and can significantly improve the accuracy and stability of the signal.

[0025] Based on any of the above embodiments, the original mass spectrometry data includes the corresponding relationship between ion signal intensity and mass-to-charge ratio. After collecting the original mass spectrometry data, it further includes: Preprocess the corresponding relationship between the ion signal intensity and the mass-to-charge ratio. The preprocessing includes mapping the data to the [0, 1] interval by using maximum-minimum normalization; And removing high-frequency noise using moving average or wavelet transform.

[0026] In the embodiments of the present invention, the dimension difference is eliminated by maximum-minimum normalization, and the ion signal intensities of different magnitudes are uniformly scaled to the interval [0, 1], avoiding the interference of numerical range differences on model training. It is applicable to mass spectrometry data under different instruments or experimental conditions, enhancing the universality of the method.

[0027] The random high-frequency fluctuations (such as spike signals) introduced by instrument electronic noise, environmental interference, etc. are removed by moving average or wavelet transform denoising; while smoothing the noise, the low-frequency characteristic peaks (such as the mass spectrometry peaks of target compounds) are protected from being damaged.

[0028] Based on any of the above embodiments, the training method of the ridge regression model includes: Step 201, dividing the historical mass spectrometry data sample set into a training set and a validation set; Step 202, initializing the regularization coefficient and the weight; Step 203, setting the model objective function, and using the gradient descent method to minimize the model objective function, and iteratively updating the weight until the output converges; In the embodiments of the present invention, the model objective function is:

[0029] Among them: the L2 regularization term is used to penalize the overly large weight, avoid overfitting, and improve the generalization ability of the model; X: input feature matrix (characteristics of preprocessed mass spectrometry data).

[0030] y: target variable (optimized K value, which needs to be determined in advance through experiments or cross-validation).

[0031] ω: model weight (regression coefficient).

[0032] α: regularization strength hyperparameter, controlling the model complexity (preventing overfitting); The regression coefficient ω is used to reflect the contribution weight of each input feature to the prediction of the K value. By analyzing ω, key features can be identified (such as the signals in certain frequency bands have a greater impact on the K value).

[0033] Step 204, using the validation set to evaluate whether the number of times of the additive noise predicted by the ridge regression model can make the decomposition effect of the ensemble empirical mode decomposition optimal, and adjusting the regularization coefficient according to the validation result until the number of times of the additive noise predicted by the ridge regression model can make the decomposition effect of the ensemble empirical mode decomposition meet the preset requirements.

[0034] In the embodiment of the present invention, the set number K of the optimized additive noise for ridge regression prediction is used to control the number of times of noise addition in EEMD, ensuring that the decomposed IMFs are purer, reducing mode mixing, dynamically adjusting the value of K to adapt to different mass spectrometry signal characteristics, and improving the decomposition accuracy.

[0035] Based on any of the above embodiments, the decomposing the original mass spectrometry data into a plurality of intrinsic mode functions includes: Step 301, adding Gaussian white noise to the original mass spectrometry data to obtain a noisy signal; Step 302, performing an empirical mode decomposition on the noisy signal once to obtain a set of intrinsic mode functions, where one empirical mode decomposition includes: finding all local maxima and minima of the noisy signal; connecting the extreme points by interpolation to form upper and lower envelopes; calculating the mean envelope; extracting candidate intrinsic mode functions according to the mean envelope, and outputting the intrinsic mode function when the candidate intrinsic mode function meets the judgment conditions; the judgment conditions include that the difference between the number of extreme points and the number of zero-crossing points does not exceed 1 and the mean of the envelope is zero; Step 303, the number of executions of the empirical mode decomposition is the set number of the optimized additive noise, and the results after decomposition of the set number of the optimized additive noise are averaged to obtain a plurality of intrinsic mode functions.

[0036] In the embodiment of the present invention, the plurality of intrinsic mode functions include high-frequency signals, characteristic signals, and trend terms; Among them, the high-frequency signal :

[0037] Characteristic signal :

[0038] Trend term :

[0039] : ; Among them, IMF1 represents the noise part in the signal; IMF2 to IMF 13 represents the characteristic signal part, ℎ is the weighting coefficient, K i is the noise adjustment parameter, R r and R l are the scale factors of the trend term respectively. Based on any of the above embodiments, the baseline correction of the reconstructed signal includes: Constructing an optimization objective for baseline correction: Among them, the i-th data point of the original signal (ion signal intensity including baseline drift), is the i-th data point of the baseline signal to be solved; is the dynamic weight, adjusted according to the difference between the signal and the baseline, is the smoothing parameter (regularization coefficient), controlling the smoothness of the baseline signal, is the m-th order difference operator (usually m = 2), used to calculate the curvature or higher-order derivative of the baseline, is the total number of signal data points.

[0040] Dynamically adjust the weight in the baseline correction optimization target according to the difference between the signal and the baseline, and solve the baseline correction optimization target after weight adjustment to obtain the optimal baseline.

[0041] In the embodiments of the present invention, In the signal peak region (y0,i > xb,i), reduce the weight to reduce the interference of the peak on the baseline fitting; in the baseline region, maintain the weight to ensure the fitting accuracy.

[0042] Based on any of the above embodiments, the mass spectrometry signal processing method provided by the embodiments of the present invention includes: Step S1: Obtain the relative intensity Y of the mass spectrometry data signal of the actual sample 2-hydroxyglutaric acid directly obtained by the mass spectrometer, and its mass number is X.

[0043] Step S2: Preprocess the original data and input it into a machine learning (ridge regression) model for training. Mine the linear patterns and hidden laws in the data through regularization constraints, and use the trained model to predict the newly input data and calculate the characteristic optimization coefficient between the prediction result and the original data.

[0044] Step S3: Use the EEMD method to decompose the original data into multiple IMFs as Figure 2 shown, distinguish the high-frequency, low-frequency and trend terms as Figure 3 shown, and calculate the correlation coefficient between each IMF and the original data.

[0045] In the EEMD method, the key parameters include Nstd and NE. Nstd is the ratio of the noise standard deviation to the signal standard deviation, which is mainly used to set the standard deviation of Gaussian white noise, so as to remove the noise in the original signal. The specific value of Nstd should be adjusted according to the noise intensity in the original signal. In the embodiment of the present invention, it is set to 0.1 through comparative experiments; NE represents the average number of signal decomposition times, which is used to set the number of noise additions. If NE is too large, it will lead to too long processing time, and if it is too small, it may lead to insufficient noise decomposition. In the embodiment of the present invention, it is set to 300 according to the noise characteristics and application environment. Through the EEMD method, the characteristic frequency components with physical significance in the mass spectrometry signal of 2-hydroxyglutaric acid can be effectively decomposed. As Figure 3 shown, these frequency components are arranged from high to low. Among them, IMF1 has the highest frequency, and IMF13, that is, the trend term (E_r) of the signal, has the lowest frequency. According to the time-frequency distribution and amplitude characteristics of the signal, generally speaking, IMF1 represents the noise part in the signal; IMF2 to IMF13 represent the characteristic signal part of 2-hydroxyglutaric acid; the last-order IMF14 represents the background component of the signal.

[0046] Step S4: The original mass spectrometry signal of 2-hydroxyglutaric acid after being processed by the optimized coefficient is reconstructed by EEMD to obtain the optimized signal Step S5: After the original mass spectrometry signal of 2-hydroxyglutaric acid is reconstructed by the EEMD method for feature enhancement, baseline drift appears. The baseline correction method is used to process the baseline drift phenomenon. The comparison of the signals before and after the baseline correction is as Figure 4 shown.

[0047] The comparison between the original mass spectrometry data and the optimized mass spectrometry data is as Figure 5 shown.

[0048] The mass spectrometry signal processing method provided by the embodiment of the present invention obtains the mass spectrometry data obtained by a small mass spectrometer, uses a machine learning algorithm to assist in optimizing the key parameter K value in EEMD to improve the decomposition accuracy and modeling performance, uses the EEMD method to decompose the original data into multiple intrinsic mode functions, distinguishes the high-frequency, low-frequency and trend terms, calculates the correlation coefficient between each IMF and the original data, recombines the signals to obtain the processed mass spectrometry data, and uses the baseline correction method to process the baseline drift phenomenon. It has the advantages of simple operation, excellent intelligence level, and high accuracy of the characteristic peak information of the mass spectrometry data.

[0049] Next, the mass spectrometry signal processing device provided by the present invention will be described. The mass spectrometry signal processing device described below can be correspondingly referred to the mass spectrometry signal processing method described above.

[0050] Figure 6The structural schematic diagram of the mass spectrometry signal processing device provided by the embodiment of the present invention is as follows Figure 5 As shown, the mass spectrometry signal processing device provided by the embodiment of the present invention includes: An acquisition module 601 for acquiring original mass spectrometry data; An optimization module 602 for optimizing key parameters in the ensemble empirical mode decomposition based on a ridge regression model and outputting the number of ensemble times of the optimized additive noise; A decomposition module 603 for decomposing the original mass spectrometry data into a plurality of intrinsic mode functions based on the number of ensemble times of the optimized additive noise and calculating the correlation coefficient between each intrinsic mode function and the original mass spectrometry data; A correction module 604 for extracting the most relevant intrinsic mode component according to the correlation coefficient, reconstructing a signal based on the most relevant intrinsic mode component, and performing baseline correction on the reconstructed signal to obtain optimized mass spectrometry data.

[0051] The mass spectrometry signal processing device provided by the embodiment of the present invention acquires original mass spectrometry data; optimizes key parameters in the ensemble empirical mode decomposition based on a ridge regression model and outputs the number of ensemble times of the optimized additive noise; decomposes the original mass spectrometry data into a plurality of intrinsic mode functions based on the number of ensemble times of the optimized additive noise and calculates the correlation coefficient between each intrinsic mode function and the original mass spectrometry data; extracts the most relevant intrinsic mode component according to the correlation coefficient, reconstructs a signal based on the most relevant intrinsic mode component, and performs baseline correction on the reconstructed signal to obtain optimized mass spectrometry data. This method combines ensemble empirical mode decomposition with machine learning, has the advantages of being less affected by human factors and having strong applicability, and can significantly improve the accuracy and stability of signals.

[0052] Figure 7 Illustrates a schematic diagram of the physical structure of a mass spectrometer, as Figure 7As shown, the mass spectrometer may include: a processor 710, a communications interface 720, a memory 730, and a communication bus 740. Among them, the processor 710, the communications interface 720, and the memory 730 complete mutual communication through the communication bus 740. The memory 730 includes computer programs, an operating system, and acquired data. The processor 710 can call the logical instructions in the memory 730 to execute a mass spectrometry signal processing method, which includes: collecting raw mass spectrometry data; optimizing key parameters in the ensemble empirical mode decomposition based on a ridge regression model, and outputting the number of ensemble times of optimized additive noise; based on the number of ensemble times of the optimized additive noise, decomposing the raw mass spectrometry data into multiple intrinsic mode functions, and calculating the correlation coefficient between each intrinsic mode function and the raw mass spectrometry data; extracting the most relevant intrinsic mode components according to the correlation coefficient, reconstructing a signal based on the most relevant intrinsic mode components, and performing baseline correction on the reconstructed signal to obtain optimized mass spectrometry data.

[0053] In addition, when the logical instructions in the above-mentioned memory 730 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the related technology, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0054] On the other hand, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the mass spectrometry signal processing method provided by the above-mentioned various methods. The method includes: collecting raw mass spectrometry data; optimizing key parameters in the ensemble empirical mode decomposition based on a ridge regression model, and outputting the number of ensemble times of optimized additive noise; based on the number of ensemble times of the optimized additive noise, decomposing the raw mass spectrometry data into multiple intrinsic mode functions, and calculating the correlation coefficient between each intrinsic mode function and the raw mass spectrometry data; extracting the most relevant intrinsic mode components according to the correlation coefficient, reconstructing a signal based on the most relevant intrinsic mode components, and performing baseline correction on the reconstructed signal to obtain optimized mass spectrometry data.

[0055] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0056] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the relevant technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for processing mass spectrometry signals, characterized in that, Including: Collecting original mass spectrometry data; Optimizing key parameters in the ensemble empirical mode decomposition based on the ridge regression model, and outputting the optimized number of ensemble times of additive noise; Based on the optimized number of ensemble times of additive noise, decomposing the original mass spectrometry data into multiple intrinsic mode functions, and calculating the correlation coefficient between each intrinsic mode function and the original mass spectrometry data; Extracting the most relevant intrinsic mode components according to the correlation coefficient, reconstructing the signal based on the most relevant intrinsic mode components, and performing baseline correction on the reconstructed signal to obtain optimized mass spectrometry data.

2. The mass spectrometry signal processing method according to claim 1, wherein The original mass spectrometry data includes the corresponding relationship between ion signal intensity and mass-to-charge ratio. After collecting the original mass spectrometry data, it further includes: Preprocessing the corresponding relationship between the ion signal intensity and the mass-to-charge ratio, and the preprocessing includes mapping the data to the [0, 1] interval by using maximum-minimum normalization; And / or, removing high-frequency noise by using moving average or wavelet transform.

3. The mass spectrometry signal processing method according to claim 1, characterized in that The training method of the ridge regression model includes: Dividing the historical mass spectrometry data sample set into a training set and a validation set; Initializing the regularization coefficient and weights; Setting the model objective function, using the gradient descent method to minimize the model objective function, and iteratively updating the weights until the output converges; Evaluating whether the number of ensemble times of additive noise predicted by the ridge regression model can make the decomposition effect of the ensemble empirical mode decomposition optimal by using the validation set, and adjusting the regularization coefficient according to the validation result until the number of ensemble times of additive noise predicted by the ridge regression model can make the decomposition effect of the ensemble empirical mode decomposition meet the preset requirements.

4. The mass spectrometry signal processing method according to claim 1, characterized in that The decomposing the original mass spectrometry data into multiple intrinsic mode functions includes: Adding Gaussian white noise to the original mass spectrometry data to obtain a noisy signal; Performing an empirical mode decomposition on the noisy signal once to obtain a set of intrinsic mode functions, where one empirical mode decomposition includes: finding all local maxima and minima of the noisy signal; connecting the extreme points by interpolation to form upper and lower envelopes; calculating the mean envelope; extracting candidate intrinsic mode functions according to the mean envelope, and outputting the intrinsic mode function when the candidate intrinsic mode function meets the judgment conditions; the judgment conditions include that the difference between the number of extreme points and the number of zero-crossing points does not exceed 1 and the envelope mean is zero; The number of executions of the empirical mode decomposition is the optimized number of ensemble times of additive noise, and averaging the results after decomposing the optimized number of ensemble times of additive noise to obtain multiple intrinsic mode functions.

5. The mass spectrometry signal processing method according to claim 1 or 4, characterized in that The multiple intrinsic mode functions include high-frequency signals, characteristic signals, and trend terms; Among them, the high-frequency signal : Characteristic signal : Trend term : ； Among them, IMF1 represents the noise part in the signal; IMF2 to IMF 13 represent the characteristic signal part, ℎ is the weighting coefficient, and K i is the noise adjustment parameter, and R r and R l are the proportionality factors of the trend term, respectively.

6. The mass spectrometry signal processing method according to claim 2, wherein, The performing baseline correction on the reconstructed signal includes: Constructing a baseline correction optimization target; Dynamically adjusting the weights in the baseline correction optimization target according to the difference between the signal and the baseline, and solving the baseline correction optimization target after weight adjustment to obtain the optimal baseline.

7. The mass spectrometry signal processing method according to claim 6, wherein The dynamically adjusting the weights in the baseline correction optimization target according to the difference between the signal and the baseline includes: In the signal peak region, reducing the weight to reduce the interference of the peak on the baseline fitting; In the baseline region, maintaining the weight to ensure the fitting accuracy.

8. A mass spectrometry signal processing device, characterized in that, Including: A collection module for collecting original mass spectrometry data; An optimization module, configured to optimize key parameters in the ensemble empirical mode decomposition based on a ridge regression model, and output the optimized number of ensemble times of additive noise; A decomposition module, configured to decompose the original mass spectrometry data into a plurality of intrinsic mode functions based on the optimized number of ensemble times of additive noise, and calculate the correlation coefficients between each intrinsic mode function and the original mass spectrometry data; A correction module, configured to extract the most relevant intrinsic mode components according to the correlation coefficients, reconstruct a signal based on the most relevant intrinsic mode components, and perform baseline correction on the reconstructed signal to obtain optimized mass spectrometry data.

9. A mass spectrometer, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the mass spectrometry signal processing method according to any one of claims 1 to 7.

10. A non-transitory readable storage medium, on which a computer program is stored, characterized in that, When the computer program is executed by a processor, it implements the mass spectrometry signal processing method according to any one of claims 1 to 7.

Images