Analysis, processing methods and noise reduction models of microspectroscopy instruments

By decomposing the base spectral data and using machine learning models to train the noise reduction model, the problem of the noise affecting measurement accuracy of microscopic instruments is solved, and high-precision spectral detection is achieved.

CN115308130BActive Publication Date: 2024-11-29XIAMEN UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202210946093.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-08-08
Publication Date
2024-11-29
Estimated Expiration
2042-08-08

AI Technical Summary

Technical Problem

Due to the influence of noise, the measurement results are low, and it is difficult for the existing microscopic spectrometer to accurately analyze and process noise data.

Method used

The substrate spectral data is obtained by detecting the signalless substrate, and the instrument background spectral component is decomposed and removed to obtain noise values. The noise reduction model is trained using machine learning models, and a high signal-to-noise ratio spectrum is generated through the spectral fitting function, and trained in combination with the instrument noise data to identify and remove noise.

Benefits of technology

Accurate analysis and processing of noise of microspectral instruments is realized, the accuracy and sensitivity of spectral detection are improved, and true and accurate noise data are obtained.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115308130B_ABST
    Figure CN115308130B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for analyzing the noise of a micro-spectroscopic instrument, comprising: detecting a signal-free substrate through a micro-spectroscopic instrument to obtain substrate spectral data; decomposing the substrate spectral data to obtain spectral components; removing the instrument background spectral components from the spectral components to obtain the noise value of the micro-spectroscopic instrument; and determining the instrument background signal components by setting a frequency threshold after Fourier transformation. The method for analyzing the noise of a micro-spectroscopic instrument in the present invention obtains substrate spectral data by detecting a signal-free substrate, extracts the instrument background spectral components from the substrate spectral data, and removes the background spectral components from the substrate spectral data, thereby obtaining the noise of the micro-spectroscopic instrument. The noise obtained is the overall noise of the micro-spectroscopic instrument, which can truly and accurately reflect the noise of the micro-spectroscopic instrument under the current test conditions and state. At the same time, the present invention also includes a method for processing the noise of a micro-spectroscopic instrument, a noise reduction model and a device.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of spectrum detection technology, and can be applied to qualitative and quantitative analysis of substances, and in particular to analysis, processing methods, noise reduction models and equipment for microscopic spectrum instrument noise. Background Art

[0002] Existing micro-spectroscopy instruments will have background noise due to the influence of noise sources such as test components, ambient light sources and induction circuits. Background noise will affect the accuracy of measurement results. In order to reduce the influence of background noise on test results, the existing technology is to infer the noise data of each noise source based on the electrical parameters of the equipment or by using testing methods, which has poor accuracy and reduces the accuracy of test results. Summary of the invention

[0003] The embodiments of the present invention provide a method for analyzing and processing the noise of a micro-spectroscopy instrument, a noise reduction model and a device, so as to solve at least one of the above-mentioned technical problems.

[0004] In a first aspect, an embodiment of the present invention provides a method for analyzing noise of a microscopic spectroscopy instrument, comprising:

[0005] Detect the signal-free substrate by a micro-spectral instrument to obtain substrate spectrum data;

[0006] Decomposing the base spectrum data to obtain the instrument background spectrum component;

[0007] The instrument background spectrum component is removed from the base spectrum data to obtain the noise value of the micro-spectroscopy instrument.

[0008] In a second aspect, an embodiment of the present invention provides a noise reduction model for a microscopic spectroscopy instrument, wherein a training method for the noise reduction model for a microscopic spectroscopy instrument comprises:

[0009] According to the noise data analysis method of the micro-spectrographic instrument of the first aspect of the present invention, the noise value of the micro-spectrographic instrument is obtained;

[0010] Generate a high signal-to-noise ratio spectrum according to the spectrum fitting function; add the instrument noise and the high signal-to-noise ratio spectrum data to generate corresponding low signal-to-noise ratio data;

[0011] The input to the machine learning model is low signal-to-noise ratio spectral data, and the output of the model is the instrument noise;

[0012] The input of the machine learning model is set as a noisy low signal-to-noise ratio spectrum, and the output is set as the instrument noise value. The training data set is used to train the machine learning model to obtain a denoising model for microscopic spectroscopy instrument noise.

[0013] In a third aspect, an embodiment of the present invention provides a method for processing noise of a microscopic spectroscopy instrument, comprising:

[0014] Using the noise reduction model of the micro-spectroscopic instrument according to the second aspect of the present invention, the noisy spectral data of the sample are identified to obtain the noise value of the micro-spectroscopic instrument;

[0015] The spectral detection value of the sample is obtained according to the noisy spectral data and the noise value of the micro-spectroscopic instrument.

[0016] In a fourth aspect, an embodiment of the present invention provides an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for analyzing the noise of the microscopic spectroscopy instrument in the present invention, or execute the steps of the method for processing the noise of the microscopic spectroscopy instrument in the present invention.

[0017] The beneficial effects of the embodiments of the present invention are:

[0018] The method for analyzing the noise of the microspectroscopy instrument in the present invention obtains the base spectrum data by detecting the base without signal, extracts the instrument background spectrum component from the base spectrum data, and removes the background spectrum component from the base spectrum data instrument, thereby obtaining the noise of the microspectroscopy instrument. The noise obtained is the overall noise of the microspectroscopy instrument, which can truly and accurately reflect the microspectroscopy instrument noise of the current microspectroscopy instrument under the current test conditions and state.

[0019] On the other hand, the noise reduction model of the microscopic spectroscopy instrument noise in the present invention obtains the noise value through the analysis method of the microscopic spectroscopy instrument noise in the present invention, and uses the noise value as the target value of the training model. A high signal-to-noise ratio spectrum is generated by a spectral fitting function as a training set. The target value of the noise reduction model is accurate and the noise value of the detection spectrum can be accurately identified.

[0020] In addition, the method for processing the noise of the micro-spectroscopic instrument in the present invention can accurately obtain the noise of the test sample and improve the sensitivity of the spectral instrument by using a pre-stored noise reduction model of the micro-spectroscopic instrument. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0022] Figure 1 The present invention is a flowchart of an embodiment of a method for analyzing noise of a micro-spectroscopy instrument.

[0023] Figure 2 This is the collected Raman spectrum of the gold film.

[0024] Figure 3 Schematic diagram of the two components of Raman spectral data after singular value decomposition.

[0025] Figure 4 This is a spectrum diagram of the gold film spectral data after removing the background spectral component.

[0026] Figure 5 Schematic diagram of the average power spectrum of instrument noise after averaging.

[0027] Figure 6 This is a schematic diagram of the instrument noise power spectrum obtained after analyzing the confocal point scanning Raman spectrometer produced by Horiba.

[0028] Figure 7 This is a schematic diagram of the instrument noise power spectrum obtained after analyzing the confocal point scanning Raman spectrometer produced by Nanophoton.

[0029] Figure 8 Schematic diagram of infrared spectrum sampling of human serum samples.

[0030] Fig. 9 Schematic diagram of fluorescence spectrum sampling of gold film samples.

[0031] Fig.10 The present invention is a flowchart of an implementation method of a training method for a noise reduction model of a micro-spectroscopy instrument.

[0032] Fig.11 Schematic diagram of high signal-to-noise ratio Raman spectrum simulated using the Voigt model.

[0033] Fig.12 The present invention is a flowchart of an embodiment of a method for processing noise of a microscopic spectroscopy instrument.

[0034] Fig.13 Schematic diagram of noisy Raman spectroscopy data obtained using a confocal Raman spectroscopy instrument.

[0035] Fig.14 Schematic diagram of the coefficients after discrete cosine transform of noisy Raman spectral data.

[0036] Fig.15 Schematic representation of the instrument noise figure predicted by the machine learning model.

[0037] Fig.16 Schematic diagram of the predicted instrument noise.

[0038] Fig.17 Schematic diagram of the noise of high signal-to-noise ratio Raman spectrum after removing instrument noise.

[0039] Fig.18 It is a schematic structural diagram of an embodiment of an electronic device of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should be noted that the embodiments in this application and the features in the embodiments can be combined with each other without conflict.

[0041] In the present invention, "module", "device", "system" and the like refer to related entities applied to computers, such as hardware, a combination of hardware and software, software or software in execution, etc. In detail, for example, an element can be, but is not limited to, a process, a processor, an object, an executable element, an execution thread, a program and / or a computer running on a processor. In addition, an application or script program running on a server, a server can all be an element. One or more elements can be in an execution process and / or thread, and an element can be localized on a computer and / or distributed between two or more computers, and can be operated by various computer-readable media. An element can also communicate through local and / or remote processes according to a signal with one or more data packets, for example, a signal from a data that interacts with another element in a local system, a distributed system, and / or a network on the Internet through a signal to interact with other systems.

[0042] Finally, it should be noted that, in this article, the terms "include" and "comprise" include not only those elements, but also other elements not explicitly listed, or elements inherent to such processes, methods, articles or equipment. In the absence of further restrictions, the elements defined by the sentence "include..." do not exclude the existence of other identical elements in the process, method, article or equipment that includes the elements.

[0043] like Figure 1 Shown is a flow chart of one embodiment of a method for analyzing noise of a micro-spectroscopy instrument according to the present invention.

[0044] The microscopic spectrometer in the present invention includes: a fluorescence spectrometer, an infrared spectrometer, a Raman spectrometer, etc. The following will take a confocal Raman spectrometer as an example to illustrate the method for analyzing the noise of the microscopic spectrometer in the present invention.

[0045] Figure 1 The method for analyzing the noise of the micro-spectroscopy instrument of the illustrated embodiment includes:

[0046] Step S101, using a microscopic spectroscopy instrument to detect a substrate without signals to obtain substrate spectrum data. The substrate may be a gold film substrate, but is not limited thereto.

[0047] In this step, a signal-free flat gold film sample is used as a substrate, and its Raman spectrum data is measured by a confocal Raman spectrometer to obtain substrate spectrum data.

[0048] In one embodiment, the substrate spectrum data may be multiple pieces of substrate spectrum data obtained by repeatedly detecting the same signal-free substrate using a microscopic spectrum instrument. Figure 2 The figure shows the collected Raman spectrum data of the gold film. In the experiment, there are 10,000 spectrum data. It is preferred that the number of spectra is greater than 1,000 (the larger the number, the more complete the statistical characteristics).

[0049] Step S102, decomposing the base spectrum data to obtain the instrument background spectrum component.

[0050] In one embodiment, the specific steps of obtaining the instrument background spectrum component include:

[0051] First, the background spectrum data is decomposed by singular value decomposition (SVD) to obtain the instrument background spectrum component. The Raman spectrum data obtained in step S101 is subjected to singular value decomposition (SVD) to obtain the decomposed spectrum components, which include: background spectrum components and instrument noise components. The above-mentioned instrument background spectrum components are low-frequency spectra with slowly changing intensity, which can also be understood as instrument background spectrum components.

[0052] Secondly, the frequency spectrum of the spectral components is obtained through Fourier transform.

[0053] Afterwards, the component whose normalized frequency corresponding to the maximum value is within the set threshold is obtained from the spectrum as the instrument background spectrum component. The set threshold is preferably less than 0.01.

[0054] The above instrument background spectrum components are power spectra, among which the components corresponding to the normalized frequency within 1% or lower values. The others can be considered as instrument noise components. The high-frequency components with violent oscillations in intensity are instrument noise components.

[0055] like Figure 3 As shown, B1 and B7 represent the spectrum diagrams of the background spectrum components of the instrument. B2, B3, B4, B5, B6, B8, B9, and B10 represent the spectrum diagrams of the noise components of the instrument.

[0056] Step S103, removing the instrument background spectrum component from the base spectrum data to obtain the noise value of the micro-spectroscopy instrument. For example, removing the instrument background spectrum component from the base spectrum data by singular value decomposition (SVD) to obtain the noise value of the micro-spectroscopy instrument.

[0057] The noise spectrum of the confocal point scanning Raman spectrometer produced by Horiba and the line scanning confocal Raman spectrometer produced by Nanophoton were analyzed using the noise analysis method of the microspectroscopy instrument. The results are as follows: Figure 6 and Figure 7 shown.

[0058] Figure 6 The acquisition conditions of the instrument noise are as follows: the sample is a flat gold film, the objective lens is 50 times (NA=0.7), the laser wavelength is 633nm, the laser power measured at the rear focal plane of the objective lens is 1mW, the scanning mode is point-by-point scanning, the integration conditions are 0.1-0.5s per pixel, and 10,000 spectra are collected under each integration condition.

[0059] Figure 7 The acquisition conditions of the instrument noise are as follows: the sample is a flat gold film, the objective lens is 50 times (NA=0.45), the laser wavelength is 533nm, the laser power measured at the rear focal plane of the objective lens is 1uW, the scanning mode is line-by-line scanning (each line contains 400 pixels), the integration conditions are 1-5s per line, and 10,000 spectra are collected under each integration condition.

[0060] The above examples show that the noise of the spectral instrument will have a strong value at a fixed position in the frequency domain, which is the corresponding instrument background spectrum component and does not change with the integration time. Further experiments show that when the objective lens and laser are replaced, the peak position of the instrument noise power spectrum will not change significantly, so the analysis method of the microscopic spectral instrument noise in the present invention has good measurement stability and will not affect the test results due to the replacement of local measurement components.

[0061] The following example illustrates that in step S102, the base spectrum data is decomposed by the singular value decomposition (SVD) method to obtain the instrument background spectrum component.

[0062] For example: Figure 2 The instrument noise shown in is arranged as a 2D matrix X (each column vector is a noise spectrum), and its decomposition form is shown in Formula 2. n is a diagonal matrix, and its diagonal elements σ1~σ n All are greater than 0, and the other elements are 0. U and V T are all orthogonal matrices, and each column vector of U (u1 … u m ) store a component respectively (i.e. Figure 3 As shown), VT The corresponding coefficients are stored in a vector.

[0063]

[0064] Formula 2 can be further written in the form of linear superposition:

[0065]

[0066] Find the u corresponding to the background component in the matrix U i ( Figure 3 B1 and B7 represent the BG curve of the instrument noise N), let the corresponding σ i =0, and the instrument noise N can be obtained by summing the other items in formula 3, which can be expressed as formula 4:

[0067]

[0068] The above singular value decomposition SVD method can be used to analyze the base spectrum data to obtain the instrument background spectrum component, that is, Figure 3 BG curve in B1 and B7.

[0069] like Figure 8 Schematic diagram of infrared spectrum sampling of human serum samples and Fig. 9 As shown in the schematic diagram of fluorescence spectrum sampling of the gold film sample, its infrared spectrum and fluorescence spectrum have the same spectral characteristics as the Raman spectrum, so the infrared spectrum and fluorescence spectrum can also be applied to the analysis method of the noise of the microscopic spectrum instrument in the present invention.

[0070] In some embodiments, when the base spectrum data includes multiple base spectrum data, the method for analyzing the noise of the micro-spectroscopy instrument includes:

[0071] Remove the instrument background spectrum component from multiple base spectrum data:

[0072] Fast Fourier transform or discrete cosine transform multiple base spectrum data after removing instrument background spectrum components to obtain transformation coefficients of multiple noise values.

[0073] The absolute values ​​of the transformation coefficients of multiple noise values ​​are averaged to obtain the average noise spectrum of the microspectroscopy instrument.

[0074] Therefore, the average noise spectrum obtained as the noise of the microspectroscopy instrument has high accuracy and reliability.

[0075] Another embodiment of the present invention provides a noise reduction model for a micro-spectroscopy instrument, wherein the noise reduction model for a micro-spectroscopy instrument is obtained by a training method for a noise reduction model for a micro-spectroscopy instrument.

[0076] like Fig.10 Shown is a flow chart of an implementation method of a training method for a noise reduction model for a micro-spectroscopy instrument in the present invention.

[0077] The training method of the noise reduction model for micro-spectroscopy instrument includes:

[0078] Step S201 : obtaining the noise value of the micro-spectroscopic instrument according to the noise data analysis method of the micro-spectroscopic instrument in the embodiment of the present invention.

[0079] Step S202: Generate a high signal-to-noise ratio spectrum according to the spectrum fitting function, and the high signal-to-noise ratio spectrum is used to construct a training data set.

[0080] In some embodiments, a high signal-to-noise ratio spectrum expressed by the following formula 1 is generated by a Voigt function.

[0081]

[0082] Wherein, N represents the number of peaks, P represents the peak position, W represents the peak width, and A represents the peak intensity.

[0083] In some embodiments, the method for obtaining a high signal-to-noise ratio spectrum includes: assuming that there are N peaks in a spectrum data, using a computer to randomly generate N groups of P, W and A values, and respectively substituting the P, W and A values ​​into Formula 1 to generate a high signal-to-noise ratio spectrum. This improves the randomness and realism of the generated high signal-to-noise ratio spectrum signal.

[0084] Step S203, setting the machine learning model input to a noisy low signal-to-noise ratio spectrum, setting the output to the instrument noise value, and using the training data set to train the machine learning model to obtain a noise reduction model for the microscopic spectroscopy instrument noise.

[0085] The key to training a noise reduction model for microscopic spectroscopy instrument noise includes four main parts: (1) machine learning model (2) loss function (3) optimization algorithm (4) training dataset.

[0086] The machine learning model is the key to the instrument noise processing method. The noise reduction model of the microscopic spectroscopy instrument noise obtained by training in the present invention is used to predict the instrument noise.

[0087] In some embodiments, the machine learning model in the present invention is not limited to convolutional neural networks, but can also use multi-layer perceptron (MLP), recurrent neural network (RNN), generative adversarial network (GAN), attention network (Transformer), etc.

[0088] The loss function and the optimization algorithm are not limited in the present invention. As an embodiment, the loss function may use mean square error (MSE), and the optimization algorithm may use common optimizers such as adaptive moment estimation (Adam) and stochastic gradient descent (SGD).

[0089] The training data set construction method includes:

[0090] (1) Method for obtaining instrument noise:

[0091] Taking Raman spectrometer as an example, the noise of other spectrometers such as fluorescence and infrared can be deduced similarly. The present invention proposes a method for obtaining instrument noise, which includes the following steps:

[0092] S1: A flat gold film sample without signal is used as the substrate (but not limited to metal substrate). The main experimental condition variables are laser power and integration time. Under each experimental condition, a batch (preferably more than 10,000) of Raman spectra are collected, such as Figure 2 shown.

[0093] S2: Perform singular value decomposition on the batch of substrate Raman spectra to obtain Figure 3 The singular value components shown in FIG. 1 and FIG. 2 are schematic diagrams of the spectrum of the background spectrum components of the instrument. B1 and B7 are schematic diagrams of the spectrum of the background spectrum components of the instrument. B2, B3, B4, B5, B6, B8, B9 and B10 are schematic diagrams of the spectrum of the noise components of the instrument.

[0094] S3: After removing the instrument background singular value components from the raw data, the instrument noise can be obtained.

[0095] (2) Method for constructing high signal-to-noise ratio spectrum:

[0096] The Voigt function is used to fit the line shape of Raman, fluorescence and infrared spectra. The function is the convolution of Gaussian function and Lorentz function. To simplify the calculation, the Voigt function is approximated using formula 5. Among them, ρ is the mixing coefficient, which takes a value of 0-1, and the empirical value here is 0.6785. μ represents the peak position, and w is the full width at half maximum (FWHM). Formula 6 represents the Gaussian function, and formula 7 represents the Lorentz function.

[0097] V(x,μ,w)=ρ×G(x,μ,w)+(1-ρ)×L(x,μ,w) Formula 5

[0098]

[0099]

[0100] To construct a high signal-to-noise ratio spectrum with multiple peaks, the following parameters need to be set: N represents the number of peaks, P represents the peak position, W represents the peak width, and A represents the peak intensity. The user only needs to give the interval range of the above parameters and use the computer random program to generate them. The high signal-to-noise ratio spectrum finally generated can be expressed by formula 1.

[0101] like Fig.11 The following are a batch of high signal-to-noise ratio Raman spectra generated using the present invention's method for analyzing the noise of a microscopic spectrometer. The corresponding low signal-to-noise ratio spectral data can be obtained by Figure 3 to Figure 5 The instrument noise in Fig.11 The high signal-to-noise ratio spectra in are summed.

[0102] In some embodiments, the training method of the noise reduction model of the micro-spectroscopy instrument can be implemented in a system including a user end and a server end:

[0103] The training method of the noise reduction model for microscopic spectroscopy instrument noise includes:

[0104] The user uploads the noise data of the micro-spectroscopy instrument to the server. The server generates a high signal-to-noise ratio spectrum based on the spectral fitting function, and the high signal-to-noise ratio spectrum is used to construct a training data set. The server sets the machine learning model input as a low signal-to-noise ratio spectrum containing noise, sets the output as the noise value, and uses the training data set to train the machine learning model to obtain a noise reduction model for the micro-spectroscopy instrument noise.

[0105] In order to improve the machine learning model's ability to learn instrument noise patterns, the present invention also proposes an online training method. The user only needs to provide Figure 5 The instrument noise data shown does not require the storage of simulated high signal-to-noise ratio spectral data in advance. During the training process, the high and low signal-to-noise ratio spectral data are randomly generated online using the above method. That is, in different training iterations, high signal-to-noise ratio signals are not repeated, which helps the machine learning model to better learn the instrument noise pattern. This makes the model insensitive to high signal-to-noise ratio signals, thus having better generalization performance.

[0106] like Fig.12 FIG. 1 is a flow chart of an embodiment of a method for processing noise of a microscopic spectroscopy instrument in the present invention. The method for processing noise of a microscopic spectroscopy instrument in the present invention comprises:

[0107] Step S301, identifying the noisy spectral data of the sample through the noise reduction model of the micro-spectroscopic instrument in the present invention, and obtaining the noise value of the micro-spectroscopic instrument.

[0108] Step S302, obtaining the spectrum detection value of the sample according to the noisy spectrum data and the noise value of the micro-spectroscopy instrument. Subtracting the instrument noise data from the low signal-to-noise ratio spectrum data to obtain a high signal-to-noise ratio spectrum, that is, the spectrum detection value of the sample.

[0109] In some embodiments, the above step S301 includes:

[0110] The Raman spectrum data of the sample is detected and obtained. The discrete cosine transform coefficients are obtained by cosine transforming the Raman spectrum data. For example: Fig.13 As shown, the noisy Raman spectral data obtained by the confocal Raman spectroscopy instrument, that is, the Raman spectral data of the sample, is subjected to discrete cosine transformation (DCT transformation) to obtain the following Fig.14 The coefficients shown.

[0111] The discrete cosine transform coefficients are identified through the noise reduction model of the micro-spectroscopy instrument, and the instrument noise coefficients of the sample are output. For example, the discrete cosine transform coefficients are input into the trained machine learning model, and the output is as follows: Fig.15 Instrument noise figure predicted by the machine learning model shown.

[0112] The noise coefficient of the instrument is inversely discrete cosine transformed to obtain the noise value of the micro-spectroscopy instrument. Fig.15 The instrument noise coefficient shown in is subjected to an inverse discrete cosine transform (IDCT) to obtain Fig.16 The predicted instrument noise is shown in .

[0113] The above step S302 includes: removing the noise value of the micro-spectrometer from the noise value of the micro-spectrometer to obtain the Raman spectrum detection value of the sample. Fig.13 The noisy Raman spectrum shown is subtracted from Fig.16 The predicted instrument noise shown in can be used to obtain a high signal-to-noise ratio Raman spectrum after removing the instrument noise, such as Fig.17 As shown, this is the Raman spectrum detection value of the sample.

[0114] In the above example, the machine learning model is explained to obtain the discrete cosine transform coefficients of the instrument noise by learning in the frequency domain (DCT transform). However, the machine learning model in the present invention is not limited to the frequency domain, but can also act on the time-frequency domain. For example: wavelet transform or Gabor transform domain, that is, for example Fig.13 The Raman spectral data shown in the figure are subjected to wavelet or Gabor transformation, and the machine learning model predicts the instrument noise coefficient a3 in the corresponding transformation domain, and the corresponding inverse transformation is performed to obtain Fig.16 The predicted instrument noise shown in . It can also be directly applied to the time domain, that is, without performing a transformation, directly according to Fig.13 The noisy Raman spectroscopy data shown is predicted as Fig.16 The predicted instrument noise is shown in .

[0115] A fourth aspect of the present invention provides an electronic device comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute any one of the methods for analyzing the noise of a microscopic spectroscopy instrument or the steps of a method for processing the noise of a microscopic spectroscopy instrument of the present invention.

[0116] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of combined actions, but those skilled in the art should be aware that the present invention is not limited by the described order of actions, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention. In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0117] In some embodiments, an embodiment of the present invention provides a non-volatile computer-readable storage medium, in which one or more programs including execution instructions are stored. The execution instructions can be read and executed by an electronic device (including but not limited to a computer, a server, or a network device, etc.) to execute any of the above-mentioned micro-spectroscopy instrument noise big data analysis methods of the present invention.

[0118] In some embodiments, an embodiment of the present invention further provides a computer program product, which includes a computer program stored on a non-volatile computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes any of the above-mentioned micro-spectroscopy instrument noise big data analysis methods.

[0119] In some embodiments, an embodiment of the present invention also provides an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute a micro-spectroscopy instrument noise big data analysis method.

[0120] In some embodiments, the embodiments of the present invention further provide a storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements a method for analyzing big data noise of a micro-spectroscopy instrument.

[0121] Fig.18 FIG. 1 is a schematic diagram of the hardware structure of an electronic device for performing a method for analyzing noise big data of a micro-spectroscopy instrument provided by another embodiment of the present application. Fig.18 As shown, the device includes:

[0122] One or more processors 410 and memory 420, Fig.18 A processor 410 is taken as an example.

[0123] The device for executing the micro-spectroscopy instrument noise big data analysis method may further include: an input device 430 and an output device 440 .

[0124] The processor 410, the memory 420, the input device 430 and the output device 440 may be connected via a bus or other means. Fig.18 The example of connecting through bus is taken in the following.

[0125] The memory 420, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules corresponding to the method for analyzing the noise big data of a micro-spectroscopy instrument in the embodiment of the present application. The processor 410 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions and modules stored in the memory 420, that is, realizing the method for analyzing the noise big data of a micro-spectroscopy instrument in the above method embodiment.

[0126] The memory 420 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required by at least one function. The data storage area may store data created according to the use of the algorithm calling device, etc. In addition, the memory 420 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 420 may optionally include a memory remotely arranged relative to the processor 410, and these remote memories may be connected to the algorithm calling device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0127] The input device 430 can receive input digital or character information and generate signals related to user settings and function control of the algorithm calling device. The output device 440 can include display devices such as display screens.

[0128] One or more modules are stored in the memory 420, and when executed by one or more processors 410, the micro-spectroscopy instrument noise big data analysis method in any of the above method embodiments is executed.

[0129] The above-mentioned product can execute the method provided in the embodiment of the present application, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not fully described in this embodiment, please refer to the method provided in the embodiment of the present application.

[0130] The electronic device of the embodiment of the present application exists in various forms, including but not limited to:

[0131] (1) Mobile communication devices: These devices are characterized by their mobile communication functions and their main purpose is to provide voice and data communications. These terminals include: smart phones (such as iPhone), multimedia phones, functional phones, and low-end phones.

[0132] (2) Ultra-mobile personal computer devices: These devices fall into the category of personal computers, have computing and processing capabilities, and generally also have mobile Internet access features. These terminals include: PDA, MID and UMPC devices, such as iPad.

[0133] (3) Portable entertainment devices: These devices can display and play multimedia content. They include audio and video players (such as iPods), handheld game consoles, e-books, smart toys, and portable car navigation devices.

[0134] (4) Server: A device that provides computing services. The server consists of a processor, hard disk, memory, system bus, etc. The server is similar to a general computer architecture, but because it needs to provide highly reliable services, it has higher requirements in terms of processing power, stability, reliability, security, scalability, and manageability.

[0135] (5) Other electronic devices with data interaction functions.

[0136] The device embodiments described above are merely illustrative, wherein 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, i.e., they may be located in one place, or may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0137] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application is described in detail with reference to the above embodiments, a person skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for analyzing noise of a microscopic spectroscopy instrument, comprising: Detecting a substrate without a signal by the micro-spectroscopy instrument to obtain substrate spectrum data; Decomposing the base spectrum data to obtain instrument background spectrum components; Removing the instrument background spectrum component from the base spectrum data to obtain a noise value of the micro-spectroscopy instrument; Decomposing the base spectrum data by a singular value decomposition method to obtain the instrument background spectrum component; Wherein, decomposing the base spectrum data to obtain the instrument background spectrum component comprises: Obtaining the frequency spectrum of the base spectrum data; Obtaining from the spectrum a component whose normalized frequency corresponding to the maximum value is within a set threshold value as an instrument background spectrum component; Wherein, the base spectrum data includes a plurality of base spectrum data; Removing the instrument background spectrum component from the plurality of base spectrum data; Fast Fourier transform or discrete cosine transform multiple base spectrum data after removing the background spectrum components of the instrument to obtain transformation coefficients of multiple noise values; The absolute values ​​of the transformation coefficients of multiple noise values ​​are averaged to obtain the average noise spectrum of the micro-spectroscopy instrument.

2. A noise reduction model for microscopic spectroscopy instrument noise, wherein: The training method of the noise reduction model of the micro-spectroscopy instrument noise includes: According to the method for analyzing noise of a micro-spectroscopy instrument according to claim 1, a noise value of the micro-spectroscopy instrument is obtained; generating a high signal-to-noise ratio spectrum according to a spectral fitting function, wherein the high signal-to-noise ratio spectrum is used to construct a training data set; The input of the machine learning model is set to a noisy low signal-to-noise ratio spectrum, the output is set to the noise value, and the training data set is used to train the machine learning model to obtain a noise reduction model for the microscopic spectroscopy instrument noise.

3. The noise reduction model for microscopic spectroscopy instrument according to claim 2, wherein: The training method can be implemented in a system including a user end and a server end. The training method of the noise reduction model of the micro-spectroscopy instrument noise includes: The user end uploads the noise value of the micro-spectroscopy instrument to the server end; The server generates a high signal-to-noise ratio spectrum according to the spectrum fitting function, and the high signal-to-noise ratio spectrum is used to construct a training data set; The server sets the machine learning model input to be a noisy low signal-to-noise ratio spectrum, sets the output to be the noise value, and uses the training data set to train the machine learning model to obtain a noise reduction model for the microscopic spectroscopy instrument noise.

4. The noise reduction model for microscopic spectroscopy instrument according to claim 2 or 3, wherein: The training data set includes: The Voigt function generates a high signal-to-noise ratio spectrum as shown in Formula 1: Where N represents the number of peaks, P represents the peak position, W represents the peak width, and A represents the peak intensity; i represents the index of a single peak in the spectrum, A(i) represents the intensity of the i-th peak, P(i) represents the position of the i-th peak, W(i) represents the width of the i-th peak, x represents the variable measured by the spectrum, and v(x, P(i), W(i)) represents the Voigt profile function of the shape of the i-th peak; S represents the total signal based on the spectral signals of all peaks, and S(i) represents the contribution of the i-th peak to the total signal.

5. The noise reduction model for microscopic spectroscopy instrument according to claim 4, wherein: The method for acquiring the high signal-to-noise ratio spectrum includes: randomly generating N groups of P, W and A values; substituting the N groups of P, W and A values ​​into Formula 1 respectively to generate multiple high signal-to-noise ratio spectra.

6. The noise reduction model for microscopic spectroscopy instrument according to claim 2 or 3, wherein: The machine learning models include: convolutional neural network model, multi-layer perceptron neural network MLP, recurrent neural network RNN, generative adversarial network GAN or Transformer neural network.

7. A method for processing noise of a microscopic spectroscopy instrument, comprising: Using the noise reduction model of the micro-spectroscopic instrument as claimed in claim 2, identifying the noisy spectral data of the sample, and obtaining the noise value of the micro-spectroscopic instrument; The spectrum detection value of the sample is obtained according to the noisy spectrum data and the noise value of the micro-spectroscopic instrument.

8. The method for processing noise of a microscopic spectrometer according to claim 7, wherein: Identifying the noisy spectral data of the sample by using the noise reduction model of the micro-spectroscopic instrument as claimed in claim 2, and obtaining the noise value of the micro-spectroscopic instrument comprises: Detecting a sample by means of the microspectroscopy instrument to obtain Raman spectrum data of the sample; Cosine transforming the Raman spectrum data to obtain discrete cosine transform coefficients; Identify the discrete cosine transform coefficients through a noise reduction model of micro-spectroscopy instrument noise, and output the instrument noise coefficient of the sample; Inverse discrete cosine transform of the instrument noise coefficient to obtain the noise value of the micro-spectroscopy instrument; The step of obtaining the spectral detection value of the sample according to the noisy spectral data and the noise value of the micro-spectroscopic instrument comprises: The noise value of the micro-spectroscopy instrument is removed from the Raman spectrum data of the sample to obtain the Raman spectrum detection value of the sample.

9. An electronic device comprising: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the method described in claim 1 or claim 7.

Citation Information

Patent Citations

  • Raman spectra pretreatment method for removing effects of background noises

    CN102998296A

  • Background dark noise subtraction method of Raman spectrometer using grating array detector

    CN110108695A