A blood testing method and system based on spectral recognition

Abstract

The present invention discloses a blood testing method and system based on spectral recognition, which relates to the field of blood component detection, and specifically comprises: obtaining ambient temperature and tester information, calculating a target wavelength range based on the ambient temperature and the tester information, collecting the tester's finger blood through visible-near infrared light in the target wavelength range, and obtaining initial spectral data; decomposing the initial spectral data through a biorthogonal Daubechies wavelet basis function to obtain low-frequency coefficients and high-frequency coefficients, smoothing and filtering the high-frequency coefficients to remove noise, calculating a mean square error value and sorting to obtain an optimal wavelet basis function and a decomposition layer number, and obtaining corresponding reconstructed spectral data; bringing the characteristic spectral data into a spectral-neutrophil quantitative regression model, and calculating the neutrophil content data of the tester's blood.

Description

Technical Field

[0001] The present invention relates to the technical field of blood component testing, and in particular to a blood testing method and system based on spectral recognition. Background Art

[0002] In recent years, the application of spectral recognition technology in the field of blood component testing has been continuously expanded and deepened by virtue of its unique advantages. Among them, visible-near infrared spectroscopy technology covers the visible light interval with a wavelength range of 400nm to 780nm and the near infrared light interval with a wavelength range of 780nm to 1100nm. The light waves in these two intervals interact significantly with various components in the blood, thereby generating measurable spectral signals. Different components have different absorption and scattering characteristics of light and produce different spectral characteristics. Using spectral recognition technology to analyze and process these characteristics can achieve qualitative and quantitative detection of blood components. Non-invasive testing through visible-near infrared spectroscopy technology avoids blood sampling for the tester, reduces the risk of infection, is relatively simple and fast to operate, and can achieve rapid detection of blood components, thereby real-time monitoring of changes in component parameters in the blood and providing timely information for clinical decision-making.

[0003] At present, the Chinese invention patent with application number CN202210035976.7 discloses a method for the operation layout of a sensor group for non-invasive blood component detection. The application scheme specifically includes the following steps: preset the parameters to be measured; preset the sensor group: configure the sensor group according to the resonance absorption condition of the component to be measured to the near-infrared spectrum; sensor group layout: according to the resonance absorption condition of the sensor group to the near-infrared spectrum, preset the layout quantity and installation strategy of the transmitting sensor, transmission and reflection receiving sensor; preset sensor response: configure the response control timing of the transmitting sensor, so that the transmitting sensor is turned on in turn by the control instruction, so that each spectral wavelength sensor group performs data collection without interfering with each other, and accurately obtains the pulse waveform formed by the photoelectric signal through each component to be measured, thereby completing the operation layout of the sensor group. The technical solution involved in this invention can conveniently and quickly collect data on the blood component concentration value of the person being measured, ensure accurate calculation data, and provide accurate data premise for subsequent test calculations. However, due to the complex composition of blood, the wide range of wavelengths of irradiation causes the absorption peaks of different blood components to overlap, resulting in interference between the spectral signals of different components, affecting the accuracy of the test results. In addition, the hemoglobin content is huge, and the absorption of visible-infrared light is the strongest, affecting the spectral data of other components. In addition, the human blood composition is complex, and it is difficult to directly extract features from the collected spectral data, and the results obtained by direct calculation have large errors. Summary of the invention

[0004] The technical problem solved by the present invention is that the traditional visible-infrared spectral identification method selects full-band scanning when testing blood components, and does not consider the interference of hemoglobin for wavelength optimization. The blood components are complex, and it is difficult to directly extract features from the collected spectral data. The neutrophil content calculated by the traditional method has a large error.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A blood testing method based on spectral recognition specifically comprises the following steps:

[0007] Step S1, obtaining the ambient temperature and the tester's information, calculating the target wavelength range based on the ambient temperature and the tester's information, collecting the tester's finger blood through visible-near infrared light in the target wavelength range, and obtaining initial spectral data;

[0008] Step S2, decomposing the initial spectral data by biorthogonal Daubechies wavelet basis function to obtain a set of coefficient pairs of low-frequency coefficients and high-frequency coefficients, performing smoothing filtering on the high-frequency coefficients to remove noise, selecting the optimal coefficient pair by calculating the mean square error value and sorting from low to high, and obtaining the optimal wavelet basis function and the optimal decomposition level based on the optimal coefficient pair;

[0009] Step S3, performing optimal wavelet decomposition on the initial spectral data based on the optimal wavelet basis function and the optimal decomposition layer number to obtain target low-frequency coefficients and target high-frequency coefficients, performing linear transformation and mean filtering on the target low-frequency coefficients to obtain characteristic low-frequency coefficients, performing Gaussian filtering on the target high-frequency coefficients to obtain characteristic high-frequency coefficients, and obtaining characteristic spectral data through wavelet reconstruction;

[0010] Step S4, bringing the characteristic spectrum data into the spectrum-neutrophil quantitative regression model to calculate the neutrophil content data of the test subject's blood.

[0011] As a preferred embodiment of the blood test method based on spectral recognition described in the present invention, the ambient temperature is measured, the tester information is counted, the tester information includes the tester's gender, age and pathological data, the blood red blood cell concentration reference range is obtained based on the tester information and the human blood red blood cell concentration reference model, the light transmittance ratio coefficient is calculated by the blood red blood cell concentration range and the hemoglobin absorption coefficient, and the target wavelength range of the near-infrared irradiation light is obtained based on the light transmittance ratio coefficient and the ambient temperature. The calculation expression is:

[0012] ;

[0013] Where Q represents the transmittance ratio coefficient, Indicated in concentration The hemoglobin light absorption coefficient under Indicated in concentration The hemoglobin light absorption coefficient under Indicates the upper limit of the reference range of red blood cell concentration. Indicates the lower limit of the reference range of red blood cell concentration. Hemoglobin accounts for the largest proportion of the main cell components of human blood. Using the traditional full-band method not only increases energy consumption and time, but is also susceptible to interference from temperature and hemoglobin absorption. By optimizing the wavelength range of hemoglobin absorption, the influence of hemoglobin on spectral data can be reduced, which is conducive to collecting spectral data that clearly shows the characteristics of neutrophils.

[0014] As a preferred solution of the blood test method based on spectral recognition described in the present invention, the spectrometer irradiation parameters are set based on the target wavelength range of visible-near infrared irradiation light, the tester's finger is clamped and kept stable, and the initial spectral data is collected. The operation is simple and fast, and the user's professional requirements are low.

[0015] As a preferred solution of the blood testing method based on spectral recognition described in the present invention, wherein: wavelet basis functions and decomposition levels are selected, the initial spectral data are wavelet decomposed respectively by functions db1 to db10 in the biorthogonal Daubechies wavelet basis functions, and two to five layers of wavelet decomposition are performed in turn by each wavelet basis function to obtain a set of coefficient pairs of high-frequency coefficients and low-frequency coefficients.

[0016] As a preferred solution of the blood testing method based on spectral recognition described in the present invention, wherein: each high-frequency coefficient in the coefficient pair set is filtered by a ratio smoothing threshold method to obtain a smoothed high-frequency coefficient, the smoothed high-frequency coefficient and the corresponding low-frequency coefficient are subjected to an inverse wavelet transform to obtain reconstructed spectral data, the mean square error values ​​of the reconstructed spectral data are calculated and sorted, the coefficient pair corresponding to the minimum mean square error value is selected as the optimal coefficient pair, and the wavelet basis function and the decomposition level number corresponding to the optimal coefficient pair are determined as the optimal wavelet basis function and the optimal decomposition level number.

[0017] As a preferred scheme of the blood testing method based on spectral recognition described in the present invention, wherein: based on the optimal wavelet basis function and the optimal decomposition layer number, the initial spectral data is optimally decomposed by wavelet to obtain the target low-frequency coefficient and the target high-frequency coefficient, the target low-frequency coefficient is linearly transformed and mean filtered to obtain the characteristic low-frequency coefficient, the target high-frequency coefficient is Gaussian filtered to obtain the characteristic high-frequency coefficient, wavelet reconstruction is performed based on the characteristic low-frequency coefficient and the characteristic high-frequency coefficient to obtain characteristic spectral data, the characteristic spectral data is brought into the spectral-neutrophil quantitative regression model to calculate the cell number prediction matrix, and the cell number prediction matrix is ​​inversely matrixed to obtain the neutrophil content data of the test subject's blood.

[0018] As a preferred embodiment of the blood test method based on spectral recognition described in the present invention, the spectrum-neutrophil quantitative regression model is calculated by an improved partial least squares method, and the calculation logic of the spectrum-neutrophil quantitative regression model includes: matrixing the characteristic spectrum data based on the corresponding wavelength sequence of the target wavelength range of the near-infrared irradiation light to obtain a sample characteristic matrix, establishing an equation group of the sample characteristic matrix and the cell number matrix based on the principal component analysis method, and decomposing the sample characteristic matrix and the cell number matrix into corresponding load matrices, score matrices and residual matrices respectively;

[0019] The vectors in the sample feature matrix are mapped to the high-dimensional matrix space through the polynomial kernel function matrix, and the weight components and the eigenvalue diagonal matrix are calculated by the eigenvalue decomposition method. The prediction score matrix corresponding to the cell number matrix is ​​obtained by projection calculation based on the weight components and the cell number matrix. The prediction score matrix corresponding to the sample feature matrix is ​​calculated based on the prediction score matrix corresponding to the cell number matrix and the eigenvalue diagonal matrix. The cell number prediction matrix is ​​calculated by the score matrix corresponding to the sample spectral matrix, the prediction score matrix corresponding to the sample spectral matrix, the prediction score matrix corresponding to the cell number matrix, and the load matrix corresponding to the cell number matrix. The calculation expression is:

[0020] ;

[0021] ;

[0022] ;

[0023] ;

[0024] Among them, X represents the sample spectrum matrix, T represents the score matrix corresponding to the sample spectrum matrix, P represents the load matrix corresponding to the sample spectrum matrix, E represents the residual matrix corresponding to the sample spectrum matrix, Y represents the cell number matrix, U represents the score matrix corresponding to the cell number matrix, Q represents the load matrix corresponding to the cell number matrix, and F represents the residual matrix corresponding to the cell number matrix. represents the weight component, represents the eigenvalue diagonal matrix , Represents the prediction score matrix corresponding to the cell number matrix, Represents the prediction score matrix corresponding to the sample spectrum matrix, represents the i-th orthogonal component corresponding to the kernel function matrix, t represents the matrix transpose, Represents the cell number prediction matrix.

[0025] A blood testing system based on spectral recognition, specifically comprising: an acquisition module, a data processing module, a power module and a terminal module;

[0026] The acquisition module is used to collect visible-near infrared spectrum data of the tester's blood and transmit it to the data processing module. The acquisition module includes a light source submodule, a receiving submodule, a temperature measurement submodule and an acquisition and data transmission submodule;

[0027] The light source submodule is used to transmit visible-near infrared light waves with a wavelength range of 400nm to 1100nm, and includes a laser diode, a fiber coupler and a transmission fiber;

[0028] The receiving submodule is used to receive the transmission wave and the reflection wave of the visible-near infrared wave, and includes a diode array detector, a finger fixture, a photoelectric converter and an analog-to-electric converter;

[0029] The temperature measurement submodule is used to measure the ambient temperature and provide ambient temperature parameters for calculating the target wavelength range, and includes a thermistor probe and a thermoelectric conversion module;

[0030] The data processing module is used to calculate the target wavelength range according to the ambient temperature and the tester's information, collect the tester's finger blood through the visible-near infrared light of the target wavelength range to obtain initial spectral data, decompose the initial spectral data through the biorthogonal Daubechies wavelet basis function to obtain a coefficient pair set of low-frequency coefficients and high-frequency coefficients, perform smoothing filtering on the high-frequency coefficients to remove noise, select the optimal coefficient pair by calculating the mean square error value and sorting from low to high, obtain the optimal wavelet basis function and the optimal decomposition layer number based on the optimal coefficient pair, perform optimal wavelet decomposition on the initial spectral data based on the optimal wavelet basis function and the optimal decomposition layer number to obtain the target low-frequency coefficient and the target high-frequency coefficient, perform linear transformation and mean filtering on the target low-frequency coefficient to obtain the characteristic low-frequency coefficient, perform Gaussian filtering on the target high-frequency coefficient to obtain the characteristic high-frequency coefficient, obtain characteristic spectral data through wavelet reconstruction, bring the characteristic spectral data into the spectrum-neutrophil quantitative regression model, and calculate the neutrophil content data of the tester's blood;

[0031] The power module is used to provide power to the acquisition module, the data processing module and the terminal module through a lithium battery, and is connected to the acquisition module, the data processing module and the terminal module through a power line;

[0032] The terminal module is used to accept and execute user input instructions, and feed back instruction execution results to the user through a display interface. Input instructions include start instructions, test instructions and query instructions. Instruction execution result feedback includes displaying system status, data acquisition results, real-time test results and historical test results to the user. The terminal module is connected to the acquisition module and the data processing module through a data line, and is connected to the power module through a power line.

[0033] Beneficial effects of the present invention: The present invention can detect the neutrophil content in blood in real time without loss, which is beneficial to judge and detect the immune status and pathological information of the tester. Taking into full consideration the fact that hemoglobin accounts for the largest proportion among the main cells in human blood, the traditional full-band scanning method is avoided, which not only reduces the energy consumption and time of spectral scanning, but also reduces the influence of hemoglobin on spectral data by optimizing the wavelength range of hemoglobin absorption, which is beneficial to collect spectral data that clearly shows the characteristics of neutrophils. At the same time, the acquisition operation is simple and fast, and the professional requirements for users are low. When processing the spectral data, it is first filtered by wavelet decomposition, which is beneficial to reduce noise errors while retaining effective information, and effectively simplify the subsequent processing of complex spectral data. Calculation by the established regression model is beneficial to quantitatively calculate the number of neutrophils from complex blood components, and ensure that the results can be output in real time while calculating high-precision results. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A basic flow chart of a blood testing method based on spectral recognition according to the present invention;

[0035] Figure 2 The figure is a schematic diagram of the framework of a blood testing system based on spectral recognition according to the present invention. DETAILED DESCRIPTION

[0036] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0037] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides a blood testing method based on spectral recognition, which specifically includes the following steps:

[0038] Step S1, obtaining the ambient temperature and the tester's information, calculating the target wavelength range based on the ambient temperature and the tester's information, collecting the tester's finger blood through visible-near infrared light in the target wavelength range, and obtaining initial spectral data;

[0039] Step S2, decomposing the initial spectral data by biorthogonal Daubechies wavelet basis function to obtain a set of coefficient pairs of low-frequency coefficients and high-frequency coefficients, performing smoothing filtering on the high-frequency coefficients to remove noise, selecting the optimal coefficient pair by calculating the mean square error value and sorting from low to high, and obtaining the optimal wavelet basis function and the optimal decomposition level based on the optimal coefficient pair;

[0040] Step S3, performing optimal wavelet decomposition on the initial spectral data based on the optimal wavelet basis function and the optimal decomposition layer number to obtain target low-frequency coefficients and target high-frequency coefficients, performing linear transformation and mean filtering on the target low-frequency coefficients to obtain characteristic low-frequency coefficients, performing Gaussian filtering on the target high-frequency coefficients to obtain characteristic high-frequency coefficients, and obtaining characteristic spectral data through wavelet reconstruction;

[0041] Step S4, bringing the characteristic spectrum data into the spectrum-neutrophil quantitative regression model to calculate the neutrophil content data of the test subject's blood.

[0042] In this embodiment, the ambient temperature is measured, and the tester information is counted, the tester information includes the tester's gender, age and pathological data, and the blood red blood cell concentration reference range is obtained based on the tester information and the human blood red blood cell concentration reference model. The light transmittance ratio coefficient is calculated by the blood red blood cell concentration range and the hemoglobin absorption coefficient. The target wavelength range of the near-infrared irradiation light is obtained based on the light transmittance ratio coefficient and the ambient temperature. The calculation expression is:

[0043] ;

[0044] Where Q represents the transmittance ratio coefficient, Indicated in concentration The hemoglobin light absorption coefficient under Indicated in concentration The hemoglobin light absorption coefficient under Indicates the upper limit of the reference range of red blood cell concentration. Indicates the lower limit of the reference range of red blood cell concentration. Hemoglobin accounts for the largest proportion of the main cell components of human blood. Using the traditional full-band method not only increases energy consumption and time, but is also susceptible to interference from temperature and hemoglobin absorption. By optimizing the wavelength range of hemoglobin absorption, the influence of hemoglobin on spectral data can be reduced, which is conducive to collecting spectral data that clearly shows the characteristics of neutrophils.

[0045] In this embodiment, the spectrometer illumination parameters are set based on the target wavelength range of the visible-near infrared illumination light, and the tester's finger is clamped and kept stable to complete the collection of initial spectral data.

[0046] In this embodiment, wavelet basis functions and decomposition levels are selected, and the initial spectral data are respectively subjected to wavelet decomposition using functions db1 to db10 in the biorthogonal Daubechies wavelet basis functions. Two to five layers of wavelet decomposition are performed in sequence using each wavelet basis function to obtain a set of coefficient pairs of high-frequency coefficients and low-frequency coefficients.

[0047] In this embodiment, each high-frequency coefficient in the coefficient pair set is filtered by a ratio smoothing threshold method to obtain a smoothed high-frequency coefficient, and the smoothed high-frequency coefficient and the corresponding low-frequency coefficient are subjected to an inverse wavelet transform to obtain reconstructed spectral data. The mean square error values ​​of the reconstructed spectral data are calculated and sorted respectively, and the coefficient pair corresponding to the minimum mean square error value is selected as the optimal coefficient pair, and the wavelet basis function and the number of decomposition layers corresponding to the optimal coefficient pair are determined as the optimal wavelet basis function and the optimal number of decomposition layers.

[0048] In this embodiment, the initial spectral data is optimally decomposed by wavelet based on the optimal wavelet basis function and the optimal number of decomposition layers to obtain target low-frequency coefficients and target high-frequency coefficients, the target low-frequency coefficients are linearly transformed and mean filtered to obtain characteristic low-frequency coefficients, the target high-frequency coefficients are Gaussian filtered to obtain characteristic high-frequency coefficients, wavelet reconstruction is performed based on the characteristic low-frequency coefficients and the characteristic high-frequency coefficients to obtain characteristic spectral data, the characteristic spectral data are entered into the spectrum-neutrophil quantitative regression model to calculate the cell number prediction matrix, and the cell number prediction matrix is ​​inversely matrixed to obtain the neutrophil content data of the test subject's blood.

[0049] Among them, the quantitative calculation of neutrophils is completed by calculating the cell number prediction matrix, which is conducive to the accurate and efficient detection of blood neutrophils.

[0050] In this embodiment, the spectrum-neutrophil quantitative regression model is calculated by an improved partial least squares method, and the calculation logic of the spectrum-neutrophil quantitative regression model includes: matrixing the characteristic spectrum data based on the corresponding wavelength sequence of the target wavelength range of the near-infrared irradiation light to obtain a sample characteristic matrix, and the logic of the matrix processing includes: arranging the wavelengths in the wavelength sequence in descending order as the rows of the matrix, and taking the corresponding measurement results of each wavelength as the columns of the matrix to obtain the sample characteristic matrix;

[0051] Based on the principal component analysis method, the equation group of the sample feature matrix X and the cell number matrix Y was established. The sample feature matrix X and the cell number matrix Y were decomposed into corresponding load matrices P and Q, score matrices T and U, and residual matrices E and F, respectively. The covariance matrix in the traditional partial least squares method was replaced by the polynomial kernel function matrix for calculation. The vectors in the sample feature matrix X were mapped to the high-dimensional matrix space through the polynomial kernel function matrix, and the weight components were calculated by the eigenvalue decomposition method. Diagonal matrix with eigenvalues , based on the weight components Projection calculation is performed with the cell number matrix Y to obtain the prediction score matrix corresponding to the cell number matrix Y , based on the prediction score matrix corresponding to the cell number matrix Diagonal matrix with eigenvalues Calculate and obtain the prediction score matrix corresponding to the sample feature matrix X , through the score matrix T corresponding to the sample spectrum matrix and the predicted score matrix corresponding to the sample spectrum matrix , the prediction score matrix corresponding to the cell number matrix Loading matrix corresponding to the cell number matrix Calculate the cell number prediction matrix , its calculation expression is:

[0052] ;

[0053] ;

[0054] ;

[0055] ;

[0056] Among them, X represents the sample spectrum matrix, T represents the score matrix corresponding to the sample spectrum matrix, P represents the load matrix corresponding to the sample spectrum matrix, E represents the residual matrix corresponding to the sample spectrum matrix, Y represents the cell number matrix, U represents the score matrix corresponding to the cell number matrix, Q represents the load matrix corresponding to the cell number matrix, and F represents the residual matrix corresponding to the cell number matrix. represents the weight component, represents the eigenvalue diagonal matrix , Represents the prediction score matrix corresponding to the cell number matrix, Represents the prediction score matrix corresponding to the sample spectrum matrix, represents the i-th orthogonal component corresponding to the kernel function matrix, t represents the matrix transpose, Represents the cell number prediction matrix.

[0057] Example 2, reference Figure 2 , as an embodiment of the present invention, provides a blood testing system based on spectrum recognition, which specifically comprises: an acquisition module, a data processing module, a power supply module and a terminal module;

[0058] The acquisition module is used to collect visible-near infrared spectrum data of the tester's blood and transmit it to the data processing module. The acquisition module includes a light source submodule, a receiving submodule, a temperature measurement submodule and an acquisition and data transmission submodule;

[0059] The light source submodule is used to transmit visible-near infrared light waves with a wavelength range of 400nm to 1100nm, and includes a laser diode, a fiber coupler and a transmission fiber;

[0060] The receiving submodule is used to receive the transmission wave and the reflection wave of the visible-near infrared wave, and includes a diode array detector, a finger fixture, a photoelectric converter and an analog-to-electric converter;

[0061] The temperature measurement submodule is used to measure the ambient temperature and provide ambient temperature parameters for calculating the target wavelength range, and includes a thermistor probe and a thermoelectric conversion module;

[0062] The data processing module is used to calculate the target wavelength range according to the ambient temperature and the tester's information, collect the tester's finger blood through the visible-near infrared light of the target wavelength range to obtain initial spectral data, decompose the initial spectral data through the biorthogonal Daubechies wavelet basis function to obtain a coefficient pair set of low-frequency coefficients and high-frequency coefficients, perform smoothing filtering on the high-frequency coefficients to remove noise, select the optimal coefficient pair by calculating the mean square error value and sorting from low to high, obtain the optimal wavelet basis function and the optimal decomposition layer number based on the optimal coefficient pair, perform optimal wavelet decomposition on the initial spectral data based on the optimal wavelet basis function and the optimal decomposition layer number to obtain the target low-frequency coefficient and the target high-frequency coefficient, perform linear transformation and mean filtering on the target low-frequency coefficient to obtain the characteristic low-frequency coefficient, perform Gaussian filtering on the target high-frequency coefficient to obtain the characteristic high-frequency coefficient, obtain characteristic spectral data through wavelet reconstruction, bring the characteristic spectral data into the spectrum-neutrophil quantitative regression model, and calculate the neutrophil content data of the tester's blood;

[0063] The power module is used to provide power to the acquisition module, the data processing module and the terminal module through a lithium battery, and is connected to the acquisition module, the data processing module and the terminal module through a power line;

[0064] The terminal module is used to accept and execute user input instructions, and feed back instruction execution results to the user through a display interface. Input instructions include start instructions, test instructions and query instructions. Instruction execution result feedback includes displaying system status, data acquisition results, real-time test results and historical test results to the user. The terminal module is connected to the acquisition module and the data processing module through a data line, and is connected to the power module through a power line.

[0065] The present invention also provides an electronic device, including a processor and a memory, wherein the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, a blood testing system based on spectral recognition described in Example 2 is run.

[0066] The present invention also provides a storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the blood testing system based on spectral recognition described in Example 2 is run.

[0067] It should be understood by those skilled in the art that the embodiments of the present invention may provide methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. Among them, the storage medium may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0068] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A blood testing method based on spectral recognition, characterized in that: The steps include: Step S1, obtaining the ambient temperature and the tester's information, calculating the target wavelength range based on the ambient temperature and the tester's information, collecting the tester's finger blood through visible-near infrared light in the target wavelength range, and obtaining initial spectral data; Step S2, decomposing the initial spectral data by biorthogonal Daubechies wavelet basis function to obtain a set of coefficient pairs of low-frequency coefficients and high-frequency coefficients, performing smoothing filtering on the high-frequency coefficients to remove noise, selecting the optimal coefficient pair by calculating the mean square error value and sorting from low to high, and obtaining the optimal wavelet basis function and the optimal decomposition level based on the optimal coefficient pair; Step S3, performing optimal wavelet decomposition on the initial spectral data based on the optimal wavelet basis function and the optimal decomposition layer number to obtain target low-frequency coefficients and target high-frequency coefficients, performing linear transformation and mean filtering on the target low-frequency coefficients to obtain characteristic low-frequency coefficients, performing Gaussian filtering on the target high-frequency coefficients to obtain characteristic high-frequency coefficients, and obtaining characteristic spectral data through wavelet reconstruction; Step S4, bringing the characteristic spectrum data into the spectrum-neutrophil quantitative regression model to calculate the neutrophil content data of the test subject's blood; The wavelet basis function and the number of decomposition layers are selected, and the initial spectral data are respectively subjected to wavelet decomposition by functions db1 to db10 in the biorthogonal Daubechies wavelet basis function, and the coefficient pair set of high-frequency coefficients and low-frequency coefficients is obtained by sequentially performing two to five layers of wavelet decomposition by each wavelet basis function; Filtering each high-frequency coefficient in the coefficient pair set by a ratio smoothing threshold method to obtain a smoothed high-frequency coefficient, performing an inverse wavelet transform based on the smoothed high-frequency coefficient and the corresponding low-frequency coefficient to obtain reconstructed spectral data, respectively calculating the mean square error values ​​of the reconstructed spectral data and sorting them from low to high, selecting the coefficient pair corresponding to the minimum mean square error value as the optimal coefficient pair, and determining the wavelet basis function and the number of decomposition layers corresponding to the optimal coefficient pair as the optimal wavelet basis function and the optimal number of decomposition layers; Based on the optimal wavelet basis function and the optimal number of decomposition layers, the initial spectral data is optimally decomposed by wavelet to obtain the target low-frequency coefficient and the target high-frequency coefficient. The target low-frequency coefficient is linearly transformed and mean filtered to obtain the characteristic low-frequency coefficient. The target high-frequency coefficient is Gaussian filtered to obtain the characteristic high-frequency coefficient. Wavelet reconstruction is performed based on the characteristic low-frequency coefficient and the characteristic high-frequency coefficient to obtain characteristic spectral data. The characteristic spectral data is brought into the spectrum-neutrophil quantitative regression model to calculate the cell number prediction matrix. The cell number prediction matrix is ​​inversely matrixed to obtain the neutrophil content data of the test subject's blood.

2. A blood testing method based on spectral recognition as claimed in claim 1, characterized in that: The ambient temperature is measured and the tester information, including the tester's gender, age and pathological data, is collected. The reference range of red blood cell concentration is obtained based on the tester information and the human red blood cell concentration reference model. The transmittance ratio coefficient is calculated by the red blood cell concentration range and the absorption coefficient of hemoglobin. The target wavelength range of near-infrared irradiation light is obtained based on the transmittance ratio coefficient and the ambient temperature.

3. A blood testing method based on spectral recognition as claimed in claim 1, characterized in that: The spectrometer irradiation parameters are set based on the target wavelength range of visible-near infrared irradiation light, and the tester's fingers are clamped and kept stable to complete the collection of initial spectral data.

4. A blood testing method based on spectral recognition as claimed in claim 1, characterized in that: The spectrum-neutrophil quantitative regression model is calculated by an improved partial least squares method. The calculation logic of the spectrum-neutrophil quantitative regression model includes: matrixing the characteristic spectrum data based on the corresponding wavelength sequence of the target wavelength range of the near-infrared irradiation light to obtain a sample characteristic matrix, establishing an equation group of the sample characteristic matrix and the cell number matrix based on the principal component analysis method, and decomposing the sample characteristic matrix and the cell number matrix into corresponding load matrices, score matrices and residual matrices respectively; The vectors in the sample feature matrix are mapped to the high-dimensional matrix space through the polynomial kernel function matrix, and the weight components and the eigenvalue diagonal matrix are calculated by the eigenvalue decomposition method. The prediction score matrix corresponding to the cell number matrix is ​​obtained by projection calculation based on the weight components and the cell number matrix. The prediction score matrix corresponding to the sample feature matrix is ​​calculated based on the prediction score matrix corresponding to the cell number matrix and the eigenvalue diagonal matrix. The cell number prediction matrix is ​​calculated by the score matrix corresponding to the sample spectral matrix, the prediction score matrix corresponding to the sample spectral matrix, the prediction score matrix corresponding to the cell number matrix, and the load matrix corresponding to the cell number matrix. The calculation expression is: ； ； ； ； Among them, X represents the sample spectrum matrix, T represents the score matrix corresponding to the sample spectrum matrix, P represents the load matrix corresponding to the sample spectrum matrix, E represents the residual matrix corresponding to the sample spectrum matrix, Y represents the cell number matrix, U represents the score matrix corresponding to the cell number matrix, Q represents the load matrix corresponding to the cell number matrix, and F represents the residual matrix corresponding to the cell number matrix. represents the weight component, represents the eigenvalue diagonal matrix , Represents the prediction score matrix corresponding to the cell number matrix, Represents the prediction score matrix corresponding to the sample spectrum matrix, represents the i-th orthogonal component corresponding to the kernel function matrix, t represents the matrix transpose, Represents the cell number prediction matrix.

5. A blood testing system based on spectral recognition for implementing the blood testing method based on spectral recognition as claimed in claim 1, characterized in that: include: Acquisition module, data processing module, power module and terminal module; The acquisition module is used to collect visible-near infrared spectrum data of the tester's blood and transmit it to the data processing module. The acquisition module includes a light source submodule, a receiving submodule, a temperature measurement submodule and an acquisition and data transmission submodule; The light source submodule is used to transmit visible-near infrared light waves with a wavelength range of 400nm to 1100nm, and includes a laser diode, a fiber coupler and a transmission fiber; The receiving submodule is used to receive the transmission wave and the reflection wave of the visible-near infrared wave, and includes a diode array detector, a finger fixture, a photoelectric converter and an analog-to-electric converter; The temperature measurement submodule is used to measure the ambient temperature and provide ambient temperature parameters for calculating the target wavelength range, and includes a thermistor probe and a thermoelectric conversion module; The data processing module is used to calculate the target wavelength range according to the ambient temperature and the tester's information, collect the tester's finger blood through the visible-near infrared light of the target wavelength range to obtain initial spectral data, decompose the initial spectral data through the biorthogonal Daubechies wavelet basis function to obtain a coefficient pair set of low-frequency coefficients and high-frequency coefficients, perform smoothing filtering on the high-frequency coefficients to remove noise, select the optimal coefficient pair by calculating the mean square error value and sorting from low to high, obtain the optimal wavelet basis function and the optimal decomposition layer number based on the optimal coefficient pair, perform optimal wavelet decomposition on the initial spectral data based on the optimal wavelet basis function and the optimal decomposition layer number to obtain the target low-frequency coefficient and the target high-frequency coefficient, perform linear transformation and mean filtering on the target low-frequency coefficient to obtain the characteristic low-frequency coefficient, perform Gaussian filtering on the target high-frequency coefficient to obtain the characteristic high-frequency coefficient, obtain characteristic spectral data through wavelet reconstruction, bring the characteristic spectral data into the spectrum-neutrophil quantitative regression model, and calculate the neutrophil content data of the tester's blood; The power module is used to provide power to the acquisition module, the data processing module and the terminal module through a lithium battery, and is connected to the acquisition module, the data processing module and the terminal module through a power line; The terminal module is used to accept and execute user input instructions, and feed back instruction execution results to the user through a display interface. Input instructions include start instructions, test instructions and query instructions. Instruction execution result feedback includes displaying system status, data acquisition results, real-time test results and historical test results to the user. The terminal module is connected to the acquisition module and the data processing module through a data line, and is connected to the power module through a power line.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the blood testing method based on spectrum recognition as claimed in any one of claims 1 to 4 are implemented.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the blood testing method based on spectrum recognition as claimed in any one of claims 1 to 4 are implemented.

Images