A method for processing image data of a neomycin spectrum based on a deep learning model

Abstract

The application relates to the technical field of image processing, in particular to a new neomycin spectrum image data processing method based on a deep learning model, which comprises the following steps: obtaining initial new neomycin spectrum data, inputting each new neomycin spectrum into a target deep learning model to obtain corresponding image feature vectors; and obtaining new neomycin classification cluster data according to the image feature vectors; it is known that, by using the time sequence relationship degree between the image features of any two initial new neomycin images, the initial new neomycin images are classified based on the time sequence relationship degree, the time relationship between the new neomycin images can be analyzed, the data of the new neomycin changing with time is established, and the data quantity of the new neomycin images is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to an image data processing method for neomycin spectrum based on a deep learning model. BACKGROUND

[0002] Neomycin is an aminoglycoside antibiotic obtained by biological fermentation, and its main components are neomycin A, neomycin B and neomycin C, among which neomycin B is the main component. The content of each component of neomycin is required in the pharmacopoeia of each country, and the content of neomycin A is not more than 2% and the content of neomycin C is not more than 15%. The product obtained by the current fermentation process basically does not exceed the standard range of neomycin A, but since neomycin C and neomycin B are stereoisomers, and neomycin B is synthesized from neomycin C by epimerase, neomycin C is a necessary precursor for the synthesis of neomycin B, and it cannot bypass neomycin C in the synthesis pathway through technical means.

[0003] Bioactivity detection found that the antibacterial activity of neomycin C is only one third of that of neomycin B, and the toxic side effects are large, so it has become one of the controlled impurities in the quality standards of each country.

[0004] With the rapid development of deep learning technology in various fields, deep learning has further promoted the analysis of proteomics in the field of chemical omics. Some studies identify markers by analyzing raw mass spectrometry using MSGF+, MaxQuant and other tools, and obtain differential markers through statistical analysis, and then use traditional machine learning methods such as decision tree and support vector machine to establish a disease diagnosis classification model; some studies identify markers and establish a disease diagnosis model based on convolutional neural network by extracting features from mass spectrometry data in the form of a graph; these classification results excessively depend on the accuracy of marker identification, and since marker identification is prone to not being completely identified or the identified markers are not highly related to the disease, which will lead to a decrease in the accuracy of the downstream diagnosis model.

[0005] However, for neomycin components, deep learning has not been used before, and at the same time, the amount of neomycin data is relatively small, so it is not possible to accurately analyze the time relationship between neomycin images when a certain amount of neomycin image data is obtained, and to establish data of neomycin changing with time. SUMMARY

[0006] According to the present application, an image data processing method for neomycin spectrum based on a deep learning model is provided, which comprises the following steps:

[0007] S100, obtaining initial neomycin spectrum data A={A1,..., An} and a deep learning model. i ,..., An} and a deep learning model.m}, A i It is the i-th initial neomycin spectrum, where i ranges from 1 to m, and m is the number of initial neomycin spectra.

[0008] S200, each A i Inputting it into the target deep learning model yields A i The corresponding image feature vector B i ={B i1 , ..., B ij , ..., B in}, B ij It is A i The j-th image feature, where j ranges from 1 to n, and n is A. i The corresponding number of image features.

[0009] S300, according to each B i Obtain the neomycin taxonomic cluster data U={U1, ..., U...} t , ..., U β}, U t Let t be the taxonomic cluster of the t-th neomycin, where t ranges from 1 to β, and β is the number of taxonomic clusters of neomycin.

[0010] The S300 step also includes the following steps:

[0011] S301, according to B i Besides B, in B i The temporal relation degree F between other image feature vectors besides [the above] i ={F i1 , ..., F iu , ..., F iη}, F iu It is B i Besides B, in B i The temporal relation degree between the u-th image feature vectors, F iu Meets the following conditions:

[0012] B 0 uj It is B except for B. i The j-th image feature in the u-th image feature vector other than B, where u ranges from 1 to η, and η is a feature vector in B other than B. i The total number of other image feature vectors besides η, where ΔB g It is the classification feature threshold for the g-th acquisition time, where △B gThe acquisition of is to first obtain the temporal correlation degree between the image feature vector of the g-th sample image and the image feature vector of the g+1-th sample image in a certain neomycin component, and then to find the mean of the temporal correlation degree between the image feature vector of the g-th sample image and the image feature vector of the g+1-th sample image of all neomycin components. max() is the maximum value function.

[0013] S302, Based on the initial inter-vector temporal relation degree matrix F and the preset temporal relation threshold ΔF, generate the target inter-vector temporal relation degree matrix F. 0 F meets the following conditions:

[0014] .

[0015] S303, based on the temporal relationship matrix F between target vectors 0 Generate neomycin classification cluster data U={U1, ..., U...} t , ..., U β}

[0016] Compared with the prior art, the present invention has at least the following beneficial effects:

[0017] The present invention discloses an image data processing method for neomycin spectral images based on a deep learning model, the method comprising: S100, acquiring initial neomycin spectral data A={A1, ..., A...} i , ..., A m}, A i This is the i-th initial neomycin spectrum, where i ranges from 1 to m, and m is the number of initial neomycin spectra; S200, each A i Inputting it into the target deep learning model yields A i The corresponding image feature vector B i ={B i1 , ..., B ij , ..., B in}, B ij It is A i The j-th image feature, where j ranges from 1 to n, and n is A. i The corresponding number of image features; S300, based on each B i Obtain the neomycin taxonomic cluster data U={U1, ..., U...} t , ..., U β}, U t Let be the taxonomic cluster of the t-th neomycin, where t ranges from 1 to β, and β is the number of neomycin taxonomic clusters; step S300 further includes the following step: S301, according to B i Besides B, in B iThe temporal relation degree F between other image feature vectors besides [the above] i ={F i1 , ..., F iu , ..., F iη}, F iu It is B i Besides B, in B i The temporal relation degree between the u-th image feature vectors, F iu Meets the following conditions:

[0018] B 0 uj It is B except for B. i The j-th image feature in the u-th image feature vector other than B, where u ranges from 1 to η, and η is a feature vector in B other than B. i The total number of other image feature vectors besides η, where ΔB g It is the classification feature threshold for the g-th acquisition time, where △B g The acquisition involves first obtaining the temporal relation degree between the image feature vector of the g-th sample image and the image feature vector of the g+1-th sample image in a certain neomycin component, and then taking the mean of the temporal relation degrees between the image feature vectors of the g-th sample image and the image feature vectors of the g+1-th sample image of all neomycin components, with max() being the maximum value function; S302, based on the initial inter-vector temporal relation degree matrix F and the preset temporal relation threshold ΔF, the target inter-vector temporal relation degree matrix F is generated. 0 S303, based on the temporal relationship matrix F between target vectors 0 Generate neomycin classification cluster data U={U1, ..., U...} t , ..., U β It can be seen that by using the temporal relationship between the image features of any two initial neomycin images, and then classifying the initial neomycin images based on the temporal relationship, the temporal relationship between neomycin images can be analyzed, establishing data on the change of neomycin over time, thus increasing the amount of neomycin image data. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1This is a flowchart illustrating an image data processing method for neomycin spectra based on a deep learning model, provided as an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] like Figure 1 As shown, this embodiment provides a method for processing neomycin spectrum image data based on a deep learning model. The method includes the following steps:

[0023] S100, Obtain initial neomycin spectral data A={A1, ..., A...} i , ..., A m}, A i It is the i-th initial neomycin spectrum, where i ranges from 1 to m, and m is the number of initial neomycin spectra.

[0024] Specifically, the initial neomycin spectrum is the initial neomycin mass spectrum image, which is obtained by a mass spectrometer. Those skilled in the art are familiar with any existing method for obtaining a neomycin mass spectrum image using a mass spectrometer, and will not elaborate further here.

[0025] Furthermore, A1 to A m Each initial neomycin mass spectrum image was inconsistent.

[0026] S200, each A i Inputting it into the target deep learning model yields A i The corresponding image feature vector B i ={B i1 , ..., B ij , ..., B in}, B ij It is A i The j-th image feature, where j ranges from 1 to n, and n is A. i The corresponding number of image features.

[0027] Specifically, step S200 also includes the following steps:

[0028] S201, Obtain neomycin sample image data C={C1, ..., C...} r , ..., C s}, C r ={Cx1 , ..., C rg , ..., C rz}, C rg The image is the g-th sample image in the r-th neomycin fraction, where r ranges from 1 to s, s is the total number of neomycin fractions, and g ranges from 1 to z, where z is the total number of sample images in each neomycin fraction. Furthermore, the sample image is a mass spectrometry image of any neomycin fraction changing over time, wherein the sample image acquired at the g-th time node and the sample image acquired at the g+1-th time node remain unchanged in different neomycin fractions. This can be understood as: the g-th sample image is the sample image acquired at the g-th acquisition time.

[0029] S202, for each C rg After processing, we obtain C. rg The corresponding intermediate image C 0 rg , where C 0 rg Including several element points, C 0 rg Each element point (x0, y0) in the array satisfies the following condition:

[0030] ;

[0031] Where kh is the horizontal kernel size, kw is the vertical kernel size, μh is the mean of the Gaussian distribution in the horizontal direction, μw is the mean of the Gaussian distribution in the vertical direction, e is the horizontal offset within the kernel, f is the vertical offset within the kernel, bh is the horizontal position offset, bw is the vertical position offset, vh is the horizontal scaling factor, vw is the vertical scaling factor, σh is the standard deviation of the horizontal Gaussian distribution, σw is the standard deviation of the vertical Gaussian distribution, x is the horizontal coordinate point in the sample image, y is the vertical coordinate point in the sample image, x0 is the horizontal coordinate point in the intermediate image, and y0 is the vertical coordinate point in the intermediate image.

[0032] S203, according to C rg For each element point (x0, y0), obtain C. rg Image feature vector D rg =(D 1 rg , ..., D j rg , ..., D n rg ), D j rg It is C rg The j-th image feature in the image.

[0033] Furthermore, step S203 also includes the following steps:

[0034] S2031, according to C rg For each element point (x0, y0), select C. rg Corresponding key element data G rg ={(XG 1 rg YG 1 rg ), ..., (XG a rg YG a rg ), ..., (XG b rg YG b rg )}, (XG a rg YG a rg ) is the a-th key element point, XG a rg It is C rg The key coordinates of the a-th point on the X-axis, YG a rg It is C rg The key coordinates of the a-th point on the Y-axis, where each key element point (XG) a rg YG a rg ) represents the element point (x0, y0) that represents the component of neomycin, where a ranges from 1 to b, and b is the number of key element points. This can be understood as: C rg The corresponding key element is C. rg The elemental points representing the components of neomycin, C rg Other element points and C rg The key elements in the image are clearly distinguishable and can be differentiated by features identified from the image, such as texture features, RGB value features, and other image features. Those skilled in the art are familiar with any existing method for extracting image features, so it will not be elaborated here.

[0035] S2032, from C rg Several peak points and degrees of change are selected from the data. Each peak point and degree of change is then inserted as an image feature into the empty set of the image feature vector to obtain D. rg , can be understood as: D rg The image features are peak points or the degree of change.

[0036] S204, all D rgThe preset deep learning model is used as a sample to train the target deep learning model. Those skilled in the art know the methods for training deep learning models in any existing technology, which will not be described in detail here; for example, the preset deep learning model is a CNN-based model.

[0037] The above-mentioned method can optimize the image features of neomycin samples using a suitable kernel to generate a target deep learning model, which is beneficial for extracting accurate image features from the initial neomycin spectrum and thus accurately classifying the relationship between the initial neomycin spectrum.

[0038] S300, according to each B i Obtain the neomycin taxonomic cluster data U={U1, ..., U...} t , ..., U β}, U t Let t be the taxonomic cluster of the t-th neomycin, where t ranges from 1 to β, and β is the number of taxonomic clusters of neomycin.

[0039] Specifically, the S300 procedure also includes the following steps:

[0040] S301, according to B i Besides B, in B i The temporal relation degree F between other image feature vectors besides [the above] i ={F i1 , ..., F iu , ..., F iη}, F iu It is B i Besides B, in B i The temporal relation degree between the u-th image feature vectors, F iu Meets the following conditions:

[0041] B 0 uj It is B except for B. i The j-th image feature in the u-th image feature vector other than B, where u ranges from 1 to η, and η is a feature vector in B other than B. i The total number of other image feature vectors besides η, where ΔB g It is the temporal relation threshold corresponding to the sample images of different neomycin components collected at the g-th time node, and max() is the maximum value function.

[0042] Furthermore, △B gThe acquisition involves first obtaining the temporal correlation degree between the image feature vector of the g-th sample image and the image feature vector of the (g+1)-th sample image in a given neomycin component, and then averaging the temporal correlation degrees between the image feature vectors of the g-th sample image and the (g+1)-th sample image for all neomycin components. For example, the temporal correlation degree between the image feature vector of the g-th sample image and the image feature vector of the (g+1)-th sample image in the r-th neomycin component satisfies the following condition: D j rg D is the j-th image feature in the g-th sample image of the r-th neomycin component. j rg+1 It is the j-th image feature in the g+1-th sample image of the r-th neomycin component.

[0043] S302, Based on the initial inter-vector temporal relation degree matrix F and the preset temporal relation threshold ΔF, generate the target inter-vector temporal relation degree matrix F. 0 F meets the following conditions:

[0044]

[0045] The above, when F iu When ≥ △F, at F 0 Retain F in China iu When F iu When < △F, at F 0 Setting F in the middle iu =0, where the timing relationship threshold can be set by those skilled in the art according to actual needs, and will not be elaborated here.

[0046] S303, based on the temporal relationship matrix F between target vectors 0 Generate neomycin classification cluster data U={U1, ..., U...} t , ..., U β}; can be understood as: F 0 Each initial neomycin spectrum with a corresponding horizontal temporal relation degree of 0 is taken as a single neomycin taxonomic cluster, and F is used as the taxonomic cluster. 0 All initial neomycin spectra with non-zero lateral temporal relation degrees are considered as single neomycin taxonomic clusters. For example, when F i1 To F iη When both are 0, the corresponding initial neomycin spectrum is taken as the taxonomic cluster of a single neomycin. Meanwhile, when F... m1 To F mη When both are 0, the corresponding initial neomycin spectrum is also a single neomycin taxonomic cluster; and when F i1 To F iη The corresponding initial neomycin spectrum and when Fm1 To F mη All other initial neomycin spectra preceding the corresponding initial neomycin spectra are treated as a single neomycin taxonomic cluster.

[0047] This embodiment provides a method for image data processing of neomycin spectral data based on a deep learning model. The method includes: S100, obtaining initial neomycin spectral data A={A1, ..., A...} i , ..., A m}, A i This is the i-th initial neomycin spectrum, where i ranges from 1 to m, and m is the number of initial neomycin spectra; S200, each A i Inputting it into the target deep learning model yields A i The corresponding image feature vector B i ={B i1 , ..., B ij , ..., B in}, B ij It is A i The j-th image feature, where j ranges from 1 to n, and n is A. i The corresponding number of image features; S300, based on each B i Obtain the neomycin taxonomic cluster data U={U1, ..., U...} t , ..., U β}, U t Let be the taxonomic cluster of the t-th neomycin, where t ranges from 1 to β, and β is the number of neomycin taxonomic clusters; step S300 further includes the following step: S301, according to B i Besides B, in B i The temporal relation degree F between other image feature vectors besides [the above] i ={F i1 , ..., F iu , ..., F iη}, F iu It is B i Besides B, in B i S302, based on the initial inter-vector temporal relation degree matrix F and the preset temporal relation threshold ΔF, generate the target vector temporal relation degree matrix F. 0 S303, based on the temporal relationship matrix F between target vectors 0 Generate neomycin classification cluster data U={U1, ..., U...} t , ..., U βIt can be seen that by using the temporal relationship between the image features of any two initial neomycin images, and then classifying the initial neomycin images based on the temporal relationship, the temporal relationship between neomycin images can be analyzed, establishing data on the change of neomycin over time, thus increasing the amount of neomycin image data.

[0048] While specific embodiments of the invention have been described in detail by way of example, those skilled in the art should understand that the examples are for illustrative purposes only and not intended to limit the scope of the invention. It should also be understood that various modifications can be made to the embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.

Claims

1. A method for image data processing of neomycin spectral maps based on a deep learning model, characterized in that, The method includes the following steps: S100, Obtain initial neomycin spectral data A={A1, ..., A...} i , ..., A m }, A i It is the i-th initial neomycin spectrum, where i ranges from 1 to m, and m is the number of initial neomycin spectra; S200, each A i Inputting it into the target deep learning model yields A i The corresponding image feature vector B i ={B i1 , ..., B ij , ..., B in }, B ij It is A i The j-th image feature, where j ranges from 1 to n, and n is A. i The corresponding number of image features; S300, according to each B i Obtain the neomycin taxonomic cluster data U={U1, ..., U...} t , ..., U β }, U t Let be the taxonomic cluster of the t-th neomycin, where t ranges from 1 to β, and β is the number of taxonomic clusters of neomycin; step S300 also includes the following steps: S301, obtain B i Except for B i The temporal relation degree F between other image feature vectors besides [the above] i ={F i1 , ..., F iu , ..., F iη }, F iu It is B i Except for B i The temporal relation degree between the u-th image feature vectors other than the u-th image feature vectors, F iu The following conditions must be met: B 0 uj It is except B i The j-th image feature in the u-th image feature vector other than B, where u ranges from 1 to η, and η is the sum of the values ​​of u and η, respectively. i The total amount of other image feature vectors besides ΔB, where ΔB g is the temporal relation threshold corresponding to the sample images collected at the g-th time node for different neomycin components, where g ranges from 1 to z, z is the number of sample images in each neomycin component, and max() is the maximum value function; △B g The acquisition process includes the following: Obtain the temporal correlation degree between the image feature vector of the g-th sample image and the image feature vector of the g+1-th sample image in the r-th neomycin component, where r ranges from 1 to s, and s is the total number of neomycin components; The mean of the temporal correlation between the image feature vector of the g-th sample image and the image feature vector of the (g+1)-th sample image of all neomycin components is obtained; The temporal relationship between the image feature vector of the g-th sample image and the image feature vector of the (g+1)-th sample image in the r-th neomycin component satisfies the following condition: D j rg D is the j-th image feature in the g-th sample image of the r-th neomycin component. j rg+1 It is the j-th image feature in the g+1-th sample image of the r-th neomycin component; S302, Based on the initial inter-vector temporal relation degree matrix F and the preset temporal relation threshold ΔF, generate the target inter-vector temporal relation degree matrix F. 0 F meets the following conditions: When F iu When ≥ △F, at F 0 Retain F in China iu When F iu When < △F, at F 0 Setting F in the middle iu =0; S303, based on the temporal relationship matrix F between target vectors 0 Generate neomycin classification cluster data U={U1, ..., U...} t , ..., U β } 2. The image data processing method for neomycin spectral maps based on a deep learning model according to claim 1, characterized in that, The initial neomycin spectrum is the initial neomycin mass spectrum image.

3. The image data processing method for neomycin spectral maps based on a deep learning model according to claim 2, characterized in that, The neomycin mass spectrometry image was obtained using a mass spectrometer.

4. The image data processing method for neomycin spectral maps based on a deep learning model according to claim 2, characterized in that, A1 to A m Each initial neomycin mass spectrum image was inconsistent.

5. The image data processing method for neomycin spectral maps based on a deep learning model according to claim 1, characterized in that, η = n - 1.

6. The image data processing method for neomycin spectral maps based on a deep learning model according to claim 1, characterized in that, Step S303 includes: F 0 Each initial neomycin spectrum with a corresponding horizontal temporal relation degree of 0 is taken as a single neomycin taxonomic cluster, and F is used as the taxonomic cluster. 0 All initial neomycin spectra with a non-zero horizontal temporal relation degree are considered as a single neomycin taxonomic cluster.

Images