A device for calculating the degree of orthokeratology lenses

Abstract

The present invention provides a device for calculating the degree of orthokeratology lenses, the device comprising: an input module, a judgment module, a first calculation module, a second calculation module and an output module; the input module is used to input a user's corneal topography or corneal topography parameters; the judgment module is used to judge whether the input module input is a corneal topography, if so, the input corneal topography is transmitted to the first calculation module, otherwise, the input corneal topography parameters are transmitted to the second calculation module; the first calculation module calculates the user's corneal curvature value according to the corneal topography; the second calculation module calculates the user's corneal curvature value according to the corneal topography parameters; the output module outputs the calculated corneal curvature value of the user. Compared with the prior art, the present invention only needs to obtain the user's corneal topography or corneal topography parameters to obtain the corresponding orthokeratology lens parameters, thereby avoiding the error caused by manual trial wearing.

Description

Technical Field

[0001] The present invention relates to the technical field of orthokeratology lens fitting, and in particular to a device for calculating the degree of orthokeratology lenses. Background Art

[0002] Currently, the way to fit orthokeratology lenses in the ophthalmic medical industry is for optometrists to manually test-fit orthokeratology lenses on users multiple times based on corneal topography parameters, and then derive appropriate orthokeratology lens parameters that the user needs to wear. Summary of the invention

[0003] In order to solve the above problems, the present invention provides a device for calculating the degree of orthokeratology lenses.

[0004] The purpose of the present invention is achieved by the following technical solutions:

[0005] A device for calculating the degree of orthokeratology lenses, the device comprising: an input module, a judgment module, a first calculation module, a second calculation module and an output module;

[0006] The input module is used to input the user's corneal topography or corneal topography parameters;

[0007] When the input is the corneal topography of the user, the judging module transmits the corneal topography of the user to the first calculating module; when the input is the corneal topography parameters of the user, the judging module transmits the corneal topography parameters of the user to the second calculating module;

[0008] The first calculation module is used to calculate the corneal curvature value of the user according to the received corneal topography;

[0009] The second calculation module is used to calculate the corneal curvature value of the user according to the received corneal topography parameters;

[0010] The output module is used to output the calculated corneal curvature value of the user.

[0011] In an optional implementation, the first calculation module includes:

[0012] A first historical data acquisition unit, configured to acquire first historical data, wherein the first historical data includes: a corneal topography of a fitted user and corresponding corneal curvature value data;

[0013] A first model building unit is used to build a neural network model for calculating corneal curvature values ​​according to color pixel values ​​of corneal topography based on the first historical data;

[0014] The first corneal curvature value calculation unit is used to input the user's corneal topography into the model to obtain the user's corneal curvature value.

[0015] In an optional implementation, the neural network model is constructed using a deep learning algorithm, and the construction process includes:

[0016] S1: performing image segmentation processing on the corneal topography of the fitted user to obtain a characteristic region image of the user's corneal topography and a characteristic region image of the color reference identification plate, and taking the characteristic region image of the user's corneal topography and its corresponding corneal curvature value as sample data;

[0017] S2: Divide all sample data into training set and test set;

[0018] S3: retaining a color reference identification plate feature area image, and establishing a correspondence between color pixel values ​​and corneal curvature values ​​according to color pixel values ​​of the color reference identification plate feature area image;

[0019] According to the correspondence between the color pixel value and the corneal curvature value, each sample data in the training set is processed to obtain the corneal curvature prediction value of each sample data, and the error value between the corneal curvature prediction value of each sample data and the corneal curvature value corresponding to the sample data is calculated, and the model parameters are adjusted according to the obtained error value so that the model reaches a stable state.

[0020] In an optional implementation, the deep learning algorithm is: a CNN convolutional neural network algorithm;

[0021] The step S3 comprises:

[0022] Performing noise reduction processing on each pixel point in the characteristic region image of the color control identification plate to obtain a color pixel value of each pixel point in the characteristic region image of the color control identification plate after noise reduction;

[0023] Establishing a corresponding relationship between the color pixel value and the corneal curvature value according to the color pixel value of each pixel point in the characteristic area image of the color control identification plate after noise reduction;

[0024] Performing noise reduction processing on each sample data in the training set to obtain the color pixel value of the pixel point in each sample data after noise reduction;

[0025] According to the established correspondence between color pixel values ​​and corneal curvature values, the predicted corneal curvature value of each sample data in the training set is determined, and the error value between the predicted corneal curvature value of each sample data and the corneal curvature value corresponding to its sample data is calculated. According to the obtained error value, the model parameters are adjusted to make the model reach a stable state.

[0026] In an optional implementation, the first calculation module further includes: an evaluation unit;

[0027] The evaluation unit is used to input each sample data of the test set into the neural network model to obtain the corneal curvature prediction value of each sample data of the test set, and evaluate the accuracy of the neural network model based on the obtained corneal curvature prediction value of each sample data of the test set and the corneal curvature value of the corresponding sample data.

[0028] In an optional implementation, the second calculation module includes:

[0029] A second historical data acquisition unit is used to acquire second historical data, wherein the second historical data includes: corneal topography parameters of the fitted user and corresponding corneal curvature true values;

[0030] A second data processing unit, used for preprocessing the second historical data to remove invalid data;

[0031] A second model building unit, configured to build, based on the second historical data, a linear regression model for calculating a corneal curvature value based on the preprocessed historical data;

[0032] The second corneal curvature value calculation unit is used to obtain corneal topography parameters of the user to be measured, and input the corneal topography parameters into the model to obtain the corneal curvature value of the user to be measured.

[0033] In an optional implementation, the linear regression model is constructed using a supervised learning algorithm, specifically including:

[0034] T1: taking the pre-processed corneal topography parameters of the user and the true value of the corneal curvature of the corresponding orthokeratology lens as a sample data; wherein the sample data includes: a plurality of independent variables and a dependent variable;

[0035] T2: Based on all the sample data obtained, a supervised learning algorithm is used to build a multivariate linear regression model;

[0036] T3: Use a stepwise regression algorithm to eliminate insignificant independent variables until the insignificant independent variables are eliminated from the multivariate linear regression model; thereby obtaining a linear regression model:

[0037] Ke=b0+b1×Kf+b2×min K+b3×Em

[0038] Where Ke is the corneal curvature value, b0 is a constant term, b1, b2, and b3 are the regression coefficients of Kf, minK, and Em, respectively, Kf is the flat meridian refractive power, that is, the horizontal curvature, minK is the minimum meridian refractive power, that is, the minimum curvature, and Em is the corneal eccentricity in the meridian direction of the minimum corneal curvature.

[0039] In an optional embodiment, in step T3, the stepwise regression algorithm is used to eliminate insignificant independent variables until the insignificant independent variables are eliminated from the multivariate linear regression model, including:

[0040] T31: Perform an F test on the introduced independent variable to determine whether the independent variable is significant;

[0041] T32: If the result is significant, the independent variable is introduced into the established multiple linear regression model, and the introduced independent variables are tested one by one. If the originally introduced independent variable is no longer significant due to the newly introduced independent variable, the originally introduced independent variable is eliminated so that the model only contains significant independent variables.

[0042] T33: According to steps T31-T32, the independent variables are introduced one by one until all insignificant independent variables are eliminated from the multiple linear regression model.

[0043] In an optional implementation, the second calculation submodule further includes: a model optimization unit:

[0044] The model optimization unit is used to optimize the linear regression model. The specific optimization process is:

[0045] Re-obtain corneal topography parameters of a new fitted user and the corresponding true value of corneal curvature;

[0046] Substituting the re-acquired corneal topography parameters into the linear regression model to calculate the corresponding corneal curvature value;

[0047] The error value between each corneal curvature value and the corresponding true value of the corneal curvature is calculated, and the model parameters are adjusted according to the obtained error value to obtain the optimization model of the linear regression model:

[0048] Ke=b0′+b1′×Kf+b2′×min K+b3′×Em

[0049] Wherein, Ke is the corneal curvature value, b0' is a constant term, b1', b2', b3' are respectively the optimized regression coefficients of Kf, minK, Em, wherein, b0'=6.3982, b1'=0.3395, b2'=0.5234, b3'=-0.9399.

[0050] The beneficial effects of the present invention are as follows: the present invention obtains the user's corneal topography or corneal topography parameters through an input module, determines through a judgment module whether the input module inputs the user's corneal topography or corneal topography parameters, and then calculates the user's corneal curvature value through a first calculation module or a second calculation module, and the user's corneal curvature value is the orthokeratology lens parameter required by the user. Compared with the prior art, the present invention only needs to obtain the user's corneal topography or corneal topography parameters to obtain the corresponding orthokeratology lens parameters, thereby avoiding the errors caused by manual trial wearing and simplifying the steps of fitting orthokeratology lenses, that is, eliminating the steps of multiple trial wearing of orthokeratology lenses, and also facilitating the work of optometrists and ensuring the eye safety of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0052] Figure 1 It is a framework structure diagram of a device for calculating the degree of orthokeratology lenses provided by an embodiment of the present invention;

[0053] Figure 2 is a framework structure diagram of the first calculation module 30 provided in an embodiment of the present invention;

[0054] Figure 3 is a framework structure diagram of the second calculation module 40 provided in an embodiment of the present invention;

[0055] Figure 4 It is a workflow diagram of the calculation device for calculating the parameters of orthokeratology lenses provided in an embodiment of the present invention.

[0056] Figure numerals: input module 10, judgment module 20, first calculation module 30, second calculation module 40, output module 50, first historical data acquisition unit 31, first model building unit 32, first corneal curvature value calculation unit 33, evaluation unit 34, second historical data acquisition unit 41, second data processing unit 42, second model building unit 43, second corneal curvature value calculation unit 44, model optimization unit 45. DETAILED DESCRIPTION

[0057] The present invention is further described in conjunction with the following examples.

[0058] Figure 1 A device for calculating the degree of orthokeratology lenses is shown, and the device comprises: an input module 10, a judgment module 20, a first calculation module 30, a second calculation module 40 and an output module 50.

[0059] The input module 10 is used to input the user's corneal topography or corneal topography parameters.

[0060] The judgment module 20 is used to judge whether the input of the input module 10 is a corneal topography map. If so, the input corneal topography map is transmitted to the first calculation module 30 . Otherwise, the input corneal topography map parameters are transmitted to the second calculation module 40 .

[0061] The first calculation module 30 is used to calculate the corneal curvature value of the user according to the received corneal topography.

[0062] The second calculation module 40 is used to calculate the corneal curvature value of the user according to the received corneal topography parameters.

[0063] The output module 50 is used to output the calculated corneal curvature value of the user.

[0064] Figure 4 A workflow diagram of the computing device for calculating orthokeratology lens parameters is shown.

[0065] The specific workflow is as follows:

[0066] At the beginning, the user's corneal topography or corneal topography parameters are input through the input module 10; the judgment module 20 judges whether the input module 10 inputs the user's corneal topography or corneal topography parameters. If the judgment module 20 determines that the input is the user's corneal topography, the user's corneal topography is transmitted to the first calculation module 30, and the first calculation module 30 calculates the user's corneal curvature value and outputs it through the output module 50; if the judgment module 20 determines that the input is the user's corneal topography parameters, the user's corneal topography parameters are transmitted to the second calculation module 40, and the second calculation module 40 calculates the user's corneal curvature value and outputs it through the output module 50.

[0067] In an alternative embodiment, see Figure 2 The first calculation module 30 includes: a first historical data acquisition unit 31, a first model construction unit 32, a first corneal curvature value calculation unit 33 and an evaluation unit 34.

[0068] The first historical data acquisition unit 31 is used to acquire first historical data, which includes: a corneal topography of a fitted user and corresponding corneal curvature value data; optionally, the corneal topography of the fitted user can be acquired by a corneal topographer.

[0069] The first model building unit 32 is used to build a neural network model for calculating the corneal curvature value according to the color pixel value of the corneal topography according to the first historical data.

[0070] In an optional implementation, the neural network model is constructed using a deep learning algorithm, and the construction process includes:

[0071] Performing image segmentation processing on the corneal topography of the fitted user to obtain a characteristic region image of the user's corneal topography and a characteristic region image of the color reference identification plate, and taking the characteristic region image of the user's corneal topography and its corresponding corneal curvature value as sample data;

[0072] Divide all sample data into training set and test set;

[0073] A color control identification plate characteristic region image is retained, and a corresponding relationship between color pixel values ​​and corneal curvature values ​​is established according to color pixel values ​​of the color control identification plate characteristic region image;

[0074] According to the correspondence between the color pixel value and the corneal curvature value, each sample data in the training set is processed to obtain the corneal curvature prediction value of each sample data, and the error value between the corneal curvature prediction value of each sample data and the corneal curvature value corresponding to the sample data is calculated, and the model parameters are adjusted according to the obtained error value so that the model reaches a stable state.

[0075] Preferably, the deep learning algorithm is: CNN convolutional neural network algorithm;

[0076] The process of constructing the neural network model using the CNN convolutional neural network algorithm includes:

[0077] Performing image segmentation processing on the corneal topography of the fitted user to obtain a characteristic region image of the user's corneal topography and a characteristic region image of the color reference identification plate, and taking the characteristic region image of the user's corneal topography and its corresponding corneal curvature value as sample data;

[0078] Divide all sample data into training set and test set;

[0079] A color control identification plate feature area image is retained. In the first step, the convolution layer is used to identify the color of each pixel in the color control identification plate feature area image to obtain the color pixel value of each pixel; in the second step, the pooling layer is used to perform denoising on the color pixel value of each pixel obtained by the convolution layer, so as to retain only the useful information in the color control identification plate feature area image and reduce unnecessary noise transmission; in the third step, the color pixel value of each pixel is summarized by the fully connected layer; in the fourth step, the classification and comparison conversion is performed. The corneal curvature value corresponding to the color in the color control identification plate feature area image is a continuous value of 0 to 101.5, which corresponds to the continuous value of the color pixel value of 0 to 255. The corneal curvature value is evenly divided into 255 parts, and then the specific corneal curvature value is corresponded to the color pixel value of 0 to 255, so that the color pixel value and the corneal curvature value in the color control identification plate feature area image form a one-to-one correspondence, and the established correspondence between the color pixel value and the corneal curvature value is further stored. After that, each sample data in the training set is processed by the convolution layer, the pooling layer and the fully connected layer in turn to obtain the color pixel value of the pixel point in each sample data in the training set. According to the established correspondence between the color pixel value and the corneal curvature value, the corneal curvature prediction value of each sample data in the training set is determined, and the error value between the corneal curvature prediction value of each sample data and the corneal curvature value corresponding to its sample data is calculated. According to the obtained error value, the model parameters are adjusted to make the model reach a stable state.

[0080] The first corneal curvature value calculation unit 33 is used to input the user's corneal topography into the above model to obtain the user's corneal curvature value.

[0081] The evaluation module 34 is used to input each sample data of the test set into the neural network model to obtain the corneal curvature prediction value of each sample data of the test set, and evaluate the accuracy of the neural network model based on the obtained corneal curvature prediction value of each sample data in the test set and the corneal curvature value of the corresponding sample data.

[0082] In order to match orthokeratology lenses suitable for users, it is necessary to obtain the user's corneal curvature value, and then match the appropriate orthokeratology lenses according to the corneal curvature value. The first calculation module 30 provided in the above embodiment of the present invention can obtain the corneal curvature value of the user to be tested based on the constructed neural network model, and then match the orthokeratology lenses suitable for the user to be tested according to the obtained corneal curvature value. The first calculation module 30 saves the steps of multiple trial wearing of orthokeratology lenses, and also facilitates the work of optometrists and ensures the eye safety of patients.

[0083] In an alternative embodiment, see Figure 3The second calculation module 40 includes: a second historical data acquisition unit 41, a second data processing unit 42, a second model construction unit 43, a second corneal curvature value calculation unit 44 and a model optimization unit 45.

[0084] A second historical data acquisition unit 41 is used to acquire second historical data, wherein the second historical data includes: corneal topography parameters of the fitted user and corresponding corneal curvature true values;

[0085] A second data processing unit 42 is used to pre-process the second historical data to remove invalid data;

[0086] A second model building unit 43 is used to build a linear regression model for calculating the corneal curvature value according to the preprocessed historical data based on the second historical data;

[0087] The second corneal curvature value calculation unit 44 is used to obtain corneal topography parameters of the user to be measured, and input the corneal topography parameters into the model to obtain the corneal curvature value of the user to be measured;

[0088] The model optimization unit 45 is used to optimize the linear regression model.

[0089] In an optional implementation, the linear regression model is constructed using a supervised learning algorithm, specifically including:

[0090] T1: taking the pre-processed corneal topography parameters of the user and the true value of the corneal curvature of the corresponding orthokeratology lens as a sample data; wherein the sample data includes: a plurality of independent variables and a dependent variable;

[0091] T2: Based on all the sample data obtained, a supervised learning algorithm is used to build a multivariate linear regression model;

[0092] T3: Use a stepwise regression algorithm to eliminate insignificant independent variables until the insignificant independent variables are eliminated from the multivariate linear regression model; thereby obtaining a linear regression model:

[0093] Ke=b0+b1×Kf+b2×min K+b3×Em

[0094] Where Ke is the corneal curvature value, b0 is a constant term, b1, b2, and b3 are the regression coefficients of Kf, minK, and Em, respectively, Kf is the flat meridian refractive power, that is, the horizontal curvature, minK is the minimum meridian refractive power, that is, the minimum curvature, and Em is the corneal eccentricity in the meridian direction of the minimum corneal curvature.

[0095] In an optional embodiment, in step T3, the stepwise regression algorithm is used to eliminate insignificant independent variables until the insignificant independent variables are eliminated from the multivariate linear regression model, including:

[0096] T31: Perform an F test on the introduced independent variable to determine whether the independent variable is significant;

[0097] T32: If the result is significant, the independent variable is introduced into the established multiple linear regression model, and the introduced independent variables are tested one by one. If the originally introduced independent variable is no longer significant due to the newly introduced independent variable, the originally introduced independent variable is eliminated so that the model only contains significant independent variables.

[0098] T33: According to steps T31-T32, the independent variables are introduced one by one until all insignificant independent variables are eliminated from the multiple linear regression model.

[0099] In an optional implementation, the specific optimization process of the model optimization unit 45 is:

[0100] Re-obtain corneal topography parameters of a new fitted user and the corresponding true value of corneal curvature;

[0101] Substituting the re-acquired corneal topography parameters into the linear regression model to calculate the corresponding corneal curvature value;

[0102] The error value between each corneal curvature value and the corresponding true value of the corneal curvature is calculated, and the model parameters are adjusted according to the obtained error value to obtain the optimization model of the linear regression model:

[0103] Ke=b0′+b1′×Kf+b2′×minK+b3′×Em

[0104] Wherein, Ke is the corneal curvature value, b0' is a constant term, b1', b2', b3' are respectively the optimized regression coefficients of Kf, minK, Em, wherein, b0'=6.3982, b1'=0.3395, b2'=0.5234, b3'=-0.9399.

[0105] A model optimization unit 45 is set up to obtain the corneal topography parameters of the new user, and substitute the parameters into the established linear regression model to obtain the corneal curvature value of the user. The obtained corneal curvature value is compared with the actual corneal curvature value obtained by the optometrist for the user, and the model is further tuned according to the comparison result to achieve higher accuracy, thereby providing the user with accurate corneal curvature value, which is convenient for the subsequent configuration of suitable orthokeratology lenses for the user.

[0106] The second calculation module 40 of the above embodiment of the present invention can obtain a linear regression model based on the corneal topography parameters of the fitted user and the corresponding corneal curvature true value, and then use the model to obtain the orthokeratology lens parameters that the user to be tested needs to wear (i.e., the user's corneal curvature value). The above embodiment of the present invention is based on the existing corneal topography parameters, and the above linear regression model can be used to calculate the user's corneal curvature value, and then configure a suitable orthokeratology lens for the user, thereby eliminating the steps of multiple trial wearing of orthokeratology lenses, facilitating the work of optometrists and ensuring the eye safety of patients.

[0107] The present invention obtains the user's corneal topography or corneal topography parameters through an input module, determines through a judgment module whether the input module inputs the user's corneal topography or corneal topography parameters, and then calculates the user's corneal curvature value through a first calculation module 30 or a second calculation module 40. The user's corneal curvature value is the orthokeratology lens parameter required by the user. Compared with the prior art, the present invention only needs to obtain the user's corneal topography or corneal topography parameters to obtain the corresponding orthokeratology lens parameters, thereby avoiding the error caused by manual trial wearing and simplifying the steps of fitting orthokeratology lenses, that is, eliminating the steps of multiple trial wearing of orthokeratology lenses, and also facilitating the work of optometrists and ensuring the eye safety of patients.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. 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 solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

Claims

1. A device for calculating the degree of orthokeratology lenses, characterized in that: include: An input module, a judgment module, a first calculation module, a second calculation module and an output module; The first calculation module includes: A first historical data acquisition unit, configured to acquire first historical data, wherein the first historical data includes: a corneal topography of a fitted user and corresponding corneal curvature value data; A first model building unit is used to build a neural network model for calculating corneal curvature values ​​according to color pixel values ​​of corneal topography using a deep learning algorithm according to the first historical data; A first corneal curvature value calculation unit, used for inputting a user's corneal topography into the model to obtain a corneal curvature value of the user; The neural network model building process includes: S1: performing image segmentation processing on the corneal topography of the fitted user to obtain a characteristic region image of the user's corneal topography and a characteristic region image of the color reference identification plate, and taking the characteristic region image of the user's corneal topography and its corresponding corneal curvature value as sample data; S2: Divide all sample data into training set and test set; S3: retaining a color reference identification plate feature area image, and establishing a correspondence between color pixel values ​​and corneal curvature values ​​according to color pixel values ​​of the color reference identification plate feature area image; Processing each sample data in the training set according to the correspondence between the color pixel value and the corneal curvature value to obtain a corneal curvature prediction value of each sample data, calculating an error value between the corneal curvature prediction value of each sample data and the corneal curvature value corresponding to the sample data, and adjusting the model parameters according to the obtained error value so that the model reaches a stable state; The second calculation module includes: A second historical data acquisition unit is used to acquire second historical data, wherein the second historical data includes: corneal topography parameters of the fitted user and corresponding corneal curvature true values; A second data processing unit, used for preprocessing the second historical data to remove invalid data; A second model building unit, configured to build, based on the second historical data, a linear regression model for calculating a corneal curvature value based on the preprocessed second historical data; A second corneal curvature value calculation unit is used to obtain corneal topography parameters of the user to be tested, and input the corneal topography parameters into the model to obtain the corneal curvature value of the user to be tested; The input module is used to input the user's corneal topography or corneal topography parameters; When the input is the corneal topography of the user, the judging module transmits the corneal topography of the user to the first calculating module; when the input is the corneal topography parameters of the user, the judging module transmits the corneal topography parameters of the user to the second calculating module; Used to determine whether the input of the input module is a corneal topography map, if so, transmit the input corneal topography map to the first calculation module, otherwise, transmit the input corneal topography map parameters to the second calculation module; The first calculation module is used to calculate the corneal curvature value of the user according to the received corneal topography; The second calculation module is used to calculate the corneal curvature value of the user according to the received corneal topography parameters; The output module is used to output the calculated corneal curvature value of the user.

2. The device for calculating the orthokeratology lens power according to claim 1, characterized in that: The deep learning algorithm is: CNN convolutional neural network algorithm; The step S3 comprises: Performing noise reduction processing on each pixel point in the characteristic region image of the color control identification plate to obtain a color pixel value of each pixel point in the characteristic region image of the color control identification plate after noise reduction; Establishing a corresponding relationship between the color pixel value and the corneal curvature value according to the color pixel value of each pixel point in the characteristic area image of the color control identification plate after noise reduction; Performing noise reduction processing on each sample data in the training set to obtain the color pixel value of the pixel point in each sample data after noise reduction; According to the established correspondence between color pixel values ​​and corneal curvature values, the predicted corneal curvature value of each sample data in the training set is determined, and the error value between the predicted corneal curvature value of each sample data and the corneal curvature value corresponding to its sample data is calculated. According to the obtained error value, the model parameters are adjusted to make the model reach a stable state.

3. The device for calculating orthokeratology lens power according to claim 1, characterized in that: The first calculation module further includes: an evaluation unit; The evaluation unit is used to input each sample data of the test set into the neural network model to obtain the corneal curvature prediction value of each sample data of the test set, and evaluate the accuracy of the neural network model based on the obtained corneal curvature prediction value of each sample data of the test set and the corneal curvature value of the corresponding sample data.

4. The device for calculating orthokeratology lens power according to claim 1, characterized in that: The linear regression model is constructed using a supervised learning algorithm, specifically including: T1: taking the pre-processed corneal topography parameters of the user and the true value of the corneal curvature of the corresponding orthokeratology lens as a sample data; wherein the sample data includes: a plurality of independent variables and a dependent variable; T2: Based on all the sample data obtained, a supervised learning algorithm is used to build a multivariate linear regression model; T3: Use a stepwise regression algorithm to eliminate insignificant independent variables until the insignificant independent variables are eliminated from the multivariate linear regression model; thereby obtaining a linear regression model: Ke=b0+b1×Kf+b2×min K+b3×Em Where Ke is the corneal curvature value, b0 is a constant term, b1, b2, and b3 are the regression coefficients of Kf, minK, and Em, respectively, Kf is the flat meridian refractive power, that is, the horizontal curvature, minK is the minimum meridian refractive power, that is, the minimum curvature, and Em is the corneal eccentricity in the meridian direction of the minimum corneal curvature.

5. The device for calculating the orthokeratology lens power according to claim 4, characterized in that: In step T3, the stepwise regression algorithm is used to eliminate insignificant independent variables until the insignificant independent variables are eliminated from the multivariate linear regression model, including: T31: Perform an F test on the introduced independent variable to determine whether the independent variable is significant; T32: If the result is significant, the independent variable is introduced into the established multiple linear regression model, and the introduced independent variables are tested one by one. If the originally introduced independent variable is no longer significant due to the newly introduced independent variable, the originally introduced independent variable is eliminated so that the model only contains significant independent variables. T33: According to steps T31-T32, the independent variables are introduced one by one until all insignificant independent variables are eliminated from the multivariate linear regression model.

6. The device for calculating orthokeratology lens power according to claim 1, characterized in that: The second calculation submodule further includes: a model optimization unit: The model optimization unit is used to optimize the linear regression model. The specific optimization process is: Re-obtain corneal topography parameters of a new fitted user and the corresponding true value of corneal curvature; Substituting the re-acquired corneal topography parameters into the linear regression model to calculate the corresponding corneal curvature value; The error value between each corneal curvature value and the corresponding true value of the corneal curvature is calculated, and the model parameters are adjusted according to the obtained error value to obtain the optimization model of the linear regression model: Ke=b0'+b1'×Kf+b2'×min K+b3'×Em Wherein, Ke is the corneal curvature value, b0' is a constant term, b1', b2', and b3' are the optimized regression coefficients of Kf, minK, and Em, respectively, wherein b0'=6.3982, b1'=0.3395, b2'=0.5234, and b3'=-0.9399.

Images