Highly myopia eye bottom grading method and system, electronic equipment and storage medium

Abstract

The invention relates to a high myopia eye bottom grading method. The method comprises the following steps: acquiring an attribute feature, a visual function feature and an eye fundus image of a target object; determining retina structure features of the target object based on the fundus image; performing principal component analysis in the attribute feature, the visual function feature and the retina structure feature to obtain a preset number of first principal component features; and inputting the first principal component feature into a fundus grading model to obtain a fundus grading result of high myopia corresponding to the target object, the fundus grading model being a preset clustering model screened from a plurality of preset clustering models and meeting a preset condition. The invention further provides a high myopia bottom grading system, electronic equipment and a storage medium. According to the method, fundus grading evaluation is carried out on high myopia through the fundus grading model, and compared with an artificial fundus grading evaluation mode, the accuracy and comprehensiveness of fundus grading evaluation can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, system, electronic device and storage medium for grading high myopia fundus. Background Art

[0002] High myopia may cause a variety of serious complications, even blindness, and the visual impairment it brings is permanent and irreversible. Currently, high myopia is considered to be the primary cause of visual impairment among people of prime working age. Therefore, regular fundus monitoring and risk assessment of high myopia through examination means, as well as early detection, early intervention, and early treatment, are of great significance for myopia prevention and control and social and economic development.

[0003] Currently, the fundus grading for high myopia mainly relies on single examination methods such as fundus color photography or optical coherence tomography, and lacks comprehensive evaluation based on multiple indicators. Although experienced experts can conduct limited fundus assessment for high myopia by combining multiple examinations, this manual grading is time-consuming and labor-intensive, easily affected by subjective factors, and has low accuracy.

[0004] Chinese patent document CN111199794A discloses a surgical intelligent decision-making system suitable for high myopia cataract and its establishment method. The establishment method includes: constructing a deep learning model, which is composed of two networks, evaluation and decision-making. The evaluation network includes a LOCSIII grading deep learning model, an Emery nuclear hardness grading deep learning model, and a fundus determination model. The decision network combines the cataract degree and fundus grading with the patient's basic information, vision, intraocular pressure, A-ultrasound / IOLmaster report, corneal endothelial count, corneal astigmatism, B-ultrasound report and other information, and establishes a linear regression model after mapping to a high-dimensional space through a kernel function; constructing a training set; the model is trained and learned using the training set, and the parameters are adjusted. The system includes a storage device, an evaluation module, a decision module, and a test evaluation adjustment module. The present invention can quickly and accurately identify the cataract degree and fundus condition of patients with high myopia cataract, provide doctors with comprehensive surgical decisions, and improve the homogeneity of prevention and treatment.

[0005] However, there is no report on a method, system, electronic device and storage medium for grading high myopia fundus. Summary of the invention

[0006] The purpose of the present invention is to provide a method, system, electronic device and storage medium for grading high myopia fundus in view of the deficiencies in the prior art.

[0007] In a first aspect, the present invention provides a method for grading fundus of high myopia, comprising:

[0008] Acquire the attribute characteristics, visual function characteristics and fundus images of the target object;

[0009] determining the retinal structural features of the target object based on the fundus image;

[0010] Performing principal component analysis on the attribute features, the visual function features, and the retinal structure features to obtain a preset number of first principal component features;

[0011] The first principal component feature is input into a fundus grading model to obtain a fundus grading result of the target object corresponding to high myopia, wherein the fundus grading model is a preset clustering model that meets preset conditions and is screened out from multiple preset clustering models.

[0012] As a preferred example, the principal component analysis is performed on the attribute features, the visual function features and the retinal structure features to obtain a preset number of first principal component features, including:

[0013] Decentralizing the attribute features, the visual function features, and the retinal structure features to obtain multidimensional features with the same scale;

[0014] Generate a principal component feature matrix corresponding to the multidimensional features;

[0015] The multidimensional features are processed by dimensionality reduction using the principal component feature matrix to obtain a preset number of first principal component features.

[0016] As another preferred example, before inputting the first principal component feature into a fundus grading model to obtain a fundus grading result of the target object corresponding to high myopia, the method further includes:

[0017] Screening a fundus grading model that meets preset conditions from multiple preset clustering models;

[0018] The screening of a fundus grading model that meets preset conditions from a plurality of preset clustering models includes:

[0019] Acquiring multidimensional features of the sample, wherein the multidimensional features of the sample include sample attribute features, sample visual function features, and sample retinal structure features;

[0020] Performing principal component analysis on the sample attribute features, the sample visual function features, and the sample retinal structure features, and screening out a preset number of first principal component features;

[0021] Inputting the first principal component feature into a plurality of preset clustering models respectively to obtain a clustering result corresponding to each of the preset clustering models;

[0022] Based on the clustering result, a preset clustering model that meets preset conditions is selected from the plurality of preset clustering models as the fundus grading model.

[0023] As another preferred example, the clustering result includes multiple classification clusters, each of which contains at least one feature point;

[0024] The step of selecting a preset clustering model that meets preset conditions from the plurality of preset clustering models based on the clustering result as the fundus grading model comprises:

[0025] According to the clustering result, calculating a clustering evaluation index of each preset clustering model in the plurality of preset clustering models, the clustering evaluation index comprising at least one of a silhouette coefficient, a variance ratio criterion and a Davidson-Botting index;

[0026] Based on the clustering evaluation index, a preset clustering model that meets preset conditions is screened from the multiple preset clustering models as the fundus grading model.

[0027] As another preferred example, before inputting the first principal component feature into the fundus grading model to obtain the fundus grading result of the target object corresponding to high myopia, the method further includes:

[0028] Acquire a plurality of classification clusters contained in the clustering result output by the fundus grading model for the first principal component feature;

[0029] Corresponding preset fundus assessment levels are respectively configured for the multiple classification clusters.

[0030] As another preferred example, the inputting the first principal component feature into a fundus grading model to obtain a fundus grading result corresponding to high myopia of the target object includes:

[0031] Inputting the first principal component feature into a fundus grading model, clustering the first principal component feature in the fundus grading model, and determining a target classification cluster to which the first principal component feature belongs from among the multiple classification clusters;

[0032] The preset fundus assessment level corresponding to the target classification cluster is determined as the fundus grading result corresponding to high myopia of the target object.

[0033] In a second aspect, the present invention provides a high myopia fundus grading system, comprising:

[0034] An acquisition module is used to acquire the attribute characteristics, visual function characteristics and fundus image of the target object; a determination module is used to determine the retinal structure characteristics of the target object based on the fundus image; an analysis module is used to perform principal component analysis on the attribute characteristics, the visual function characteristics and the retinal structure characteristics to obtain a preset number of first principal component characteristics;

[0035] An evaluation module is used to input the first principal component feature into a fundus grading model to obtain a fundus grading result corresponding to high myopia of the target object, wherein the fundus grading model is a preset clustering model that meets preset conditions and is screened out from multiple preset clustering models.

[0036] In a third aspect, the present invention provides an electronic device, comprising:

[0037] A processor and a memory, wherein the memory is used to store a computer program, and the processor is used to call and run the computer program stored in the memory to execute the method described in the method for grading the fundus of high myopia.

[0038] In a fourth aspect, the present invention provides a computer-readable storage medium for storing a computer program, wherein the computer program enables a computer to execute the method described in the method for grading the fundus of high myopia.

[0039] The advantages of the present invention are as follows: the present invention first obtains the attribute characteristics and visual function characteristics of the target object, and determines the retinal structure characteristics according to the fundus image of the target object, and then performs principal component analysis on the attribute characteristics, visual function characteristics and retinal structure characteristics to obtain a preset number of first principal component characteristics. By analyzing the principal component characteristics, the feature dimension can be reduced, the main features therein can be retained, and the efficiency of fundus grading for high myopia can be improved. The first principal component characteristics can then be input into the fundus grading model to obtain the fundus grading result of the target object corresponding to high myopia. In view of the fact that the fundus grading model is a preset clustering model that meets the preset conditions selected from multiple preset clustering models, the fundus grading model is used to perform fundus grading assessment on high myopia, which can improve the accuracy and comprehensiveness of fundus grading assessment compared to the manual fundus grading assessment method. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 An application scenario diagram provided for an embodiment of the present application;

[0041] Figure 2 A schematic diagram of a process for grading a high myopia fundus provided in an embodiment of the present application;

[0042] Figure 3 A schematic flow chart of a method for grading fundus of high myopia provided in another embodiment of the present application;

[0043] Figure 4 A schematic diagram of the structure of a high myopia fundus grading system provided in an embodiment of the present application;

[0044] Figure 5 A schematic diagram of the structure of a high myopia fundus grading system provided by another embodiment of the present application;

[0045] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content of the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0047] Example 1

[0048] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0049] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0050] High myopia may cause a variety of serious complications, even blindness, and the resulting visual impairment is permanent and irreversible. Currently, high myopia is considered the primary cause of visual impairment in people of prime working age. Therefore, regular fundus monitoring and risk assessment of high myopia through examination methods, early detection, early intervention, and early treatment are of great significance for myopia prevention and control and socioeconomic development. At present, the fundus grading of high myopia mainly relies on a single examination method such as fundus color photography or optical coherence tomography, and lacks a comprehensive evaluation based on multiple indicators. Although experienced experts can combine multiple examinations to perform limited high myopia grading and evaluation, manual grading is time-consuming and labor-intensive, easily interfered by subjective factors, and has low accuracy.

[0051] In order to solve the above technical problems, the technical solution of the present invention is to combine the attribute characteristics, visual function characteristics and retinal structure characteristics of the target object so as to comprehensively evaluate the risk of high myopia from multiple feature dimensions. Furthermore, the principal component characteristics can be analyzed in the above multi-dimensional characteristics. By analyzing the principal component characteristics, the feature dimensions can be reduced, and the main features therein can be retained, thereby improving the efficiency of fundus grading for high myopia. The principal component characteristics can then be input into the fundus grading model to obtain the fundus grading results of the target object corresponding to high myopia. By using the fundus grading model to perform fundus grading assessment for high myopia, the accuracy of fundus grading assessment can be improved compared to manual fundus grading assessment. It should be understood that the technical solution of the present invention can be applied to the following scenarios, but is not limited to:

[0052] In some possible implementations, Figure 1 An application scenario diagram provided by an embodiment of the present invention, such as Figure 1 As shown, the application scenario may include an electronic device 110 and a network device 120. The electronic device 110 may establish a connection with the network device 120 via a wired network or a wireless network.

[0053] Exemplarily, the electronic device 110 may be a desktop computer, a laptop computer, a tablet computer, etc., but is not limited thereto. The network device 120 may be a terminal device or a server, but is not limited thereto. In one embodiment of the present invention, the electronic device 110 may send a request message to the network device 120, and the request message may be used to request to obtain the attribute characteristics, visual function characteristics, and fundus image of the target object. Further, the electronic device 110 may receive a response message sent by the network device 120, and the response message includes the attribute characteristics, visual function characteristics, and fundus image of the target object.

[0054] also, Figure 1 An electronic device 110 and a network device 120 are provided as an example, but other numbers of electronic devices and network devices may be included in practice, and the present invention is not limited thereto.

[0055] In other possible implementations, the technical solution of the present invention may also be executed by the electronic device 110, or the technical solution of the present invention may also be executed by the network device 120, and the present invention does not limit this. After introducing the application scenarios of the embodiments of the present invention, the technical solution of the present invention will be described in detail below:

[0056] Figure 2 A flowchart of a method for grading high myopia fundus provided by an embodiment of the present invention, the method can be as follows Figure 1 The electronic device 110 shown in the figure performs, but is not limited to this. Figure 2 As shown, the method may include the following steps:

[0057] Step 210, obtaining the attribute characteristics, visual function characteristics and fundus image of the target object. The target object may be an object to be evaluated for high myopia risk; the attribute characteristics may include the age, gender and axial length of the target object; the visual function characteristics may include contrast sensitivity and the best corrected visual acuity of the target object; the fundus image is a standard objective diagnostic image obtained by ophthalmology, which mainly reflects the characteristics of the posterior segment of the eye such as the retina, choroid and optic nerve. The fundus image can be obtained by at least one of the following methods: fundus photography, magnetic resonance imaging, optical coherence tomography (OCT), etc.

[0058] Step 220: Determine the retinal structural features of the target object based on the fundus image.

[0059] The retinal structural features may include the level of myopic macular degeneration (MMD) corresponding to the target object, the area of ​​beta-parapapillary atrophy (β-PPA) and retinal thickness.

[0060] In a specific application scenario, the retinal structural features of the target object can be obtained by performing feature recognition on the fundus image based on the fundus image. For the disclosed embodiment, for the fundus image of the target object, the MMD level of the target object can be determined based on the international META-PM classification system; the β-PPA area can be measured using ImageJ software; and the retinal thickness can be obtained through OCT.

[0061] Step 230: Perform principal component analysis on the attribute features, visual function features, and retinal structure features to obtain a preset number of first principal component features.

[0062] For the disclosed embodiment, the principal component analysis (PCA) algorithm can be used to perform principal component analysis on the attribute features, visual function features and retinal structure features to obtain a preset number of principal component features. Among them, the preset number of principal component features corresponding to the screening (i.e., the feature dimension k) can be set according to the actual application scenario. For example, if the total number of features included in the clinical application is 12, k can be 1-12. If the number of linearly related features is 6, k can be set to 6. If it can be set to 6, no specific limitation is made here.

[0063] In specific application scenarios, attribute features, visual function features, and retinal structure features can simultaneously contain relevant features and irrelevant features corresponding to fundus grading assessment. By performing principal component feature analysis on attribute features, visual function features, and retinal structure features, interference of irrelevant features in fundus grading assessment can be avoided. Principal component feature analysis can achieve data dimensionality reduction, reduce the amount of data processing in subsequent risk classification models, and thus improve the efficiency of fundus grading for high myopia.

[0064] Step 240: input the first principal component feature into a fundus grading model to obtain a fundus grading result of the target object corresponding to high myopia, wherein the fundus grading model is a preset clustering model that is selected from multiple preset clustering models and meets preset conditions.

[0065] In specific application scenarios, a large amount of sample data can be used to evaluate the clustering effects of multiple preset clustering models, and according to the quality of the clustering pattern and the variance explanation ratio, the preset clustering model with the best model performance (i.e., meeting the preset conditions) is selected from multiple preset clustering models as the fundus grading model. The preset clustering models may include K-means model, Gaussian mixture model, etc.

[0066] The above-mentioned multiple preset clustering models can be unsupervisedly trained using the multidimensional features of the samples, and according to the clustering results of each preset clustering model for the multidimensional features of the samples, the preset clustering model with the best clustering effect is screened out from the multiple preset clustering models as the fundus grading model.

[0067] In summary, according to the high myopia fundus grading method provided by the present invention, the attribute characteristics and visual function characteristics of the target object can be first obtained, and the retinal structure characteristics can be determined according to the fundus image of the target object, and then the principal component analysis is performed in the attribute characteristics, visual function characteristics and retinal structure characteristics to obtain a preset number of first principal component characteristics. By analyzing the principal component characteristics, the feature dimension can be reduced, the main features can be retained, and the efficiency of fundus grading for high myopia can be improved. Afterwards, the first principal component characteristics can be input into the fundus grading model to obtain the fundus grading result of the target object corresponding to high myopia. In view of the fact that the fundus grading model is a preset clustering model that meets the preset conditions selected from multiple preset clustering models, the fundus grading model is used to perform fundus grading assessment on high myopia, which can improve the accuracy of fundus grading assessment compared to the manual fundus grading assessment method.

[0068] based on Figure 2 The embodiment shown is a refinement and extension of the above embodiment. In order to fully illustrate the specific implementation process of the method of this embodiment, this embodiment provides the following Figure 3 The specific method shown. Figure 3 based on Figure 2As shown in 3, the method comprises the following steps:

[0069] Step 310: Acquire the attribute characteristics, visual function characteristics and fundus image of the target object. For the embodiment of the present disclosure, the specific implementation process can refer to the relevant description in step 210 of the embodiment, which will not be repeated here.

[0070] Step 320: Determine the retinal structural features of the target object based on the fundus image.

[0071] For the embodiment of the present disclosure, the specific implementation process can refer to the relevant description in step 220 of the embodiment, which will not be repeated here.

[0072] Step 330: Perform principal component analysis on the attribute features, visual function features, and retinal structure features to obtain a preset number of first principal component features.

[0073] For the embodiments of the present disclosure, the steps of the embodiments may include: performing decentralized processing on attribute features, visual function features and retinal structure features (decentralization is an important step in the principal component analysis process, which is to eliminate the influence of the average value of each feature and make the features of each dimension have zero mean in the data set. Usually, each feature is subtracted from its own average value to obtain a decentralized data set) to obtain multidimensional features with the same scale; generating a principal component feature matrix corresponding to the multidimensional features; using the principal component feature matrix to perform feature dimensionality reduction processing on the multidimensional features to obtain a preset number of first principal component features. Among them, the purpose of decentralized processing is to make each feature have the same weight in terms of value, so as to avoid dominating the analysis results due to the excessively large numerical range of some features. Multidimensional features include attribute features, visual function features and retinal structure features with the same scale after decentralized processing.

[0074] Step 340: Screen a fundus grading model that meets preset conditions from a plurality of preset clustering models.

[0075] Accordingly, for the embodiment of the present disclosure, the screening of the fundus grading model that meets the preset conditions from multiple preset clustering models in step 340 may include the following steps:

[0076] Step 340 - 1 , obtaining sample multidimensional features, where the sample multidimensional features include sample attribute features, sample visual function features, and sample retinal structure features.

[0077] In specific application scenarios, multiple high myopia patients can be used as sample objects, and corresponding sample multi-dimensional features can be collected for the sample objects. Sample attribute features may include the age, gender, and axial length of the sample object; sample visual function features may include the contrast sensitivity and best corrected visual acuity of the sample object; sample retinal structure features may include the MMD level, β-PPA area, and retinal thickness of the sample object.

[0078] Step 340-2, perform principal component analysis on the sample attribute features, sample visual function features and sample retinal structure features (the number of multidimensional features included in the example is 10, of which the number of linearly correlated features is 4, and the number of second principal component features is preset to 6, so that the data set is mapped to a six-dimensional space, and 6 second principal component features are obtained, which will be used for subsequent clustering model analysis), and screen out a preset number of second principal component features.

[0079] For the embodiment of the present disclosure, the screening of the second principal component feature can refer to the screening process of the first principal component feature in steps 130 and 330 of the embodiment, which will not be repeated here.

[0080] Step 340 - 3 : Input the second principal component features into a plurality of preset clustering models respectively to obtain a clustering result corresponding to each preset clustering model.

[0081] The preset clustering model may include a K-means model, a Gaussian mixture model, etc. The clustering result includes a plurality of classification clusters divided by the preset clustering model for the second principal component feature, and each classification cluster includes at least one feature point.

[0082] Step 340 - 4 : Based on the clustering result, a preset clustering model that meets preset conditions is selected from a plurality of preset clustering models as a fundus grading model.

[0083] For the embodiments of the present disclosure, the embodiment steps may include: calculating a clustering evaluation index for each preset clustering model in multiple preset clustering models based on the clustering results, the clustering evaluation index including at least one of a silhouette coefficient, a variance ratio criterion and a Davidson-Borgin index; based on the clustering evaluation index, screening a preset clustering model that meets preset conditions from multiple preset clustering models as a fundus grading model.

[0084] Accordingly, the preset conditions include but are not limited to at least one of the following: the silhouette coefficient is greater than the first preset threshold; the Calinski-Harabasz Index is greater than the second preset threshold; the Davies-Bouldin Index is less than the third preset threshold. Among them, the first preset threshold is a value between [-1,1] and close to 1, and the specific value can be set according to the actual application scenario; the second preset threshold and the third preset threshold can be set according to the actual application scenario.

[0085] Step 350: Acquire multiple classification clusters contained in the clustering result output by the fundus grading model for the second principal component feature, and configure corresponding preset fundus grades for each of the multiple classification clusters.

[0086] In specific application scenarios, the above-mentioned multiple preset clustering models can be trained unsupervisedly using the multidimensional features of the samples. According to the clustering results of each preset clustering model for the multidimensional features of the samples, the preset clustering model with the best clustering effect is selected from the multiple preset clustering models as the fundus grading model.

[0087] For the disclosed embodiment, the clustering result of the fundus grading model output for the second principal component feature can be obtained. After experimental verification, the clustering result can include multiple classification clusters of high myopia, each classification cluster corresponding to a different feature cluster, reflecting the different pathological risks of the fundus with high myopia.

[0088] Exemplarily, when the clustering results output by the fundus grading model for the second principal component feature included 4 classification clusters, among the 4 classification clusters, patients in category 1 were significantly younger, while patients in category 4 were significantly older; category 1 had the highest contrast sensitivity, followed by category 2, then category 3, and category 4 had the lowest contrast sensitivity. Compared with category 1, category 2 had a higher percentage of MMD2 and a thinner parapapillary retinal nerve fiber layer. The proportion of MMD ≥ 3 in category 3 and category 4 eyes was significantly higher, the β-PPA area was larger, and compared with category 3, category 4 had a higher MMD grade and a thinner macular thickness (P<0.05).

[0089] Accordingly, for this embodiment, the above four classification clusters can be used as standard division results, and corresponding preset fundus assessment levels are configured for multiple classification clusters according to the classification characteristics corresponding to each classification cluster, such as the preset fundus assessment levels of category 1 to category 4 are 1, 2, 3, and 4, respectively. The larger the preset fundus assessment level, the higher the degree of visual function and fundus structure damage of high myopia.

[0090] Step 360: input the first principal component feature into the fundus grading model, cluster the first principal component feature in the fundus grading model, determine the target classification cluster to which the first principal component feature belongs among multiple classification clusters, and determine the preset fundus assessment level corresponding to the target classification cluster as the fundus grading result corresponding to high myopia of the target object.

[0091] For the disclosed embodiment, after determining the first principal component feature corresponding to the target object, the first principal component feature can be input into the fundus grading model to obtain the predicted score of the first principal component feature for each of the multiple classification clusters, and the classification cluster with the highest corresponding predicted score among the multiple classification clusters is determined as the target classification cluster to which the first principal component feature belongs. Then, the preset fundus assessment level corresponding to the target classification cluster can be determined as the fundus grading result of the target object corresponding to high myopia.

[0092] For example, when the first principal component feature is input into the fundus grading model and the target classification cluster is determined to be the second category (i.e., the second classification cluster) of the two classification clusters mentioned in step 350 of the embodiment, the preset fundus assessment level corresponding to the second category can be determined as the fundus grading result of the target object corresponding to high myopia, that is, the fundus grading result of the target object corresponding to high myopia is level 2. Furthermore, the high myopia characteristics of the second classification cluster can be obtained, which is convenient for monitoring and early warning of myopic fundus lesion functions based on the high myopia characteristics.

[0093] In summary, the technical solution in the present application can first obtain the attribute characteristics and visual function characteristics of the target object, and determine the retinal structure characteristics based on the fundus image of the target object, and then perform principal component analysis on the attribute characteristics, visual function characteristics and retinal structure characteristics to obtain a preset number of first principal component characteristics. By analyzing the principal component characteristics, the feature dimension can be reduced, the main features can be retained, and the efficiency of fundus grading for high myopia can be improved. The first principal component characteristics can then be input into the fundus grading model to obtain the fundus grading result of the target object corresponding to high myopia. In view of the fact that the fundus grading model is a preset clustering model that meets the preset conditions selected from multiple preset clustering models, the high myopia is graded and evaluated by the fundus grading model. Compared with the manual grading evaluation method, the accuracy and comprehensiveness of the high myopia grading evaluation can be improved.

[0094] Based on the above Figure 2 , Figure 3 Provide a detailed description of the method for grading high myopia, such as Figure 4 As shown, Figure 4 FIG. 1 is a block diagram of a high myopia fundus grading system according to an exemplary embodiment. Figure 4 As shown, the device comprises:

[0095] An acquisition module 41, which can be used to acquire attribute characteristics, visual function characteristics and fundus images of a target object;

[0096] A determination module 42, which can be used to determine the retinal structural features of the target object based on the fundus image;

[0097] An analysis module 43, which can be used to perform principal component analysis on the attribute features, visual function features and retinal structure features to obtain a preset number of first principal component features;

[0098] The evaluation module 44 can be used to input the first principal component feature into the fundus grading model to obtain the fundus grading result of the target object corresponding to high myopia, wherein the fundus grading model is a preset clustering model that meets preset conditions and is screened out from multiple preset clustering models.

[0099] In some embodiments of the present application, the analysis module 43 can be specifically used to decentralized attribute features, visual function features and retinal structure features to obtain multidimensional features with the same scale; generate a principal component feature matrix corresponding to the multidimensional features; use the principal component feature matrix to perform feature dimensionality reduction processing on the multidimensional features to obtain a preset number of first principal component features.

[0100] In some embodiments of the present application, the analysis module 43 may also be used to screen a fundus grading model that meets preset conditions from a plurality of preset clustering models;

[0101] Correspondingly, when the analysis module 43 screens a fundus grading model that meets preset conditions from multiple preset clustering models, it can be specifically used to obtain sample multidimensional features, which include sample attribute features, sample visual function features, and sample retinal structure features; principal component analysis is performed on the sample attribute features, the sample visual function features, and the sample retinal structure features to screen out a preset number of second principal component features; the second principal component features are respectively input into multiple preset clustering models to obtain clustering results corresponding to each preset clustering model; based on the clustering results, a preset clustering model that meets the preset conditions is screened from multiple preset clustering models as a fundus grading model.

[0102] In some embodiments of the present application, the clustering results include multiple classification clusters, each classification cluster contains at least one feature point; based on the clustering results, a preset clustering model that meets the preset conditions is screened from multiple preset clustering models as a fundus grading model, the analysis module 43 can also be used to calculate the clustering evaluation index of each preset clustering model in the multiple preset clustering models according to the clustering results, the clustering evaluation index includes at least one of the silhouette coefficient, the variance ratio criterion and the Davidson-Borgin index; based on the clustering evaluation index, a preset clustering model that meets the preset conditions is screened from multiple preset clustering models as a fundus grading model.

[0103] In some embodiments of the present application, Figure 5 As shown, the device may further include: a configuration module 45; an acquisition module 41, which may also be used to acquire a plurality of classification clusters contained in the clustering result output by the fundus grading model for the second principal component feature;

[0104] The configuration module 45 can be used to configure corresponding preset fundus assessment levels for the multiple classification clusters respectively.

[0105] In some embodiments of the present application, when the first principal component feature is input into the fundus grading model to obtain the fundus grading result of the target object corresponding to high myopia, the evaluation module 44 can be specifically used to input the first principal component feature into the fundus grading model, cluster the first principal component feature in the fundus grading model, and determine the target classification cluster to which the first principal component feature belongs among multiple classification clusters; and determine the preset fundus evaluation level corresponding to the target classification cluster as the fundus grading result of the target object corresponding to high myopia.

[0106] Regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0107] In the embodiments of the present application, the attribute characteristics and visual function characteristics of the target object can be first obtained, and the retinal structure characteristics can be determined according to the fundus image of the target object, and then the principal component analysis can be performed on the attribute characteristics, visual function characteristics and retinal structure characteristics to obtain a preset number of first principal component characteristics. By analyzing the principal component characteristics, the feature dimension can be reduced, and the main features can be retained, thereby improving the efficiency of fundus grading for high myopia. The first principal component characteristics can then be input into the fundus grading model to obtain the fundus grading result of the target object corresponding to high myopia. In view of the fact that the fundus grading model is a preset clustering model that meets the preset conditions selected from multiple preset clustering models, the fundus grading model is used to grade and evaluate high myopia, which can improve the accuracy and comprehensiveness of the fundus grading evaluation compared to the manual grading evaluation method.

[0108] The above text describes the high myopia fundus grading system of the embodiment of the present invention from the perspective of the functional module in conjunction with the accompanying drawings. It should be understood that the functional module can be implemented in hardware form, can be implemented by instructions in software form, and can also be implemented by a combination of hardware and software modules. Specifically, the steps of the high myopia fundus grading method embodiment in the embodiment of the present invention can be completed by the hardware integrated logic circuit and / or software form instructions in the processor, and the steps of the high myopia fundus grading method applied in conjunction with the embodiment of the present invention can be directly embodied as a hardware decoding processor to perform, or a combination of hardware and software modules in the decoding processor to perform. Optionally, the software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory, and completes the steps in the above-mentioned high myopia fundus grading method embodiment in conjunction with its hardware.

[0109] Figure 6 is a schematic block diagram of an electronic device 600 according to an embodiment of the present invention.

[0110] like Figure 6 As shown, the electronic device 600 may include:

[0111] The memory 610 and the processor 620, the memory 610 is used to store the computer program and transmit the program code to the processor 620. In other words, the processor 620 can call and run the computer program from the memory 610 to implement the method in the embodiment of the present invention.

[0112] For example, the processor 620 may be configured to execute the above method embodiments according to instructions in the computer program.

[0113] In some embodiments of the present invention, the processor 620 may include but is not limited to:

[0114] General-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc.

[0115] In some embodiments of the present invention, the memory 610 includes but is not limited to:

[0116] Volatile memory and / or non-volatile memory. Among them, the non-volatile memory can be read-only memory (ROM), programmable ROM (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM) or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM) and direct memory bus random access memory (Direct Rambus RAM, DR RAM).

[0117] In some embodiments of the present invention, the computer program may be divided into one or more modules, which are stored in the memory 610 and executed by the processor 620 to complete the method provided by the present invention. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in the controller.

[0118] like Figure 6 As shown, the electronic device 600 may further include:

[0119] The transceiver 630 may be connected to the processor 620 or the memory 610 .

[0120] The processor 620 may control the transceiver 630 to communicate with other devices, specifically, to send data to other devices or receive data sent by other devices. The transceiver 630 may include a transmitter and a receiver. The transceiver 630 may further include an antenna, and the number of antennas may be one or more.

[0121] It should be understood that the various components in the electronic device are connected via a bus system, wherein the bus system includes not only a data bus but also a power bus, a control bus and a status signal bus.

[0122] The present invention also provides a computer storage medium having a computer program stored thereon, which, when executed by a computer, enables the computer to perform the method of the above method embodiment. In other words, an embodiment of the present invention also provides a computer program product containing instructions, which, when executed by a computer, enables the computer to perform the method of the above method embodiment.

[0123] When software is used for implementation, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function according to the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable systems. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (Digital Subscriber Line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integration. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a digital video disc (Digital Video Disc, DVD)), or a semiconductor medium (e.g., a solid-state drive (Solid State Disk, SSD)), etc.

[0124] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments applied herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0125] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or modules, which can be electrical, mechanical or other forms.

[0126] The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. For example, each functional module in each embodiment of the present application may be integrated into a processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0127] The above is only a preferred embodiment of the present invention. It should be pointed out that ordinary technicians in this technical field can make several improvements and supplements without departing from the method of the present invention. These improvements and supplements should also be regarded as the scope of protection of the present invention.

Claims

1. A method for grading high myopia fundus, characterized in that: include: Acquire the attribute characteristics, visual function characteristics and fundus images of the target object; determining the retinal structural features of the target object based on the fundus image; Performing principal component analysis on the attribute features, the visual function features, and the retinal structure features to obtain a preset number of first principal component features; The first principal component feature is input into a fundus grading model to obtain a fundus grading result of the target object corresponding to high myopia, wherein the fundus grading model is a preset clustering model that meets preset conditions and is screened out from multiple preset clustering models.

2. The method according to claim 1, characterized in that The principal component analysis is performed on the attribute feature, the visual function feature and the retinal structure feature to obtain a preset number of first principal component features, including: Decentralizing the attribute features, the visual function features, and the retinal structure features to obtain multidimensional features with the same scale; Generate a principal component feature matrix corresponding to the multidimensional features; The multidimensional features are processed by dimensionality reduction using the principal component feature matrix to obtain a preset number of first principal component features.

3. The method according to claim 1, characterized in that Before inputting the first principal component feature into a fundus grading model to obtain a fundus grading result corresponding to high myopia of the target object, the method further includes: Screening a fundus grading model that meets preset conditions from multiple preset clustering models; The screening of a fundus grading model that meets preset conditions from a plurality of preset clustering models includes: Acquiring multidimensional features of the sample, wherein the multidimensional features of the sample include sample attribute features, sample visual function features, and sample retinal structure features; Performing principal component analysis on the sample attribute features, the sample visual function features, and the sample retinal structure features, and screening out a preset number of first principal component features; Inputting the first principal component feature into a plurality of preset clustering models respectively to obtain a clustering result corresponding to each of the preset clustering models; Based on the clustering result, a preset clustering model that meets preset conditions is selected from the plurality of preset clustering models as the fundus grading model.

4. The method according to claim 3, characterized in that The clustering result includes a plurality of classification clusters, each of which includes at least one feature point; The step of selecting a preset clustering model that meets preset conditions from the plurality of preset clustering models based on the clustering result as the fundus grading model comprises: According to the clustering result, calculating a clustering evaluation index of each preset clustering model in the plurality of preset clustering models, the clustering evaluation index comprising at least one of a silhouette coefficient, a variance ratio criterion and a Davidson-Botting index; Based on the clustering evaluation index, a preset clustering model that meets preset conditions is screened from the multiple preset clustering models as the fundus grading model.

5. The method according to claim 3, characterized in that: Before inputting the first principal component feature into the fundus grading model to obtain the fundus grading result of the target object corresponding to high myopia, the method further includes: Acquire a plurality of classification clusters contained in the clustering result output by the fundus grading model for the first principal component feature; Corresponding preset fundus assessment levels are respectively configured for the multiple classification clusters.

6. The method according to claim 5, characterized in that The step of inputting the first principal component feature into a fundus grading model to obtain a fundus grading result corresponding to high myopia of the target object includes: Inputting the first principal component feature into a fundus grading model, clustering the first principal component feature in the fundus grading model, and determining a target classification cluster to which the first principal component feature belongs from among the multiple classification clusters; The preset fundus assessment level corresponding to the target classification cluster is determined as the fundus grading result corresponding to high myopia of the target object.

7. A high myopia fundus grading system, characterized in that: include: An acquisition module, used to acquire attribute characteristics, visual function characteristics and fundus images of a target object; A determination module, configured to determine the retinal structural features of the target object based on the fundus image; An analysis module, configured to perform principal component analysis on the attribute features, the visual function features and the retinal structure features to obtain a preset number of first principal component features; An evaluation module is used to input the first principal component feature into a fundus grading model to obtain a fundus grading result corresponding to high myopia of the target object, wherein the fundus grading model is a preset clustering model that meets preset conditions and is screened out from multiple preset clustering models.

8. An electronic device, characterized in that: include: A processor and a memory, the memory being used to store a computer program, and the processor being used to call and run the computer program stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein the computer program enables a computer to execute the method according to any one of claims 1 to 6.

Images