Wide-angle fundus image analysis method and device, electronic device, and computer-readable storage medium

Abstract

The present application relates to a wide-angle fundus image analysis method and device, electronic equipment, and computer-readable storage medium. The method includes: preprocessing the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image; inputting the wide-angle fundus preprocessed image into the optic disc macula precision positioning model to obtain the optic disc macula recognition result; inputting the wide-angle fundus preprocessed image into the fundus change feature recognition model to obtain the fundus change feature recognition result; processing the optic disc macula recognition result and the fundus change feature recognition result to obtain wide-angle fundus analysis data. It can quantitatively analyze the fundus changes related to high myopia in wide-angle fundus photography, improve the efficiency, accuracy and stability of fundus image analysis, and has the advantages of wide application range, high precision, flexibility and simplicity.

Description

Technical Field

[0001] The present application relates to the fields of bioinformatics and deep learning, and specifically to a wide-angle fundus image analysis method and device, an electronic device, and a computer-readable storage medium. Background Art

[0002] The prevalence of myopia has increased significantly worldwide, and some people will develop high myopia, which is the main cause of irreversible visual impairment worldwide. Fundus photography is currently a commonly used ophthalmological examination for screening fundus lesions due to its easy availability. Fundus photography can be used to detect early pathological changes in high myopia and predict the progression of myopia, so as to carry out early intervention. At present, the research on fundus lesions in high myopia is mainly based on the following: Figure 1A The conventional fundus image shown is established, but due to its limited field of view, it can only cover 45° of the fundus field of view, making it difficult to record the outer edge of the progressively expanding chorioretinal atrophic lesion in each direction.

[0003] In recent years, wide-angle fundus photography has attracted more and more attention. Figure 1B The wide-angle fundus image shown can cover up to 200° of fundus field, so wide-angle fundus photography is helpful to record peripheral retinal lesions, which are more common in the fundus with longer axial length. At present, the research on fundus lesions in high myopia mainly focuses on leopard spots and blood vessels. The quantitative analysis of leopard spot density, optic disc morphology, atrophic arc, and vascular morphology in fundus photography explores the influencing factors of high myopia and better reveals the occurrence and development of high myopia.

[0004] Globally, there is a shortage of well-trained myopia ophthalmologists. It is difficult to diagnose early pathological changes in the fundus of high myopia in general ophthalmology institutions, and it is difficult to conduct effective and long-term follow-up. The doctor's observation results are affected by personal experience, skills and subjective judgment, which may lead to inconsistent results; the results of direct observations may vary greatly at different times or between different doctors, which affects the reliability of the results; the accuracy and stability of the judgment and analysis may decrease due to excessive workload and fatigue. Summary of the invention

[0005] The present application mainly provides a wide-angle fundus image analysis method and device, an electronic device, and a computer-readable storage medium.

[0006] The first aspect of the present application provides a wide-angle fundus image analysis method, comprising: preprocessing the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image; inputting the wide-angle fundus preprocessed image into an optic disc macula precision positioning model to obtain an optic disc macula recognition result; inputting the wide-angle fundus preprocessed image into a fundus change feature recognition model to obtain a fundus change feature recognition result; processing the optic disc macula recognition result and the fundus change feature recognition result to obtain wide-angle fundus analysis data.

[0007] Optionally, the optic disc and macula recognition results include: optic disc recognition results and macula recognition results; the fundus change feature recognition results include: leopard spot recognition results and retinal blood vessel recognition results; the wide-angle fundus analysis data include: average leopard spot density, leopard spot density in the optic disc area, leopard spot density in the macula area, blood vessel density, average blood vessel diameter, average blood vessel length, average blood vessel curvature, and blood vessel fractal dimension.

[0008] Optionally, the step of preprocessing the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image includes: for the original wide-angle fundus image, removing the surrounding non-fundus structure area using a fundus area segmentation model trained based on a semantic segmentation fully convolutional network to obtain a region of interest image; for the region of interest image, denoising using a low-pass filter, performing brightness normalization correction, and performing image enhancement to obtain the wide-angle fundus preprocessed image.

[0009] Optionally, the optic disc and macula precise positioning model is obtained by training the optic disc and macula features based on the YOLOv8 detection network.

[0010] Optionally, the method includes the step of training the optic disc and macula detection network, including: inputting the fundus image of the original image size with the optic disc and macula positions marked into the YOLOv8 detection network for initial training, thereby obtaining a coarse positioning detection model for the optic disc and macula; based on the optic disc center point and the macula center point detected by the coarse positioning detection model, calculating the distance from the optic disc center point to the macula center point, and cropping the fundus image into a square with the optic disc center point as the center and X times the spacing as the side length, wherein 2.5≤X≤3.5; accurately labeling the macula and optic disc in the cropped square image and inputting it into the YOLOv8 detection network for re-training, thereby obtaining a fine positioning model for the optic disc and macula.

[0011] Optionally, the fundus change feature recognition model is trained based on the TransUNet network model.

[0012] Optionally, the method includes a step of training the fundus change feature recognition model, including: performing multi-block processing on the labeled wide-angle fundus image and inputting it into the TransUNet network as a training sample.

[0013] The wide-angle fundus image analysis device provided in the second aspect of the present application is used to execute the wide-angle fundus image analysis method of the present application, and the device includes: a preprocessing unit, used to preprocess the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image; a first recognition unit, used to input the wide-angle fundus preprocessed image into an optic disc macula precision positioning model to obtain an optic disc macula recognition result; a second recognition unit, used to input the wide-angle fundus preprocessed image into a fundus change feature recognition model to obtain a fundus change feature recognition result; a data processing unit, used to process the optic disc macula recognition result and the fundus change feature recognition result to obtain wide-angle fundus analysis data.

[0014] The electronic device provided in the third aspect of the present application includes: a memory storing computer-readable instructions; and a processor executing the computer-readable instructions stored in the memory to implement the wide-angle fundus image analysis method of the present application.

[0015] The fourth aspect of the present application provides a computer-readable storage medium, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor, the wide-angle fundus image analysis method of the present application is implemented.

[0016] The technical solution of the present application can perform quantitative analysis on fundus changes related to high myopia in wide-angle fundus photography, improve the efficiency, accuracy and stability of fundus image analysis, and has the advantages of wide application range, high precision, flexibility and simplicity. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] For purposes of illustration and not limitation, the present application will now be described according to its preferred embodiments, particularly with reference to the accompanying drawings, in which:

[0018] Figure 1A This is a schematic diagram of traditional fundus photography. Figure 1B It is a schematic diagram of wide-angle fundus photography;

[0019] Figure 2 is a flow chart of a wide-angle fundus image analysis method according to an exemplary embodiment of the present application;

[0020] Figure 3 is a preprocessing schematic diagram of a wide-angle fundus image analysis method according to an exemplary embodiment of the present application;

[0021] Figure 4 is a schematic diagram of optic disc macula detection in a wide-angle fundus image analysis method according to an exemplary embodiment of the present application;

[0022] Figure 5 is a structural block diagram of a wide-angle fundus image analysis device according to an exemplary embodiment of the present application;

[0023] Figure 6 is a structural block diagram of an electronic device according to an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0024] The following is a description of exemplary embodiments of the present application in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered to be merely exemplary. Therefore, it should be appreciated by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description. It should be noted that, in the absence of conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0025] like Figure 2 As shown, the wide-angle fundus image analysis method of the exemplary embodiment of the present application mainly includes the following steps A to D.

[0026] Step A: preprocessing the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image.

[0027] The specific process of step A can be found in Figure 3 The invention relates to an understanding of the present invention, including: (1) for the original wide-angle fundus image, the peripheral non-fundus structure area is removed by using a fundus region segmentation model based on semantic segmentation full convolutional network training to obtain a region of interest image, wherein the region of interest represents the effective information area in the fundus image, that is, the part containing fundus information; (2) the region of interest image is denoised by using a low-pass filter, brightness normalization correction is performed, and image enhancement is performed to obtain a wide-angle fundus preprocessed image.

[0028] Step B: Input the wide-angle fundus preprocessed image into the optic disc macula precise positioning model to obtain the optic disc macula recognition result. The optic disc macula recognition result may specifically include the optic disc recognition result and the macula recognition result.

[0029] Among them, the optic disc macula precision positioning model can be obtained by training the optic disc macula features based on the YOLOv8 detection network. The YOLOv8 detection network can be structurally divided into four parts: input, backbone network, neck, and detection head. It can capture features at different levels, has extremely high monitoring performance for targets of different sizes, and has extensive generalization ability and robustness, and can generate confidence and location information for detected targets.

[0030] The wide-angle fundus image analysis method of the present application may further include the step of training an optic disc and macula detection network, which may specifically include: inputting the fundus image of the original image size with the optic disc and macula positions marked into the YOLOv8 detection network for initial training, thereby obtaining a rough positioning detection model for the optic disc and macula; based on the optic disc center point and the macula center point detected by the rough positioning detection model, calculating the distance from the optic disc center point to the macula center point, and square cropping the fundus image with the optic disc center point as the center and X times the distance as the side length (such as Figure 4 As shown), where 2.5≤X≤3.5; the cropped square image is accurately labeled with the macula and optic disc and then input into the YOLOv8 detection network for retraining, thereby obtaining the optic disc and macula precise positioning model.

[0031] Step C: Input the wide-angle fundus preprocessed image into the fundus change feature recognition model to obtain the fundus change feature recognition result. The fundus change feature recognition result may specifically include the leopard spot recognition result and the retinal blood vessel recognition result.

[0032] Among them, the fundus change feature recognition model can be trained based on the TransUNet network model. The TransUNet network model combines the advantages of Transformer and U-Net, using Transformer as an encoder to capture global context information, and based on U-Net to perform precise feature extraction and image detail restoration, thereby enhancing the ability to understand global information while maintaining the model's high sensitivity to local details.

[0033] The steps of training the fundus change feature recognition model include performing multi-block processing on the labeled wide-angle fundus images and inputting them into the TransUNet network as training samples. On the one hand, this can better utilize the rich information in the high-resolution wide-angle fundus images, and on the other hand, it can expand the number of training samples.

[0034] Step D: Process the optic disc macula identification results and the fundus change feature identification results to obtain wide-angle fundus analysis data. The wide-angle fundus analysis data may specifically include: average leopard spot density, leopard spot density in the optic disc area, leopard spot density in the macula area, blood vessel density, average blood vessel diameter, average blood vessel length, average blood vessel curvature, and blood vessel fractal dimension, etc. The definitions of some of the data are as follows.

[0035] Average leopard spot density: the ratio of the total area of ​​the fundus leopard spot area to the area of ​​the region of interest. The leopard spot area represents the exposed area of ​​the choroidal large blood vessels caused by atrophy of the retinal pigment epithelium and choriocapillaris, which appears as a leopard-like texture on the fundus image.

[0036] Leopard spot density in the optic disc area: the average leopard spot density within the circle with the center of the optic disc as the center and a radius of twice the diameter of the optic disc.

[0037] Leopard spot density in the macular area: the average leopard spot density within the circle with the fovea centralis as the center and twice the diameter of the optic disc as the radius.

[0038] In order to enable those skilled in the art to better understand, specific embodiments are listed below to illustrate the wide-angle fundus image analysis method of the present application.

[0039] (1) Preprocessing: For each fundus image to be processed, the fundus region segmentation model based on semantic segmentation Fully Convolutional Networks (FCN) training is first used to remove the non-fundus structure area around the image to obtain the region of interest (ROI). The fundus region segmentation model uses 400 labeled samples for model training and 100 samples for testing. The segmentation accuracy of the test model is 99.3%. After removing the non-fundus structure, a low-pass filter is further used to reduce the image high-frequency noise, thereby reducing the interference of image noise information on subsequent feature recognition. At the same time, in order to make the texture features of the image more obvious, the Gamma algorithm is used to normalize the image brightness, and then the image is enhanced based on the contrast limited adaptive histogram equalization (CLAHE) method, thereby improving the difference in the internal features of the fundus structure on the image, which is conducive to the improvement of the information interaction ability between different channels of the image during subsequent feature recognition and reducing computational redundancy.

[0040] (2) Optic disc and macula identification: The optic disc and macula identification model based on the YOLOv8 detection network is used for identification. The model training stage can be divided into two stages: coarse positioning and fine positioning. In the coarse positioning stage, the original large and small fundus images with the optic disc and macula positions marked are input into the YOLOv8 detection network for training, thereby obtaining the optic disc and macula coarse positioning detection model. Based on the optic disc and macula detected by the coarse positioning detection model, the distance from the center point of the optic disc to the center point of the macula is calculated. The image is cropped into a square with the optic disc center as the center and a side length of 3 times the distance. The cropped image is accurately labeled with the macula and optic disc and then input into the YOLOv8 detection network for training, thereby obtaining the optic disc and macula fine positioning model. 200 images are used to verify the optic disc and macula detection results, and the accuracy, sensitivity and specificity are calculated, and the results are 99.9%, 87.6% and 99.9% respectively.

[0041] (3) Fundus change feature recognition: The fundus change feature recognition model based on the TransUNet network was used to identify the retinal vascular features and leopard spot features. First, the threshold segmentation method was used to preliminarily annotate the retinal vascular and leopard spot features, and then two doctors performed corrections and verifications respectively, which can improve the accuracy of annotation. Since the pixel resolution of wide-angle fundus images is high, if large-size images are input into the network, the network feature extraction will be insufficient, affecting the final segmentation accuracy. The nine-square image block method is preferably used to improve the accuracy, and the image is divided into multiple small images of 3×3 according to the length and width of the image. In order to avoid the loss of image information, there is overlap between the cropping of multiple small images. After the cropping operation, the number of retinal vascular annotated samples and leopard spot annotated samples was expanded from the original 31 and 24 to 279 and 216. The expanded labeled samples were divided into training set and test set according to 8:2. The samples in the training set were input into the TransUnet network model for training. The data in the test set were used to evaluate the accuracy of the model segmentation, and the accuracy, sensitivity and specificity were calculated respectively. The verification results of blood vessel segmentation were 98.5%, 74.2%, and 99.5%, and the verification results of leopard spot segmentation were 96.3%, 73.2%, and 98.2%.

[0042] (4) Calculation of wide-angle fundus analysis data: Based on the identified retinal blood vessels and leopard spot features, the density of leopard spots, leopard spot density in the optic disc and macula, blood vessel density, average blood vessel diameter, blood vessel length, average blood vessel curvature, and blood vessel fractal dimension are calculated. These data can provide a relatively comprehensive description of the leopard spots and retinal blood vessel features.

[0043] The wide-angle fundus image analysis device of the exemplary embodiment of the present application is used in the wide-angle fundus image analysis method of the exemplary embodiment of the present application, such as Figure 5 As shown, it mainly includes: a preprocessing unit 100, a first recognition unit 200, a second recognition unit 300 and a data processing unit 400. The preprocessing unit 100 is used to preprocess the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image. The first recognition unit 200 is used to input the wide-angle fundus preprocessed image into the optic disc macula precision positioning model to obtain the optic disc macula recognition result. The second recognition unit 300 is used to input the wide-angle fundus preprocessed image into the fundus change feature recognition model to obtain the fundus change feature recognition result. The data processing unit 400 is used to process the optic disc macula recognition result and the fundus change feature recognition result to obtain the wide-angle fundus analysis data.

[0044] In summary, the wide-angle fundus image analysis method and device of the embodiments of the present application have the following advantages: they can perform quantitative analysis on fundus changes related to high myopia in wide-angle fundus photography, thereby improving the efficiency, accuracy and stability of fundus image analysis.

[0045] Combine the following Figure 6 The electronic device of the implementation mode of the present application is introduced in detail. The electronic device can be any intelligent terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), a car computer, etc. The electronic device may include: the processor 1110 can be implemented by a general central processing unit (CPU), a microprocessor, an application specific integrated circuit (ASIC), or one or more integrated circuits, etc., for executing related programs to implement the technical solution provided in the embodiment of the present application; the memory 1120 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device or a random access memory (RAM). The memory 1120 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 1120, and the processor 1110 is called to execute the wide-angle fundus image analysis method of the embodiment of the present application; the input / output interface 1130 is used to realize information input and output; the communication interface 1140 is used to realize the communication interaction between this device and other devices, and communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.); the bus 1150 transmits information between the various components of the device (such as the processor 1110, the memory 1120, the input / output interface 1130 and the communication interface 1140); wherein the processor 1110, the memory 1120, the input / output interface 1130 and the communication interface 1140 realize the communication connection between each other within the device through the bus 1150.

[0046] The embodiment of the present application also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by the processor, the wide-angle fundus image analysis method of the embodiment of the present application is implemented. The storage medium may include a high-speed random access memory, and may also include a non-transitory memory, such as at least one U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, and other media that can store programs.

[0047] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0048] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0049] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0050] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0051] The terms "first", "second", etc. in the specification 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 in appropriate circumstances, so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units need not be 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 equipment. It should be understood that in the present application, "at least one" refers to one or more, and "a plurality of" refers to two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships can exist.

[0052] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation, such as multiple units 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 devices or units, which can be electrical, mechanical or other forms.

[0053] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0054] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0055] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions for enabling an electronic device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present application.

[0056] The preferred embodiments of the embodiments of the present application are described above with reference to the accompanying drawings, but the scope of rights of the embodiments of the present application is not limited thereby. Any modifications, equivalent substitutions and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall be within the scope of rights of the embodiments of the present application. The above is only a preferred implementation mode of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application, and these improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A wide-angle fundus image analysis method, characterized in that: include: Preprocessing the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image; Inputting the wide-angle fundus preprocessed image into the optic disc macula precise positioning model to obtain the optic disc macula recognition result; Inputting the wide-angle fundus preprocessed image into a fundus change feature recognition model to obtain a fundus change feature recognition result; The optic disc macula recognition result and the fundus change feature recognition result are processed to obtain wide-angle fundus analysis data.

2. The method according to claim 1, characterized in that The optic disc and macula identification results include: optic disc identification results and macula identification results; The fundus change feature recognition result includes: leopard spot recognition result and retinal blood vessel recognition result; The wide-angle fundus analysis data include: average leopard spot density, leopard spot density in the optic disc area, leopard spot density in the macula area, blood vessel density, average blood vessel diameter, average blood vessel length, average blood vessel curvature, and blood vessel fractal dimension.

3. The method according to claim 1 or 2, characterized in that: The step of preprocessing the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image comprises: For the original wide-angle fundus image, a fundus region segmentation model trained based on a semantic segmentation fully convolutional network is used to remove the surrounding non-fundus structure area to obtain an image of the region of interest; The image of the region of interest is subjected to denoising using a low-pass filter, brightness normalization correction, and image enhancement to obtain the wide-angle fundus preprocessing image.

4. The method according to claim 1 or 2, characterized in that: The optic disc macula precise positioning model is obtained based on the YOLOv8 detection network after training the optic disc macula features.

5. The method according to claim 3, characterized in that: The method comprises the steps of training the optic disc macula detection network, comprising: The fundus image with the original size marked with the optic disc and macula positions is input into the YOLOv8 detection network for initial training, thereby obtaining the optic disc and macula coarse positioning detection model; Based on the optic disc center and the macula center detected by the coarse positioning detection model, the distance from the optic disc center to the macula center is calculated, and the fundus image is cut into a square with the optic disc center as the center and X times the distance as the side length, where 2.5≤X≤3.5; The cropped square image is accurately labeled with the macula and optic disc and then input into the YOLOv8 detection network for retraining to obtain the optic disc and macula precise positioning model.

6. The method according to claim 1 or 2, characterized in that: The fundus change feature recognition model is obtained based on the training of the TransUNet network model.

7. The method according to claim 6, characterized in that The method comprises the step of training the fundus change feature recognition model, including: performing multi-block processing on the labeled wide-angle fundus image and inputting the image as a training sample into the TransUNet network.

8. A wide-angle fundus image analysis device, used to execute the wide-angle fundus image analysis method according to any one of claims 1 to 7, characterized in that: The device comprises: A preprocessing unit, used for preprocessing the original wide-angle fundus image to obtain a wide-angle fundus preprocessed image; A first recognition unit is used to input the wide-angle fundus preprocessed image into the optic disc macula precise positioning model to obtain an optic disc macula recognition result; A second recognition unit is used to input the wide-angle fundus preprocessed image into a fundus change feature recognition model to obtain a fundus change feature recognition result; The data processing unit is used to process the optic disc macula recognition result and the fundus change feature recognition result to obtain wide-angle fundus analysis data.

9. An electronic device, characterized in that: The electronic device comprises: a memory storing computer-readable instructions; and A processor executes computer-readable instructions stored in the memory to implement the wide-angle fundus image analysis method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the wide-angle fundus image analysis method according to any one of claims 1 to 7 is implemented.

Images