Method, device, computer and storage medium for segmenting optic cup and optic disc

Abstract

The present invention relates to artificial intelligence and digital medical technology, and an embodiment thereof provides a method, device, computer, and storage medium for optic cup and optic disc segmentation. The method comprises the following steps: acquiring an original fundus image; performing feature region extraction on the original fundus image to obtain a candidate region image; inputting the candidate region image into an optic disc detection network branch, a cup and disc segmentation network branch, and a quality scoring network branch for training to obtain a target neural network module, wherein the optic disc detection network branch is used to determine optic disc position information in the candidate region image, the cup and disc segmentation network branch is used to segment the candidate region image and output a cup and disc segmentation image, and the quality scoring network branch is used to perform quality scoring on the cup and disc segmentation image; and inputting the original fundus image into the target neural network module to obtain a target cup and disc segmentation image output by the target neural network module, thereby optimizing the segmentation process and effectively improving the accuracy of segmentation.

Description

Technical Field

[0001] The embodiments of the present invention relate to, but are not limited to, the field of artificial intelligence digital medical technology, and in particular to a method, device, computer, and storage medium for segmenting an optic cup and optic disc. Background Art

[0002] Glaucoma is the second leading cause of blindness worldwide. Its early symptoms are often subtle, leading to late detection and potentially irreparable damage, resulting in permanent blindness. Measuring parameters such as the cup-to-disc ratio (the ratio of the vertical diameter or area of the optic cup to the optic disc, as well as the relative position of the cup and disc rim) in fundus images is crucial for the diagnosis of eye diseases like glaucoma. Therefore, accurately segmenting the optic cup and disc from fundus images is crucial for the precise quantification of these parameters.

[0003] The current fundus image segmentation method mainly focuses on single optic disc tissue segmentation, and the only optic cup and optic disc segmentation model adopts a two-stage method. First, the optic disc area is detected using a detection network, the optic disc area is cropped out, and then sent to the segmentation network for optic cup and optic disc segmentation. The training steps of this type of method are relatively cumbersome, and the overall application is time-consuming. Summary of the Invention

[0004] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0005] The main purpose of the embodiments of the present invention is to provide a method, device, computer and storage medium for optic cup and optic disc segmentation, which can optimize the segmentation process and effectively improve the accuracy of segmentation.

[0006] In a first aspect, an embodiment of the present invention provides a method for segmenting an optic cup and an optic disc, comprising:

[0007] Acquire an original fundus image, wherein the original fundus image is a fundus image including fusion image data of the optic cup and the optic disc;

[0008] Performing feature region extraction processing on the original fundus image to obtain a candidate region image;

[0009] Inputting the candidate region image into the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch for training processing, respectively, to obtain a target neural network module, wherein the optic disc detection network branch is used to determine the optic disc position information in the candidate region image, the cup-disc segmentation network branch is used to segment the candidate region image and output the optic cup-optic disc segmentation image, and the quality scoring network branch is used to perform quality scoring on the optic cup-optic disc segmentation image output by the cup-disc segmentation network branch;

[0010] The original fundus image is input into the target neural network module to obtain the target optic cup and optic disc segmentation image output by the target neural network module.

[0011] In one embodiment, the candidate region image is respectively input into the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch for training to obtain a target neural network module, including:

[0012] Inputting the candidate area image into a visor detection network branch to determine visor location information, wherein the visor location information includes category information and bounding box coordinate information;

[0013] Calculating the category information and the bounding box coordinate information using a first loss function and a second loss function respectively to obtain a first loss function value and a second loss function value;

[0014] Inputting the candidate area image into the cup-disc segmentation network branch for segmentation processing to obtain a cup-disc segmentation image, and calculating the cup-disc segmentation image through a third loss function to obtain a third loss function value;

[0015] Inputting the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate region image and the cup-disc segmentation image into the quality scoring network branch to obtain a quality score value, and calculating the quality score value through a fourth loss function to obtain a fourth loss function value;

[0016] Inputting the first loss function value, the second loss function value, the third loss function value and the fourth loss function value into a fifth loss function to calculate a comprehensive loss function value;

[0017] The optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch are iteratively trained according to the comprehensive loss function value to obtain the target neural network module.

[0018] In one embodiment, the step of calculating the cup disc segmentation image through a third loss function to obtain a third loss function value includes:

[0019] The optic cup region and the optic disc region in the cup-optic disc segmentation image are respectively subjected to calculation processing with the optic cup mask and the optic disc mask to obtain a third loss function value.

[0020] In one embodiment, inputting the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate region image and the cup-disc segmentation image into the quality scoring network branch to obtain a quality score value, and calculating the quality score value through a fourth loss function to obtain a fourth loss function value includes:

[0021] Inputting the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate area image and the cup-disc segmentation image into the quality scoring network branch to obtain an optic disc score value and a optic cup score value;

[0022] The optic disc score value, the optic cup score value, the optic cup area, the optic disc area, the optic cup mask, and the optic disc mask are calculated using a fourth loss function to obtain a fourth loss function value.

[0023] In one embodiment, the iterative training of the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch according to the comprehensive loss function value to obtain the target neural network module includes:

[0024] Iteratively training the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch respectively, and determining the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch corresponding to the minimum value of the comprehensive loss function during the iterative training as the target optic disc detection network branch, the target cup-disc segmentation network branch, and the target quality scoring network branch;

[0025] The target neural network module is obtained according to the target optic disc detection network branch, the target cup-disc segmentation network branch and the target quality scoring network branch.

[0026] In one embodiment, inputting the original fundus image into the target neural network module to obtain the target optic cup and optic disc segmentation image output by the target neural network module includes:

[0027] Inputting the original fundus image into the target optic disc detection network branch in the target neural network module to determine the target optic disc position information;

[0028] Inputting the original fundus image into the target cup-disc segmentation network branch in the target neural network module to obtain a cup-disc segmentation image;

[0029] Positioning processing is performed on the cup-disc segmentation image according to the target optic disc position information to obtain a target optic cup-disc segmentation image.

[0030] In one embodiment, the performing feature region extraction processing on the original fundus image to obtain the candidate region image includes:

[0031] Performing a first extraction process on the original fundus image by a feature extraction module to obtain an underlying feature image;

[0032] The region extraction module performs a second extraction process on the underlying feature image to obtain a candidate region image.

[0033] In a second aspect, an embodiment of the present invention provides an optic cup and optic disc segmentation device, comprising:

[0034] An acquisition module, configured to acquire an original fundus image, wherein the original fundus image is a fundus image including fusion image data of the optic cup and the optic disc;

[0035] An extraction module, configured to perform feature region extraction processing on the original fundus image to obtain a candidate region image;

[0036] a training module, configured to input the candidate region image into the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch for training processing, respectively, to obtain a target neural network module, wherein the optic disc detection network branch is configured to determine optic disc position information in the candidate region image, the cup-disc segmentation network branch is configured to segment the candidate region image and output the optic cup-optic disc segmentation image, and the quality scoring network branch is configured to perform quality scoring on the optic cup-optic disc segmentation image output by the cup-disc segmentation network branch;

[0037] The output module is used to input the original fundus image into the target neural network module to obtain the target optic cup and optic disc segmentation image output by the target neural network module.

[0038] In one embodiment, the extraction module is further configured to perform a first extraction process on the original fundus image through the feature extraction module to obtain an underlying feature image; and perform a second extraction process on the underlying feature image through the region extraction module to obtain a candidate region image.

[0039] In one embodiment, the training module is further used to input the candidate area image into the optic disc detection network branch to determine the optic disc position information, wherein the optic disc position information includes category information and bounding box coordinate information; the category information and the bounding box coordinate information are calculated by a first loss function and a second loss function respectively to obtain a first loss function value and a second loss function value; the candidate area image is input into the cup-disc segmentation network branch for segmentation processing to obtain a cup-optic disc segmentation image, and the cup-optic disc segmentation image is calculated by a third loss function to obtain a third loss function value; the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate area image and the cup-optic disc segmentation image are input into the quality scoring network branch to obtain a quality score value, and the quality score value is calculated by a fourth loss function to obtain a fourth loss function value; the first loss function value, the second loss function value, the third loss function value and the fourth loss function value are input into a fifth loss function to obtain a comprehensive loss function value; and the optic disc detection network branch, the cup-disc segmentation network branch and the quality scoring network branch are iteratively trained according to the comprehensive loss function value to obtain the target neural network module.

[0040] In one embodiment, the training module is further configured to perform calculations on the optic cup region and the optic disc region in the cup-optic disc segmentation image, the optic cup mask and the optic disc mask, respectively, to obtain a third loss function value.

[0041] In one embodiment, the training module is further used to input the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate area image and the cup-disc segmentation image into the quality scoring network branch to obtain an optic disc score value and an optic cup score value; and calculate the optic disc score value, the optic cup score value, the optic cup area, the optic disc area, the optic cup mask and the optic disc mask through a fourth loss function to obtain a fourth loss function value.

[0042] In one embodiment, the training module is further used to iteratively train the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch, respectively, and determine the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch corresponding to the minimum value of the comprehensive loss function value during the iterative training process as the target optic disc detection network branch, the target cup-disc segmentation network branch, and the target quality scoring network branch; and obtain the target neural network module based on the target optic disc detection network branch, the target cup-disc segmentation network branch, and the target quality scoring network branch.

[0043] In one embodiment, the output module is further used to input the original fundus image into the target optic disc detection network branch in the target neural network module to determine the target optic disc position information; input the original fundus image into the target cup-disc segmentation network branch in the target neural network module to obtain a cup-disc segmentation image; and perform positioning processing on the cup-disc segmentation image according to the target optic disc position information to obtain a target optic cup-optic disc segmentation image.

[0044] In a third aspect, an embodiment of the present invention provides a computer, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for segmenting the optic cup and optic disc as described in the first aspect is implemented.

[0045] In a fourth aspect, a computer-readable storage medium stores computer-executable instructions, wherein the computer-executable instructions are used to execute the optic cup and optic disc segmentation method according to the first aspect.

[0046] The present invention includes a method for optic cup and optic disc segmentation, comprising the following steps: obtaining an original fundus image, the original fundus image being a fundus image including fused image data of the optic cup and optic disc; performing feature region extraction processing on the original fundus image to obtain a candidate region image; inputting the candidate region image into an optic disc detection network branch, a cup and disc segmentation network branch, and a quality scoring network branch for training processing to obtain a target neural network module, wherein the optic disc detection network branch is used to determine optic disc position information in the candidate region image, and the quality scoring network branch is used to perform quality scoring on the cup and disc segmentation image output by the cup and disc segmentation network branch; inputting the original fundus image into the target neural network module to obtain a target cup and disc segmentation image output by the target neural network module. An improved neural network is used to simultaneously implement optic disc region detection and cup and disc segmentation, and the quality of the cup and disc segmentation is scored and evaluated by the quality scoring branch to control the output standard, thereby optimizing the cup and disc segmentation process and improving the accuracy of the segmentation.

[0047] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a schematic diagram of a system architecture platform for executing an optic cup and optic disc segmentation method provided by one embodiment of the present invention;

[0049] Figure 2 is a flow chart of a method for segmenting an optic cup and optic disc provided by one embodiment of the present invention;

[0050] Figure 3 This is a flowchart of a network training method in an optic cup and optic disc segmentation method provided by an embodiment of the present invention;

[0051] Figure 4 is a schematic diagram of a network block diagram for performing an optic cup and optic disc segmentation method provided by one embodiment of the present invention;

[0052] Figure 5 Schematic diagram of generating a target optic cup and optic disc segmentation image according to a optic cup and optic disc segmentation method provided by one embodiment of the present invention;

[0053] Figure 6 Schematic diagram of an optic cup and optic disc segmentation device provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0055] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and the like in the specification, claims, or accompanying drawings are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0056] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0057] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0058] Digital medicine is a new, modern approach to healthcare that applies modern computer and information technology to the entire medical process. It represents the development direction and management goal of public healthcare. The emergence of digital medical equipment has greatly enriched the content and capacity of medical information. From the original visualization of one-dimensional information, such as important electrophysiological information like electrocardiograms (ECGs) and electroencephalograms (EEGs), to two-dimensional information, such as medical imaging information from CT, MRI, color ultrasound, and digital radiography (DR), and finally to three-dimensional visualization, even four-dimensional information can be obtained, such as a real-time dynamic display of a 3D heart. This information has greatly enriched doctors' diagnostic techniques and ushered medicine into a new era of visual information. Digital medicine is more than just a simple collection of digital medical devices; it is a new, modern approach that applies contemporary computer and information technology to the entire medical process. In digital medicine, patients can complete their medical consultations with minimal steps, doctors' diagnostic accuracy is significantly improved, and patient records record all current and historical patient health information. This greatly facilitates both doctors' diagnosis and patient self-examination, and enables the retrieval of comprehensive patient data required for remote consultations, providing fast and efficient service.

[0059] Currently, glaucoma is the second leading cause of blindness worldwide. Its early symptoms are often subtle, leading to late detection and potentially irreparable damage, resulting in permanent blindness. Measuring parameters such as the cup-to-disc ratio (the ratio of the vertical diameter or area of the optic cup to the optic disc, as well as the relative position of the cup and disc rim) in fundus images is crucial for the diagnosis of eye diseases like glaucoma. Therefore, accurately segmenting the optic cup and disc from fundus images is crucial for the accurate quantification of these parameters.

[0060] The relevant fundus image segmentation methods mainly focus on single optic disc tissue segmentation, and the only optic cup and optic disc segmentation model adopts a two-stage method. First, the optic disc area is detected using a detection network, and then the optic disc area is cropped out and sent to the segmentation network for optic cup and optic disc segmentation. The training steps of this type of method are relatively cumbersome, and the overall application is time-consuming.

[0061] To address the above-mentioned problems, embodiments of the present invention provide a method, apparatus, computer, and storage medium for cup and disc segmentation. The method comprises the following steps: obtaining an original fundus image, which includes fused image data of the optic cup and optic disc; performing feature region extraction on the original fundus image to obtain a candidate region image; inputting the candidate region image into an optic disc detection network branch, a cup and disc segmentation network branch, and a quality scoring network branch for training to obtain a target neural network module. The optic disc detection network branch is used to determine optic disc location information in the candidate region image, and the quality scoring network branch is used to perform quality scoring on the cup and disc segmentation image output by the cup and disc segmentation network branch; and inputting the original fundus image into the target neural network module to obtain a target cup and disc segmentation image output by the target neural network module. An improved neural network is used to simultaneously implement optic disc region detection and cup and disc segmentation. The quality scoring branch is used to score and evaluate the cup and disc segmentation quality and control the output standard. This not only optimizes the cup and disc segmentation process but also improves segmentation accuracy.

[0062] The embodiments of the present invention are further described below with reference to the accompanying drawings.

[0063] like Figure 1 As shown, Figure 1 FIG. 1 is a schematic diagram of a system architecture platform 100 for executing an optic cup and optic disc segmentation method according to an embodiment of the present invention.

[0064] exist Figure 1 In the example of FIG, the system architecture platform 100 is provided with a processor 110 and a memory 120, wherein the processor 110 and the memory 120 can be connected via a bus or other means. Figure 1 The bus connection is taken as an example.

[0065] The memory 120 is a non-transitory computer-readable storage medium that can be used to store non-transitory software programs and non-transitory computer executable programs. In addition, the memory 120 may include a high-speed random access memory and may also include a non-transitory memory, such as at least one disk storage device, a flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory 120 may optionally include a memory remotely located relative to the processor 110, and these remote memories may be connected to the system architecture platform via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0066] Those skilled in the art will appreciate that the system architecture platform can be applied to existing communication network systems and subsequently evolved mobile communication network systems, etc., and this embodiment does not specifically limit this.

[0067] It will be understood by those skilled in the art that Figure 1 The system architecture platform shown in the figure does not constitute a limitation on the embodiments of the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0068] The system architecture platform 100 can be an independent system architecture platform, or it can be a cloud system architecture platform 100 that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0069] Based on the above system architecture platform, various embodiments of the optic cup and optic disc segmentation method of the present invention are proposed below.

[0070] like Figure 2 As shown, Figure 2 This is a flow chart of a method for optic cup and optic disc segmentation provided by an embodiment of the present invention. The method is applied to the above-mentioned architecture platform and includes but is not limited to step S100, step S200, step S300 and step S400.

[0071] Step S100 : acquiring an original fundus image, wherein the original fundus image is a fundus image including fused image data of the optic cup and the optic disc.

[0072] Specifically, the original fundus image is the fusion image data of the optic cup and the optic disc, which is an unprocessed image of the fundus. The original fundus image can be a fundus image in a historical database or a fundus image disclosed on a public network. This embodiment does not specifically limit it, and the number and type of original fundus images can be increased or decreased according to actual training needs.

[0073] Step S200 , performing feature region extraction processing on the original fundus image to obtain a candidate region image.

[0074] Specifically, the original fundus image can be first subjected to a first extraction process through a feature extraction module, that is, the category to which the original fundus image belongs is identified through a feature extraction module composed of multiple convolutional layers, thereby obtaining an underlying feature image, and then the underlying feature image is subjected to a second extraction process through a region extraction module to obtain a candidate region image, that is, region extraction is performed on the underlying feature image to obtain a candidate region image of interest for subsequent training.

[0075] In step S300, the candidate region image is input into the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch for training processing to obtain a target neural network module, wherein the optic disc detection network branch is used to determine the optic disc position information in the candidate region image, the cup-disc segmentation network branch is used to segment the candidate region image and output a optic cup-optic disc segmentation image, and the quality scoring network branch is used to perform quality scoring on the optic cup-optic disc segmentation image output by the cup-disc segmentation network branch.

[0076] Specifically, after the candidate area image is extracted, the candidate area image is respectively input into the optic disc detection network branch, the cup-disc segmentation network branch and the quality scoring network branch for training processing. According to the training results, the parameters in the optic disc detection network branch, the cup-disc segmentation network branch and the quality scoring network branch are adjusted and processed respectively to obtain the target neural network module. The optic disc detection network branch is used to determine the optic disc position information in the candidate area image, the cup-disc segmentation network branch is used to segment the candidate area image and output the optic cup-optic disc segmentation image, and the quality scoring network branch is used to perform quality scoring on the optic cup-optic disc segmentation image output by the cup-disc segmentation network branch.

[0077] Step S400: inputting the original fundus image into the target neural network module to obtain the target optic cup and optic disc segmentation image output by the target neural network module.

[0078] In one embodiment, a raw fundus image requiring optic cup and disc segmentation is received. The raw fundus image includes fused image data of the optic cup and disc. The purpose of this technical solution is to segment the optic cup and disc in the raw fundus image. First, feature region extraction is performed on the raw fundus image to obtain candidate region images. The candidate region images are then input into the optic disc detection network branch, the cup and disc segmentation network branch, and the quality scoring network branch for training, respectively, to obtain a target neural network module. The raw fundus image is then input into the target neural network module to obtain a target cup and disc segmentation image output by the target neural network module. In the technical solution of this embodiment, an improved neural network is used to simultaneously perform optic disc region detection and cup and disc segmentation. The quality of the cup and disc segmentation is scored and evaluated by the quality scoring branch to control the output standard, thereby optimizing the cup and disc segmentation process and improving the accuracy of the segmentation.

[0079] Reference Figure 3 In one embodiment, step S300 includes but is not limited to step S310, step S320, step S330, step S340, step S350 and step S360.

[0080] Step S310: Input the candidate region image into the optic disc detection network branch to determine optic disc location information, where the optic disc location information includes category information and bounding box coordinate information;

[0081] Step S320 , calculating the category information and the bounding box coordinate information using a first loss function and a second loss function respectively to obtain a first loss function value and a second loss function value;

[0082] Step S330, inputting the candidate region image into the cup-disc segmentation network branch for segmentation processing to obtain a cup-disc segmentation image, and calculating the cup-disc segmentation image through a third loss function to obtain a third loss function value;

[0083] Step S340: Inputting the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate region image and the cup-disc segmentation image into the quality scoring network branch to obtain a quality score value, and calculating the quality score value using a fourth loss function to obtain a fourth loss function value;

[0084] Step S350: Input the first loss function value, the second loss function value, the third loss function value, and the fourth loss function value into the fifth loss function to calculate a comprehensive loss function value;

[0085] Step S360 , iteratively training the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch according to the comprehensive loss function value to obtain a target neural network module.

[0086] In one embodiment, referring to Figure 4 , Figure 4This is the improved Mask-RCNN network framework diagram. The Mask-RCNN network framework includes three network branches: the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch. The optic disc detection network branch consists of convolutional layers and fully connected layers and is used to determine the optic disc's location. The cup-disc segmentation network branch consists of multiple convolutional layers and is used to accurately segment the optic cup and optic disc in the candidate region image. The quality scoring network branch is composed of multiple convolutional layers and multiple fully connected layers. The quality scoring network branch is used to score the feature data output by the first convolutional layer of the cup-disc segmentation branch and the cup-disc segmentation image, and outputs the cup segmentation results and the quality score of the optic disc segmentation results. This Mask-RCNN network can simultaneously achieve optic disc region detection, cup-disc segmentation, and quality score assessment. This technical solution can break through the existing two-stage training method and adopt a network framework to realize end-to-end training of optic disc detection and cup-disc segmentation. That is, the target neural network module is obtained by adopting a network framework to realize end-to-end training of optic disc detection and cup-disc segmentation, which effectively simplifies the overall process of cup-disc segmentation and ensures the credibility of the segmentation results through quality scoring network branch evaluation, so that the output target optic cup and optic disc segmentation image can help guide the subsequent auxiliary diagnosis of glaucoma.

[0087] It should be noted that the video disc detection network branch may include 1 convolutional layer and 2 fully connected layers, or may include 2 convolutional layers and 2 fully connected layers. This embodiment does not limit the number of convolutional layers and fully connected layers, and can be set according to actual needs.

[0088] It should be noted that the cup-and-disc segmentation network branch may include four or five convolutional layers. This embodiment does not limit the number of convolutional layers and can be set according to actual needs. When the cup-and-disc segmentation network branch has four convolutional layers, the data input to the quality scoring network branch can be the combination of the feature data output by the first convolutional layer of the cup-and-disc segmentation network branch after the candidate region image passes through the four convolutional layers of the cup-and-disc segmentation network branch, and the cup-and-disc segmentation image output by the candidate region image after passing through the four convolutional layers of the cup-and-disc segmentation network branch.

[0089] It should be noted that the quality scoring network branch may include 2 convolutional layers and 2 fully connected layers, or may include 3 convolutional layers and 2 fully connected layers. This embodiment does not limit the number of convolutional layers and fully connected layers, and can be set according to actual needs.

[0090] An embodiment of step S310 and step S320, referring to Figure 4The optic disc detection network branch of the candidate area image is input into the optic disc detection network branch to determine the optic disc position information. The optic disc position information includes category information and bounding box coordinate information. For example, the output optic disc position information can be [x, y, w, h, c], where [x, y] represents the coordinates of the upper left corner of the bounding box, [w, h] represents the width and height of the bounding box, and c represents the category (0 represents non-optical disc, 1 represents optic disc). The category information and bounding box coordinate information are respectively passed through the first loss function L cls and the second loss function L box Calculate the first loss function value and the second loss function value, where L cls Indicates whether it is correctly identified as a optic disc, using binary cross entropy loss, L box Indicates the accuracy of bounding box positioning, using L1 regression loss function. cls and L box The mathematical formula is as follows:

[0091]

[0092]

[0093] Among them, [x,y,w,h,c] represents the real label, Represents the output of the model.

[0094] It should be noted that the first loss function may adopt a binary cross entropy loss function or other loss functions, and this embodiment does not make a sole limitation thereto.

[0095] It should be noted that the second loss function may adopt the L1 regression loss function or other loss functions, and this embodiment does not make a sole limitation thereto.

[0096] An embodiment of step S330, referring to Figure 4 The optic cup is the white area, and the optic disc is the sum of the gray area and the white area. It can be seen that the optic cup is located inside the optic disc. This branch inputs the candidate area image into the cup-disc segmentation network branch for segmentation processing to obtain the cup-disc segmentation image, and calculates the cup-disc segmentation image through the third loss function to obtain the third loss function value. The third loss function L mask The Dice Loss function is used for calculation. mask The mathematical formula is as follows:

[0097] L mask =0.5*(1-Dice_coeff(pred 视盘 ,mask 视盘 ))+0.5*(1-Dice_coeff(pred 视杯 ,mask视杯 ))

[0098] Among them, pred 视盘 and pred 视杯 is the output of the model, mask 视盘 and mask 视杯 is the real label.

[0099] An embodiment of step S340, referring to Figure 4 The feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate region image and the cup-disc segmentation image are input into the quality scoring network branch to obtain the quality score value, and the quality score value is converted into the quality score value by the fourth loss function L score The fourth loss function value of this branch is calculated. score Using binary cross entropy, the input of the fourth loss function is the quality score of the network output and the Dice score between the mask output by the segmentation branch and the standard mask. score The mathematical formula is as follows:

[0100] L score =0.5*Cross_entropy(score 视盘 ,Dice_coeff(pred 视盘 ,mask 视盘 ))+Cross_entropy(score 视杯 ,Dice_coeff(pred 视杯 ,mask 视杯 ))

[0101] In one embodiment of step S350, the three branches of the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch are optimized simultaneously during the training phase. The fifth loss function L used in the training process is final The first loss function value, the second loss function value, the third loss function value and the fourth loss function value are calculated to obtain the comprehensive loss function value, and the fifth loss function L final for:

[0102] L final =L cls +L box +L mask +L score

[0103] In one embodiment of step S360, the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch are iteratively trained using a stochastic gradient descent optimization algorithm. The optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch corresponding to the minimum value of the comprehensive loss function during the iterative training process are determined as the target optic disc detection network branch, the target cup-disc segmentation network branch, and the target quality scoring network branch; and the target neural network module is obtained based on the target optic disc detection network branch, the target cup-disc segmentation network branch, and the target quality scoring network branch. That is, training data is obtained from a public dataset, and the model can be iteratively trained using SGD optimization. The comprehensive loss function L is selected. final The smallest model is taken as the optimal model, L final The smaller it is, the better the overall accuracy in classification, positioning, segmentation, etc.

[0104] It is understandable that during the training phase or in actual application, cup-disc segmentation results with a quality score greater than 0.5 are considered to have high credibility, and the calculated cup-disc ratio can be used to assist in the subsequent diagnosis of glaucoma.

[0105] Reference Figure 5 In one embodiment, step S400 includes but is not limited to step S510, step S520 and step S530.

[0106] Step S510, inputting the original fundus image into the target optic disc detection network branch in the target neural network module to determine the target optic disc position information;

[0107] Step S520, inputting the original fundus image into the target cup-disc segmentation network branch in the target neural network module to obtain a cup-disc segmentation image;

[0108] Step S530 : performing positioning processing on the cup-disc segmentation image according to the target optic disc position information to obtain the target optic cup-optic disc segmentation image.

[0109] Specifically, after the optic disc detection network branch, cup-disc segmentation network branch, and quality scoring network branch are trained according to the comprehensive loss function to obtain the target neural network module, there is no need to adjust the parameters in the optic disc detection network branch, cup-disc segmentation network branch, and quality scoring network branch. At this time, the original fundus image can be input into the target optic disc detection network branch in the target neural network module to determine the target optic disc position information. At the same time, the original fundus image can be input into the target cup-disc segmentation network branch in the target neural network module to obtain a cup-disc segmentation image. Finally, the cup-disc segmentation image is positioned according to the target optic disc position information to obtain a target cup-disc segmentation image. Since the target neural network module is obtained by implementing end-to-end training of optic disc detection and cup-disc segmentation using a network framework, the overall process of cup-disc segmentation is effectively simplified, and the credibility of the segmentation result is ensured by evaluation through the quality scoring network branch, so that the output target cup-disc segmentation image can help guide subsequent auxiliary diagnosis of glaucoma.

[0110] Based on the above-mentioned optic cup and optic disc segmentation method, various embodiments of the optic cup and optic disc segmentation device, computer, and computer-readable storage medium of the present invention are respectively proposed below.

[0111] Reference Figure 6 One embodiment of the present invention further provides an optic cup and optic disc segmentation device, comprising:

[0112] An acquisition module 610 is configured to acquire an original fundus image, wherein the original fundus image is a fundus image including fusion image data of the optic cup and the optic disc;

[0113] Extraction module 620, used for performing feature region extraction processing on the original fundus image to obtain a candidate region image;

[0114] a training module 630 for inputting the candidate region image into the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch for training processing to obtain a target neural network module, wherein the optic disc detection network branch is used to determine the optic disc position information in the candidate region image, and the quality scoring network branch is used to perform quality scoring on the optic cup and optic disc segmentation image output by the cup-disc segmentation network branch;

[0115] The output module 640 is used to input the original fundus image into the target neural network module to obtain the target optic cup and optic disc segmentation image output by the target neural network module.

[0116] In one embodiment, the extraction module 620 is further configured to perform a first extraction process on the original fundus image through the feature extraction module to obtain an underlying feature image; and perform a second extraction process on the underlying feature image through the region extraction module to obtain a candidate region image.

[0117] In one embodiment, the training module 630 is further used to input the candidate area image into the optic disc detection network branch to determine the optic disc position information, where the optic disc position information includes category information and bounding box coordinate information; the category information and bounding box coordinate information are calculated using a first loss function and a second loss function respectively to obtain a first loss function value and a second loss function value; the candidate area image is input into the cup-disc segmentation network branch for segmentation processing to obtain a cup-optic disc segmentation image, and the cup-optic disc segmentation image is calculated using a third loss function to obtain a third loss function value; the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate area image and the cup-optic disc segmentation image are input into the quality scoring network branch to obtain a quality score value, and the quality score value is calculated using a fourth loss function to obtain a fourth loss function value; the first loss function value, the second loss function value, the third loss function value and the fourth loss function value are input into the fifth loss function to obtain a comprehensive loss function value; and the optic disc detection network branch, the cup-disc segmentation network branch and the quality scoring network branch are iteratively trained according to the comprehensive loss function value to obtain a target neural network module.

[0118] In one embodiment, the training module 630 is further configured to perform calculations on the optic cup region and the optic disc region in the cup-optic disc segmentation image, the optic cup mask and the optic disc mask, respectively, to obtain a third loss function value.

[0119] In one embodiment, the training module 630 is further used to input the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate area image and the cup-disc segmentation image into the quality scoring network branch to obtain the optic disc score value and the optic cup score value; the optic disc score value, the optic cup score value, the optic cup area, the optic disc area, the optic cup mask and the optic disc mask are calculated through the fourth loss function to obtain the fourth loss function value.

[0120] In one embodiment, the training module 630 is further used to perform iterative training on the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch, respectively, and determine the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch corresponding to the minimum value of the comprehensive loss function value during the iterative training process as the target optic disc detection network branch, the target cup-disc segmentation network branch, and the target quality scoring network branch; and obtain the target neural network module based on the target optic disc detection network branch, the target cup-disc segmentation network branch, and the target quality scoring network branch.

[0121] In one embodiment, the output module 640 is further used to input the original fundus image into the target optic disc detection network branch in the target neural network module to determine the target optic disc position information; input the original fundus image into the target cup-disc segmentation network branch in the target neural network module to obtain a cup-disc segmentation image; and perform positioning processing on the cup-disc segmentation image according to the target optic disc position information to obtain a target cup-optic disc segmentation image.

[0122] It should be noted that the technical means used, technical problems solved, and technical effects achieved in the various embodiments of the above-mentioned optic cup and optic disc segmentation device are consistent with those in the embodiments of the optic cup and optic disc segmentation method, and are not described in detail here. For details, please refer to the embodiments of the optic cup and optic disc segmentation method.

[0123] In addition, one embodiment of the present invention provides a computer, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor.

[0124] The processor and the memory may be connected via a bus or other means.

[0125] It should be noted that the computer in this embodiment may include: Figure 1 The memory and processor in the embodiment shown can constitute Figure 1 Part of the system architecture platform in the illustrated embodiment, both belong to the same inventive concept, so both have the same implementation principles and beneficial effects, and will not be described in detail here.

[0126] The non-transitory software program and instructions required to implement the optic cup and optic disc segmentation method of the above embodiment are stored in the memory. When executed by the processor, the optic cup and optic disc segmentation method of the above embodiment is executed, for example, the above-described Figure 2 Method steps S100 to S400, Figure 3 Method steps S310 to S360 and Figure 5 Method steps S510 to S530 in .

[0127] The computer includes components such as a radio frequency (RF) circuit, a memory, an input unit, a display unit, a sensor, an audio circuit, a wireless fidelity (WiFi) module, a processor, and a power supply. Those skilled in the art will appreciate that this embodiment does not limit the structure of the computer to a single component, and the computer may include more or fewer components than those in this embodiment, or may combine certain components or arrange the components differently.

[0128] RF circuits can be used to receive and send signals during information transmission or calls. In particular, after receiving the downlink information from the base station, it is sent to the processor for processing; in addition, the designed uplink data is sent to the base station. Generally, the RF circuit includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (LNA), a duplexer, etc. In addition, the RF circuit can also communicate with the network and other devices through wireless communication. The above-mentioned wireless communication can use any communication standard or protocol, including but not limited to the Global System of Mobile communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, Short Messaging Service (SMS), etc.

[0129] The memory can be used to store software programs and modules. The processor executes the various functional applications and data processing of the computer by running the software programs and modules stored in the memory. The memory can mainly include a program storage area and a data storage area. The program storage area can store an operating system, at least one application required for a function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created based on the use of the computer (such as audio data, a phone book, etc.). In addition, the memory can include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0130] The input unit can be used to receive input digital or character information, and to generate key signal input related to the computer's settings and function control. Specifically, the input unit may include a touch panel and other input devices. A touch panel, also known as a touch screen, can collect touch operations on or near it (such as operations performed on or near the touch panel using any suitable object or accessory such as a finger, stylus, etc.) and drive the corresponding connection device according to a pre-set program. Optionally, the touch panel may include two parts: a touch detection device and a touch controller. The touch detection device detects the touch direction and detects the signal generated by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device and converts it into touch point coordinates, which are then sent to the processor, and can receive and execute commands sent by the processor. In addition, the touch panel can be implemented using a variety of types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch panel, the input unit may also include other input devices. Specifically, other input devices may include, but are not limited to, one or more of a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, a joystick, etc.

[0131] The display unit can be used to display input information or provided information and various menus of the computer. The display unit may include a display panel, and optionally, the display panel may be configured in the form of a liquid crystal display (Liquid Crystal Display, LCD), an organic light-emitting diode (Organic Light-Emitting Diode, OLED), etc. Further, the touch panel may cover the display panel, and when the touch panel detects a touch operation on or near it, it is transmitted to the processor to determine the category of the touch event, and then the processor provides a corresponding visual output on the display panel according to the category of the touch event. Although the touch panel and the display panel are two independent components to realize the input and output functions of the computer, in some embodiments, the touch panel and the display panel can be integrated to realize the input and output functions of the computer.

[0132] The computer may also include at least one sensor, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor may adjust the brightness of the display panel according to the brightness of the ambient light, and the proximity sensor may turn off the display panel and / or backlight when the computer is moved to the ear. As a type of motion sensor, the accelerometer sensor can detect the magnitude of acceleration in all directions (generally three axes), and can detect the magnitude and direction of gravity when stationary. It can be used for applications that segment computer posture (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration segmentation related functions (such as pedometer, tapping), etc.; as for other sensors that the computer can also be configured with, such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., they will not be described here.

[0133] Audio circuits, speakers, and microphones provide an audio interface. The audio circuit converts received audio data into electrical signals and transmits them to the speaker, which then converts them into sound signals for output. The microphone, on the other hand, converts collected sound signals into electrical signals, which are then received by the audio circuit and converted into audio data. This audio data is then processed by the processor and then transmitted via the RF circuit to, for example, another computer, or stored in a memory for further processing.

[0134] WiFi is a short-range wireless transmission technology. Computers can use WiFi modules to send and receive emails, browse the web, and access streaming media, providing wireless broadband internet access. WiFi modules are not essential components of computers and can be omitted as needed without changing the essence of the invention.

[0135] The processor is the control center of a computer, connecting all parts of the computer using various interfaces and circuits. By running or executing software programs and / or modules stored in memory and accessing data stored in memory, it performs various computer functions and processes data, thereby providing overall computer monitoring. Optionally, the processor may include one or more processing units; preferably, the processor may integrate an application processor and a modem processor, where the application processor primarily handles the operating system, user interface, and application programs, while the modem processor primarily handles wireless communications. It is understood that the modem processor may not be integrated into the processor.

[0136] The computer also includes a power source (such as a battery) for supplying power to various components. Preferably, the power source can be logically connected to the processor through a power management system, thereby managing charging, discharging, and power consumption through the power management system.

[0137] Although not shown, the computer may also include a camera, a Bluetooth module, etc., which will not be described in detail here.

[0138] In this embodiment, the processor included in the terminal device is capable of executing the pedestrian re-segmentation method of the previous embodiment.

[0139] In addition, an embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are used to execute the above-mentioned terminal-side optic cup and optic disc segmentation method, for example, executing the above-described Figure 2 Method steps S100 to S400, Figure 3 Method steps S310 to S360 and Figure 5 Method steps S510 to S530 in .

[0140] Those skilled in the art will appreciate that all or some of the steps and systems in the method disclosed above can be implemented as software, firmware, hardware, and appropriate combinations thereof. Some physical components or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, and the computer-readable medium can include computer storage media (or non-transitory media) and communication media (or temporary media). As known to those skilled in the art, the term computer storage media is included in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data) and is volatile and non-volatile, removable, and non-removable. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory, or other memory technology, CD-ROM, digital versatile disks (DVD), or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage, or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, as is well known to those skilled in the art, communication media typically includes computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0141] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above implementation. Those skilled in the art can also make various equivalent modifications or substitutions under the shared conditions that do not violate the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A method for segmenting an optic cup and an optic disc, characterized in that: include: Acquire an original fundus image, wherein the original fundus image is a fundus image including fusion image data of the optic cup and the optic disc; Performing feature region extraction processing on the original fundus image to obtain a candidate region image; Inputting the candidate region image into the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch for training processing, respectively, to obtain a target neural network module, wherein the optic disc detection network branch is used to determine the optic disc position information in the candidate region image, the cup-disc segmentation network branch is used to segment the candidate region image and output a optic cup-optic disc segmentation image, and the quality scoring network branch is used to perform quality scoring on the optic cup-optic disc segmentation image output by the cup-disc segmentation network branch; Inputting the original fundus image into the target neural network module to obtain a target optic cup and optic disc segmentation image output by the target neural network module; The candidate area image is respectively inputted into the optic disc detection network branch, the cup-disc segmentation network branch and the quality scoring network branch for training processing to obtain a target neural network module, including: Inputting the candidate area image into a visor detection network branch to determine visor location information, wherein the visor location information includes category information and bounding box coordinate information; Calculating the category information and the bounding box coordinate information using a first loss function and a second loss function respectively to obtain a first loss function value and a second loss function value; Inputting the candidate area image into the cup-disc segmentation network branch for segmentation processing to obtain a cup-disc segmentation image, and calculating the cup-disc segmentation image through a third loss function to obtain a third loss function value; Inputting the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate region image and the cup-disc segmentation image into the quality scoring network branch to obtain a quality score value, and calculating the quality score value through a fourth loss function to obtain a fourth loss function value; Inputting the first loss function value, the second loss function value, the third loss function value and the fourth loss function value into a fifth loss function to calculate a comprehensive loss function value; The optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch are iteratively trained according to the comprehensive loss function value to obtain the target neural network module.

2. The optic cup and optic disc segmentation method according to claim 1, characterized in that: The step of calculating the cup disc segmentation image through a third loss function to obtain a third loss function value includes: The optic cup region and the optic disc region in the cup-optic disc segmentation image are respectively subjected to calculation processing with the optic cup mask and the optic disc mask to obtain a third loss function value.

3. The optic cup and optic disc segmentation method according to claim 2, characterized in that: The step of inputting the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate region image and the cup-disc segmentation image into the quality scoring network branch to obtain a quality score value, and calculating the quality score value through a fourth loss function to obtain a fourth loss function value includes: Inputting the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate area image and the cup-disc segmentation image into the quality scoring network branch to obtain an optic disc score value and a optic cup score value; The optic disc score value, the optic cup score value, the optic cup area, the optic disc area, the optic cup mask, and the optic disc mask are calculated using a fourth loss function to obtain a fourth loss function value.

4. The optic cup and optic disc segmentation method according to claim 1, characterized in that: The iterative training of the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch according to the comprehensive loss function value to obtain the target neural network module includes: Iteratively training the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch respectively using a stochastic gradient descent optimization algorithm, and determining the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch corresponding to the minimum value of the comprehensive loss function during the iterative training as the target optic disc detection network branch, the target cup-disc segmentation network branch, and the target quality scoring network branch; The target neural network module is obtained according to the target optic disc detection network branch, the target cup-disc segmentation network branch and the target quality scoring network branch.

5. The optic cup and optic disc segmentation method according to claim 4, characterized in that: The step of inputting the original fundus image into the target neural network module to obtain a target optic cup and optic disc segmentation image output by the target neural network module comprises: Inputting the original fundus image into the target optic disc detection network branch in the target neural network module to determine the target optic disc position information; Inputting the original fundus image into the target cup-disc segmentation network branch in the target neural network module to obtain a cup-disc segmentation image; Positioning processing is performed on the cup-disc segmentation image according to the target optic disc position information to obtain a target optic cup-disc segmentation image.

6. The optic cup and optic disc segmentation method according to claim 1, characterized in that: The step of performing feature region extraction processing on the original fundus image to obtain a candidate region image includes: Performing a first extraction process on the original fundus image by a feature extraction module to obtain an underlying feature image; The region extraction module performs a second extraction process on the underlying feature image to obtain a candidate region image.

7. A device for separating optic cup and optic disc, characterized in that: include: An acquisition module, configured to acquire an original fundus image, wherein the original fundus image is a fundus image including fusion image data of the optic cup and the optic disc; An extraction module, configured to perform feature region extraction processing on the original fundus image to obtain a candidate region image; a training module, configured to input the candidate region image into the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch for training processing, respectively, to obtain a target neural network module, wherein the optic disc detection network branch is configured to determine optic disc position information in the candidate region image, the cup-disc segmentation network branch is configured to segment the candidate region image and output a optic cup-optic disc segmentation image, and the quality scoring network branch is configured to perform quality scoring on the optic cup-optic disc segmentation image output by the cup-disc segmentation network branch; The candidate area image is respectively inputted into the optic disc detection network branch, the cup-disc segmentation network branch and the quality scoring network branch for training processing to obtain a target neural network module, including: Inputting the candidate area image into a visor detection network branch to determine visor location information, wherein the visor location information includes category information and bounding box coordinate information; Calculating the category information and the bounding box coordinate information using a first loss function and a second loss function respectively to obtain a first loss function value and a second loss function value; Inputting the candidate area image into the cup-disc segmentation network branch for segmentation processing to obtain a cup-disc segmentation image, and calculating the cup-disc segmentation image through a third loss function to obtain a third loss function value; Inputting the feature data output by the first convolutional layer of the cup-disc segmentation network branch of the candidate region image and the cup-disc segmentation image into the quality scoring network branch to obtain a quality score value, and calculating the quality score value through a fourth loss function to obtain a fourth loss function value; Inputting the first loss function value, the second loss function value, the third loss function value and the fourth loss function value into a fifth loss function to calculate a comprehensive loss function value; Iteratively training the optic disc detection network branch, the cup-disc segmentation network branch, and the quality scoring network branch according to the comprehensive loss function value to obtain the target neural network module; The output module is used to input the original fundus image into the target neural network module to obtain the target optic cup and optic disc segmentation image output by the target neural network module.

8. A computer comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for segmenting the optic cup and optic disc according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the optic cup and optic disc segmentation method according to any one of claims 1 to 6.

Images