30 results about "Optic cup/disc ratio" patented technology

The invention discloses a self-adaption eye fundus image processing method and device based on a passive field. The method comprises the following steps: acquiring an eye fundus image with a label in a source domain and an eye fundus image without a label in a target domain, and preprocessing the eye fundus image; training a teacher model and optimizing model parameters; using teacher model parameters to initialize two student models with the same structure; inputting the target domain unmarked fundus image into the teacher model and the student model for pixel-level learning; constructing category prototypes of optic cups and optic disks based on high-confidence features extracted by the teacher model, and guiding student model features to be aligned with the category prototypes; calculating the prediction difference of the teacher model and the student model in the boundary region, and correcting the segmentation result of the fuzzy boundary; constructing a total loss function, and iteratively updating teacher model parameters to obtain a fundus image segmentation model; and performing optic cup and optic disk segmentation on the target domain eye fundus image by using the eye fundus image segmentation model, and performing quantitative analysis. According to the method, the generalization and segmentation performance of the model on the target domain is remarkably improved.

A multivariable artificial intelligence-based monitoring system for early detection of glaucoma and method thereof. The multivariable artificial intelligence system comprises a computing device having a control unit and one or more non-transitory storage devices for storing instructions to be executed by the control unit. The computing device is in communication with an application server via a network. The computing device includes an input module, an image enhancing module, a feature extraction module, a post image processing module, and a parameter selection module. The proposed multivariable artificial intelligence system provides a deep learning architecture to segment the optic disc and optic cup in two ways using two different networks termed Multi Spatial Attention Feature Fusion Network (MSAFF-Net) and Multi Dilated Edge Extraction Network (MDEE-Net) respectively.

The invention discloses a difficulty perception-based passive domain adaptive fundus image segmentation method and device, and belongs to the technical field of computer vision processing, and the method comprises the steps: obtaining a target domain fundus image, and carrying out the preprocessing of the target domain fundus image; initializing a teacher model and two student models; performing weak enhancement processing on the preprocessed image, inputting the processed image into a teacher model, obtaining a prediction probability, calculating an entropy value of each sample, and dividing the samples into an easy sample set and a difficult sample set according to the entropy values; strong enhancement processing is carried out on the easy sample set and the difficult sample set, and then the easy sample set and the difficult sample set are respectively input into student models for training; external information is injected into the two student models respectively, and the training process is optimized; after each iteration period is finished, the parameters of the teacher model are updated based on the parameters of the two student models until the teacher model converges; and inputting a to-be-segmented eye fundus image into the converged teacher model to obtain a segmentation result of the eye fundus image. According to the invention, high-precision segmentation of the optic cup and optic disc of the eye fundus image is realized.

The invention discloses a passive domain field adaptive eye fundus image segmentation method based on clustering filter pseudo label optimization. The method comprises the following steps: step 1, student network eye fundus optic cup and optic disc segmentation; 2, segmenting a teacher network fundus optic cup and an optic disk; step 3, clustering a filter; step 4, performing pseudo label optimization based on a clustering filter; 5, calculating a student network segmentation loss function and updating parameters; step 6, updating teacher network parameters; and step 7, calculating a weighted Dice loss function. According to the method, the false label generated by the teacher model is optimized, the stability of the false label is enhanced, a weighted Dice loss function is introduced, the model is forced to pay attention to a sparse but important target area, the deviation caused by class imbalance is reduced, and higher-precision fundus optic disc and optic cup segmentation is realized.

The disclosure provides a disc optic cup segmentation method, a model training method, an apparatus, a device and a medium, relating to the technical fields of artificial intelligence and intelligent medicine. The specific implementation scheme is as follows: the disc optic cup segmentation method comprises the following steps: obtaining a fundus image; based on an attention mechanism, the disc and the optic cup of the fundus image are segmented to obtain a segmented image of the disc and the optic cup. According to the disclosure, the disc optic cup segmentation efficiency can be effectively improved. In addition, the disclosure also provides a training method of a disc optic cup segmentation model, which can effectively improve the accuracy of the trained disc optic cup segmentation model.

The invention discloses a pathological myopia eye bottom image data amplification method based on deep learning, and the method comprises the steps: generating high-quality eye bottom image data through an improved positioning strategy and a vector quantization variational auto-encoder (VQ-VAE) model; the method comprises the steps of using a UNet model to position an optic cup, an optic disk and a macular of a healthy fundus image, calculating relative positions between a focus and the optic cup and the optic disk, and accurately pasting a focus area to a corresponding position of the healthy fundus image. A fundus image is coded and decoded through a VQ-VAE model, a high-quality synthetic image is generated, brand new focus features are generated in combination with a GPT model, and the diversity of a data set is enhanced. According to the method, the quality and diversity of pathological myopia eye bottom image data amplification can be effectively improved, and diagnosis and treatment of pathological myopia are supported.

The invention discloses an optic disc and optic cup segmentation method based on semi-supervised learning, belongs to the field of image processing and artificial intelligence, and aims to solve the problems of instability of an optic disc and optic cup segmentation model caused by few labeled eye fundus images and poor precision caused by noise pollution. In the first stage, a double-branch improved DeepLab v3 + model is constructed to complete optic disc segmentation, after full supervision and anti-noise training are carried out by utilizing an annotated fundus image, the double-branch model is used for processing an unannotated fundus image, the anti-noise capability of the model is evaluated by counting the consistency of pseudo-label pixels before and after noise, and the model with strong anti-noise performance is used for supervising the training of the other model; in the second stage, a SegFormer model is constructed and trained to complete optic cup segmentation; and finally connecting the two models in series to realize two-stage segmentation of the optic disc and the optic cup. According to the method, unlabeled data is fully utilized for semi-supervised training, the noise immunity and segmentation precision of the model are improved, and technical support is provided for glaucoma screening work.

This invention provides a fundus image classification system and storage medium for chronic kidney disease detection, addressing the technical problems of low efficiency and large errors in existing fundus image classification methods. The system includes: acquiring and preprocessing fundus images to generate standardized fundus images; dividing the processed fundus images into training and testing sets, and augmenting the training set; preprocessing the training and testing sets to ensure uniform class distribution in each training batch input to the model during training; integrating learning strategies and training branched neural network models; and fusing the models to obtain the final detection model, achieving fundus image classification. This invention offers high efficiency and excellent automation; it extracts features such as the optic cup / disc ratio and arteriovenous ratio from fundus images and combines them with machine learning methods for classification. This invention also boasts high computational speed and low computer resource consumption during operation.

This invention presents a passive domain adaptive fundus segmentation method based on evidence representation and marginal screening. Under the condition that the source domain training images and their pixel-level annotations are inaccessible, a Student network and a Teacher network are constructed to learn segmentation of the target domain fundus image. The network outputs positive foreground evidence and negative background evidence for each segmentation channel, and constructs a pixel-level Beta distribution. Based on the Beta distribution, foreground probability, evidence margin, and evidence strength are calculated. Reliable positive pseudo-labels are obtained through joint screening, and a weighted segmentation loss is constructed by combining pseudo-label weights. Simultaneously, regularization loss and evidence strength constraint loss are introduced to optimize the Student network. Finally, the exponential moving average coefficient is dynamically adjusted based on the amount of evidence to update the Teacher network parameters. This method can effectively suppress the propagation of pseudo-label noise and improve the accuracy, stability, and cross-domain adaptability of optic cup and optic disc segmentation in the target domain fundus image.

The disclosure describes a neural network-based glaucoma recognition device and method, which includes an input unit, a preprocessing unit, a segmentation unit, a feature extraction unit, and a classification unit. The input unit is used to receive a first image. The preprocessing unit is used to preprocess the first image to obtain a first preprocessed image. The segmentation unit is used to input the first preprocessed image into an artificial neural network based on deep learning to generate an optic disc region image and an optic cup region image. The feature extraction unit obtains multiple glaucoma features based on the optic disc region image and the optic cup region image corresponding to the first preprocessed image. The classification unit is used to input feature information including glaucoma features corresponding to the first preprocessed image into a classifier based on machine learning for classification to obtain a glaucoma classification result. According to the present scheme, the accuracy of glaucoma recognition can be improved.

The present application provides an eye state evaluation method and an electronic device. The method comprises: obtaining an optic disc image region from a first fundus image, and generating a plurality of optic cup disc ratio evaluation results based on the optic disc image region by a plurality of first models; obtaining a first evaluation result of the eye based on the optic cup disc ratio evaluation results; performing a plurality of data augmentation operations on the first fundus image to generate a plurality of second fundus images; generating a plurality of optic nerve fiber layer defect evaluation results based on the second fundus images by a plurality of second models; obtaining a second evaluation result of the eye based on the optic nerve fiber layer defect evaluation results; and obtaining an optic nerve evaluation result of the eye based on the first evaluation result and the second evaluation result. Thus, the present application can provide a reference for doctors when evaluating the eye state of patients, thereby helping doctors to give appropriate evaluation results for the eyes of patients.

The invention relates to the technical field of eye detection, in particular to a fundus data determination method. The method comprises the following steps: performing layering processing on first scanning data containing an optic nerve head, determining an identifiable layering and an inner boundary membrane, determining an optic disc edge of the optic nerve head based on the identifiable layering, and determining the optic disc edge of the optic nerve head based on a surface area formed by a closed curve extending from the optic disc edge to the inner boundary membrane. And determining a closed curve when the surface area meets a preset condition as a target curve, determining an optic cup edge of the optic nerve head based on the target curve, and determining fundus data according to the optic disk edge and the optic cup edge. By adopting the method, the accuracy can be improved.

The invention relates to the technical field of medical image segmentation, in particular to a retina fundus image optic disc optic cup self-adaptive segmentation method and a retina fundus image optic disc optic cup self-adaptive segmentation system. According to the invention, multiple random forward propagation is carried out on the first U-Net to obtain a coarse segmentation mean image and a variance image; before the second U-Net is calculated, performing the following adaptive judgment on the area of the roughly segmented optic disc: if the area is greater than or equal to a preset threshold value, inputting the fundus image and the roughly segmented mean image into the second U-Net to obtain an optic disc and optic cup segmentation result; if the area is smaller than a preset threshold value, a weighted uncertainty graph needs to be further calculated, the graph is weighted superposition of the roughly-segmented optic disc and the high-variance graph, and the weight considers the distance between the roughly-segmented optic disc and the high-variance region and also considers the shapes of the optic disc and the high-variance region. And inputting the weighted uncertainty map, the fundus image and the coarse segmentation mean image into a second U-Net to obtain an optic disc and optic cup segmentation result. Through adaptive judgment, the optic disc and optic cup segmentation precision is effectively improved.

The invention discloses a fundus image optic cup and optic disc segmentation method and device based on edge enhancement and visual features, and belongs to the technical field of image segmentation, and the method comprises the steps: employing an optic disc ROI coarse positioning model to carry out the positioning of an obtained to-be-segmented fundus image, and generating an optic disc ROI image; a Laplace feature map and a semantic vector are extracted from an optic disk ROI image by using an optic cup and optic disk segmentation model, multilayer feature coding is performed on the semantic vector to obtain a coding feature map, edge feature enhancement is performed on the coding feature map of each level according to the Laplace feature map and the decoding feature map of the previous level, and an edge enhancement feature map of each level is generated. And decoding each edge enhanced feature map, a fusion feature map of the coding feature map of the same level and the decoding feature map of the previous level to obtain a decoded feature map, and finally performing pixel classification on the decoded feature map to obtain an optic cup and optic disk segmentation result. According to the method, the segmentation precision of the optic cup and the optic disc is improved while the relatively high calculation efficiency is maintained.

Provided is an image processing system and method for analyzing an optic cup and optic disc. The system and method include the following steps. A processor obtains a fundus image. The processor recognizes the optic disc and optic cup in the fundus image using an image recognition model and marks the outlines of the optic disc and optic cup. An interpretation mode is provided so that interpretation data may be generated for risk assessment according to the outlines of the optic disc and optic cup as marked. The image recognition model is a deep learning model trained by a large amount of pre-collected fundus maps. The outlines of the optic cups and optic discs in the fundus images have been marked in advance.

The present invention proposes a method for glaucoma optic cup and optic disc segmentation with contour injection content, the method comprising: inputting a fundus retinal color image into a coding module for encoding; using a contour decoder to decode the contours of the optic cup and optic disc in stages for multi-level features; using a content decoder to decode the content segmentation of the optic cup and optic disc for multi-level features; using a contour injection module to accept contour decoding features corresponding to different stages of the optic cup and optic disc to perform contour injection on the content decoding features of the first stage of the optic cup and optic disc, and iteratively guide the content decoding of the next stage content decoder to obtain a optic cup and optic disc content segmentation mask map. The present invention is based on an encoder and decoder with a large-size convolution kernel, which is different from the design paradigm of a traditional convolutional network based on a 3×3 window, effectively expanding the receptive field, and can obtain more global features with a computational amount less than that of global self-attention.

The present invention discloses a method for jointly segmenting the optic cup and optic disc in retinal fundus images, belonging to the field of medical image processing. The present invention combines the elliptical morphological features of the optic cup and optic disc to construct a two-stage optic cup and optic disc joint segmentation network model from the perspective of ellipse detection. In the first stage, the optic cup is located within the optic disc region, and the Paired-Box RPN is introduced to achieve coupled detection of the minimum bounding rectangle of the optic cup and optic disc. In the second stage, the five parameters of the ellipse in the bounding box region are learned. Based on the elliptical morphological features and spatial geometric constraints of the optic cup and optic disc, the optic cup and optic disc are jointly segmented using the optic cup and optic disc joint segmentation network model. This method can solve the problem of uneven segmentation edges and achieve accurate optic cup and optic disc segmentation.

The application discloses a kind of dense lesion semi-automatic labeling method for fundus image, method includes constructing and training dense lesion segmentation network, macula fovea positioning network and optic cup optic disc segmentation positioning network;Utilize network to process to obtain dense lesion prediction contour area, macular area and optic cup optic disc area respectively;Three labelers are assigned to fundus image, each labeler selects key area and other area from macular area and optic cup optic disc area and carries out labeling operation, to obtain first round labeling result;According to first round labeling result, the lesion labeling consistency index between every two labelers is calculated to determine two labelers to execute second round lesion labeling, and second round labeling result is obtained;Second round labeling result is audited to obtain final labeling result.Therefore, the problem of large difficulty in dense lesion labeling and large difference in labeling by different personnel is solved, and the labeling efficiency is improved, the labeling time is reduced, and the labeling cost is reduced.

The invention relates to the field of eye fundus color photo optic cup and optic disc segmentation, and provides an eye fundus color photo optic cup and optic disc segmentation method and device, equipment, a medium and a product, and the method comprises the steps: obtaining a to-be-segmented eye fundus color photo; inputting the fundus photo to be segmented into the trained fundus photo optic cup and optic disc segmentation model to obtain a corresponding optic cup and optic disc segmentation result; the fundus color photo optic cup and optic disc segmentation model is constructed by replacing an encoder in a DPT architecture with a DINOv2 encoder. The training process comprises a process of diffusion training of a DINOv2 encoder, a consistency training process combining image-level spatial transformation and feature-level hidden space disturbance, and a process of adaptive optimization of pseudo labels in the consistency training process based on category prototype guidance. According to the method, the utilization rate of limited annotation data is improved, the medical image annotation cost is remarkably reduced, meanwhile, the image segmentation precision is ensured, and reliable technical support can be provided for eye disease diagnosis.

A multivariable artificial intelligence-based monitoring system for early detection of glaucoma and method thereof. The multivariable artificial intelligence system comprises a computing device having a control unit and one or more non-transitory storage devices for storing instructions to be executed by the control unit. The computing device is in communication with an application server via a network. The computing device includes an input module, an image enhancing module, a feature extraction module, a post image processing module, and a parameter selection module. The proposed multivariable artificial intelligence system provides a deep learning architecture to segment the optic disc and optic cup in two ways using two different networks termed multi spatial attention feature fusion network (MSAFF-Net) and multi dilated edge extraction network (MDEE-Net) respectively.

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.

Error correction method for determination of the boundaries of the optic disc and optic cup in fundus and its confirmation terminal, which relates to the technical field of recognition and processing of retinal fundus images. It solves the technical problem that the error of manually segmenting and measuring the boundaries of the optic disc and optic cup is large at present, resulting in a large error in the finally output feature data. The method includes: collecting the retinal fundus image of a patient through a fundus image camera; then, two or more measurement personnel use independent clients at different times to manually annotate the contours of the optic disc and optic cup presented in the same image at different placement angles, so that the contours of the optic disc and optic cup form a closed area; after each annotation is completed, the image analysis algorithm of the adaptive model will analyze the marked data and output the corresponding vertical cup-to-disc ratio, cup-to-disc area ratio, and minimum disc margin distance feature data; finally, the optimal output value is comprehensively determined according to the vertical cup-to-disc ratio, cup-to-disc area ratio, and minimum disc margin distance feature data output each time. It effectively overcomes the deficiency of large error in single-person single-time annotation and improves the accuracy of the determination of the boundaries of the optic disc and optic cup in fundus.

The invention provides a multi-modal retinal disease intelligent auxiliary diagnosis system, which comprises an image preprocessing module used for processing a binocular fundus image uploaded by a user; the disease classification module is used for extracting global features and local lesion details by adopting a dual-channel DINOv2 model, initializing weights in a data set through transfer learning, performing fine adjustment on the fundus image data set, and performing fundus multi-label classification; and the batch processing module is used for disassembling batch binocular eye fundus image processing tasks into independent sub-tasks based on a distributed task scheduling and dynamic resource allocation technology, and distributing the independent sub-tasks to a plurality of edge computing nodes for parallel processing. According to the method, a multi-label independent classifier is designed, eight independent three-layer MLP classifiers are adopted, and each classifier focuses on single pathological feature modeling. Through parameter space decoupling design, gradient conflicts among multiple labels are avoided, and accurate capture of heterogeneity pathologies such as diabetes microvascular leakage characteristics and glaucoma optic cup morphological parameters is ensured.

The invention discloses a method for performing joint segmentation on an optic cup and an optic disk of an eye fundus image based on an MTS-Net neural network combining a VisionTransform and MedNext structure, relates to the field of deep learning and medical image processing, and comprises data processing and segmentation prediction of a visual result network. The overall process is as shown in Figure 1. The method comprises the following steps: step 1, the preprocessing network firstly carries out center cutting, edge cutting and normalization operation on a collected picture; 2, constructing a segmentation model, wherein the segmentation model comprises a deep convolution module, a multi-head self-attention module and a decoding module; wherein the deep convolution module comprises four sub-modules, namely a Depthwise deep convolution module, a feature expansion layer, a channel reduction module and a point-by-point convolution module, the deep convolution module is used as an encoder module of the MTS-Net network, and the deep convolution module extracts features in three stages; the VisionTransform module is used for making up the defect of global feature capture of convolution in the MedNext module in the feature extraction process, and the decoding module uses a structure symmetrical to the encoding module and is mainly used for gradually recovering spatial resolution in three steps and carrying out feature fusion, so that image segmentation is carried out in the later period. And step 3, realizing the eye fundus image optic cup and optic disc segmentation task through the MTS-Net network trained in the step 4. The method is used for solving the problems of prediction errors and too slow training speed of a balance network model in an optic cup and optic disk proportion estimation task in glaucoma screening at the present stage.

The invention discloses a fundus image optic cup and optic disc segmentation method based on improved U-net, and belongs to the technical field of medical image processing. Comprises: acquiring a fundus image data set; an improved U-net network is constructed; a CrossFormer layer is added between the last layer of the encoder and a decoding structure; an ASPP detection head is embedded at the last end of the decoder; the method comprises the following steps of: constructing a compound type loss function in which Dice Loss and Cross Entry Loss are combined; training an improved U-net network by using the fundus image data set; and inputting a fundus image to be segmented into the trained improved U-net network for segmentation to obtain an optic cup and optic disc segmentation result. Based on the improved U-net, the problems that blood vessels penetrate through the optic cup and optic disc area, the edge is fuzzy, the proportion is small, and accurate segmentation is difficult are solved.

The invention discloses a medical image segmentation method, system and device based on an evidence theory, and a storage medium, and belongs to the technical field of medical image segmentation. The method comprises: acquiring a target fundus image; inputting the target fundus image into a pre-trained image segmentation model to obtain a final segmentation result; wherein the training process of the image segmentation model comprises the following steps: acquiring a fundus image data set which comprises fundus images and annotation data of a plurality of experts on an optic disc and an optic cup of the same fundus image; and inputting the fundus image data set into a pre-constructed image segmentation model for training to obtain a final segmentation result of the optic disc and the optic cup. According to the method, through an evidence theory framework and consistency constraint learning, the segmentation performance and the uncertainty quantification capability are remarkably improved.

The invention relates to a passive domain eye fundus image segmentation method based on multi-scale features and stepped attention, and the method specifically comprises the following steps: obtaining an optic cup image and a test disc image to be segmented to construct an image data set, and inputting the image data set to a pseudo tag generation unit and an image segmentation unit; the pseudo-label generation unit selects a DeepLab network as a pre-training model, outputs of the pre-training model pass through a multi-scale feature fusion module, and a fused image segmentation result is used as a pseudo-label of a to-be-detected image; the image segmentation unit adopts an improved DeepLab network, a step attention fusion module is introduced between an encoder and a decoder of the DeepLab network, migration training is carried out on a pre-training model in cooperation with a generated pseudo tag, and a more accurate image segmentation result is generated. According to the invention, migration training is carried out on the DeepLab network based on the generated pseudo tag and the introduced stepped attention fusion module, so that the overall and detail features of the to-be-segmented image can be better learned, and the accuracy of domain adaptation is improved.

The present invention discloses a glaucoma diagnosis method based on deep learning. The method preprocesses an input raw fundus image data set to obtain image data that can be recognized and utilized by model 1; trains model 1 so that it can accurately segment the optic disc and optic cup areas based on the fundus image and outputs a key feature image showing the optic disc and optic cup segmentation; extracts the key feature image of model 1 in the process of processing the fundus image as an input parameter of model 2, and preprocesses the image data to be input into model 2 and converts it into data in a unified format; after receiving the input data of model 1, model 2 is trained and then outputs a target result, which is: glaucoma or non-glaucoma.

The application discloses a kind of based on double-path self-supervision's cross-domain view cup view disc automatic segmentation method and device, method includes: using source domain-target domain and target domain-source domain's two-way fundus image domain transformation, respectively generate target domain style source domain image and source domain style target domain image, the source domain image has segmentation label, target domain image is no segmentation label;Using DeepLabV3+ image segmentation network as framework, MobileNetV2 feature extractor is matched, constructs double-path fundus image segmentation network;The prediction segmentation result of the unsupervised image output by double-path network is fused, and the fusion pseudo label is obtained, and the further training of no label image is guided using fusion pseudo label;Using a cross-domain contrast constraint mechanism, the similarity of image features before and after domain transformation is optimized, so that the information related to image structure is retained in features.The device comprises a memory and a processor.The application improves the accuracy of target domain image prediction segmentation result.

The invention discloses an optic cup optic disk field generalization segmentation method, which comprises the following steps: converting a color space of an eye fundus image into a YUV color space, and carrying out feature extraction to obtain a first low-level feature; performing homomorphic curvature filtering on the first low-level feature to obtain a second low-level feature; performing internal mixing of a mean value and a standard deviation on the first low-level feature to generate a third low-level feature; obtaining a first high-level feature according to the third low-level feature; performing feature extraction and merging on the first advanced features under different scales to obtain second advanced features; adjusting the second advanced feature by using the multi-domain feature and the code of the domain knowledge base to obtain a third advanced feature; and combining the third high-level feature with the second high-level feature and the second low-level feature to generate a final output feature. According to the method, feature extraction is carried out in the YUV color space, feature mixing is carried out on the low-level features, and the low-level features are individually enhanced, so that the diversity of the features is enhanced, detail features of the optic cup and optic disc are focused, and the generalization ability of the model is improved.