13 results about "Retina eye" patented technology

The invention discloses a retina fundus image generation method based on a diffusion model. Firstly, a training data set and a regularization data set are constructed; then, constructing a diffusion model which comprises a variational auto-encoder, a U-Net denoising network and a text encoder; the variational auto-encoder comprises a variational encoder and a variational decoder, and the U-Net denoising network is embedded between the variational encoder and the variational decoder; the structured text prompt passes through a text encoder to obtain a text embedding vector, and the vector is injected into the U-Net denoising network as condition information; the variational encoder compresses the retina fundus image into potential features, the U-Net denoising network carries out denoising on the potential features under the guidance of condition information, and the denoised potential features are reconstructed into a high-resolution retina fundus image through the variational encoder; and finally, generating a retina fundus image based on the pre-trained diffusion model. The controllability of image generation is improved, and the generated image is highly consistent between the focus form and the medical description.

The invention discloses a retina eye movement tracking and OCT (Optical Coherence Tomography) scanning compensation method and system, and the method comprises the steps: firstly dividing a fundus SLO image into a plurality of sub-regions, and calculating the stability index weight of each sub-region, thereby adaptively selecting a stable sub-region which is most suitable for being used as a tracking feature; and performing cross-correlation matching on the selected sub-regions to estimate local displacements, and performing weighted fusion on a plurality of local displacements based on a stability index weight to obtain a global eye motion vector of the current frame. According to the global eye motion vector, the drop point position of an OCT scanning light beam is adjusted in real time, and dynamic compensation of B-scan scanning is achieved. According to the method, stable and accurate retina positioning can be kept under the conditions of eye movement, illumination change and local image degradation, the alignment quality and the scanning definition of the OCT image are improved, automatic positioning of the same retina position can be achieved for examination of the same user at different moments, and the accuracy of retina positioning is improved. Regional repeated scanning can be directly carried out to observe focus changes at the same position, and the optical coherence tomography device is suitable for high-speed OCT imaging and clinical fundus examination.

In order to solve the problems of limited labeled data, differences between blood vessels and interference of lesion area, the present application discloses a retinal blood vessel image segmentation method based on a multi-scale dilated convolution residual network, which is still a challenge problem for accurately segmenting retinal blood vessels, especially fine blood vessels, on a retinal fundus image. The present application designs a multi-scale residual input and output module to make up for the loss of part of the blood vessel structure information due to down-sampling, combines the advantages of dilated convolution and DropBlock to relieve network overfitting and reduce the influence of the lesion area on blood vessel feature extraction, and further introduces a multi-scale mean pooling module to obtain high-level features and retain context information. Finally, by improving the way of jump connection, the dilated convolution is effectively used to improve the information transmission capacity of the jump connection. Compared with other algorithms, the present application can more accurately segment the fine blood vessels in the retinal image under complex conditions and has better robustness.

The present application relates to the technical field of medical image segmentation, and particularly relates to a retinal fundus picture optic disc and cup adaptive segmentation method and system.The present application carries out multiple random forward propagation in the first U-Net to obtain a rough segmentation mean graph and a variance graph.Before calculating the second U-Net, the following adaptive judgment is made on the area of the rough segmentation optic disc: if the area is greater than or equal to a preset threshold, the fundus picture and the rough segmentation mean graph are input into the second U-Net to obtain the optic disc and cup segmentation result;if the area is less than the preset threshold, the weighted uncertainty graph needs to be further calculated, which is the weighted superposition of the rough segmentation optic disc and the high variance graph, and the weight considers both the distance of the rough segmentation optic disc and the high variance area and the shape of the optic disc and the high variance area.The weighted uncertainty graph, the fundus picture and the rough segmentation mean graph are input into the second U-Net to obtain the optic disc and cup segmentation result.Through adaptive judgment, the present application effectively improves the optic disc and cup segmentation precision.

The invention discloses a color retina fundus image enhancement method, which belongs to the technical field of medical image processing, and comprises the following steps: carrying out decomposition processing on an input color retina fundus image by using guide filtering, and separating a noise layer from a structural layer; further decomposing the structural layer into a base layer and a detail layer by using guide filtering again; performing color space conversion on the base layer, converting an RGB space into an HSV space, performing local brightness adaptive method adjustment on a V channel, and converting the HSV space after adjustment of the V channel into the RGB space; carrying out weighted enhancement processing on the detail layer based on a Weber coefficient; and fusing the enhanced detail layer and the base layer after brightness adjustment, and outputting an enhanced color retina fundus image. According to the invention, the color authenticity of the color retina fundus image can be maintained, and the local contrast and detail visibility of the image can be effectively improved.

An image processing system to improve the quality of images via a context-aware OT framework. The system combines a representation of each one of a set of unlabeled low-quality images and corresponding high-quality images as a collection of feature vectors in a deep layer of a neural network, then obtains contextual information, abstracted in the deep layer of the neural network, associated with the representation of each one of low-quality images and the high-quality images. The system then derives a context-aware optimal transport method based on the obtained contextual information, and trains the OT framework, during its training phase, to learn, via application of the context-aware optimal transport method, a mapping between each one of the low-quality images and high-quality images. Thereafter, a low-quality image can be translated, via the trained OT framework, during its inference phase, to a corresponding high-quality image.

The present application relates to a retinal blood vessel segmentation method based on neural ordinary differential equation, which solves the defects of poor scale adaptability and insufficient extraction of fine blood vessel features compared with the prior art. The present application comprises the following steps: obtaining a retinal fundus image and pre-processing; constructing a retinal blood vessel segmentation network; training the retinal blood vessel segmentation network; obtaining the retinal blood vessel segmentation result of the fundus image. The encoder and decoder paths are stacked by the neural ordinary differential equation module, which uses its continuous dynamic characteristics to adaptively adjust the effective receptive field by calculating the feature derivative to adapt to different scale blood vessel features; the mixed attention perception module is located at the jump connection, which adopts the "asymmetric convolution decomposition" architecture, combines channel mixing and multi-scale spatial attention, and accurately captures fine linear blood vessel features while suppressing background noise.

An AI-based retinal image analysis system for the automated detection of diabetic retinopathy, comprising the following: an image acquisition unit consisting of an optical arrangement with a coaxial illumination source, a multi-lens arrangement and an image sensor for capturing retinal images; a preprocessing unit that is operationally coupled with the image acquisition unit and is configured to receive the retinal fundus images and perform image normalization including illumination correction, noise reduction and contrast enhancement; a feature extraction unit comprising at least one processor configured to perform a variety of folding operations on the preprocessed retinal fundus images to generate hierarchical feature representations corresponding to the anatomical structures and pathological regions of the retina; a classification unit comprising at least one processor configured to process the hierarchical feature representations and produce a severity classification output corresponding to the stages of diabetic retinopathy; a storage unit that is operationally coupled to the feature extraction unit and the classification unit and stores the trained parameters assigned to the feature extraction unit and the classification unit; and a display unit configured to show the user the severity classification output and the corresponding diagnostic information.

The application discloses an eye age difference index construction method and system based on fundus images and electronic equipment, and relates to the technical field of image processing. The application trains the constructed double-channel fusion attention network model by adopting a training set and a pairing matrix loss function, can reduce the prediction error between samples in addition to the mean square error, and makes the prediction accuracy of the trained double-channel fusion attention network model further improved. The trained network model is used as a fundus age prediction model to analyze the left and right eye retina fundus images of a to-be-measured object, and the characteristics are fused, so that more accurate fundus age can be obtained, so as to accurately construct the eye age difference index, and then provide a favorable basis for judging the aging track degree of human health.

The invention discloses a microaneurysm segmentation method based on grouping attention and multi-scale depth fusion, and belongs to the technical field of medical image processing. Comprises: obtaining a retina fundus image; the method comprises the following steps: constructing a network model based on packet attention and multi-scale deep fusion, taking a U-Net network as a basis, using ResNet50 as a backbone network of an encoder, embedding a packet multi-scale attention module GMA in a bottleneck block of the ResNet50, and introducing a multi-scale feature transfer module MSDF between the encoder and a decoder to optimize feature transfer; and inputting a to-be-segmented retina fundus image into the trained network model based on grouped attention and multi-scale depth fusion for segmentation to obtain a microaneurysm segmentation result. According to the method, grouping attention and multi-scale depth are fused, and the problems that an MA target is likely to be confused with the background, the edge is fuzzy, the proportion is small, and accurate segmentation is difficult are solved.

The application discloses a kind of based on histogram matching and weighted retinal fundus image splicing fusion method, first the retinal fundus image of different regions of the same mode obtained is histogram matched, for reducing the brightness and contrast difference between image, then the weight matrix and affine matrix of image are sequentially obtained, and the histogram matched image and weight matrix are carried out affine transformation, finally based on the image and weight matrix after affine transformation sequentially carry out the splicing fusion of image.The scheme of the application, when histogram matching is carried out, the pixel distribution histogram of all images is counted as the target of histogram matching, reduces the brightness contrast difference between image, while avoiding the image distortion caused by using specific target;While using weight control the value of pixel when fusion image, using the way of segmentation carries out weighted summation, effectively reduces the difference at image joint, and the image fusion effect is good.

This invention discloses a retinal vessel segmentation method based on orientation field guidance and graph neural network optimization, belonging to the field of image analysis and medical image processing technology. The invention first acquires retinal fundus images and their corresponding vessel annotation maps, and preprocesses the raw retinal fundus images. Then, it constructs a retinal vessel segmentation model based on orientation field guidance and graph neural network optimization. This model includes a backbone segmentation model based on a UNet++ network architecture and a post-processing refinement module based on a visual graph neural network. The preprocessed image is input into the backbone segmentation model to obtain a preliminary vessel segmentation probability map, which is then processed by the post-processing refinement module to obtain the final vessel segmentation result. Experiments show that this invention effectively addresses the shortcomings of existing retinal vessel segmentation methods in terms of small vessel extraction, structural connectivity preservation, and background noise suppression.

A method of estimating optical coherence tomography (OCT) parameters includes: obtaining a retinal fundus image; and segmenting the retina fundus image for generating a 2D image of the target region. The method further comprises the steps of inputting the 2D image of the target area into a neural network; estimated OCT parameters are generated using a neural network.