84 results about "Retinal image" patented technology

The invention discloses an OCT (Optical Coherence Tomography) retina image denoising method based on a high-frequency enhanced diffusion model, which is characterized in that a double-branch diffusion model network constructed based on frequency perception Fourier transform attention (FFTA) is used for separating different frequency domains, enhancing specific frequency domain features and fusing the specific frequency domain features into a spatial domain feature map to realize stronger frequency domain perception and processing capability; the invention relates to a time step T coding method for a module of frequency domain attention. The time step T of the diffusion model is used as a parameter to be coded into a frequency domain attention module to participate in self-attention matrix calculation, so that frequency domain feature processing is aligned with the iteration step number of the diffusion model, and different frequency domain information is pointedly processed at different time steps T; the frequency selective hopping mechanism uses pooling operation to obtain a high / low frequency characteristic pattern, so that the network has adaptive frequency domain retention capability for different local parts.

The present application belongs to the field of medical image segmentation, and particularly relates to a retinal blood vessel image segmentation method. In view of the problems of low segmentation accuracy, insufficient segmentation ability of small blood vessels at the edge of eyeball, fracture at the blood vessel branch, and excessive interference of image noise in the existing retinal blood vessel image segmentation, the method comprises the steps of retinal image preprocessing and establishment of a retinal blood vessel segmentation model, wherein the preprocessing comprises converting a color retinal image into a gray image by giving different weights to the RGB three channels of the color retinal image; using a normalized and contrast-limited adaptive histogram equalization method to improve the image; using a local adaptive gamma change algorithm to adjust the retinal image; using translation, rotation, and noise increase to expand the data set; and the model establishment comprises feature extraction, feature fusion, and retinal blood vessel image segmentation.

The present disclosure is particularly directed to contact lens devices and / or methods for myopia treatment. The present disclosure is directed to modifying the incoming light by a contact lens, using a stop signal to slow the rate of progression of myopia. More particularly, the present disclosure is directed to a purposeful configuration of a non-circular, non-transparent aperture stop on the original base single-vision optic zone of a contact lens that can facilitate the redistribution of light energy into the oblique frequencies of the retinal image to provide an optical stop signal that prevents, reduces or controls the progression of refractive error of progressive myopia.

The invention provides a fundus retina image recognition method based on an improved lightweight network, and belongs to the technical field of image processing, and the method comprises the steps: obtaining pathological and non-pathological myopia fundus retina image data sets, carrying out the preprocessing and dividing, replacing an SE module with an ECA module based on MobileNetV3-Small, and optimizing the structure to obtain a lightweight model; during training, unbalance is processed by using an inverse frequency weighting method, and a recall rate optimal model is stored; and finally, the loading model carries out reasoning on the input image, and outputs and stores a recognition result. According to the fundus retina image recognition method based on the improved lightweight network, the problems that a traditional pathological myopia processing method depends on expert experience, diagnosis is difficult in a region lacking medical resources, and an existing model is large in parameter quantity, low in reasoning speed and difficult to deploy in a basic medical institution are solved.

Provided is a visual field estimation apparatus using a machine learning model that has been machine-trained so that, with respect to a plurality of eyes to be examined, at least one of image information of retina and three-dimensional structure information of retina is acquired at each of a plurality of time points, visual field related information relating to a visual field of the eye to be examined is acquired at a corresponding time point, at least one of the image information of the retina and the three-dimensional structure information of the retina regarding each of the eyes to be examined is input to the machine learning model as input information, visual field change information representing change of the visual field and obtained based on the visual field related information at the plurality of time points with regard to the corresponding eye to be examine is used as training information, and visual field change information estimated on the basis of the input information is output from the machine learning model.

The invention discloses an OCT retina image classification method based on an improved ResNet-34 network, and relates to the technical field of retina medical image classification, and the method comprises the steps: obtaining a retina OCT image data set, and carrying out the preprocessing of an image; an improved ResNet-34 network model is constructed, a CBAM attention module is embedded in a residual block of the improved ResNet-34 network model, a CBAM-Block structure is formed, a channel attention module and a space attention module are combined in a series connection mode, channel feature response of a focus area is self-adaptively enhanced, and a key space position is focused; dynamically adjusting floating point calculation precision in a network training process by adopting an automatic mixing precision training technology; pre-training weight initialization network parameters based on an ImageNet data set are loaded, and the model is finely adjusted through transfer learning; and training a network by using the preprocessed data set, and outputting a retina pathological state classification result. According to the method, through a triple collaborative optimization mechanism of a CBAM-Block residual structure, automatic mixing precision training and transfer learning, the precision and efficiency of retina OCT image classification are remarkably improved.

To provide image retention and stitching for minimal flash eye disease diagnosis.SOLUTION: Systems and methods for minimizing retinal exposure to flash during image acquisition for diagnosis are provided herein. In certain embodiments, the system captures multiple retinal images of different retinal regions. The system determines that a first portion of the first image does not satisfy a criterion while a second portion of the first image satisfies the criterion, identifies a portion of the retina depicted in the first portion that does not satisfy the criterion, and determines whether the portion of the retina is depicted in a third portion of the second image and whether the third portion satisfies the criterion. In response to determining that the third portion satisfies the criteria, the system performs a diagnosis. In response to determining that the portion of the retina is not depicted in the second image, the system captures an additional image of the retinal region.SELECTED DRAWING: Figure 6

A method of training a convolutional neural network model to predict diabetes from an image of a retina is provided. The method of training a convolutional neural network includes processing a first dataset, wherein processing the first dataset comprises: extracting a circular region from a retinal image, resizing the circular region, cropping the circular region, and placing the circular region onto a black background; training an initial model using a second dataset to yield a first model; training the first model using a third dataset to yield a second model; and training the second model using the first dataset to yield a third model.

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.

Systems for customized visual field (VF) testing use machine learning models (15) trained on retinal images (12A, 12C, 12D) including optical coherence tomography (OCT), optical coherence tomography angiography (OCTA), fundus, and / or fluorescein angiography images. In operation, when a particular VF test (13) is to be prepared for a patient, a retinal image of the patient is submitted to the current machine model, which responds by synthesizing a VF for the patient. The synthesized VF can be used to optimize the particular VF test prior to performing the particular VF test on the patient.

A system includes system housing, an eccentric radiation source, and a radiation sensor. The radiation produced by the eccentric radiation source can be collected by the radiation sensor to generate images of retinas for a patient. The system also includes a vision screening device connected with the eccentric radiation source and the radiation sensor via the system house that can control and synchronize actions for the eccentric radiation source and the radiation sensor. The vision screening device further analyzes the images generated by the radiation sensor via neural network algorithms to determine spherical error slopes, refractive errors, and recommendations for the patient.

A computer-implemented method of prediction of a cardiovascular, CV, event of a patient, based on a retinal image from the patient, is provided. The method comprises extracting a feature data set from the retinal image using a neural network, the neural network being pretrained to output feature data sets based on retinal images; inputting the extracted feature data set to a machine learning model, ML model, the ML model being pretrained to output a CV event indicator, the pretraining being based on the following data: training feature data sets extracted from retinal images; and training CV event indicators, one training CV event indicator per each one of the training feature data sets; and obtaining a CV event indicator of the patient from the ML model. A computer-implemented method of training a deep learning model based on retinal images and CV event indicators wherein the deep learning model comprises a neural network, NN, and a machine learning, ML, model, is provided. A data processing system, a training data processing system and a computer program product are also provided.

Systems, devices and methods to overcome IOL deficiencies are disclosed by providing at least one of a phakic or aphakic IOL to provide correction of defocus and astigmatism, reduce higher order monochromatic and chromatic aberrations and provide extended depth of field to improve visual quality. The IOL includes a virtual aperture integrated into the IOL. The structure and design allow light rays that intersect the virtual aperture and are widely dispersed across the retina, effectively preventing the light rays from reaching a detectable level on the retina. The virtual aperture helps to eliminate monochromatic and chromatic aberrations, resulting in a high-definition retinal image. The depth of field is increased on larger diameter optic zone IOLs for a given definition of acceptable vision.

This invention discloses a method and system for grading myopic macular degeneration (ATN) based on multimodal medical imaging. The system includes an image acquisition module for acquiring multimodal retinal images to be graded; a preprocessing module for preprocessing the multimodal images to obtain image sequences that meet preset grading requirements; and a grading module for inputting the image sequences into a pre-stored multimodal multi-task grading model. The model performs feature extraction, multimodal fusion, and multi-task classification on the image sequences, outputting a lesion grade prediction result for each grading task. This invention significantly improves the model's sensitivity to pathological regions by performing pixel-level segmentation of the input images and locating key anatomical structures. In subsequent steps, spatial weighting is applied to CFP image modalities, and specific regions of OCT and OCTA are enhanced, effectively improving the accuracy and robustness of lesion diagnosis and localization.

The invention relates to an eye age determination method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps of obtaining an original fundus image of a user, and performing image processing on the original fundus image to obtain a target fundus image; inputting the target fundus image into a pre-trained eye age recognition model to obtain eye age information of the user output by the eye age recognition model; wherein the eye age identification model is obtained by training based on a deformation field algorithm and a retina image age causal feature extraction algorithm based on a double-task learning framework. By adopting the method, the accuracy of determining the eye age can be improved.

The application provides a zero-reference retinal image enhancement method, system, device and medium, wherein the method comprises the following steps: acquiring an atmospheric light map component of a to-be-processed retinal image in an atmospheric scattering model; adding a random noise map to a gray-scale map of the image to obtain a transmission map component in the model; performing deblurring processing on the to-be-processed image first, and generating a first image under the guidance of a transmission map generation module; at the same time, extracting a V channel of the to-be-processed image, improving a low-light problem through learning of an input low-light enhancement curve, and generating a second image; then, fusing the first image and the second image to obtain an enhanced image; finally, using an atmospheric scattering model formula, combining the atmospheric light map, the transmission map and the enhanced image to calculate a quality degradation image, and constraining the quality degradation image and the to-be-processed image through a reconstruction loss, and finally realizing efficient image enhancement. The method can realize the effect of simultaneously optimizing multiple distortions.

The invention discloses a retina image abnormity intelligent identification system based on ophthalmology department auxiliary diagnosis, which belongs to the technical field of medical image processing analysis, and is characterized in that a plurality of groups of images of the same examined eye are acquired to form an analysis sequence, and an anatomical structure feature map and a tissue texture feature map of each frame of image in the sequence are extracted in parallel; performing conditional normal texture construction and intra-group comparison, multi-frame texture consistency verification and anatomical structure deviation quantitative analysis on the anatomical structure feature graph and the tissue texture feature graph to generate a texture abnormal intensity graph, a texture spatial-temporal variation graph, a texture region stability classification graph and a structure deviation graph; carrying out adaptive weighted fusion and grading judgment on the various types of results; according to the method, through a collaborative technical route of multi-modal imaging, double-feature decoupling extraction, conditional normal comparison, space-time consistency verification and adaptive fusion decision, dependence on labeled data is reduced, and meanwhile, high-robustness intelligent recognition of structure and texture anomalies in the retina image is achieved.

A technique and apparatus for performing a visual field test on an eye includes sequentially presenting visual stimuli to the eye on a display. A first fixation target is presented on display in a first position offset from a center of the display, such as while presenting first visual stimuli on the display for measuring a visual function sensitivity of the eye contemporaneously with presenting the first fixation target. A response of the eye to the first visual stimuli is registered. A second fixation target is presented on the display in a second position offset from the center of the display, wherein the second position is different than the first position, such as while presenting second visual stimuli on the display for measuring a visual function sensitivity of the eye contemporaneously with presenting the second fixation target. A response of the eye to the second visual stimuli is registered.

The invention discloses a high dynamic range display effect optimization method based on human visual characteristics, and the method comprises the steps: carrying out the standardized nonlinear coding and decoding and HDR post-processing of an SDR image, and obtaining the linear brightness distribution of an HDR image after the SDR image is converted; linear brightness distribution of the HDR image is used as input, EOTF and brightness normalization are carried out, a glare spread function filter convolution kernel is calculated, retinal HDR image perception simulation is carried out, a retinal contrast distribution result is obtained, an evaluation index of a contrast peak signal-to-noise ratio is established, and high dynamic range display effect optimization is carried out. Theoretical support is provided for the HDR adjustment strategy conforming to the human visual characteristics, and practical support is provided for improving the display quality of the display device.

Previous work on normalization / calibration methods for fundus imaging captured by multispectral or hyperspectral retinal cameras may be insufficient to accurately correct retinal images, for example to identify subtle spatial-spectral features indicative of a disease. In particular, previous techniques may not account for several important factors that affect the accuracy of a captured cube of image data. Accordingly, techniques described herein include improved methods for multispectral or hyperspectral image data correction.

The invention discloses a retina OCT image denoising method and system based on an improved diffusion model, and is applied to the technical field of medical image processing. The method comprises the following steps of obtaining retina OCT image data, preprocessing the retina OCT image data to obtain a training data set composed of noise-truth value image pairs, performing structure improvement on a basic diffusion model based on ConvNeXt Block, a dual efficient attention module DEAB and a diffusion model solver DPM-Solver, and obtaining a training data set composed of noise-truth value image pairs. And performing iterative training and evaluation on the improved denoising diffusion model through the training data set to obtain a trained retina OCT image denoising model. According to the method, the details and the hierarchical structure of the retina OCT image can be kept while noise interference is effectively suppressed, and compared with a traditional denoising method, the method has the advantages of being excellent in denoising performance, high in image quality and high in clinical application value.

A method comprising: receiving (i) a first OCT scan, and (ii) a second, subsequent, OCT scan, each comprising one or more two-dimensional (2D) scan images of an eye of a subject; receiving a first and second retinal images of the eye of the subject, associated with the first and second OCT scans; matching corresponding first and second pixel column patches, each comprising one or more pixel columns, associated respectively with the scan images of the first and second OCT scans, based, at least in part, on performing image registration between the scan images of the first and second OCT scans; and using the matched corresponding first and second pixel column patches as input for a trained machine learning model, wherein an output of the trained machine learning model classifies the second pixel column patches as representing retinal atrophy or not representing retinal atrophy.

This invention discloses a feature fusion-based OCT image classification method for tuberculous retinal diseases, comprising the following steps: Step 1: acquiring tuberculous retinal OCT images; Step 2: image preprocessing; Step 3: constructing multiple baseline neural networks; Step 4: adding a multi-scale gated channel attention mechanism to the multiple baseline networks, utilizing the learning of the neural network to obtain the importance between feature channels, thereby strengthening features beneficial to the classification task and suppressing useless features; Step 5: training and testing: finding the optimal model under the optimal parameters of each network structure through multiple comparative experiments; Step 6: model fusion; This invention utilizes a multi-layer convolutional neural network and a gated attention mechanism to construct a basic learner that acts as a feature extractor, then fuses the image features extracted by different models, and inputs the fused features into a support vector machine for further classification and prediction, thereby further improving the accuracy of tuberculous retinal OCT image classification.

Disclosed are systems, devices, and methods that overcome limitations of IOLs at least by providing a phakic or aphakic IOL that provides correction of defocus and astigmatism, decreases higher-order monochromatic and chromatic aberrations, and provides an extended depth of field to improve vision quality. The IOL includes a virtual aperture integrated into the IOL. The construction and arrangement permit optical rays which intersect the virtual aperture and are widely scattered across the retina, causing the light to be virtually prevented from reaching detectable levels on the retina. The virtual aperture helps remove monochromatic and chromatic aberrations, yielding high-definition retinal images. For a given definition of acceptable vision, the depth of field is increased over a larger diameter optical zone IOL.

The present invention relates to an inspection apparatus of the ocular fundus. The apparatus according to the invention comprises a first arrangement of components for acquiring images of the retina through line scanning ophthalmoscopy, a second arrangement of components for acquiring images of the retina by optical coherence tomography and a control arrangement to control the operation of said first and second arrangement of components.

The invention discloses a field sequential display color separation suppression method and system based on human eye tracking, and belongs to the technical field of display. The method comprises the following steps: acquiring a fixation point coordinate and an eyeball movement velocity vector of an observer in real time through an eye movement tracking device, and synchronously analyzing a sub-field time sequence signal of a display system; constructing a motion prediction model containing a system delay parameter, performing time domain extrapolation on an eyeball trajectory by using an algorithm, and accurately calculating a fixation point prediction position of each monochromatic sub-field at an actual lightening moment; on the basis of a retina imaging stabilization principle, taking the starting moment of the current frame as a reference anchor point to generate a reverse geometric translation compensation amount; the original image of each sub-field is reconstructed, and the displacement vacancy is eliminated in combination with the overscanning edge compensation technology. And finally, the display terminal is driven strictly according to the time sequence, so that the sub-field rays lightened at different time are projected to the same physical area of the retina to realize imaging superposition. According to the method, color separation artifacts are inhibited from a physical mechanism, and the visual quality is remarkably improved.

The application discloses a kind of based on frequency domain enhancement's retinal blood vessel segmentation method, it is related to retinal blood vessel segmentation technical field, the method includes: the input retinal image is preprocessed to enhance blood vessel structure information;The spatial domain feature of image is extracted by multilevel encoder;Dynamic frequency domain enhancement processing is carried out to spatial domain feature, including using learnable Gaussian high-pass filter and K groups learnable complex Fourier base weighted fusion to generate adaptive frequency domain mask to enhance blood vessel edge high-frequency component;Frequency domain enhancement feature and spatial domain feature are input into cross-domain routing attention module and are fused;Key feature position is filtered using double-layer routing mechanism and sparse attention is calculated;Blood vessel segmentation result is output by the decoder upsampling fusion feature and output blood vessel segmentation result.The application is enhanced by learnable Gaussian high-pass filter adaptive blood vessel edge frequency domain feature, combined with spatial domain feature to carry out cross-domain routing attention fusion, realizes high-precision, high-efficiency blood vessel segmentation under the improvement U-Net architecture.

A system for coronary artery calcium (CAC) estimation includes a processor and a memory, including instructions stored thereon, which when executed by the processor, cause the system to: extract a plurality of features from patient data using a first machine learning (ML) model, the patient data including a retinal image, sociodemographic data, and / or clinical data; refine the extracted plurality of features using an attention layer, the attention layer configured to highlight retinal structures with an indication of calcification risk; combine a subset of the plurality of features into a data vector using a second ML model; provide a CAC estimation based on the data vector; determine that the provided CAC estimation exceeds a predetermined threshold; and generate an output indicating a calcification risk level, based on the CAC estimation.

The present invention relates to an AI-based blinding disease detection system and method, the system comprising: one or more cameras for photographing the eyes of a patient; a device management module for collecting retinal images and monitoring the state of each camera; a database for storing the retinal images; an AI management module for evaluating the risk of blinding diseases by analyzing the retinal images on the basis of an AI algorithm; a result provision module for providing, to a customer terminal, the result of evaluating the risk of blinding diseases; and a third-party connecting module linked to a hospital information system (IHS) or a cloud-based external system so as to connect to an external tool for additional analysis or report generation.