80 results about "Retinal vessel" patented technology

The invention discloses a fundus image retinal vessel segmentation method and a fundus image retinal vessel segmentation system based on deep learning. The fundus image retinal vessel segmentation method comprises the steps of performing data amplification on a training set, enhancing an image, training a convolutional neural network by using the training set, segmenting the image by using a convolutional neural network segmentation model to obtain a segmentation result, training a random forest classifier by using features of the convolutional neural network, extracting a last layer of convolutional layer output from the convolutional neural network, using the convolutional layer output as input of the random forest classifier for pixel classification to obtain another segmentation result, and fusing the two segmentation results to obtain a final segmentation image. Compared with the traditional vessel segmentation method, the fundus image retinal vessel segmentation method uses the deep convolutional neural network for feature extraction, the extracted features are more sufficient, and the segmentation precision and efficiency are higher.

The invention relates to a U-shaped retinal vessel segmentation method adaptive to scale information. The U-shaped retinal vessel segmentation method comprises the following steps: preprocessing a retinal vessel image; And constructing a retinal vessel segmentation model 2. The method can effectively solve the problems that adjacent blood vessels are easy to connect, microvessels are too wide, tiny blood vessels are easy to break, segmentation at the intersection of the blood vessels is insufficient, image noise is too sensitive, a target and a background gray value are crossed, and a visual disc and a focus are segmented by mistake. According to the method, multiple network models are fused under the condition of low complexity, an excellent segmentation result is obtained on a DRIVE dataset, and the accuracy rate and the sensitivity of the method are 97.48% and 85.78% respectively. And the ROC curve value reaches 98.72% and reaches the current medical practical application level.

The multi-aperture system of the present invention provides a retinal oximetry apparatus for determining the level of oxygen saturation in retinal vessels using a lenslet array comprising at least seven lenses for the simultaneous measurement of reflected light with at least three wavelengths and at least four polarization states. The multi-aperture system of the present invention further provides an apparatus for determining the level of oxygen saturation in retinal vessels using a lenslet array comprising at least ten lenses for the simultaneous measurement of reflected light with at least three wavelengths for oxygen measurement, at least three wavelengths for melanin content, and at least four polarization states. Methods of operating the same are also provided.

The invention discloses a retinal vessel segmentation method based on combination of deep learning and a traditional method and relates to the fields of computer vision and mode recognition. According to the method, two grayscale images are both used as training samples of a network, corresponding data amplification, including elastic deformation, smooth filtering, etc., is done against the problem of less retinal image data, and wide applicability of the method is improved. According to the method, an FCN-HNED retinal vessel segmentation deep network is constructed, an autonomous learning process is realized to a great extent through the network, convolutional features of a whole image can be shared, feature redundancy can be reduced, the category of multiple pixels can be recovered from the abstract features, a CLAHE graph and a gauss matched filtering graph of the retinal vessel image are input into the network, an obtained vessel segmentation graph is subjected to weighted average, and therefore a better and more intact retinal vessel segmentation probability graph is obtained. Through the processing mode, the robustness and accuracy of vessel segmentation are improved to a great extent.

Systems and methods are described providing for the use of ultrasound energy to effect the dislodging of one or more blood clots inside blood vessels. Such clots can include those inside retinal vessels, especially in patients with central retinal vein occlusion. Embodiments of the present disclosure may be used for any retinal arterial or venous occlusion. In exemplary embodiments, a small probe can be inserted into the eye of a patient and placed over the retinal vessels. Acoustic streaming created by the probe can be directed to an area or region including targeted blood vessels, resulting in increased flow in one or more retinal veins and facilitating or effecting mechanical dislodging of one or more blood clots in the targets blood vessels. Exemplary embodiments can utilize ultrasonic energy produced at a frequency of approximately 44 MHz to 46 MHz with pulse repetition frequencies of approximately 100 Hz to 100 kHz.

The present invention discloses a dynamic scale distribution-based retinal vessel extraction method and system. The method includes the following steps of: retinal image preprocessing: contrast enhancement is performed on the green channel component of a color retinal image; image segmentation: the preprocessed retinal image is segmented into a set number of sub-images; vessel classification: the vessels of each sub-image are divided into three categories, namely, a large category, a medium category and a small category; dynamic scale allocation: filters of different scales are dynamically selected to enhance vessels of different widths; multi-scale matched filtering; threshold processing: vascular structures are extracted, nonvascular structure was removed, the extraction results of all the sub-images are re-spliced, so that a retinal vessel network binary image can be obtained; and post-processing: post-processing is carried out, so that a high-segmentation accuracy retinal vessel network image can be obtained. With the method and system of the invention adopted, the vessel extraction of the retinal image can be realized; excessive estimation of the widths of the vessels can be avoided when complex nonvascular structures are removed; and simpler and more accurate retinal vessel extraction can be realized.

The invention discloses a retinal vessel segmentation method of a fundus image based on a classification and regression tree and AdaBoost. The method comprises the step of: constructing a 36-dimensinal feature vector including a local feature, a morphological feature and a pixel vector field divergence feature for each pixel point in the fundus image, so as to determine whether the pixel point is on a vessel. During classified calculation, the classification and regression tree is used as a weak classifier, so as to classify a sample set, then an AdaBoost classifier is trained, so as to obtain a strong classifier, and thus, the classified determination of each pixel point is completed, so as to obtain final segmentation results. The method has the advantages that a vessel trunk is preferably extracted, great advantages are taken to treat high-brightness focal areas, later treatment is facilitated and visual results are provided for main vessel diseases, and the method is suitable for computer aided quantitative analysis of the fundus image and disease diagnosis and has obvious clinical significance in auxiliary diagnosis of related diseases.

The invention relates to a level set retinal vessel image segmentation method with shape prior being fused. The method comprises the following steps: 1) enhancing a retinal vessel image by utilizing a morphological operator and through Gaussian convolution; 2) roughly segmenting the retinal vessel image through anisotropic property of a Hessian matrix and an improved vessel response function, and serving the images as shape constrains and initial information; and 3) constructing a retinal vessel segmentation level set model comprising a local area energy fitting item, a shape constraint item, a level set function regularity maintenance item, a length punishment item and a weight area restraint item by utilizing shape prior and retinal image data. The segmentation method is high in segmentation result accuracy, can replace manual segmentation, and can play an important helping role in diagnosis and treatment of clinical related eye diseases, and has a higher clinical application value.

Systems and methods are described providing for the use of ultrasound energy to effect the dislodging of one or more blood clots inside blood vessels. Such clots can include those inside retinal vessels, especially in patients with central retinal vein occlusion. Embodiments of the present disclosure may be used for any retinal arterial or venous occlusion. In exemplary embodiments, a small probe can be inserted into the eye of a patient and placed over the retinal vessels. Acoustic streaming created by the probe can be directed to an area or region including targeted blood vessels, resulting in increased flow in one or more retinal veins and facilitating or effecting mechanical dislodging of one or more blood clots in the targets blood vessels. Exemplary embodiments can utilize ultrasonic energy produced at a frequency of approximately 44 MHz to 46 MHz with pulse repetition frequencies of approximately 100 Hz to 100 kHz.

The present invention discloses an ELM-based fundus image retinal vessel segmentation method. According to the method, a 39-dimensional feature vector including a Hessian matrix feature, a local feature, a gradient field feature and a morphological feature is constructed for each pixel in a fundus image so as to be used for judging whether each pixel belongs to pixels on a blood vessel; and training samples are adopted to train an ELM, so that a classifier can be obtained, classification judgment of each pixel on the image to be tested is completed, and final segmentation results are obtained. The method has the advantages of short training time, higher segmentation speed of the fundus image to be tested and better extraction effects of the trunk parts of blood vessels, and is advantageous in the processing of high-brightness lesion areas, is suitable for post-processing, provides intuitive results for the lesions of main blood vessels, is suitable for computer-aided quantitative analysis and disease diagnosis of fundus images and has significant clinical significance for the auxiliary diagnosis of related diseases.

The invention discloses a method and a device of segmenting retinal vessels in a fundus image. The method comprises steps that the fundus image is acquired; the fundus image is processed based on a Hessian Matrix to acquire a first retinal vessel image; the binarization processing of the first retinal vessel image is carried out to acquire a second retinal vessel image; broken vessels in the second retinal vessel image are reconstructed to acquire a third retinal vessel image. The technical problems of the prior art such as high complexity and inaccurate segmentation results during the segmentation of the retinal vessels in the fundus image are solved.

The invention belongs to the field of medical image processing and application, and provides a retinal vascular tortuosity calculation method based on an ophthalmoscope image and an application thereof. Firstly, a digital ophthalmoscope is used for obtaining an eye fundus image for screening the crowd, and a non-subsample discrete wavelet transform (UDWT) is then used for enhancing the image; secondly, local entropy texture of a retinal gray image is extracted, and a fuzzy C-mean clustering (FCM) method is used for segmenting a retinal vessel; and thirdly, the segmented vessel is skeletonized, topological levels of the skeleton are calculated, and a tortuosity calculation model in the invention is used for tortuosity calculation of the vascular skeleton. The method provided by the invention is easy to implement, accurate, reliable and convenient in clinical application.

Retinal vessel extraction on color images of eyes are disclosed. In one embodiment, the method comprises performing alpha matting on a multi-dimensional feature set derived from image data of a first image and performing retinal vessel extraction, including performing segmentation on the image by separating foreground and background image data from the first image using an output from the alpha matting.

Provided is a method of imaging of in vivo retina haemodynamics and measuring of the absolute flow velocity. On the basis of conventional FD.OCT, an OCT angiography technology and a Doppler OCT imaging technology are integrated, and measurement of the retinal vessel absolute flow velocity and the haemodynamics is achieved. According to the method of imaging of the in vivo retina haemodynamics and measuring of the absolute flow velocity, distribution of a three-dimensional blood vessel geometric structure is rebuilt through the OCT angiography technology, an included angle of a blood vessel and a light beam is calculated and serves as a Doppler angle according to the geometric structure of a three-dimensional blood vessel, a Doppler frequency shift of a blood flow is obtained through a Doppler OCT technology, correction is conducted according to a calculated Doppler angle, and the absolute flow velocity is obtained, and the problem of the retinal vessel Doppler angle is solved. According to the method of imaging of the in vivo retina haemodynamics and measuring of the absolute flow velocity, the three-dimensional blood vessel structure rebuilt through the OCT angiography technology is combined with the absolute flow velocity value of a particular location, and measurement of the in vivo retina haemodynamics can be achieved by simulating the flow of the blood in the blood vessel.

The invention discloses an arteriovenous retinal vessel segmentation method for an eye fundus image. The method comprises the following steps that binarization processing is carried out on a pre-processed eye fundus image according to a preset image-binary threshold value, the central line and the edge in the eye fundus image after the binarization processing are extracted, and a blood vessel tree is obtained; disconnection process is carried out on the branching portions of the blood vessel tree to obtain blood vessel segments, line segmentation is carried out on all the blood vessel segments to obtain blood vessels, and an original blood vessel set is obtained; mis-segmented blood vessels are determined and removed from the original blood vessel set to obtain a global blood vessel set. According to the method, after the original blood vessel set is obtained, the mis-segmented blood vessels are further determined through the background and the shapes of the blood vessels, mis-segmented blood vessels which are caused by annular light reflection of photograph, non-blood-vessel step edges around an optic disc, porphyritic lesions, bleeding lesions and other reasons can be removed effectively, and the segmentation precision of the blood vessels is improved.

The invention relates to the technical field of medical science, in particular to a retinal vessel recognition method and a retinal vessel recognition device. The method comprises the following steps that a retinal grey-scale map in a green path is extracted from a retinal ocular fundus image in an RGB format; a plurality of comparison scales are set; pixel points on the retinal grey-scale map are subjected to contrast quantification in different directions under each comparison scale; retinal binary contrast maps are obtained, wherein pixels in the retinal binary contrast maps are divided into retinal vessel pixels and non-retinal-vessel pixels; the obtained retinal binary contrast maps are subjected to difference fusion; noise and artifacts in the images obtained after the fusion are removed, and a retinal fusion image is obtained; a center reflecting region is determined in the retinal fusion map; the center reflecting region is filled; and a retinal vessel distribution diagram is obtained. The method and the device provided by the invention have the advantage that the accuracy and the reliability of the retinal vessel recognition are improved.

The invention discloses a combined retinal vessel segmentation system and method based on the fractal dimension number and Gauss filtering. The system comprises an image pre-processing unit used for improving retina image contrast and improving image quality to reduce interference of imaging problems on the subsequent processing process, a local detection unit used for extracting detailed portions of the retina image after pre-processing to improve fine vessel extraction accuracy, a global detection unit used for extracting a main portion of the retina image after pre-processing to highlight the global information of the retina vessel, a retina image fusion unit used for fusing the detailed portions of the retina image extracted by the local detection unit and the main portion of the retina image extracted by the global detection unit and improving system segmentation accuracy, and a retina image post-processing unit. The system is advantaged in that the fractal dimension number and Gauss filtering are combined, the system is suitable for segmentation of vessels of retina eyeground images, and segmentation accuracy is substantially improved.

The invention relates to a dual-wavelength retinal vessel blood oxygen measuring system based on a fundus camera. The system comprises two parts, namely, a retinal image acquisition subsystem and a blood oxygen calculation subsystem, wherein the retinal image acquisition subsystem forms a retinal image acquired by the fundus camera to infinity through an imaging objective, and a telecentric beam path in the image space is formed. A right-angled reflecting prism splits a light beam into two ways, the light beam is reflected by reflectors and transmits interference filters with specific wavelengths to be filtered into imaging light with required wavelengths, the imaging light is imaged on a CCD (charge coupled device) target surface simultaneously after passing through the imaging objective, and dual-wavelength simultaneous retinal imaging is realized; the blood oxygen calculation subsystem finishes calculation of the retinal vessel blood oxygen saturation after performing a series of operations such as image processing, vessel extraction, light density calculation and the like on an acquired dual-wavelength retinal image. With the adoption of the system, retinal vessel function information is acquired while the retinal structure image is acquired, and a powerful tool can be provided for life science related research and diagnosis of fundus related diseases.

The invention belongs to the field of medical image processing and application, and provides a retinal vessel three-dimensional reconstruction method based on a portable digital fundus camera. The method comprise steps that firstly, a portable digital fundus image is used, the camera is used for two fundus images by camera displacement, both images are require to contain retinal optic nerve disc structures, and a certain amount of offset is generated; and then, the retinal vessel is segmented, and the feature points are extracted, and based on retinal vessel anatomical hierarchy characteristicpoint matching and a three-dimensional vessel reconstruction model, the retinal vessel is subjected to three-dimensional reconstruction. The retinal vessel three-dimensional reconstruction method based on the portable digital fundus camera can provide new supplement to an existing fundus blood vessel observation and analysis based on two-dimension.

The invention discloses a retinal image segmentation algorithm based on a residual error U-NET network, which comprises the following steps: A, downloading a color fundus retina image, and carrying out sample expansion on the downloaded image; B, preprocessing the original image; C, obtaining an Improved residual U-NET network by adding a residual structure to the U-NET network; D, taking the processed retina image as input, and carrying out pre-training on the training sample to obtain Initial parameters of the residual U-NET network model; E, Segmenting test samples using a trained residualU-NET neural network model to obtain a final retinal vessel image segmentation map; the residual U-NET neural network model iss adopted to segment the diabetic retina image, the processing effect isgood, the method can be widely applied to the field of diabetic retina diagnosis, and powerful theoretical and technical support is provided for pathological diagnosis of the diabetic retina image.

The invention discloses a diabetic retinal vessel image segmentation method and belongs to the technical field of image processing. The method comprises steps as follows: step 1, a preprocessed eye fundus image is subjected to binaryzation processing according to preset binaryzation threshold values, a center line and the edge of the eye fundus image subjected to binaryzation processing are extracted, and a vessel tree is obtained; step 2, bifurcation of the vessel tree is subjected to disconnection processing, vessel sections are obtained, each vessel section is linearly segmented, vessels are obtained, and an original vessel set is obtained; step 3, mistakenly segmented vessels are determined and removed from the original vessel set, and a global vessel set is obtained. After the original vessel set is obtained, the mistakenly segmented vessels are further determined by use of the background and the shape of vessels, and the mistakenly segmented vessels caused by annular light reflection due to shooting, non-vascular step edges on the periphery of the optic disk, porphyritic lesion, bleeding lesion and other causes can be effectively removed.

The invention discloses a retinal vessel segmentation method based on dense convolution and depth separable convolution. The method comprises the following steps: preprocessing an original image of aretinal vessel map; carrying out data enhancement on the data set; constructing a full convolutional neural network based on combination of dense convolution and depth separable convolution, and training the training set by using a loss function with weight; and testing and obtaining a final segmentation result graph. Encoding-decoding symmetric network is used as a backbone network. The additiondepth can separate convolution, so that the model parameter quantity is greatly reduced; dense convolution blocks are used, all layers are connected on a channel, information transmission is enhanced,and characteristic values of all scales are effectively utilized; in the image preprocessing process, self-adaptive gamma correction is carried out on the image, different gamma values are used for correcting different feature areas, and background noise is weakened while the feature contrast ratio is improved; during training, a loss function with weight is used to enhance the proportion of theblood vessel to be segmented during training.

The invention discloses an arteriovenous retinal vessel optic disk positioning method of an eye fundus image. The method comprises the following steps: acquiring a global vessel set of the eye fundus image, wherein the global vessel set is a set of all vessels in the eye fundus image; acquiring a convergence region of each vessel by using a fuzzy convergence algorithm aiming at each vessel in the global vessel set; counting the number of the convergence regions to which each pixel point of the eye fundus image belongs to serve as voting value of the pixel point; selecting the previous n pixel points with large voting values, applying an eight connection-based region communication algorithm to the selected n pixel points to obtain a plurality of communication regions, and taking a central coordinate of the communication region with the maximum area as the optic disk positioning information. According to the optic disk positioning method, a multiple positioning voting mode is adopted, the method has robustness and behaves well under the complex condition of more diseases in the eye fundus image, finally a template is added for matching to serve as positioning and non-convergent correction, and the rejection is less compared with the method that the condition is not treated.

The invention discloses a method for screening ocular fundus images. According to a tree network structure and grayscale distribution features of retinal vessels, a vascular skeleton is checked from multi-scale discrete gauss through a Morlet wavelet and gauss matched filtering partitioning mode, and therefore the contrast between micro vessels and a background region is increased; meanwhile, combined with a Gabor wavelet algorithm, wavelet features of the algorithm and green channel grayscale information are utilized to form feature vectors; and artery and vein vessels and ocular features of intersections, arches, yellow spots, black spots, vitiligos, etc. of the vessels are subjected to marking learning training through the deep neural network technology. Through the method, screening and classification of the ocular fundus images can be completed, the diagnosis speed of ocular fundus diseases is increased, the problems of large manual consumption and low efficiency are solved, and finally effort is made for improving the health level of the entire society.

The invention belongs to the fields of diabetes mellitus and medicinal chemistry, and relates to the application of icariin or icarisid II in preparation of a medicine for preventing and curing diabetic retinopathy, in particular to the application of the icariin, an icariin-containing extract, the icarisid II or an icarisid II-containing extract in preparation of a medicine for preventing and / or curing the diabetic retinopathy. The icariin, the icariin-containing extract, the icarisid II or the icarisid II-containing extract have effective effects on repair or prevention and cure of the diabetic retinopathy, can improve the retinal vessel endothelial cell function remarkably, and can promote the reparative regeneration of optic nerve cells for diabetes mellitus.

The invention discloses a marine fish image recognition method based on deep learning, and belongs to the field of computer vision and deep learning. According to the invention, a method based on a deep learning network is utilized to realize recognition of marine fish images, and various marine fishes can also be identified even if biological oceans are not understood. Cost is reduced, fishes ofvarious varieties can be recognized, and compared with manual recognition, economic cost is reduced. And the segmentation precision is high: when the test set D is identified, the average recognitionprecision reaches 0.91. And the segmentation speed is high: the data loading time is not calculated, and the time for simply segmenting one image does not exceed 0.4 second. And the model not only canbe used for marine fish image recognition, but also can be used in the field of medical image segmentation, such as fundus retinal vessel segmentation and hippocampus segmentation.