CN114334124A - Pathological myopia detection system based on deep neural network - Google Patents
Pathological myopia detection system based on deep neural network Download PDFInfo
- Publication number
- CN114334124A CN114334124A CN202111391392.5A CN202111391392A CN114334124A CN 114334124 A CN114334124 A CN 114334124A CN 202111391392 A CN202111391392 A CN 202111391392A CN 114334124 A CN114334124 A CN 114334124A
- Authority
- CN
- China
- Prior art keywords
- pathological
- pathological myopia
- lesion
- fundus
- network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 208000001309 degenerative myopia Diseases 0.000 title claims abstract description 96
- 230000004340 degenerative myopia Effects 0.000 title claims abstract description 96
- 238000001514 detection method Methods 0.000 title claims abstract description 64
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 17
- 230000003902 lesion Effects 0.000 claims abstract description 77
- 230000001575 pathological effect Effects 0.000 claims abstract description 18
- 230000002159 abnormal effect Effects 0.000 claims abstract description 6
- 238000012549 training Methods 0.000 claims description 24
- 238000000034 method Methods 0.000 claims description 15
- 230000036285 pathological change Effects 0.000 claims description 12
- 231100000915 pathological change Toxicity 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 8
- 208000001491 myopia Diseases 0.000 claims description 7
- 230000004379 myopia Effects 0.000 claims description 7
- 206010003694 Atrophy Diseases 0.000 claims description 6
- 230000037444 atrophy Effects 0.000 claims description 6
- 210000004220 fundus oculi Anatomy 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 241000282485 Vulpes vulpes Species 0.000 claims description 5
- 239000004922 lacquer Substances 0.000 claims description 5
- 238000000137 annealing Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 201000010099 disease Diseases 0.000 description 9
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 9
- 208000005590 Choroidal Neovascularization Diseases 0.000 description 4
- 206010060823 Choroidal neovascularisation Diseases 0.000 description 4
- 241000282373 Panthera pardus Species 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000003745 diagnosis Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000003759 clinical diagnosis Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000004438 eyesight Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000003703 image analysis method Methods 0.000 description 1
- 208000002780 macular degeneration Diseases 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 208000015122 neurodegenerative disease Diseases 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Landscapes
- Eye Examination Apparatus (AREA)
Abstract
The invention provides a pathological myopia detection system based on a deep neural network, which comprises a computer memory, a computer processor and an executable program, wherein the executable program is used for receiving a fundus image, transmitting the fundus image into a trained pathological myopia detection network model and finally outputting a result; the pathological myopia detection network model comprises a focus detector, a pathological degree classifier and a pathological myopia discriminator; the lesion degree classifier is used for judging the lesion grade of the fundus image; if the image lesion grade is abnormal, the fundus image is sent to a lesion detector; the focus detector is used for detecting the focus type and position in the fundus image; and the pathological myopia discriminator judges whether the fundus image has pathological myopia according to the output results of the pathological degree classifier and the focus detector. The system can predict, classify and position the whole pathological degree of the eyeground and judge whether the eyeground suffers from pathological myopia or not.
Description
[ technical field ] A method for producing a semiconductor device
The invention relates to the technical field of ophthalmology imaging, in particular to a pathological myopia detection system based on a deep neural network.
[ background of the invention ]
Pathological myopia generally refers to diseases with refractive power greater than-6D and/or ocular axis greater than 26.5mm, and may be accompanied by degenerative diseases such as scleral retrouveitis. About 3% of the population worldwide suffers from pathological myopia, which is one of three major blind diseases in the world, especially is most common in Asians and most common in young and middle-aged people. Pathological myopia generally has more characteristic eyeground pathological changes, and the pathological myopic eyeground pathological changes are classified into 5 grades by combining the previous research: normal, stigmatic fundus oculi, diffuse choroidal atrophy, macular atrophy. Pathological myopia is generally accompanied by the development of complications, among which Choroidal Neovascularization (CNV), lacquer cracks (lacquer cracks), Fuchs plaques (Fuchs spot) have a greater impact on vision. In clinical diagnosis, the presence of these lesions is an important factor in determining the level of pathological myopia in a patient.
At present, fundus camera imaging is the most common ophthalmologic clinical examination means, but doctors have the defects of poor repeatability, different judgment standards and the like when diagnosing fundus images. Therefore, many studies are being made to assist a doctor in diagnosis by processing fundus images using a machine learning method such as deep learning.
For example, chinese patent publication No. CN111046835A discloses a fundus oculi illumination multiple disease detection system based on regional feature set neural network, which includes a semantic segmentation sub-network and a plurality of classifiers. The system operation steps are as follows: the method comprises the steps of obtaining an original fundus picture to be detected, inputting the original fundus picture into a multi-fundus disease detection network model, and respectively carrying out classification on various fundus diseases by the network model to judge whether the fundus diseases are ill or not. Chinese patent publication No. CN110163839A discloses a leopard streak fundus image recognition method that acquires a fundus image and classifies the fundus image by a machine learning model to obtain a classification result indicating the degree of significance of a leopard streak feature of the fundus image.
The existing fundus image analysis method based on machine learning algorithm such as deep learning mostly focuses on the whole fundus and predicts the lesion degree or lesion type, and the prediction on the position and type information of a specific lesion is lacked. Meanwhile, a mature detection method based on deep learning is lacking for fundus diseases with pathological myopia which are gradually emphasized in recent years.
[ summary of the invention ]
The invention aims to solve the problems in the prior art and provides a pathological myopia detection system based on a deep neural network.
In order to achieve the above purpose, the present invention provides a pathological myopia detection system based on a deep neural network, which includes a computer memory, a computer processor capable of communicating with the computer memory, and an executable program stored in the computer memory and executable on the computer processor, wherein the executable program is used for receiving a fundus image, transmitting the fundus image into a trained pathological myopia detection network model, and finally outputting a result; the pathological myopia detection network model comprises a focus detector, a pathological degree classifier and a pathological myopia discriminator;
the lesion degree classifier is used for judging the lesion grade of the fundus image; if the image lesion grade is abnormal, the fundus image is sent to a lesion detector;
the focus detector is used for detecting the focus type and position in the fundus image;
and the pathological myopia discriminator judges whether the fundus image has pathological myopia according to the output results of the pathological degree classifier and the focus detector.
Preferably, the specific training process of the pathological myopia detection network model comprises the following steps:
s1, collecting fundus images and marking the fundus images to be used as a training data set, wherein the data set comprises a fundus lesion degree data set and a focus detection data set.
S2, preprocessing the fundus image in the data set to enable the fundus image to meet network input requirements;
s3, training the pathological myopia detection network model, wherein the training process is divided into two stages:
in the first stage, a pathological change degree classifier is trained by using a fundus pathological change degree data set;
in the second stage, a lesion detector is trained using a lesion detection dataset.
Preferably, the first stage comprises the following specific steps:
s31, randomly initializing network parameters of the lesion degree classifier;
s32, setting the initial learning rate to be 0.001, and attenuating the learning rate by adopting cosine annealing;
and S33, adopting a random gradient descent algorithm, completing one round of training of all data in the training set through one network, fixing the part of parameters after the network converges, and completing the training.
Preferably, the second stage comprises the following specific steps:
s34, initializing parameters of a backbone network of the focus detector, wherein a lesion degree classifier network finished in the first stage of training is initialized, and network parameters of the rest part of the focus detector are initialized randomly;
s35, after network convergence, obtaining a complete pathological myopia detection network model;
and S36, inputting the collected eye fundus images into a pathological myopia detection network model, predicting whether the eye fundus images are suffered from pathological myopia or not by the pathological myopia detection network model, and if so, outputting the position and the type of a focus by the pathological myopia detection network model to assist a doctor in diagnosis.
Preferably, the fundus lesion degree data set is labeled as a fundus image lesion degree, and comprises five lesion grades: level 0, level 1, level 2, level 3, level 4, respectively, represent: normal, leopard-streak fundus oculi, diffuse choroidal atrophy, macular atrophy, the labeling of the lesion detection dataset includes a lesion position and a lesion type, the lesion position includes a center point coordinate, a width and a height of the lesion, and the lesion type includes CNV, lacquer cracks and Fuchs spots.
Preferably, the specific prediction process of the pathological myopia detection network model comprises the following steps:
a. preprocessing the fundus image to be predicted to enable the fundus image to meet the input requirement of the pathological myopia detection network model;
b. the fundus image is firstly input into a pathological change degree classifier network to predict the pathological change degree, and if the prediction result is normal: if the level is 0, the prediction is finished, and the pathological myopia detection network model outputs non-PM (0);
c. if the prediction result is abnormal: grade 1, grade 2, grade 3, grade 4, then the eyeground picture enters the focus detector network, the focus detected by the focus detector network will be marked, and the focus type and prediction probability are marked;
d. the pathological myopia discriminator synthesizes output information of the focus detector network and the pathological degree classifier network, and judges whether the fundus is suffered from pathological myopia;
e. the pathological myopia detection network model outputs the pathological degree, the focus position and the type of the input fundus image, and whether the fundus image belongs to pathological myopia.
Preferably, in the step d, the specific method for judging whether the fundus has pathological myopia is as follows: if the fundus lesion degree is normal: grade 0, or fundus lesions degree leopard streak fundus: grade 1, if no focus is detected, the myopia is judged to be non-pathological myopia, and if the focus is not detected, the myopia is judged to be pathological myopia.
Preferably, the lesion degree classifier comprises a depth residual error network, a full connection layer and a Softmax classification layer.
Preferably, the lesion detector comprises a backbone network, an RPN network, a ROI-Pooling layer and a classifier.
Preferably, the pathological myopia discriminator is constituted by a logical judgment sentence.
The invention has the beneficial effects that:
1. compared with the existing fundus image analysis model based on the neural network, the method can predict the overall lesion degree or type of the fundus image, and can also predict the specific lesion position and type.
2. The invention can predict the lesion grade of the fundus image, detect the lesion position of the fundus image and judge whether the sample is pathologically myopic by combining the information of the fundus image and the lesion position.
The features and advantages of the present invention will be described in detail by way of examples.
[ detailed description ] embodiments
The invention relates to a pathological myopia detection system based on a deep neural network, which comprises the following parts: a computer memory, a computer processor, and an executable program. The executable program receives the fundus image, transmits the fundus image into the trained pathological myopia detection network model, and finally outputs a result.
The pathological myopia detection network model comprises a focus detector, a pathological degree classifier and a pathological myopia discriminator. The lesion degree classifier is used for judging the lesion grade of the fundus image (the specific grade division is shown in a table 1); if the lesion grade of the fundus image is not 0, sending the fundus image to a lesion detector, wherein the lesion detector is used for detecting the type and the position of a lesion in the fundus image; and the pathological myopia discriminator judges whether the fundus image has pathological myopia according to the output results of the pathological degree classifier and the focus detector.
TABLE 1
The lesion degree classifier comprises a depth residual error network (ResNet-101) and a full connection Layer (FC Layer), and a Softmax classification Layer. The structure of the ResNet-101 comprises a first convolution layer, a second convolution layer group, a third convolution layer group, a fourth convolution layer group and a fifth convolution layer group.
The focus detector borrows a fast-RCNN target detection network structure, and specifically comprises a backbone network (ResNet-101), an RPN network, an ROI-Powing layer and a classifier.
The pathological myopia discriminator consists of a logic judgment statement.
The pathological myopia detection network model is specifically trained as follows:
(1) collecting fundus images and marking the fundus images as a training data set, wherein the data set is divided into two parts, namely a fundus lesion degree data set, and the marking of the fundus lesion degree data set is fundus image lesion degree (0, 1, 2, 3, 4); secondly, focus detection data set, the labeling of which includes: lesion location (center coordinates, width, height) and lesion type (CNV, lacquer crack, Fuchs spot).
(2) And preprocessing the fundus images in the data set to enable the fundus images to meet the network input requirement.
(3) Training the pathological myopia detection network model, wherein the training process is divided into two stages:
the first stage is as follows: a lesion degree classifier is trained using the fundus lesion degree data set.
The partial network parameters are initialized randomly.
The initial learning rate was set to 0.001 and the cosine annealing learning rate decay was used.
And (3) adopting a random gradient descent algorithm, completing one round of training of all data in the training set through a network once, fixing the part of parameters after the network converges, and completing the training.
And a second stage: a lesion detection dataset is used to train a lesion detector.
In the lesion detector network, parameters of a backbone network ResNet101 initialize a lesion degree classifier network finished from stage one training. And randomly initializing the rest network parameters.
The training method and the selection of the penalty function are referred to in the literature (S.ren, K.He, R.Girshick and J.Sun, "Faster R-CNN: Towards read-Time Object Detection with Region pro-posal Networks," in IEEE Transactions on Pattern Analysis and Machine Analysis, vol.39, No.6, pp.1137-1149,1June 2017, doi: 10.1109/TPAMI.2016.2577031).
And after the network convergence, obtaining a complete pathological myopia detection network model.
The collected fundus images are input into a pathological myopia detection network model, the network predicts whether the fundus images suffer from pathological myopia, and if the fundus images suffer from pathological myopia, the network also outputs the position and the type of a focus to assist a doctor in diagnosis.
The pathological myopia detection network model specifically predicts the process as follows:
and preprocessing the fundus image to be predicted to enable the fundus image to meet the input requirement of the network model.
The fundus image is firstly input into a pathological change degree classifier network to predict the pathological change degree, if the prediction result is normal (0), the prediction is finished, and the pathological myopia detection network model outputs non-PM (0)
If the prediction result is abnormal (1, 2, 3, 4), the fundus image enters a focus detection network, the focus detected by the network is framed by a frame, and the focus type and the prediction probability are marked.
The pathological myopia discriminator synthesizes the output information of the focus detector network and the pathological degree classifier network to judge whether the eyeground suffers from pathological myopia, and the method comprises the following steps: if the degree of fundus lesions is normal (0) or the degree of fundus lesions is leopard-streak fundus (1) but no lesion is detected, the myopia is judged to be non-pathological myopia, and if the degree of fundus lesions is not normal, the myopia is judged to be pathological myopia.
Finally, the pathological myopia detection network outputs the pathological degree, the focus position and the type of the input fundus image and whether the fundus image belongs to pathological myopia.
The pathological myopia detection network model of the invention realizes grading of myopia macular degeneration, has high grading efficiency and high precision, can provide powerful AI technical auxiliary support for diseases which are concerned on the day of pathological myopia in clinic, and helps doctors to quickly and effectively screen and diagnose the diseases.
The above embodiments are illustrative of the present invention, and are not intended to limit the present invention, and any simple modifications of the present invention are within the scope of the present invention.
Claims (10)
1. A deep neural network based pathological myopia detection system comprising a computer memory, a computer processor in communication with the computer memory, and an executable program stored in the computer memory and executable on the computer processor, characterized in that: the executable program is used for receiving the fundus image, transmitting the fundus image into the trained pathological myopia detection network model and finally outputting a result; the pathological myopia detection network model comprises a focus detector, a pathological degree classifier and a pathological myopia discriminator;
the lesion degree classifier is used for judging the lesion grade of the fundus image; if the image lesion grade is abnormal, the fundus image is sent to a lesion detector;
the focus detector is used for detecting the focus type and position in the fundus image;
and the pathological myopia discriminator judges whether the fundus image has pathological myopia according to the output results of the pathological degree classifier and the focus detector.
2. The system of claim 1, wherein the system comprises: the specific training process of the pathological myopia detection network model comprises the following steps:
s1, collecting fundus images and marking the fundus images to be used as a training data set, wherein the data set comprises a fundus lesion degree data set and a focus detection data set.
S2, preprocessing the fundus image in the data set to enable the fundus image to meet network input requirements;
s3, training the pathological myopia detection network model, wherein the training process is divided into two stages:
in the first stage, a pathological change degree classifier is trained by using a fundus pathological change degree data set;
in the second stage, a lesion detector is trained using a lesion detection dataset.
3. The system of claim 2, wherein the deep neural network based pathological myopia detection system comprises: the first stage comprises the following specific steps:
s31, randomly initializing network parameters of the lesion degree classifier;
s32, setting the initial learning rate to be 0.001, and attenuating the learning rate by adopting cosine annealing;
and S33, adopting a random gradient descent algorithm, completing one round of training of all data in the training set through one network, fixing the part of parameters after the network converges, and completing the training.
4. The system of claim 2, wherein the deep neural network based pathological myopia detection system comprises: the second stage comprises the following specific steps:
s34, initializing parameters of a backbone network of the focus detector, wherein a lesion degree classifier network finished in the first stage of training is initialized, and network parameters of the rest part of the focus detector are initialized randomly;
s35, after network convergence, obtaining a complete pathological myopia detection network model;
and S36, inputting the collected eye fundus images into a pathological myopia detection network model, predicting whether the eye fundus images are suffered from pathological myopia or not by the pathological myopia detection network model, and outputting the position and the type of a focus by the pathological myopia detection network model if the eye fundus images are suffered from the pathological myopia.
5. The system of claim 2, wherein the deep neural network based pathological myopia detection system comprises: the fundus oculi lesion degree data set is marked as fundus oculi image lesion degree, and comprises five lesion grades: level 0, level 1, level 2, level 3, level 4, respectively, represent: normal, leopard-streak fundus oculi, diffuse choroidal atrophy, macular atrophy, the labeling of the lesion detection dataset includes a lesion position and a lesion type, the lesion position includes a center point coordinate, a width and a height of the lesion, and the lesion type includes CNV, lacquer cracks and Fuchs spots.
6. The system of any one of claims 1 to 5, wherein: the specific prediction process of the pathological myopia detection network model comprises the following steps:
a. preprocessing the fundus image to be predicted to enable the fundus image to meet the input requirement of the pathological myopia detection network model;
b. the fundus image is firstly input into a pathological change degree classifier network to predict the pathological change degree, and if the prediction result is normal: grade 0, the prediction is finished, the network model for detecting the pathological myopia outputs non PM (0)
c. If the prediction result is abnormal: grade 1, grade 2, grade 3, grade 4, then the eyeground picture enters the focus detector network, the focus detected by the focus detector network will be marked, and the focus type and prediction probability are marked;
d. the pathological myopia discriminator synthesizes output information of the focus detector network and the pathological degree classifier network, and judges whether the fundus is suffered from pathological myopia;
e. the pathological myopia detection network model outputs the pathological degree, the focus position and the type of the input fundus image, and whether the fundus image belongs to pathological myopia.
7. The system of claim 6, wherein the deep neural network based pathological myopia detection system comprises: in the step d, the specific method for judging whether the fundus suffers from pathological myopia is as follows: if the pathological degree of the fundus is normal or the pathological degree of the fundus is leopard-streak fundus, but no focus is detected, the myopia is judged to be non-pathological myopia, and if the other conditions are not detected, the myopia is judged to be pathological myopia.
8. The deep neural network-based pathological myopia detection system of claim 1 or 2, wherein: the lesion degree classifier comprises a depth residual error network, a full connection layer and a Softmax classification layer.
9. The deep neural network-based pathological myopia detection system of claim 1 or 2, wherein: the lesion detector comprises a backbone network, an RPN network, an ROI-Powing layer and a classifier.
10. The deep neural network-based pathological myopia detection system of claim 1 or 2, wherein: the pathological myopia discriminator consists of a logic judgment statement.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111391392.5A CN114334124A (en) | 2021-11-23 | 2021-11-23 | Pathological myopia detection system based on deep neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111391392.5A CN114334124A (en) | 2021-11-23 | 2021-11-23 | Pathological myopia detection system based on deep neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114334124A true CN114334124A (en) | 2022-04-12 |
Family
ID=81045871
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111391392.5A Pending CN114334124A (en) | 2021-11-23 | 2021-11-23 | Pathological myopia detection system based on deep neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114334124A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115049655A (en) * | 2022-08-15 | 2022-09-13 | 汕头大学·香港中文大学联合汕头国际眼科中心 | Mouse model retina focus distribution analysis method |
| CN116491892A (en) * | 2023-06-28 | 2023-07-28 | 依未科技(北京)有限公司 | Myopia fundus change assessment method and device and electronic equipment |
-
2021
- 2021-11-23 CN CN202111391392.5A patent/CN114334124A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115049655A (en) * | 2022-08-15 | 2022-09-13 | 汕头大学·香港中文大学联合汕头国际眼科中心 | Mouse model retina focus distribution analysis method |
| CN115049655B (en) * | 2022-08-15 | 2022-11-11 | 汕头大学·香港中文大学联合汕头国际眼科中心 | Mouse model retina focus distribution analysis method |
| US11839428B1 (en) | 2022-08-15 | 2023-12-12 | Joint Shantou International Eye Center of Shantou University and the Chinese University of Hong Kong | Method for analyzing distribution of retinal lesions in mouse model |
| CN116491892A (en) * | 2023-06-28 | 2023-07-28 | 依未科技(北京)有限公司 | Myopia fundus change assessment method and device and electronic equipment |
| CN116491892B (en) * | 2023-06-28 | 2023-09-22 | 依未科技(北京)有限公司 | Myopia fundus change assessment method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bilal et al. | Diabetic retinopathy detection and classification using mixed models for a disease grading database | |
| CN110010219B (en) | Intelligent detection system and detection method for retinopathy by optical coherence tomography | |
| Xiong et al. | Multimodal machine learning using visual fields and peripapillary circular OCT scans in detection of glaucomatous optic neuropathy | |
| CN109635862B (en) | Sorting method for retinopathy of prematurity plus lesion | |
| CN109726743B (en) | Retina OCT image classification method based on three-dimensional convolutional neural network | |
| CN109948719B (en) | Automatic fundus image quality classification method based on residual dense module network structure | |
| CN111046835A (en) | Eyeground illumination multiple disease detection system based on regional feature set neural network | |
| CN114334124A (en) | Pathological myopia detection system based on deep neural network | |
| CN113160119A (en) | Diabetic retinopathy image classification method based on deep learning | |
| Consejo et al. | Detection of subclinical keratoconus with a validated alternative method to corneal densitometry | |
| Chen et al. | Dual-path and multi-scale enhanced attention network for retinal diseases classification using ultra-wide-field images | |
| Parra et al. | Automatic diagnosis of ocular toxoplasmosis from fundus images with residual neural networks | |
| Antal et al. | Evaluation of the grading performance of an ensemble-based microaneurysm detector | |
| Tang et al. | Applications of deep learning for detecting ophthalmic diseases with ultrawide-field fundus images | |
| Thanh et al. | A real-time classification of glaucoma from retinal fundus images using AI technology | |
| CN113273959B (en) | Portable diabetic retinopathy diagnosis and treatment instrument | |
| CN113284140B (en) | Binocular keratoconus diagnosis method based on multi-modal data | |
| Srivastava et al. | Role of the choroid in automated age-related macular degeneration detection from optical coherence tomography images | |
| Hussein et al. | Automatic classification of AMD in retinal images | |
| Poranki et al. | Performance evaluation of ai assisted automotive diabetic retinopathy classification systems | |
| WO2023140469A1 (en) | Device and method for diagnosis of pathologic myopia | |
| Chalakkal | Automatic Retinal Image Analysis to Triage Retinal Pathologies | |
| Aouf et al. | Glaucoma detection from fundus camera image | |
| CN116168255A (en) | Retina OCT (optical coherence tomography) image classification method with robust long tail distribution | |
| Ananya et al. | Glaucoma Detection using HOG based features with Feed-forward Neural Networks and Extreme Learning Machine |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |