CN114898172A - Diabetic retinopathy classification modeling method based on multi-feature DAG network - Google Patents
Diabetic retinopathy classification modeling method based on multi-feature DAG network Download PDFInfo
- Publication number
- CN114898172A CN114898172A CN202210362909.6A CN202210362909A CN114898172A CN 114898172 A CN114898172 A CN 114898172A CN 202210362909 A CN202210362909 A CN 202210362909A CN 114898172 A CN114898172 A CN 114898172A
- Authority
- CN
- China
- Prior art keywords
- image
- retinal
- feature
- features
- characteristic
- 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.)
- Granted
Links
- 206010012689 Diabetic retinopathy Diseases 0.000 title claims abstract description 19
- 238000000034 method Methods 0.000 title claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 23
- 230000002207 retinal effect Effects 0.000 claims abstract description 18
- 210000001525 retina Anatomy 0.000 claims description 21
- 210000004204 blood vessel Anatomy 0.000 claims description 14
- 208000037111 Retinal Hemorrhage Diseases 0.000 claims description 10
- 206010046996 Varicose vein Diseases 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 6
- 230000036285 pathological change Effects 0.000 claims description 6
- 231100000915 pathological change Toxicity 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 230000004044 response Effects 0.000 claims description 6
- 210000001210 retinal vessel Anatomy 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 208000017442 Retinal disease Diseases 0.000 claims description 3
- 206010038923 Retinopathy Diseases 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000004256 retinal image Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 208000027185 varicose disease Diseases 0.000 claims description 2
- 230000002008 hemorrhagic effect Effects 0.000 abstract description 6
- 238000013145 classification model Methods 0.000 abstract description 5
- 206010029113 Neovascularisation Diseases 0.000 abstract description 4
- 238000011156 evaluation Methods 0.000 abstract description 4
- 230000003902 lesion Effects 0.000 abstract description 3
- 206010012601 diabetes mellitus Diseases 0.000 description 6
- 238000000605 extraction Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000002062 proliferating effect Effects 0.000 description 2
- 201000004569 Blindness Diseases 0.000 description 1
- 206010060891 General symptom Diseases 0.000 description 1
- WQZGKKKJIJFFOK-GASJEMHNSA-N Glucose Natural products OC[C@H]1OC(O)[C@H](O)[C@@H](O)[C@@H]1O WQZGKKKJIJFFOK-GASJEMHNSA-N 0.000 description 1
- 208000032843 Hemorrhage Diseases 0.000 description 1
- 206010020710 Hyperphagia Diseases 0.000 description 1
- 208000009857 Microaneurysm Diseases 0.000 description 1
- 208000034698 Vitreous haemorrhage Diseases 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 210000000416 exudates and transudate Anatomy 0.000 description 1
- 208000030533 eye disease Diseases 0.000 description 1
- 239000008103 glucose Substances 0.000 description 1
- 238000011369 optimal treatment Methods 0.000 description 1
- 206010036067 polydipsia Diseases 0.000 description 1
- 208000022530 polyphagia Diseases 0.000 description 1
- 238000004393 prognosis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 210000002700 urine Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B3/00—Apparatus for testing the eyes; Instruments for examining the eyes
- A61B3/10—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions
- A61B3/12—Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for looking at the eye fundus, e.g. ophthalmoscopes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Ophthalmology & Optometry (AREA)
- Biophysics (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a diabetic retinopathy classification modeling method based on a multi-feature DAG network, which comprises the steps of firstly, extracting index features of diabetic retinopathy by using different methods, wherein the index features comprise hemorrhagic spot features, fundus neovascularization features and retinal varicosis features; secondly, an optimized DAG network is built, and a training scheme is continuously changed to train the network, so that the extracted features are fused in a multi-feature mode, and complex local or global features are formed through local features, so that an object is restored; finally, normal and lesion classification is performed by the softmax classifier. The performance evaluation of the classification model is carried out by using the DIARETDB1 data set and the third people's hospital data in the large connected city, and the result shows that the classification accuracy of the classification model is 98.7% and 98.5% respectively for the DIARETDB1 data set image and the hospital data.
Description
Technical Field
The invention relates to the field of medical image processing, in particular to a diabetic retinopathy classification modeling method based on a multi-feature DAG network.
Background
Diabetic retinopathy is a relatively serious eye disease causing blindness at present. Diabetic retinopathy is classified into stage VI, the first three stages are non-proliferative stages, and the last three stages are proliferative stages. Except general symptoms such as polydipsia, polyphagia, urine glucose and blood sugar rise, the diabetic patients have fundus changes which are mainly characterized by punctate hemorrhage, neovascularization and varicose blood vessels on retinas of two eyes, so that the characteristics of each period of the diabetic retinas have great significance for the diagnosis and estimation prognosis of diabetes. In the past, ophthalmologists manually evaluate fundus images according to diabetic retinopathy characteristics to detect whether retinas are diseased or not, but because some patients are in the early stage of diabetic retinopathy and the characteristics are not obvious, the problem that the optimal treatment time is missed due to inaccurate evaluation easily occurs. Therefore, a fast and accurate diabetic retinopathy image automatic identification and classification system is needed, and the diabetic retinopathy images with unobvious features are also identified.
At present, classification of retinal images is realized by adopting a characteristic extraction mode, such as extraction of microaneurysm characteristics, exudate characteristics, vitreous hemorrhage characteristics and the like, and extraction of diabetic retinal hemorrhage spot characteristics, fundus neovascularization characteristics and retinal varicosis characteristics is not performed. Meanwhile, the existing research only extracts one or two kinds of feature information, so that the classification model cannot learn the features carefully and comprehensively, and the classification accuracy is low.
Disclosure of Invention
The invention aims to solve the technical problems in the prior art and provides a diabetic retinopathy classification modeling method based on a multi-feature DAG network.
The technical solution of the invention is as follows: a classification modeling method for diabetic retinopathy based on a multi-feature DAG network is carried out according to the following steps:
step 1: preprocessing each retina image in the training set to obtain a characteristic image training set
Step 1.1 obtaining the characteristic image of the retinal hemorrhage spot
Extracting a green channel image in an RGB color mode of the retina image, graying and dividing the green channel image into a plurality of sub-imagesBlock, counting the cumulative distribution histogram of each sub-block, setting a finite threshold value T in the histogram c ：
T c ＝max(1,T d ×h×w/S)
In the formula, T d Is the iterative adaptive soft threshold, S is the total pixels of the image, h and w are the length and width of the image;
comparing the gray value in the histogram with a set limited threshold value T c Comparing the histogram with a threshold T c The gray value areas are uniformly distributed below the histogram and the total area of the histogram is ensured to be unchanged, and finally, the optimization processing is carried out by using a linear interpolation method to make the retinal hemorrhagic spot characteristic prominent, namely, a retinal hemorrhagic spot characteristic image is obtained;
step 2.2, acquiring the characteristic image of the eyeground new blood vessel
Firstly, 8 masks M are used for each pixel point in the retina image q 1, 2.., 7 take the derivatives of the convolution, each mask M q Making maximum responses to 8 specific edge directions, and taking the maximum value G of the maximum responses as the output of an edge amplitude image:
G＝max{|M 0 |,|M 1 |,|M 2 |,|M 3 |,|M 4 |,|M 5 |,|M 6 |,|M 7 |}
finally, carrying out binarization processing on the image according to a self-adaptive soft threshold value to enable the characteristics of the new blood vessels to be prominent, and obtaining a fundus new blood vessel characteristic image;
step 3.3 obtaining retinal varicose characteristic image
Dividing the retina image into N sub-blocks, and iteratively calculating the clustering center C of retinal blood vessels in the retinopathy image v And corresponding degree of membership D pv ：
Wherein N is the total number of subblocks, C is the number of clusters of the cluster, x p P 1, 2., N denotes the p-th sub-block, | | x | | | denotes a metric of arbitrary distance, and m ∈ [1, ∞) belongs to a weighted index; c k A cluster center representing a kth class; when the membership degree meets the following iteration termination condition, stopping iteration and calculating a local optimal value J m Making the retinal varicose feature prominent to obtain a retinal varicose feature image;
wherein l is the number of iteration steps, and epsilon is the error threshold;
step 3.4, the obtained retina characteristic image set is used as a characteristic image training set;
step 2: inputting the characteristic images in the characteristic image training set into a DAG network for training
Step 2.1, establishing an optimized DAG network, wherein the optimized DAG network is composed of a trunk, branches, an add layer, a pooling layer avpool and a Full connection layer Full connect, the trunk is divided into five groups, each group is composed of a convolution layer Conv, a normalization layer BN and an activation function layer relu, the branches are convolution layers skipConv, and the trunk and the branches are simultaneously linked with the add layer;
step 2.2, inputting the images of the feature image training set into the optimized DAG network, realizing multi-feature fusion and learning, and outputting a multi-feature fusion result F i add ：
F i add ＝(X i +Y i +Z i )*K＝X i *K+Y i *K+Z i *K
Wherein K represents convolution kernel, X represents convolution i Indicating the characteristics of retinal hemorrhage plaques, Y i Representing the characteristics of the fundus new blood vessels; z i Representing retinal varicose vein features;
and step 3: fusing the multi-feature results F i add Sending the data to a softmax classifier, calculating the prediction probability of normal or pathological changes according to the following formula, and realizing effective classification of diabetic retinopathy and normal;
wherein, R is a prediction probability, i is 1,2, M represents the ith image, M is the total number of retinal images, and e is a parameter;
when Y belongs to (0,0.5), the image is judged to be normal, and when Y belongs to [0.5,1), the image is judged to be pathological change.
Firstly, extracting index characteristics of diabetic retinopathy by using different methods, wherein the index characteristics comprise hemorrhagic spot characteristics, fundus neovascular characteristics and retinal varicose blood vessel characteristics; secondly, an optimized DAG network is built, and a training scheme is continuously changed to train the network, so that the extracted features are fused in a multi-feature mode, and complex local or global features are formed through local features, so that an object is restored; finally, normal and lesion classification is performed by the softmax classifier. The performance evaluation of the classification model is carried out by using the DIARETDB1 data set and the third people's hospital data in the large connected city, and the result shows that the classification accuracy of the classification model is 98.7% and 98.5% respectively for the DIARETDB1 data set image and the hospital data.
Drawings
FIG. 1 is a graph showing the result of extracting retinal hemorrhage spot characteristics according to the embodiment of the present invention.
Fig. 2 is a result chart of the embodiment of the invention for extracting the characteristics of the fundus neovascularization.
Fig. 3 is a diagram illustrating the result of extracting the retinal varicose vein feature according to the embodiment of the invention.
Fig. 4 is a diagram of a DAG network structure optimized according to an embodiment of the present invention.
FIG. 5 is an overall flow chart of an embodiment of the present invention.
Detailed Description
The invention discloses a diabetic retinopathy classification modeling method based on a multi-feature DAG network, which is shown in FIG. 5 and comprises the following steps:
step 1: taking DIARETDB1 data set and third people hospital data set (hospital data for short) in Dalian city, dividing the data sets into a training set and a testing set, preprocessing each diabetic retina image in the training set, and obtaining a characteristic image training set
Step 1.1 obtaining the characteristic image of the retinal hemorrhage spot
Extracting a green channel image in an RGB color mode of a retina image, graying the green channel image and dividing the green channel image into a plurality of sub-blocks, counting an accumulative distribution histogram of each sub-block, and setting a limited threshold value T in the histogram c ：
T c ＝max(1,T d ×h×w/S)
In the formula, T d Is the iterative adaptive soft threshold, S is the total pixels of the image, h and w are the length and width of the image;
comparing the gray value in the histogram with a set limited threshold value T c Comparing the histogram with a threshold T c The gray value regions are uniformly distributed below the histogram and the total area of the histogram is ensured to be unchanged, and finally, the transition problem of each sub-block is optimized and processed by using a linear interpolation method, so that the retinal hemorrhagic spot features are highlighted, that is, a retinal hemorrhagic spot feature image is obtained, as shown in fig. 1;
step 2.2, acquiring the characteristic image of the eyeground new blood vessel
Firstly, 8 masks M are used for each pixel point in the retina image q 1, 2.., 7 take the derivatives of the convolution, each mask M q Making maximum responses to 8 specific edge directions, and taking the maximum value G of the maximum responses as the output of an edge amplitude image:
G＝max{|M 0 |,|M 1 |,|M 2 |,|M 3 |,|M 4 |,|M 5 |,|M 6 |,|M 7 |}
finally, carrying out binarization processing on the image according to a self-adaptive soft threshold value to enable the characteristics of the new blood vessels to be prominent, and obtaining a fundus new blood vessel characteristic image as shown in figure 2;
step 3.3 obtaining retinal varicose characteristic image
Dividing the retina image into N subblocks, and iteratively calculating the clustering center C of retinal blood vessels in the retinopathy image v And corresponding degree of membership D pv ：
Wherein N is the total number of subblocks, C is the number of clusters of the cluster, x p N denotes the pth sub-block, | | denotes a metric of arbitrary distance, and m ∈ [1, ∞) belongs to a weighted index; c k A cluster center representing a kth class; when the membership degree meets the following iteration termination condition, stopping iteration and calculating a local optimal value J m Making the retinal vessel varicosity characteristic prominent to obtain a retinal vessel varicosity characteristic image, as shown in figure 3;
wherein l is the number of iteration steps, and epsilon is the error threshold;
step 3.4, the obtained retina characteristic image set is used as a characteristic image training set;
step 2: inputting the characteristic images in the characteristic image training set into a DAG network for training
Step 2.1, establishing an optimized DAG network, wherein the optimized DAG network is composed of a main trunk, two branches, an add layer, a pooling layer avpool and a Full connection layer Full connect (fc) as shown in FIG. 4, the main trunk is divided into five groups, each group is composed of a convolution layer Conv, a normalization layer BN and an activation function layer relu, the five groups are composed of a convolution layer Conv1-5, a normalization layer BN1-5 and an activation function layer relu1-5, the two branches are convolution layer skipConv-1 and skipConv-2, and the main trunk and the two branches are simultaneously linked with the add layer;
step 2.2, inputting the images of the feature image training set into a DAG network, realizing multi-feature fusion and learning, and outputting a multi-feature fusion result F i add ：
F i add ＝(X i +Y i +Z i )*K＝X i *K+Y i *K+Z i *K
Wherein K represents convolution kernel, X represents convolution i Indicating the characteristics of retinal hemorrhage plaques, Y i Representing the characteristics of the fundus new blood vessels; z i Representing retinal varicose veins characteristics;
the training parameters and values of the training parameters are shown in table 1.
TABLE 1
And 3, step 3: fusing the results F with multiple features i add Sending the data to a softmax classifier, calculating the prediction probability of normal or pathological changes according to the following formula, and realizing effective classification of diabetic retinopathy and normal;
wherein, R is a prediction probability, i is 1,2, M represents the ith image, M is the total number of diabetic retina images, and e is a parameter;
when Y belongs to (0,0.5), the image is judged to be normal, and when Y belongs to [0.5,1), the image is judged to be pathological change.
Experiment:
the DIARETDB1 data set and the test set image in the third people hospital data set (hospital data set for short) in Dalian city are input into the model established by the embodiment of the invention, and the model identifies and classifies the retina image of the data set into two categories, namely a normal fundus image and a lesion fundus image. Aiming at the two-classification problem of the retina images, the evaluation indexes of the two-classification problem are the classification accuracy (accuracy), Precision (Precision), Recall (Recall), Specificity (Specificity) and F1-score obtained after the data test set images enter the model to jointly evaluate the performance of the model, so that the model is more stable and reliable. The correlation formula is as follows:
in the formula, TP represents the number of correctly classified positive samples; TN represents the number of correctly classified negative examples; FP represents the number of false positive samples in the negative samples; FN represents the number of false marks as negative samples in the positive samples.
In order to prove the importance of the extracted features, the invention uses hospital data to respectively carry out self algorithms of not extracting the features and only extracting one or two of the features to carry out experimental result comparison. The comparative results are shown in Table 2.
TABLE 2
Meanwhile, in order to verify the performance of the model of the present invention, the same data set DIARETDB1 was selected for performance comparison between the conventional model and the model of the present invention, and the comparison result is shown in Table 3.
TABLE 3
Note: '- -' represents the deficiency.
Claims (1)
1. A classification modeling method for diabetic retinopathy based on a multi-feature DAG network is characterized by comprising the following steps:
step 1: preprocessing each retina image in the training set to obtain a characteristic image training set
Step 1.1 obtaining the characteristic image of the retinal hemorrhage spot
Extracting a green channel image in an RGB color mode of a retina image, graying and dividing the green channel image into a plurality of sub-blocks, counting an accumulative distribution histogram of each sub-block, and setting a limited threshold value T in the histogram c ：
T c ＝max(1,T d ×h×w/S)
In the formula, T d Is the iterative adaptive soft threshold, S is the total pixels of the image, h and w are the length and width of the image;
comparing the gray value in the histogram with a set limited threshold value T c Comparing the histogram with a threshold T c The gray value areas are uniformly distributed below the histogram and the total area of the histogram is ensured to be unchanged, and finally, the optimization processing is carried out by using a linear interpolation method to make the retinal bleeding spot characteristic prominent, namely, a retinal bleeding spot characteristic image is obtained;
step 2.2, acquiring the characteristic image of the eyeground new blood vessel
Firstly, 8 masks M are used for each pixel point in the retina image q 1, 2.., 7 take the derivatives of the convolution, each mask M q Making maximum responses to 8 specific edge directions, and taking the maximum value G of the maximum responses as the output of an edge amplitude image:
G＝max{|M 0 |,|M 1 |,|M 2 |,|M 3 |,|M 4 |,|M 5 |,|M 6 |,|M 7 |}
finally, carrying out binarization processing on the image according to a self-adaptive soft threshold value to enable the characteristics of the new blood vessels to be prominent, and obtaining a fundus new blood vessel characteristic image;
step 3.3 obtaining retinal varicose characteristic image
Dividing the retina image into N sub-blocks, and iteratively calculating the clustering center C of retinal blood vessels in the retinopathy image v And corresponding degree of membership D pv ：
Wherein N is the total number of subblocks, C is the number of clusters of the cluster, x p P 1, 2., N denotes the p-th sub-block, | | x | | | denotes a metric of arbitrary distance, and m ∈ [1, ∞) belongs to a weighted index; c k A cluster center representing a kth class; when the membership degree meets the following iteration termination condition, stopping iteration and calculating a local optimal value J m Making the retinal varicose feature prominent to obtain a retinal varicose feature image;
wherein l is the number of iteration steps, and epsilon is the error threshold;
step 3.4, the obtained retina characteristic image set is used as a characteristic image training set;
step 2: inputting the characteristic images in the characteristic image training set into a DAG network for training
Step 2.1, establishing an optimized DAG network, wherein the optimized DAG network is composed of a trunk, branches, an add layer, a pooling layer avpool and a Full connection layer Full connect, the trunk is divided into five groups, each group is composed of a convolution layer Conv, a normalization layer BN and an activation function layer relu, the branches are convolution layers skipConv, and the trunk and the branches are simultaneously linked with the add layer;
step 2.2, inputting the images of the feature image training set into the optimized DAG network, realizing multi-feature fusion and learning, and outputting a multi-feature fusion result F i add ：
F i add ＝(X i +Y i +Z i )*K＝X i *K+Y i *K+Z i *K
Wherein K represents convolution kernel, X represents convolution i Indicating the characteristics of retinal hemorrhage plaques, Y i Representing the characteristics of the fundus new blood vessels; z i Representing retinal varicose veins characteristics;
and step 3: fusing the results F with multiple features i add Sending the data to a softmax classifier, calculating the prediction probability of normal or pathological changes according to the following formula, and realizing effective classification of diabetic retinopathy and normal;
wherein, R is a prediction probability, i is 1,2, M represents the ith image, M is the total number of retinal images, and e is a parameter;
when Y belongs to (0,0.5), the image is judged to be normal, and when Y belongs to [0.5,1), the image is judged to be pathological change.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210362909.6A CN114898172B (en) | 2022-04-08 | 2022-04-08 | Multi-feature DAG network-based diabetic retinopathy classification modeling method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210362909.6A CN114898172B (en) | 2022-04-08 | 2022-04-08 | Multi-feature DAG network-based diabetic retinopathy classification modeling method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114898172A true CN114898172A (en) | 2022-08-12 |
| CN114898172B CN114898172B (en) | 2024-04-02 |
Family
ID=82715928
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210362909.6A Active CN114898172B (en) | 2022-04-08 | 2022-04-08 | Multi-feature DAG network-based diabetic retinopathy classification modeling method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114898172B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116864109A (en) * | 2023-07-13 | 2023-10-10 | 中世康恺科技有限公司 | Medical image artificial intelligence auxiliary diagnosis system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108021916A (en) * | 2017-12-31 | 2018-05-11 | 南京航空航天大学 | Deep learning diabetic retinopathy sorting technique based on notice mechanism |
| CN110210570A (en) * | 2019-06-10 | 2019-09-06 | 上海延华大数据科技有限公司 | The more classification methods of diabetic retinopathy image based on deep learning |
| WO2019196268A1 (en) * | 2018-04-13 | 2019-10-17 | 博众精工科技股份有限公司 | Diabetic retina image classification method and system based on deep learning |
-
2022
- 2022-04-08 CN CN202210362909.6A patent/CN114898172B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108021916A (en) * | 2017-12-31 | 2018-05-11 | 南京航空航天大学 | Deep learning diabetic retinopathy sorting technique based on notice mechanism |
| WO2019196268A1 (en) * | 2018-04-13 | 2019-10-17 | 博众精工科技股份有限公司 | Diabetic retina image classification method and system based on deep learning |
| CN110210570A (en) * | 2019-06-10 | 2019-09-06 | 上海延华大数据科技有限公司 | The more classification methods of diabetic retinopathy image based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 李琼;柏正尧;刘莹芳;: "糖尿病性视网膜图像的深度学习分类方法", 中国图象图形学报, no. 10, 31 October 2018 (2018-10-31), pages 1594 - 1603 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116864109A (en) * | 2023-07-13 | 2023-10-10 | 中世康恺科技有限公司 | Medical image artificial intelligence auxiliary diagnosis system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114898172B (en) | 2024-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bilal et al. | Diabetic retinopathy detection and classification using mixed models for a disease grading database | |
| CN109635862B (en) | Sorting method for retinopathy of prematurity plus lesion | |
| Tian et al. | Multi-path convolutional neural network in fundus segmentation of blood vessels | |
| CN109726743B (en) | Retina OCT image classification method based on three-dimensional convolutional neural network | |
| CN114926477B (en) | Brain tumor multi-mode MRI image segmentation method based on deep learning | |
| CN110751636B (en) | Fundus image retinal arteriosclerosis detection method based on improved coding and decoding network | |
| CN112837805B (en) | Eyelid topological morphology feature extraction method based on deep learning | |
| Raja et al. | Analysis of vasculature in human retinal images using particle swarm optimization based Tsallis multi-level thresholding and similarity measures | |
| CN112419248B (en) | Ear sclerosis focus detection and diagnosis system based on small target detection neural network | |
| KR102537470B1 (en) | Diagnosis method for Graves’orbitopathy using orbital computed tomography (CT) based on neural network-based algorithm | |
| CN113786185B (en) | Static brain network feature extraction method and system based on convolutional neural network | |
| CN115035127B (en) | Retina blood vessel segmentation method based on generation type countermeasure network | |
| CN113610118A (en) | Fundus image classification method, device, equipment and medium based on multitask course learning | |
| CN114038564A (en) | Noninvasive risk prediction method for diabetes | |
| Aranha et al. | Deep transfer learning strategy to diagnose eye-related conditions and diseases: An approach based on low-quality fundus images | |
| Zeng et al. | Automated detection of diabetic retinopathy using a binocular siamese-like convolutional network | |
| CN114898172B (en) | Multi-feature DAG network-based diabetic retinopathy classification modeling method | |
| Bali et al. | Analysis of deep learning techniques for prediction of eye diseases: A systematic review | |
| Elsawy et al. | A novel network with parallel resolution encoders for the diagnosis of corneal diseases | |
| CN112102234B (en) | Ear sclerosis focus detection and diagnosis system based on target detection neural network | |
| CN110610480A (en) | MCASPP neural network eyeground image optic cup optic disc segmentation model based on Attention mechanism | |
| CN117058467A (en) | Gastrointestinal tract lesion type identification method and system | |
| CN117496217A (en) | Premature infant retina image detection method based on deep learning and knowledge distillation | |
| Sharma et al. | Automatic glaucoma diagnosis in digital fundus images using convolutional neural network | |
| Balakrishnan et al. | A hybrid PSO-DEFS based feature selection for the identification of diabetic retinopathy |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |