CN106530295A - Fundus image classification method and device of retinopathy - Google Patents
Fundus image classification method and device of retinopathy Download PDFInfo
- Publication number
- CN106530295A CN106530295A CN201610978501.6A CN201610978501A CN106530295A CN 106530295 A CN106530295 A CN 106530295A CN 201610978501 A CN201610978501 A CN 201610978501A CN 106530295 A CN106530295 A CN 106530295A
- Authority
- CN
- China
- Prior art keywords
- eye fundus
- fundus image
- image
- eye
- diabetic retinopathy
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30041—Eye; Retina; Ophthalmic
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Eye Examination Apparatus (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a fundus image classification method and device of retinopathy. The method comprises a step of obtaining the fundus image of an eye detection patient, a step of preprocessing the fundus image and obtaining a corresponding grey fundus image, a step of extracting a preset image characteristic in the grey fundus image, a step of analyzing the preset image characteristic according to the diabetic retinopathy classification model of the fundus image, and obtaining the classification result of the corresponding fundus image, a step of extracting the preset image characteristic in the preprocessed grey fundus image, and carrying out prediction classification on the above preset image characteristic through the diabetic retinopathy classification model established according to a support vector machine and a deep neural network so as to obtain a final classification result. The artificial participation is not needed in the whole process, the workload of doctors is reduced, the high standard requirement of the medical qualification level of the doctor is reduced, the accuracy of classification is improved, and the method and the device have great significance for specific applications and medical research.
Description
Technical field
The present invention relates to technical field of image processing, classifies in particular to a kind of eye fundus image of PVR
Method and apparatus.
Background technology
Eye disease patient before eye examination is carried out was required to shoot its fundus photograph, and then, oculist passes through
The readability of the main portions on eyeground in fundus photograph is manually checked and analyzed, and thus tentatively judges patient whether with DR
The order of severity of (Diabetic Retinopathy, diabetic retinopathy) and suffered from DR.
Specifically, oculist obtains the substantial amounts of fundus photograph image of patient and to above-mentioned substantial amounts of fundus photograph image
Checked, then above-mentioned fundus photograph image is analyzed by personal experience, to carry out to above-mentioned fundus photograph image
Whether classification, judge patient with DR (Diabetic Retinopathy, diabetic retina finally according to above-mentioned classification situation
Disease) and suffered from DR the order of severity.
But, in above-mentioned diagnostic method, needs manually carry out classification process to substantial amounts of fundus photograph image so that doctor
Workload it is larger and higher to the doctor of doctor money level requirement to above-mentioned diagnostic method.
The content of the invention
In view of this, the purpose of the embodiment of the present invention be provide a kind of PVR eye fundus image sorting technique and
Device, it is possible to increase the accuracy classified by the eye fundus image of patient is detected to eye.
In a first aspect, embodiments providing a kind of eye fundus image sorting technique of PVR, methods described
Including：
Obtain the eye fundus image that eye detects patient；
The eye fundus image is pre-processed, the corresponding gray scale eye fundus image of the eye fundus image is obtained；
Extract the pre-set image feature in the gray scale eye fundus image；The pre-set image feature at least includes following characteristics
In one or more：Wavelet character and textural characteristics；
The pre-set image feature is analyzed according to the diabetic retinopathy disaggregated model of the eye fundus image
Process, obtain the classification results of the corresponding eye fundus image.
With reference in a first aspect, embodiments provide the first possible embodiment of first aspect, wherein, institute
State and the eye fundus image is pre-processed, obtain the corresponding gray scale eye fundus image of the eye fundus image, including：
Extract the G channel images in the RGB color space of the eye fundus image；
Calculating process is carried out to the G channel images by height cap algorithm, obtain target object and background contrasts compared with
High G channel images, and/or, partial histogram equalization process is carried out to the G channel images, target object is obtained with the back of the body
The higher G channel images of scape contrast；
Dividing processing is carried out to the target object and the higher G channel images of background contrasts, selected target thing is extracted
The target image of body.
With reference to the first possible embodiment of first aspect, second of first aspect is embodiments provided
Possible embodiment, wherein, the pre-set image feature extracted in the gray scale eye fundus image, including：
Wavelet transform process is carried out to the target image, wavelet character is obtained；
And/or,
Texture analysis process is carried out to the target image, the textural characteristics in the target image are extracted；The texture
Feature includes：Brightness, gray level co-occurrence matrixes and intermediate value run-length matrix.
With reference to second possible embodiment of first aspect, embodiments provide first aspect the third
Possible embodiment, wherein, the diabetic retinopathy disaggregated model according to the eye fundus image is to described default
Characteristics of image is analyzed process, obtains the classification results of the corresponding eye fundus image, including：
Extract all characteristic vectors in the pre-set image feature；
Obtain the corresponding weight system of each characteristic vector in the diabetic retinopathy disaggregated model of the eye fundus image
Number；
All characteristic vectors and the corresponding weight coefficient of each characteristic vector are calculated, and by result of calculation institute
The corresponding classification results of threshold range as the eye fundus image classification results.
With reference in a first aspect, embodiments provide the 4th kind of possible embodiment of first aspect, wherein, institute
The diabetic retinopathy disaggregated model for stating eye fundus image is generated according to following method：
The eye fundus image containing preset kind is chosen as training sample；The preset kind includes：Normal eye's base map
Picture, proliferative diabetic retinopathy PDR eye fundus images and non-appreciation type diabetic retinopathy NPDR eye fundus images；
Calculating is trained to the training sample by SVMs and deep neural network, eye fundus image is set up
Diabetic retinopathy disaggregated model.
Second aspect, the embodiment of the present invention additionally provide a kind of eye fundus image sorter of PVR, the dress
Put including：
Acquisition module, detects the eye fundus image of patient for obtaining eye；
Pretreatment module, for pre-processing to the eye fundus image, obtains the corresponding gray scale eye of the eye fundus image
Base map picture；
Extraction module, for extracting the pre-set image feature in the gray scale eye fundus image；The pre-set image feature is extremely
Less including one or more in following characteristics：Wavelet character and textural characteristics；
Analysis and processing module, for the diabetic retinopathy disaggregated model according to the eye fundus image to described default
Characteristics of image is analyzed process, obtains the classification results of the corresponding eye fundus image.
With reference to second aspect, the first possible embodiment of second aspect is embodiments provided, wherein, institute
Pretreatment module is stated, including：
First extraction unit, for extracting the G channel images in the RGB color space of the eye fundus image；
Height cap algorithm computing unit, for carrying out calculating process to the G channel images by height cap algorithm, obtains
Target object and the higher G channel images of background contrasts；And/or, partial histogram equalization processing unit, for described
G channel images carry out partial histogram equalization process, obtain target object and the higher G channel images of background contrasts；
Dividing processing unit, for carrying out segmentation portion to the target object and the higher G channel images of background contrasts
Reason, extracts the target image of selected target object.
With reference to the first possible embodiment of second aspect, second of second aspect is embodiments provided
Possible embodiment, wherein, the extraction module, including：
Wavelet transform process unit, for carrying out wavelet transform process to the target image, obtains wavelet character；
Second extraction unit, for texture analysis process is carried out to the target image, extracts in the target image
Textural characteristics；The textural characteristics include：Brightness, gray level co-occurrence matrixes and intermediate value run-length matrix.
With reference to second possible embodiment of second aspect, embodiments provide second aspect the third
Possible embodiment, wherein, the analysis and processing module, including：
3rd extraction unit, for extracting all characteristic vectors in the pre-set image feature；
Acquiring unit, for obtaining each characteristic vector in the diabetic retinopathy disaggregated model of the eye fundus image
Corresponding weight coefficient；
Computing unit, for calculating to all characteristic vectors and the corresponding weight coefficient of each characteristic vector,
And using result of calculation place threshold range corresponding classification results as the eye fundus image classification results.
With reference to second aspect, the 4th kind of possible embodiment of second aspect is embodiments provided, wherein, institute
The eye fundus image sorter of PVR is stated, is also included：
Sample chooses module, for choosing the eye fundus image containing preset kind as training sample；The preset kind
Including：Normal person's eye fundus image, proliferative diabetic retinopathy PDR eye fundus images and non-appreciation type diabetic retina
Pathology NPDR eye fundus image；
Model training module, based on being trained to the training sample by SVMs and deep neural network
Calculate, set up the diabetic retinopathy disaggregated model of eye fundus image.
A kind of eye fundus image sorting technique and device of PVR provided in an embodiment of the present invention, including：Obtain eye
Detect the eye fundus image of patient in portion；Above-mentioned eye fundus image is pre-processed, corresponding gray scale eye fundus image is obtained；Extract the ash
Pre-set image feature in degree eye fundus image；According to the diabetic retinopathy disaggregated model of eye fundus image to above-mentioned default figure
As feature is analyzed process, the classification results of corresponding eye fundus image are obtained, with diagnostic method of the prior art, need people
Work carries out classification process to substantial amounts of fundus photograph image so that the workload of doctor is larger to be compared, and which passes through from pretreatment
Pre-set image feature, and the diabetes by setting up according to SVMs and deep neural network are extracted in gray scale eye fundus image
PVR disaggregated model is predicted classification to above-mentioned pre-set image feature, to obtain final classification result；It is above-mentioned whole
Process reduces the workload of doctor without the need for manually participating in, and reduces the high standards of the doctor's money level to doctor, and improves
The accuracy of classification, has great significance to concrete application and medical research.
To enable the above objects, features and advantages of the present invention to become apparent, preferred embodiment cited below particularly, and coordinate
Appended accompanying drawing, is described in detail below.
Description of the drawings
In order to be illustrated more clearly that the technical scheme of the embodiment of the present invention, below by to be used attached needed for embodiment
Figure is briefly described, it will be appreciated that the following drawings illustrate only certain embodiments of the present invention, thus be not construed as it is right
The restriction of scope, for those of ordinary skill in the art, on the premise of not paying creative work, can be with according to this
A little accompanying drawings obtain other related accompanying drawings.
Fig. 1 shows a kind of flow process of the eye fundus image sorting technique of PVR that the embodiment of the present invention is provided
Figure；
Fig. 2 shows the schematic diagram of eye fundus image in the embodiment of the present invention；
Fig. 3 shows and the eye fundus image pre-processed, and obtains the corresponding gray scale eye fundus image of the eye fundus image
Flow chart；
Fig. 4 is shown according to the diabetic retinopathy disaggregated model of the eye fundus image to the pre-set image feature
Process is analyzed, the flow chart for obtaining the classification results of the corresponding eye fundus image；
Fig. 5 shows that a kind of structure of the eye fundus image sorter of PVR that the embodiment of the present invention is provided is shown
It is intended to；
Fig. 6 is shown in a kind of eye fundus image sorter of PVR that the embodiment of the present invention is provided at analysis
The structural representation of reason module；
Fig. 7 shows the structure of the eye fundus image sorter of another kind of PVR provided by the embodiment of the present invention
Schematic diagram.
Major Symbol explanation：
10th, acquisition module；20th, pretreatment module；30th, extraction module；40th, analysis and processing module；50th, sample chooses mould
Block；60th, model training module；401st, the 3rd extraction unit；402nd, acquiring unit；403rd, computing unit.
Specific embodiment
To make purpose, technical scheme and the advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
Middle accompanying drawing, is clearly and completely described to the technical scheme in the embodiment of the present invention, it is clear that described embodiment is only
It is a part of embodiment of the invention, rather than the embodiment of whole.The present invention generally described and illustrated in accompanying drawing herein is real
The component for applying example can be arranged and be designed with a variety of configurations.Therefore, below to the present invention's that provides in the accompanying drawings
The detailed description of embodiment is not intended to limit the scope of claimed invention, but is merely representative of the selected reality of the present invention
Apply example.Based on embodiments of the invention, the institute obtained on the premise of creative work is not made by those skilled in the art
There is other embodiment, belong to the scope of protection of the invention.
At present, oculist is required to obtain the substantial amounts of fundus photograph image of patient and to above-mentioned in eye examination
Substantial amounts of fundus photograph image is checked, then above-mentioned fundus photograph image is analyzed by personal experience, with to upper
State fundus photograph image to be classified, judge patient whether with DR (Diabetic finally according to above-mentioned classification situation
Retinopathy, diabetic retinopathy) and suffered from DR the order of severity.
And computer based eye examination method, the computer research people of image procossing is done accordingly for each class
For member, its focus is to focus on the process of image itself, and few people see the Jing of eye fundus image based on medical practitioner
The diagnostic classification division work for testing to do eye fundus image, and see the experience of eye fundus image to make eyeground picture based on medical practitioner
Diagnostic classification division work has great significance to concrete application and medical research.
In view of in above-mentioned diagnostic method, needs manually carry out classification process to substantial amounts of eye fundus image so that doctor's
Larger and higher to doctor's money level requirement of doctor to the above-mentioned diagnostic method problem of workload.Based on this, the embodiment of the present invention
There is provided the eye fundus image sorting technique and device of a kind of PVR, which is the eyeground figure based on computer to eye patient
As being classified, on the one hand can mitigate workload for doctor, improve operating efficiency；On the other hand can be to DR recovery feelings
Condition is done and is estimated, patient and doctor can predict postoperative patient vision restoration situation in advance.
At present, the increase with aging population and people to own health attention degree, shoots eye fundus image number
Correspondingly can increase, future has more eye fundus images to be needed to be classified, and this eye fundus image database little by little can be sent out
Transform into as big data, this automatic grading system can greatly reduce human resources, can also make operating efficiency improve a lot.
Below by eye fundus image sorting technique and device of the embodiment to PVR provided in an embodiment of the present invention
It is described.
With reference to Fig. 1, a kind of eye fundus image sorting technique of PVR, methods described bag are embodiments provided
Include：
S101, the eye fundus image for obtaining eye detection patient.
In the embodiment of the present invention, eye can be shot by medical science eye dedicated video camera and detect that patient (hereinafter referred to as suffers from
Person) eye fundus image；Specifically, patient can arrive the spot for photography (such as hospital) of specialty, and under the arrangement of staff, shoot
Corresponding eye fundus image.
For each eye detects patient, the two of patient eyes can all be checked, it is corresponding, need to obtain respectively
The left eye and right eye for taking eye detection patient distinguishes corresponding eye fundus image；Any one eyes of patient can also be examined
Look into, it is corresponding, the corresponding eye fundus image of the eyes of the corresponding inspection of eyes to be checked of patient is obtained, with reference to figure
2, Fig. 2 schematic diagrames for showing the eye fundus image described in the embodiment of the present invention.
In the embodiment of the present invention, during classifying to eye fundus image, the eyeground figure of each eye to be checked
As being one, or multiple；If multiple, then the classification results for obtaining majority are used as final classification results.
S102, the eye fundus image is pre-processed, obtain the corresponding gray scale eye fundus image of the eye fundus image.
Specifically, the eye fundus image for obtaining in step 101 is coloured image, and which includes three passages, the embodiment of the present invention
By pre-processing to the color graphics, and then the part of selection gray image therein, that is, select G passages figure therein
Picture, in order to follow-up preferably to extract pre-set image feature.
S103, the pre-set image feature extracted in the gray scale eye fundus image；The pre-set image feature at least include with
One or more in lower feature：Wavelet character and textural characteristics.
Wherein, above-mentioned specific textural characteristics are to include 1 brightness, 14 gray level co-occurrence matrixes features and 4
Value run-length matrix feature.
S104, the pre-set image feature is carried out according to the diabetic retinopathy disaggregated model of the eye fundus image
Analyzing and processing, obtains the classification results of the corresponding eye fundus image.
In the embodiment of the present invention, the diabetic retinopathy disaggregated model of above-mentioned eye fundus image is given birth to according to following method
Into：
The eye fundus image containing preset kind is chosen as training sample；The preset kind of above-mentioned training sample includes：Just
Ordinary person's eye fundus image, PDR (Physician ' s Desk Reference, proliferative diabetic retinopathy) eye fundus image
With NPDR (Nonproliferative Diabetic Retinopathy, nonproliferative diabetic retinopathy) eyeground figure
Picture；Then, calculating is trained to above-mentioned training sample by SVMs and deep neural network, sets up eye fundus image
Diabetic retinopathy disaggregated model.
Then, according to set up above-mentioned eye fundus image diabetic retinopathy disaggregated model to extract pre-set image
Feature is predicted classification, to obtain corresponding prediction classification results, wherein, prediction classification results include：Normally, PDR and
NPDR.Wherein, in the mistake for being predicted classification according to diabetic retinopathy disaggregated model to the pre-set image feature extracted
Journey includes：Ballot point is carried out to three kinds of prediction classification results by the assembled classifier in diabetic retinopathy disaggregated model
Class, using the votes using assembled classifier most classification results as the final corresponding classification results of the eye fundus image, with
Determine the classification of eye fundus image.
The eye fundus image sorting technique of a kind of PVR provided in an embodiment of the present invention, with diagnosis of the prior art
Method, needs manually carry out classification process to substantial amounts of fundus photograph image so that the workload of doctor is larger to be compared, and which passes through
Pre-set image feature is extracted from the gray scale eye fundus image of pretreatment, and by being built according to SVMs and deep neural network
Vertical diabetic retinopathy disaggregated model is predicted classification to above-mentioned pre-set image feature, to obtain final classification knot
Really；Above-mentioned whole process reduces the workload of doctor, reduces the high standard of the doctor's money level to doctor without the need for manually participating in
Require, and improve the accuracy of classification, have great significance to concrete application and medical research.
Specifically, the eye fundus image of acquisition is coloured image, and which includes RGB color space, and the RGB color space is usual
Including three passages, in the embodiment of the present invention, G channel images therein, gray scale eye of the G channel images for eye fundus image are chosen
Base map picture；It is corresponding, with reference to Fig. 3, in the embodiment of the present invention, in above-mentioned steps 102, the eye fundus image is pre-processed,
The corresponding gray scale eye fundus image of the eye fundus image is obtained, including：
G channel images in S201, the RGB color space of the extraction eye fundus image；
S202, calculating process is carried out to the G channel images by height cap algorithm, obtain target object and background contrast
The higher G channel images of degree, and/or, partial histogram equalization process is carried out to the G channel images, target object is obtained
The G channel image higher with background contrasts；
S203, the G channel image higher to the target object and background contrasts carry out dividing processing, extract and select mesh
The target image of mark object.
With reference to step 201～step 203, the G channel images in RGB color space, purpose in the embodiment of the present invention, are chosen
Be in order to subsequently from G channel images extract pre-set image feature and according to the pre-set image feature to obtain eyeground figure enter
Row classification.
And partial histogram equalization process is carried out to G channel images, or by height cap algorithm to the G passages figure
As the purpose for carrying out calculating process is for increasing the contrast of background and target object in G channel images, to split
Go out selected target object therein, be easy to preferably extract the pre-set image feature of the target object；Wherein, gray scale eye fundus image
In target object can have multiple, such as ocular angiogenesis, eye macula lutea etc.；Above-mentioned segmentation selected target object can be above-mentioned
Anticipate one or more target objects, or all of target object.
And in the embodiment of the present invention, the G channel images can be calculated separately through height cap algorithm, it is also possible to
Partial histogram equalization process is carried out to G channel images individually；Can also, local histogram is carried out to G channel images first
Equalization processing, is then carrying out calculating process to equalization processing result by height cap algorithm, then to calculating knot
Fruit carries out partial histogram equalization process；It is also possible that calculating process is carried out to G channel images by height cap algorithm first,
Then partial histogram equalization process is being carried out to calculating result.
After target object and the higher G channel images of background contrasts are obtained according to above-mentioned any one mode, to
To G channel images carry out dividing processing, extract the target image of target object interested, such as blood vessel etc..
In the embodiment of the present invention, it is the gray scale eye fundus image to be carried out according to the pre-set image feature in gray scale eye fundus image
Classification, to determine the eye conditions of patient；It is corresponding, in the embodiment of the present invention, in above-mentioned steps 103, extract the gray scale eye
Pre-set image feature in base map picture, including：
Wavelet transform process is carried out to the target image, wavelet character is obtained；And/or, the target image is carried out
Texture analysis is processed, and extracts the textural characteristics in the target image；The textural characteristics include：Brightness, gray scale symbiosis
Matrix and intermediate value run-length matrix.
In the embodiment of the present invention, wavelet transform process is carried out to target image by dual-tree complex wavelet transform method, due to
Coefficient distribution of the different classes of eye fundus image after dual-tree complex wavelet transform is different, therefore, the dual-tree complex wavelet transform side
Method can be used to be used for eye fundus image of classifying as feature.
In embodiments of the present invention, pretreated G channel images are analyzed, are extracted using two kinds of different methods
The pre-set image feature of the G channel images：
First, wavelet transform process is carried out to the target image, wavelet character is obtained：
The G channel images in RGB color space are extracted in image pre-processing phase, then by height cap algorithm to described
G channel images carry out calculating process, obtain target object and the higher G channel images of background contrasts, and/or, to G passage figures
Operation is processed as carrying out partial histogram equalization, the higher G channel images of target object and the background contrasts to obtaining are entered
Row dividing processing, extracts the target image of selected target object, then using dual-tree complex wavelet transform method to pretreated
Target image carries out wavelet transform process, obtains wavelet character.As different classes of eye fundus image entered dual-tree complex wavelet change
Coefficient distribution after changing is different, therefore, above-mentioned dual-tree complex wavelet transform method can be used to be used for eyeground figure of classifying as feature
Picture.
Second, texture analysis process is carried out to the target image, the textural characteristics in the target image are extracted：
Mainly include the G channel images extracted in RGB color space in image pre-processing phase and G channel images are adopted
Processed with improved height cap algorithm, and/or, partial histogram equalization is carried out to G channel images and processes operation, obtained
Target object and the higher G channel images of background contrasts, to the target object and the higher G passage figures of background contrasts that obtain
As carrying out dividing processing, the target image of selected target object is extracted, above-mentioned target image is extracted using the method for texture analysis
In textural characteristics, the textural characteristics include 19 features, including 1 brightness, 14 gray level co-occurrence matrixes features and 4
Intermediate value run-length matrix feature.
With reference to Fig. 4, in the embodiment of the present invention, in above-mentioned steps 104, according to the diabetic retinopathy of the eye fundus image
Become disaggregated model and process is analyzed to the pre-set image feature, obtain the classification results of the corresponding eye fundus image, wrap
Include：
S301, all characteristic vectors extracted in the pre-set image feature.
The corresponding power of each characteristic vector in S302, the diabetic retinopathy disaggregated model of the acquisition eye fundus image
Weight coefficient.
S303, all characteristic vectors and the corresponding weight coefficient of each characteristic vector are calculated, and will be calculated
As a result classification results of the threshold range corresponding classification results in place as the eye fundus image.
Specifically, based on 301～step 303 of above-mentioned steps, corresponding all features in pre-set image feature are extracted first
Vector, then obtains the corresponding weight system of each characteristic vector in the diabetic retinopathy disaggregated model of eye fundus image
Then above-mentioned all of characteristic vector and the corresponding weight coefficient of each characteristic vector are updated to diabetic retina by number
In the corresponding formula of lesion classification model, you can calculate corresponding result；And when corresponding result=- 1, then judge eyeground
The classification results of image are PDR；When corresponding result=1, then judge that the classification results of eye fundus image are NPDR.
In a kind of eye fundus image sorting technique of the PVR for providing in an embodiment of the present invention, first to obtaining
Eye fundus image carry out pretreatment operation, pretreated fundus photograph is analyzed, is extracted using two kinds of different methods
Two stack features, for above-mentioned two stack features that extract, respectively using SVMs and deep neural network by eyeground figure
As being divided into normal, PDR and tri- kinds of NPDR, the final classification of eye fundus image is finally obtained using the ballot method in assembled classifier,
The most prediction classification results of poll are obtained as final classification results using the classifying and selecting of deep neural network, which is to tool
Body application and medical research have great significance.
The present invention implements the eye fundus image sorting technique of the PVR for proposing and (is referred to as based on assembled classifier
Eye fundus image sorting technique), after pre-processing to eye fundus image, by wavelet transformation and texture analysis from pretreatment
Extract two stack features in eye fundus image afterwards respectively, classification be predicted to this two stack features secondly by deep neural network,
To obtain final classification result, the accuracy of classification is improved, the use requirement for accuracy is better met.
The eye fundus image sorting technique of a kind of PVR provided in an embodiment of the present invention, with diagnosis of the prior art
Method, needs manually carry out classification process to substantial amounts of fundus photograph image so that the workload of doctor is larger to be compared, and which passes through
Pre-set image feature is extracted from the gray scale eye fundus image of pretreatment, and by being built according to SVMs and deep neural network
Vertical diabetic retinopathy disaggregated model is predicted classification to above-mentioned pre-set image feature, to obtain final classification knot
Really；Above-mentioned whole process reduces the workload of doctor, reduces the high standard of the doctor's money level to doctor without the need for manually participating in
Require, and improve the accuracy of classification, have great significance to concrete application and medical research.
The embodiment of the present invention additionally provides a kind of eye fundus image sorter of PVR, and described device is used to perform
The eye fundus image sorting technique of above-mentioned PVR, with reference to Fig. 5, described device includes：
Acquisition module 10, detects the eye fundus image of patient for obtaining eye；
Pretreatment module 20, pre-processes to the eye fundus image, obtains the corresponding gray scale eyeground of the eye fundus image
Image；
Extraction module 30, for extracting the pre-set image feature in the gray scale eye fundus image；The pre-set image feature
At least including one or more in following characteristics：Wavelet character and textural characteristics；
Analysis and processing module 40, for according to the diabetic retinopathy disaggregated model of the eye fundus image to described pre-
If characteristics of image is analyzed process, the classification results of the corresponding eye fundus image are obtained.
The eye fundus image sorter of a kind of PVR provided in an embodiment of the present invention, with diagnosis of the prior art
Method, needs manually carry out classification process to substantial amounts of fundus photograph image so that the workload of doctor is larger to be compared, and which passes through
Pre-set image feature is extracted from the gray scale eye fundus image of pretreatment, and by being built according to SVMs and deep neural network
Vertical diabetic retinopathy disaggregated model is predicted classification to above-mentioned pre-set image feature, to obtain final classification knot
Really；Above-mentioned whole process reduces the workload of doctor, reduces the high standard of the doctor's money level to doctor without the need for manually participating in
Require, and improve the accuracy of classification, have great significance to concrete application and medical research.
Specifically, the eye fundus image of acquisition includes RGB color space, and the RGB color space generally includes three and leads to
Road, in the embodiment of the present invention, chooses G channel images therein, wherein, G channel images are the corresponding gray scale eyeground of eye fundus image
Image；It is corresponding, in the embodiment of the present invention, pretreatment module 20, including：
First extraction unit, for extracting the G channel images in the RGB color space of the eye fundus image；
Height cap algorithm computing unit, for carrying out calculating process to the G channel images by height cap algorithm, obtains
Target object and the higher G channel images of background contrasts；And/or, partial histogram equalization processing unit, for described
G channel images carry out partial histogram equalization process, obtain target object and the higher G channel images of background contrasts；
Dividing processing unit, for carrying out segmentation portion to the target object and the higher G channel images of background contrasts
Reason, extracts the target image of selected target object.
In the embodiment of the present invention, it is right to the gray scale eyeground figure institute according to the pre-set image feature in gray scale eye fundus image
The state of an illness answered is classified, to determine which kind of disease corresponding eye belongs to；It is corresponding, in the embodiment of the present invention, extraction module
30, including：
Wavelet transform process unit, for carrying out wavelet transform process to the target image, obtains wavelet character；
Second extraction unit, for texture analysis process is carried out to the target image, extracts in the target image
Textural characteristics；The textural characteristics include：Brightness, gray level co-occurrence matrixes and intermediate value run-length matrix.
Further, with reference to Fig. 6, in the eye fundus image sorter of above-mentioned PVR, analysis and processing module 40, bag
Include：
3rd extraction unit 401, for extracting all characteristic vectors in multigroup pre-set image feature；
Acquiring unit 402, for obtaining each feature in the diabetic retinopathy disaggregated model of the eye fundus image
The corresponding weight coefficient of vector；
Computing unit 403, based on carrying out to all characteristic vectors and the corresponding weight coefficient of each characteristic vector
Calculate, and using result of calculation place threshold range corresponding classification results as the eye fundus image classification results.
Further, with reference to Fig. 7, the eye fundus image sorter of PVR provided in an embodiment of the present invention is also wrapped
Include：
Sample chooses module 50, for choosing the eye fundus image containing preset kind as training sample；The default class
Type includes：Normal person's eye fundus image, proliferative diabetic retinopathy PDR eye fundus images and non-increment patients with type Ⅰ DM view
Film pathology NPDR eye fundus image；
Model training module 60, for being trained to the training sample by SVMs and deep neural network
Calculate, set up the diabetic retinopathy disaggregated model of eye fundus image.
The present invention implements the eye fundus image sorter based on assembled classifier for proposing, is carrying out pre- place to eye fundus image
After reason, many stack features are extracted from pretreated eye fundus image respectively by wavelet transformation and texture analysis, secondly by
Deep neural network is predicted classification to many stack features, to obtain final classification result, improves the accuracy of classification, preferably
Meet for the use requirement of accuracy.
In a kind of eye fundus image sorter of the PVR for providing in an embodiment of the present invention, first by pre-
The eye fundus image of 20 pairs of acquisitions of processing module carries out pretreatment operation, then pretreated fundus photograph is analyzed, is adopted
Two stack features are extracted with two kinds of different methods, for above-mentioned two stack features for extracting, respectively using SVMs and
Eye fundus image is divided into two kinds of PDR and NPDR by deep neural network, finally obtains eyeground using the ballot method in assembled classifier
The final classification of image, i.e., obtain the most prediction classification results of poll as final using the classifying and selecting of deep neural network
Classification results, which has great significance to concrete application and medical research.
The present invention implements the eye fundus image sorter of the PVR for proposing and (is referred to as based on assembled classifier
Eye fundus image sorter), after pre-processing to eye fundus image, by wavelet transformation and texture analysis from pretreatment
Extract two stack features in eye fundus image afterwards respectively, classification be predicted to this two stack features secondly by deep neural network,
To obtain final classification result, the accuracy of classification is improved, the use requirement for accuracy is better met.
The eye fundus image sorter of the PVR provided by the embodiment of the present invention can be specific hard on equipment
Part or the software being installed on equipment or firmware etc..The device provided by the embodiment of the present invention, which realizes principle and generation
Technique effect is identical with preceding method embodiment, is brief description, and device embodiment part does not refer to part, refers to aforementioned side
Corresponding contents in method embodiment.Those skilled in the art can be understood that, for convenience and simplicity of description, aforementioned
The specific work process of the system, device and unit of description, may be referred to the corresponding process in said method embodiment, here
Repeat no more.
In embodiment provided by the present invention, it should be understood that disclosed apparatus and method, other sides can be passed through
Formula is realized.Device embodiment described above is only schematic, and for example, the division of the unit, only one kind are patrolled
Volume function is divided, and can have other dividing mode when actually realizing, and for example, multiple units or component can with reference to or can
To be integrated into another system, or some features can be ignored, or not perform.It is another, it is shown or discussed each other
Coupling or direct-coupling or communication connection can be INDIRECT COUPLING or communication link by some communication interfaces, device or unit
Connect, can be electrical, mechanical or other forms.
The unit as separating component explanation can be or may not be it is physically separate, it is aobvious as unit
The part for showing can be or may not be physical location, you can local to be located at one, or can also be distributed to multiple
On NE.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in the embodiment that the present invention is provided can be integrated in a processing unit, also may be used
Being that unit is individually physically present, it is also possible to which two or more units are integrated in a unit.
If the function is realized using in the form of SFU software functional unit and as independent production marketing or when using, can be with
It is stored in a computer read/write memory medium.Based on such understanding, technical scheme is substantially in other words
The part contributed to prior art or the part of the technical scheme can be embodied in the form of software product, the meter
Calculation machine software product is stored in a storage medium, is used including some instructions so that a computer equipment (can be individual
People's computer, server, or network equipment etc.) perform all or part of step of each embodiment methods described of the invention.
And aforesaid storage medium includes：USB flash disk, portable hard drive, read-only storage (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), magnetic disc or CD etc. are various can be with the medium of store program codes.
It should be noted that：Similar label and letter represent similar terms in following accompanying drawing, therefore, once a certain Xiang Yi
It is defined in individual accompanying drawing, then in subsequent accompanying drawing which further need not be defined and is explained, additionally, term " the
One ", " second ", " the 3rd " etc. are only used for distinguishing description, and it is not intended that indicating or implying relative importance.
Finally it should be noted that：Embodiment described above, specific embodiment only of the invention, to illustrate the present invention
Technical scheme, rather than a limitation, protection scope of the present invention is not limited thereto, although with reference to the foregoing embodiments to this
It is bright to be described in detail, it will be understood by those within the art that：Any those familiar with the art
The invention discloses technical scope in, which still can be modified to the technical scheme described in previous embodiment or can be light
Change is readily conceivable that, or equivalent is carried out to which part technical characteristic；And these modifications, change or replacement, do not make
The essence of appropriate technical solution departs from the spirit and scope of embodiment of the present invention technical scheme.The protection in the present invention should all be covered
Within the scope of.Therefore, protection scope of the present invention should be defined by the scope of the claims.
Claims (10)
1. a kind of eye fundus image sorting technique of PVR, it is characterised in that methods described includes：
Obtain the eye fundus image that eye detects patient；
The eye fundus image is pre-processed, the corresponding gray scale eye fundus image of the eye fundus image is obtained；
Extract the pre-set image feature in the gray scale eye fundus image；During the pre-set image feature at least includes following characteristics
One or more：Wavelet character and textural characteristics；
Process is analyzed to the pre-set image feature according to the diabetic retinopathy disaggregated model of the eye fundus image,
Obtain the classification results of the corresponding eye fundus image.
2. the eye fundus image sorting technique of PVR according to claim 1, it is characterised in that described to the eye
Base map picture is pre-processed, and obtains the corresponding gray scale eye fundus image of the eye fundus image, including：
Extract the G channel images in the RGB color space of the eye fundus image；
Calculating process is carried out to the G channel images by height cap algorithm, target object and the higher G of background contrasts is obtained
Channel image, and/or, partial histogram equalization process is carried out to the G channel images, target object and background contrast is obtained
The higher G channel images of degree；
Dividing processing is carried out to the target object and the higher G channel images of background contrasts, selected target object is extracted
Target image.
3. the eye fundus image sorting technique of PVR according to claim 2, it is characterised in that described in the extraction
Pre-set image feature in gray scale eye fundus image, including：
Wavelet transform process is carried out to the target image, wavelet character is obtained；
And/or,
Texture analysis process is carried out to the target image, the textural characteristics in the target image are extracted；The textural characteristics
Including：Brightness, gray level co-occurrence matrixes and intermediate value run-length matrix.
4. the eye fundus image sorting technique of PVR according to claim 3, it is characterised in that described in the basis
The diabetic retinopathy disaggregated model of eye fundus image is analyzed process to the pre-set image feature, obtains corresponding institute
The classification results of eye fundus image are stated, including：
Extract all characteristic vectors in the pre-set image feature；
Obtain the corresponding weight coefficient of each characteristic vector in the diabetic retinopathy disaggregated model of the eye fundus image；
All characteristic vectors and the corresponding weight coefficient of each characteristic vector are calculated, and by threshold that result of calculation is located
Classification results of the corresponding classification results of value scope as the eye fundus image.
5. the eye fundus image sorting technique of PVR according to claim 1, it is characterised in that the eye fundus image
Diabetic retinopathy disaggregated model be according to following method generate：
The eye fundus image containing preset kind is chosen as training sample；The preset kind includes：Normal person's eye fundus image, increasing
Grow patients with type Ⅰ DM PVR PDR eye fundus images and non-appreciation type diabetic retinopathy NPDR eye fundus images；
Calculating is trained to the training sample by SVMs and deep neural network, the glycosuria of eye fundus image is set up
Sick PVR disaggregated model.
6. the eye fundus image sorter of a kind of PVR, it is characterised in that described device includes：
Acquisition module, detects the eye fundus image of patient for obtaining eye；
Pretreatment module, for pre-processing to the eye fundus image, obtains the corresponding gray scale eyeground figure of the eye fundus image
Picture；
Extraction module, for extracting the pre-set image feature in the gray scale eye fundus image；The pre-set image feature is at least wrapped
Include one or more in following characteristics：Wavelet character and textural characteristics；
Analysis and processing module, for according to the diabetic retinopathy disaggregated model of the eye fundus image to the pre-set image
Feature is analyzed process, obtains the classification results of the corresponding eye fundus image.
7. the eye fundus image sorter of PVR according to claim 6, it is characterised in that the pretreatment mould
Block, including：
First extraction unit, for extracting the G channel images in the RGB color space of the eye fundus image；
Height cap algorithm computing unit, for carrying out calculating process to the G channel images by height cap algorithm, obtains target
Object and the higher G channel images of background contrasts；And/or, partial histogram equalization processing unit, for logical to the G
Road image carries out partial histogram equalization process, obtains target object and the higher G channel images of background contrasts；
Dividing processing unit, for carrying out dividing processing to the target object and the higher G channel images of background contrasts, carries
Take the target image of selected target object.
8. the eye fundus image sorter of PVR according to claim 7, it is characterised in that the extraction mould
Block, including：
Wavelet transform process unit, for carrying out wavelet transform process to the target image, obtains wavelet character；
Second extraction unit, for texture analysis process is carried out to the target image, extracts the texture in the target image
Feature；The textural characteristics include：Brightness, gray level co-occurrence matrixes and intermediate value run-length matrix.
9. the eye fundus image sorter of PVR according to claim 8, it is characterised in that the analyzing and processing
Module, including：
3rd extraction unit, for extracting all characteristic vectors in the pre-set image feature；
Acquiring unit, for obtaining each characteristic vector correspondence in the diabetic retinopathy disaggregated model of the eye fundus image
Weight coefficient；
Computing unit, for calculating to all characteristic vectors and the corresponding weight coefficient of each characteristic vector, and will
Classification results of the threshold range corresponding classification results in result of calculation place as the eye fundus image.
10. the eye fundus image sorter of PVR according to claim 6, it is characterised in that also include：
Sample chooses module, for choosing the eye fundus image containing preset kind as training sample；The preset kind includes：
Normal person's eye fundus image, proliferative diabetic retinopathy PDR eye fundus images and non-appreciation type diabetic retinopathy
NPDR eye fundus images；
Model training module, for calculating being trained to the training sample by SVMs and deep neural network,
Set up the diabetic retinopathy disaggregated model of eye fundus image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610978501.6A CN106530295A (en) | 2016-11-07 | 2016-11-07 | Fundus image classification method and device of retinopathy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610978501.6A CN106530295A (en) | 2016-11-07 | 2016-11-07 | Fundus image classification method and device of retinopathy |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106530295A true CN106530295A (en) | 2017-03-22 |
Family
ID=58350146
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610978501.6A Pending CN106530295A (en) | 2016-11-07 | 2016-11-07 | Fundus image classification method and device of retinopathy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106530295A (en) |
Cited By (38)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107045720A (en) * | 2017-05-04 | 2017-08-15 | 深圳硅基智能科技有限公司 | Artificial neural network and system for recognizing eye fundus image lesion |
| CN107229937A (en) * | 2017-06-13 | 2017-10-03 | 瑞达昇科技(大连)有限公司 | A kind of retinal vessel sorting technique and device |
| CN107423571A (en) * | 2017-05-04 | 2017-12-01 | 深圳硅基仿生科技有限公司 | Diabetic retinopathy identifying system based on eye fundus image |
| CN107729929A (en) * | 2017-09-30 | 2018-02-23 | 百度在线网络技术（北京）有限公司 | For obtaining the method and device of information |
| CN107945870A (en) * | 2017-12-13 | 2018-04-20 | 四川大学 | Retinopathy of prematurity detection method and device based on deep neural network |
| CN108185984A (en) * | 2017-12-28 | 2018-06-22 | 中山大学 | The method that eyeground color picture carries out eyeground lesion identification |
| CN108416772A (en) * | 2018-03-07 | 2018-08-17 | 汕头大学 | A kind of strabismus detection method based on concatenated convolutional neural network |
| CN108416371A (en) * | 2018-02-11 | 2018-08-17 | 艾视医疗科技成都有限公司 | A kind of diabetic retinopathy automatic testing method |
| CN108470359A (en) * | 2018-02-11 | 2018-08-31 | 艾视医疗科技成都有限公司 | A kind of diabetic retinal eye fundus image lesion detection method |
| WO2018201647A1 (en) * | 2017-05-05 | 2018-11-08 | 平安科技（深圳）有限公司 | Method for detecting retinopathy degree level, device and storage medium |
| CN108876776A (en) * | 2018-06-13 | 2018-11-23 | 东软集团股份有限公司 | A kind of method of generating classification model, eye fundus image classification method and device |
| CN109003252A (en) * | 2017-05-30 | 2018-12-14 | 正凯人工智能私人有限公司 | Image processing method and system |
| CN109191451A (en) * | 2018-09-11 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | Method for detecting abnormality, device, equipment and medium |
| CN109273081A (en) * | 2018-09-15 | 2019-01-25 | 湖南自兴智慧医疗科技有限公司 | A kind of eyeground scanning automatic identifying system |
| CN109348732A (en) * | 2017-05-30 | 2019-02-15 | 正凯人工智能私人有限公司 | Image processing method and system |
| CN109544540A (en) * | 2018-11-28 | 2019-03-29 | 东北大学 | A kind of diabetic retina picture quality detection method based on image analysis technology |
| CN109829921A (en) * | 2018-12-14 | 2019-05-31 | 清影医疗科技（深圳）有限公司 | A kind of processing method and its system, equipment, storage medium of the CT image on head |
| CN110037651A (en) * | 2018-01-15 | 2019-07-23 | 江威 | The method of quality control and device of eye fundus image |
| CN110097966A (en) * | 2018-01-30 | 2019-08-06 | 中国移动通信有限公司研究院 | A kind of information prompting method, device and terminal device |
| CN110176298A (en) * | 2019-05-31 | 2019-08-27 | 南通大学 | Core principle component for the classification of diabetic keratopathy eye fundus image composes hash method |
| CN110327013A (en) * | 2019-05-21 | 2019-10-15 | 北京至真互联网技术有限公司 | Eye fundus image detection method, device and equipment and storage medium |
| CN110335269A (en) * | 2018-05-16 | 2019-10-15 | 腾讯医疗健康(深圳)有限公司 | The classification recognition methods of eye fundus image and device |
| CN110378864A (en) * | 2018-04-10 | 2019-10-25 | 中南大学 | A kind of glaucoma detection method indicated based on Radon characteristic of field |
| WO2019206208A1 (en) * | 2018-04-26 | 2019-10-31 | 上海鹰瞳医疗科技有限公司 | Machine learning-based eye fundus image detection method, device, and system |
| CN110399929A (en) * | 2017-11-01 | 2019-11-01 | 腾讯科技（深圳）有限公司 | Eye fundus image classification method, device and computer readable storage medium |
| CN110459299A (en) * | 2019-07-10 | 2019-11-15 | 中山大学 | A kind of retina color fundus photograph image screening technique |
| CN111028226A (en) * | 2019-12-16 | 2020-04-17 | 北京百度网讯科技有限公司 | Method and device for algorithm transplantation |
| WO2020093563A1 (en) * | 2018-11-07 | 2020-05-14 | 哈尔滨工业大学（深圳） | Medical image processing method, system, device, and storage medium |
| CN111292286A (en) * | 2018-11-21 | 2020-06-16 | 福州依影健康科技有限公司 | Method and system for analyzing change of characteristic data of fundus oculi of sugar net and storage device |
| CN111316282A (en) * | 2017-06-09 | 2020-06-19 | 萨里大学 | Method and apparatus for processing retinal images |
| CN111374632A (en) * | 2018-12-29 | 2020-07-07 | 张国明 | Retinopathy detection method, device and computer-readable storage medium |
| CN111640498A (en) * | 2020-06-08 | 2020-09-08 | 博奥生物集团有限公司 | Information processing method and device based on eye image data |
| CN111767929A (en) * | 2019-03-14 | 2020-10-13 | 上海市第一人民医院 | Method and system for constructing sub-macular neovascularization model |
| CN112232448A (en) * | 2020-12-14 | 2021-01-15 | 北京大恒普信医疗技术有限公司 | Image classification method and device, electronic equipment and storage medium |
| CN112542241A (en) * | 2020-04-07 | 2021-03-23 | 徐敬媛 | Cloud storage type color feature analysis platform and method |
| CN112734701A (en) * | 2020-12-28 | 2021-04-30 | 深圳市优必选科技股份有限公司 | Fundus focus detection method, fundus focus detection device and terminal equipment |
| WO2022041396A1 (en) * | 2020-08-25 | 2022-03-03 | 复旦大学 | Ocular surface features-based novel coronavirus pneumonia (covid-19) disease patient risk screening deep learning system |
| US11361192B2 (en) | 2017-11-01 | 2022-06-14 | Tencent Technology (Shenzhen) Company Limited | Image classification method, computer device, and computer-readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070230795A1 (en) * | 2006-04-03 | 2007-10-04 | Abramoff Michael D | Methods and Systems for Optic Nerve Head Segmentation |
| CN103870838A (en) * | 2014-03-05 | 2014-06-18 | 南京航空航天大学 | Eye fundus image characteristics extraction method for diabetic retinopathy |
| CN104346503A (en) * | 2013-07-23 | 2015-02-11 | 广州华久信息科技有限公司 | Human face image based emotional health monitoring method and mobile phone |
| CN104881683A (en) * | 2015-05-26 | 2015-09-02 | 清华大学 | Cataract eye fundus image classification method based on combined classifier and classification apparatus |
| CN105069803A (en) * | 2015-08-19 | 2015-11-18 | 西安交通大学 | Classifier for micro-angioma of diabetes lesion based on colored image |
| CN105513077A (en) * | 2015-12-11 | 2016-04-20 | 北京大恒图像视觉有限公司 | System for screening diabetic retinopathy |
-
2016
- 2016-11-07 CN CN201610978501.6A patent/CN106530295A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070230795A1 (en) * | 2006-04-03 | 2007-10-04 | Abramoff Michael D | Methods and Systems for Optic Nerve Head Segmentation |
| CN104346503A (en) * | 2013-07-23 | 2015-02-11 | 广州华久信息科技有限公司 | Human face image based emotional health monitoring method and mobile phone |
| CN103870838A (en) * | 2014-03-05 | 2014-06-18 | 南京航空航天大学 | Eye fundus image characteristics extraction method for diabetic retinopathy |
| CN104881683A (en) * | 2015-05-26 | 2015-09-02 | 清华大学 | Cataract eye fundus image classification method based on combined classifier and classification apparatus |
| CN105069803A (en) * | 2015-08-19 | 2015-11-18 | 西安交通大学 | Classifier for micro-angioma of diabetes lesion based on colored image |
| CN105513077A (en) * | 2015-12-11 | 2016-04-20 | 北京大恒图像视觉有限公司 | System for screening diabetic retinopathy |
Cited By (70)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018201632A1 (en) * | 2017-05-04 | 2018-11-08 | 深圳硅基仿生科技有限公司 | Artificial neural network and system for recognizing lesion in fundus image |
| CN107045720A (en) * | 2017-05-04 | 2017-08-15 | 深圳硅基智能科技有限公司 | Artificial neural network and system for recognizing eye fundus image lesion |
| CN107358605A (en) * | 2017-05-04 | 2017-11-17 | 深圳硅基智能科技有限公司 | For identifying the deep neural network and system of diabetic retinopathy |
| CN107358606A (en) * | 2017-05-04 | 2017-11-17 | 深圳硅基智能科技有限公司 | For identifying the artificial neural network and system of diabetic retinopathy |
| CN107423571A (en) * | 2017-05-04 | 2017-12-01 | 深圳硅基仿生科技有限公司 | Diabetic retinopathy identifying system based on eye fundus image |
| CN108553079B (en) * | 2017-05-04 | 2020-06-02 | 深圳硅基智能科技有限公司 | Pathological change recognition system based on fundus image |
| CN111481166A (en) * | 2017-05-04 | 2020-08-04 | 深圳硅基智能科技有限公司 | Automatic identification system based on eye ground screening |
| CN108172291A (en) * | 2017-05-04 | 2018-06-15 | 深圳硅基仿生科技有限公司 | Diabetic retinopathy identifying system based on eye fundus image |
| CN111493814A (en) * | 2017-05-04 | 2020-08-07 | 深圳硅基智能科技有限公司 | Recognition system for fundus lesions |
| CN107423571B (en) * | 2017-05-04 | 2018-07-06 | 深圳硅基仿生科技有限公司 | Diabetic retinopathy identifying system based on eye fundus image |
| CN107358606B (en) * | 2017-05-04 | 2018-07-27 | 深圳硅基仿生科技有限公司 | The artificial neural network device and system and device of diabetic retinopathy for identification |
| CN108771530A (en) * | 2017-05-04 | 2018-11-09 | 深圳硅基仿生科技有限公司 | Eyeground pathological changes screening system based on deep neural network |
| CN111493814B (en) * | 2017-05-04 | 2021-11-12 | 深圳硅基智能科技有限公司 | Recognition system for fundus lesions |
| CN111481166B (en) * | 2017-05-04 | 2021-11-26 | 深圳硅基智能科技有限公司 | Automatic identification system based on eye ground screening |
| CN107358605B (en) * | 2017-05-04 | 2018-09-21 | 深圳硅基仿生科技有限公司 | The deep neural network apparatus and system of diabetic retinopathy for identification |
| CN108553079A (en) * | 2017-05-04 | 2018-09-21 | 深圳硅基仿生科技有限公司 | Lesion identifying system based on eye fundus image |
| WO2018201633A1 (en) * | 2017-05-04 | 2018-11-08 | 深圳硅基仿生科技有限公司 | Fundus image-based diabetic retinopathy identification system |
| US11210789B2 (en) * | 2017-05-04 | 2021-12-28 | Shenzhen Sibionics Technology Co., Ltd. | Diabetic retinopathy recognition system based on fundus image |
| CN108172291B (en) * | 2017-05-04 | 2020-01-07 | 深圳硅基智能科技有限公司 | Diabetic retinopathy recognition system based on fundus images |
| CN107045720B (en) * | 2017-05-04 | 2018-11-30 | 深圳硅基仿生科技有限公司 | The processing system of identification eye fundus image lesion based on artificial neural network |
| WO2018201647A1 (en) * | 2017-05-05 | 2018-11-08 | 平安科技（深圳）有限公司 | Method for detecting retinopathy degree level, device and storage medium |
| CN109348732A (en) * | 2017-05-30 | 2019-02-15 | 正凯人工智能私人有限公司 | Image processing method and system |
| CN109003252A (en) * | 2017-05-30 | 2018-12-14 | 正凯人工智能私人有限公司 | Image processing method and system |
| CN111316282A (en) * | 2017-06-09 | 2020-06-19 | 萨里大学 | Method and apparatus for processing retinal images |
| CN107229937A (en) * | 2017-06-13 | 2017-10-03 | 瑞达昇科技(大连)有限公司 | A kind of retinal vessel sorting technique and device |
| US10755410B2 (en) | 2017-09-30 | 2020-08-25 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for acquiring information |
| CN107729929A (en) * | 2017-09-30 | 2018-02-23 | 百度在线网络技术（北京）有限公司 | For obtaining the method and device of information |
| CN110399929A (en) * | 2017-11-01 | 2019-11-01 | 腾讯科技（深圳）有限公司 | Eye fundus image classification method, device and computer readable storage medium |
| CN110399929B (en) * | 2017-11-01 | 2023-04-28 | 腾讯科技（深圳）有限公司 | Fundus image classification method, fundus image classification apparatus, and computer-readable storage medium |
| US11361192B2 (en) | 2017-11-01 | 2022-06-14 | Tencent Technology (Shenzhen) Company Limited | Image classification method, computer device, and computer-readable storage medium |
| CN107945870A (en) * | 2017-12-13 | 2018-04-20 | 四川大学 | Retinopathy of prematurity detection method and device based on deep neural network |
| CN107945870B (en) * | 2017-12-13 | 2020-09-01 | 四川大学 | Method and device for detecting retinopathy of prematurity based on deep neural network |
| CN108185984A (en) * | 2017-12-28 | 2018-06-22 | 中山大学 | The method that eyeground color picture carries out eyeground lesion identification |
| CN110037651A (en) * | 2018-01-15 | 2019-07-23 | 江威 | The method of quality control and device of eye fundus image |
| CN110097966B (en) * | 2018-01-30 | 2021-09-14 | 中国移动通信有限公司研究院 | Information reminding method and device and terminal equipment |
| CN110097966A (en) * | 2018-01-30 | 2019-08-06 | 中国移动通信有限公司研究院 | A kind of information prompting method, device and terminal device |
| CN108470359A (en) * | 2018-02-11 | 2018-08-31 | 艾视医疗科技成都有限公司 | A kind of diabetic retinal eye fundus image lesion detection method |
| CN108416371A (en) * | 2018-02-11 | 2018-08-17 | 艾视医疗科技成都有限公司 | A kind of diabetic retinopathy automatic testing method |
| CN108416772A (en) * | 2018-03-07 | 2018-08-17 | 汕头大学 | A kind of strabismus detection method based on concatenated convolutional neural network |
| CN110378864A (en) * | 2018-04-10 | 2019-10-25 | 中南大学 | A kind of glaucoma detection method indicated based on Radon characteristic of field |
| CN110378864B (en) * | 2018-04-10 | 2021-07-16 | 中南大学 | Glaucoma detection method based on Radon domain feature representation |
| WO2019206208A1 (en) * | 2018-04-26 | 2019-10-31 | 上海鹰瞳医疗科技有限公司 | Machine learning-based eye fundus image detection method, device, and system |
| US20210042912A1 (en) * | 2018-04-26 | 2021-02-11 | Shanghai Eaglevision Medical Technology Co., Ltd. | Method, Apparatus and System for Detecting Fundus Image Based on Machine Learning |
| US11501428B2 (en) * | 2018-04-26 | 2022-11-15 | Shanghai Eaglevision Medical Technology Co., Ltd. | Method, apparatus and system for detecting fundus image based on machine learning |
| CN110335269A (en) * | 2018-05-16 | 2019-10-15 | 腾讯医疗健康(深圳)有限公司 | The classification recognition methods of eye fundus image and device |
| CN108876776A (en) * | 2018-06-13 | 2018-11-23 | 东软集团股份有限公司 | A kind of method of generating classification model, eye fundus image classification method and device |
| CN109191451B (en) * | 2018-09-11 | 2020-10-23 | 百度在线网络技术（北京）有限公司 | Abnormality detection method, apparatus, device, and medium |
| CN109191451A (en) * | 2018-09-11 | 2019-01-11 | 百度在线网络技术（北京）有限公司 | Method for detecting abnormality, device, equipment and medium |
| CN109273081B (en) * | 2018-09-15 | 2021-10-12 | 湖南自兴智慧医疗科技有限公司 | Automatic identification system for fundus scanning |
| CN109273081A (en) * | 2018-09-15 | 2019-01-25 | 湖南自兴智慧医疗科技有限公司 | A kind of eyeground scanning automatic identifying system |
| WO2020093563A1 (en) * | 2018-11-07 | 2020-05-14 | 哈尔滨工业大学（深圳） | Medical image processing method, system, device, and storage medium |
| CN111292286A (en) * | 2018-11-21 | 2020-06-16 | 福州依影健康科技有限公司 | Method and system for analyzing change of characteristic data of fundus oculi of sugar net and storage device |
| CN109544540B (en) * | 2018-11-28 | 2020-12-25 | 东北大学 | Diabetic retina image quality detection method based on image analysis technology |
| CN109544540A (en) * | 2018-11-28 | 2019-03-29 | 东北大学 | A kind of diabetic retina picture quality detection method based on image analysis technology |
| CN109829921B (en) * | 2018-12-14 | 2023-10-31 | 清影医疗科技（深圳）有限公司 | Method and system for processing CT image of head, equipment and storage medium |
| CN109829921A (en) * | 2018-12-14 | 2019-05-31 | 清影医疗科技（深圳）有限公司 | A kind of processing method and its system, equipment, storage medium of the CT image on head |
| CN111374632A (en) * | 2018-12-29 | 2020-07-07 | 张国明 | Retinopathy detection method, device and computer-readable storage medium |
| CN111767929A (en) * | 2019-03-14 | 2020-10-13 | 上海市第一人民医院 | Method and system for constructing sub-macular neovascularization model |
| CN110327013A (en) * | 2019-05-21 | 2019-10-15 | 北京至真互联网技术有限公司 | Eye fundus image detection method, device and equipment and storage medium |
| CN110327013B (en) * | 2019-05-21 | 2022-02-15 | 北京至真互联网技术有限公司 | Fundus image detection method, device and equipment and storage medium |
| CN110176298A (en) * | 2019-05-31 | 2019-08-27 | 南通大学 | Core principle component for the classification of diabetic keratopathy eye fundus image composes hash method |
| CN110459299A (en) * | 2019-07-10 | 2019-11-15 | 中山大学 | A kind of retina color fundus photograph image screening technique |
| CN110459299B (en) * | 2019-07-10 | 2023-06-27 | 中山大学 | Retina fundus color photograph image screening method |
| CN111028226A (en) * | 2019-12-16 | 2020-04-17 | 北京百度网讯科技有限公司 | Method and device for algorithm transplantation |
| CN112542241A (en) * | 2020-04-07 | 2021-03-23 | 徐敬媛 | Cloud storage type color feature analysis platform and method |
| CN111640498A (en) * | 2020-06-08 | 2020-09-08 | 博奥生物集团有限公司 | Information processing method and device based on eye image data |
| WO2022041396A1 (en) * | 2020-08-25 | 2022-03-03 | 复旦大学 | Ocular surface features-based novel coronavirus pneumonia (covid-19) disease patient risk screening deep learning system |
| CN112232448B (en) * | 2020-12-14 | 2021-04-23 | 北京大恒普信医疗技术有限公司 | Image classification method and device, electronic equipment and storage medium |
| CN112232448A (en) * | 2020-12-14 | 2021-01-15 | 北京大恒普信医疗技术有限公司 | Image classification method and device, electronic equipment and storage medium |
| CN112734701A (en) * | 2020-12-28 | 2021-04-30 | 深圳市优必选科技股份有限公司 | Fundus focus detection method, fundus focus detection device and terminal equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106530295A (en) | Fundus image classification method and device of retinopathy | |
| CN108021916B (en) | Deep learning diabetic retinopathy sorting technique based on attention mechanism | |
| CN107423571B (en) | Diabetic retinopathy identifying system based on eye fundus image | |
| CN108771530B (en) | Fundus lesion screening system based on deep neural network | |
| CN113011485B (en) | Multi-mode multi-disease long-tail distribution ophthalmic disease classification model training method and device | |
| CN109508644B (en) | Facial paralysis grade evaluation system based on deep video data analysis | |
| Niemeijer et al. | Image structure clustering for image quality verification of color retina images in diabetic retinopathy screening | |
| CN109635862B (en) | Sorting method for retinopathy of prematurity plus lesion | |
| CN108257135A (en) | The assistant diagnosis system of medical image features is understood based on deep learning method | |
| CN112017185B (en) | Focus segmentation method, device and storage medium | |
| CN112132817A (en) | Retina blood vessel segmentation method for fundus image based on mixed attention mechanism | |
| CN110211087A (en) | The semi-automatic diabetic eyeground pathological changes mask method that can share | |
| CN112686855B (en) | Information association method of eye image and symptom information | |
| CN110751637A (en) | Diabetic retinopathy detection system, method, equipment and training system | |
| CN110338763A (en) | A kind of intelligence Chinese medicine examines the image processing method and device of survey | |
| KR20190087681A (en) | A method for determining whether a subject has an onset of cervical cancer | |
| CN114937502A (en) | Method and system for evaluating osteoporotic vertebral compression fracture based on deep learning | |
| CN112869697A (en) | Judgment method for simultaneously identifying stage and pathological change characteristics of diabetic retinopathy | |
| CN112712122A (en) | Corneal ulcer classification detection method and system based on neural network model | |
| CN115035127A (en) | Retinal vessel segmentation method based on generative confrontation network | |
| Vamsi et al. | Early Detection of Hemorrhagic Stroke Using a Lightweight Deep Learning Neural Network Model. | |
| Yadav et al. | Automatic Cataract Severity Detection and Grading Using Deep Learning | |
| CN113393445B (en) | Breast cancer image determination method and system | |
| Akram et al. | Microaneurysm detection for early diagnosis of diabetic retinopathy | |
| CN112801238B (en) | Image classification method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170322 |
|
| RJ01 | Rejection of invention patent application after publication |