CN110309862A - DME prognosis information forecasting system and its application method based on ensemble machine learning - Google Patents
DME prognosis information forecasting system and its application method based on ensemble machine learning Download PDFInfo
- Publication number
- CN110309862A CN110309862A CN201910502749.9A CN201910502749A CN110309862A CN 110309862 A CN110309862 A CN 110309862A CN 201910502749 A CN201910502749 A CN 201910502749A CN 110309862 A CN110309862 A CN 110309862A
- Authority
- CN
- China
- Prior art keywords
- machine learning
- module
- image
- dme
- deep learning
- 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
- 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
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/03—Recognition of patterns in medical or anatomical images
Abstract
A kind of DME prognosis information forecasting system and its application method based on ensemble machine learning provided by the invention, including preprocessing module pre-process data；Characteristic extracting module carries out image characteristics extraction to image after pretreatment using three kinds of deep learning models；The building of network struction module progress deep learning network；Fusion Features module merges obtained characteristics of image；The text data that data processing module generates image co-registration feature and deep learning network is handled, generating probability distribution map；Prediction module generates the predicted value of CFT and BCVA according to probability distribution graph.System provided by the invention and its application method, deep learning network processes OCT image is constructed by network struction module, clinical variable is added, obtain the text feature of blending image feature and clinical variable, the prediction of CFT and BCVA are finally made by prediction module, objective predicted value is accurately provided, precision of prediction is effectively improved, gets rid of drawback present in traditional prediction method.
Description
Technical field
The present invention relates to clinical medicine ophthalmology and computer engineering field, more particularly to a kind of based on integrated machine
The DME prognosis information forecasting system of study, further relates to the systematic difference method.
Background technique
Diabetic macular edema DME patient's vision is decreased obviously, shown as on morphology retina from central fovea of macula to
Optic nerve thickens.The operation of intravitreal anti-vascular endothelial growth factor VEGF drug is the main hand of current treatment DME
Section, but different DME patients have various reactions to anti vegf agents, wherein about one third DME patient's reaction is invalid (reacts invalid
It is less than 50 μm or best to be defined as the 1st month central fovea of macula central thickness CFT range of decrease after rule in March injects anti vegf agents
The BCVA that corrects defects of vision, which improves, is less than 0.1logMAR).Anti-vegf drug effect only maintains about January, needs repeatedly to inject repeatedly, at present not yet
It is included in medical insurance and expensive, there is certain financial burden for patient.Therefore, preoperative prediction DME subject anti-VEGF treatment
Eyesight afterwards is promoted and edema extinction situation, reacts bad patient for anti vegf agents and takes other treatment method (such as glass
Glass body cavity injection of hormone), there is very big help to the burden for mitigating patient or even society.
Studies have shown that the prognosis information of DME can be obtained from means of optical coherence tomography (OCT) image before treatment
: the OCT image parting of first, DME.It is best to anti-vegf drug response to diffuse edema type DME patient, Cystoid macular edema type
It, and the reaction of serous detachment of retina type is worst.Second, internal layer --- outer retina articulamentum and external limiting membrane it is complete
Property.Retain complete internal layer-outer retina articulamentum and external limiting membrane and CFT and BCVA after treatment before anti-vegf treatment
It is more preferable to improve correlation.Third, subfoveal choroidal thickness.Eye with the concave train of thought film thickness in thicker base center has more
Good short-term dissection and functional response, i.e., the improvement of CFT decline and BCVA by a larger margin.In addition, the prognosis information of DME can be with
Obtained from the clinical variable before treatment: male, the age is small, diabetic duration is short, the DME course of disease is short, glycosylated hemoglobin is normal,
The patients such as no full endolaser photocoagulation history react more preferable to anti-vegf therapeutic scheme.
According to above traditional prediction technique, CFT, the BCVA of oculist after DME patient's anti vegf agents is treated
The performance of prediction field is general.The factor that its main cause is that traditional prediction technique subjectivity is very strong, comprehensively considers compares
It is more and complicated, and significant portion depends on oculist's clinical experience and know-how.Therefore, for the year of shortage clinical experience
For light doctor and the doctor of the relatively low community's basic hospital of medical level, making accurately postoperative prediction is to exist very
Big difficulty.But at present still without DME patient's anti-vegf post-operative recovery that is quick, accurate, versatile, having a wide range of application
Evaluation measures are for clinic.
Summary of the invention
The present invention is to overcome above-mentioned traditional DME prognosis information prediction technique there are subjectivities strong, and significant portion depends on
In oculist's clinical experience and the technological deficiency of know-how, a kind of DME prognosis information based on ensemble machine learning is provided
Forecasting system.
The present invention also provides the systematic difference methods.
In order to solve the above technical problems, technical scheme is as follows:
DME prognosis information forecasting system based on ensemble machine learning, including preprocessing module, characteristic extracting module, net
Network constructs module, Fusion Features module, data processing module and prediction module；Wherein:
The preprocessing module pre-processes to OCT image, clinical variable text data, and processing result is sent to
The characteristic extracting module；
The characteristic extracting module carries out characteristics of image to pretreated OCT image using three kinds of deep learning models and mentions
It takes；
The network struction module carries out the building of deep learning network according to the characteristic extracting module；
The Fusion Features module merges the characteristics of image that the characteristic extracting module obtains；
The text data that the data processing module generates image co-registration feature and deep learning network is handled, raw
At probability distribution graph；
The prediction module generates the predicted value of CFT and BCVA according to probability distribution graph, completes the pre- of DME prognosis information
It surveys.
Wherein, the OCT image is tiff format, and the clinical variable text data includes but are not limited to: gender,
Age, BCVA, the research ETDRS scoring of early treatment diabetic retinopathy, intraocular pressure, diabetes type, diabetic duration,
Before whether Zeng Hangquan endolaser photocoagulation, serum casual glucose, serum baseline glycosylated hemoglobin value HbA1c, be
No adjoint hypertension, the anti vegf agents type of injection.
Wherein, described that OCT image, clinical variable text data are pre-processed specifically: the OCT image is located in advance
Reason, including for the removal of image saturated pixel value, denoising and edge deletion；The clinical variable text data, including from equipment
Upper reading clinical data is simultaneously screened, and it is spare to filter out valuable clinical data.
Wherein, the characteristic extracting module include three kinds of deep learning models include AlexNet deep learning model,
Vgg16 deep learning model and ResNet18 deep learning model.
A kind of application method of the DME prognosis information forecasting system based on ensemble machine learning, comprising the following steps:
S1: collecting the OCT image and clinical variable text data for being diagnosed as DME patient, to OCT image, clinical variable text
Notebook data is pre-processed；
S2: carrying out image characteristics extraction to pretreated OCT image using three kinds of deep learning models, each by merging
Characteristics of image constructs deep learning network；
S3: building ensemble machine learning model, the blending image feature obtained to deep learning network and pretreated
Clinical variable text data is handled, generating probability distribution map；
S4: generating the predicted value of CFT and BCVA according to probability distribution graph, completes the prediction of DME prognosis information.
Wherein, the detailed process of the step S2 are as follows:
S21: creating and finely tunes three kinds of deep learning models of the pre-training on ImageNet data set, including freezes net
Network shallow-layer weight and alternative networks task layer；
S22: three kinds of deep learning models carry out image characteristics extraction to pretreated OCT image respectively；
S23: the image characteristics extraction extracted merge simultaneously dimensionality reduction, forms deep learning network.
Wherein, the step S3 specifically:
S31: the blending image feature that deep learning network is obtained and pretreated clinical variable text data carry out
Fusion, obtains fusion feature；
S32: ensemble machine learning model is constructed using four machine learning models, fusion feature is handled, is obtained
CFT predicted value and BCVA predicted value；
S33: CFT probability distribution graph and BCVA probability distribution graph are generated according to CFT predicted value and BCVA predicted value.
Wherein, four machine learning models include SVM machine learning model, lasso machine learning model,
Decision Tree machine learning model and Random Forest machine learning model.
Wherein, in the step S4, by choosing three regions that numeric distribution is most concentrated in CFT probability distribution graph,
The average value for falling in region interior prediction value is calculated, the predicted value of CFT is obtained；By choosing numeric distribution in BCVA probability distribution graph
Two regions most concentrated calculate the average value for falling in region interior prediction value, obtain the predicted value of BCVA.
Compared with prior art, the beneficial effect of technical solution of the present invention is:
A kind of DME prognosis information forecasting system and its application method based on ensemble machine learning provided by the invention is led to
It crosses network struction module building deep learning network processes OCT image and characteristics of image is extracted by characteristic extracting module, addition is faced
Bed variable, obtains the text feature of blending image feature and clinical variable, finally utilizes ensemble machine learning mould by prediction module
Type makes the prediction of CFT and BCVA, accurately provides objective predicted value, effectively improves precision of prediction, gets rid of tradition
Drawback present in prediction technique.
Detailed description of the invention
Fig. 1 is system connection schematic diagram of the invention；
Fig. 2 is method flow schematic diagram of the invention；
Fig. 3 is specific implementation process schematic diagram of the present invention；
Wherein: 1, preprocessing module；2, characteristic extracting module；3, network struction module；4, Fusion Features module；5, data
Processing module；6, prediction module.
Specific embodiment
The attached figures are only used for illustrative purposes and cannot be understood as limitating the patent；
In order to better illustrate this embodiment, the certain components of attached drawing have omission, zoom in or out, and do not represent actual product
Size；
To those skilled in the art, it is to be understood that certain known features and its explanation, which may be omitted, in attached drawing
's.
The following further describes the technical solution of the present invention with reference to the accompanying drawings and examples.
Embodiment 1
As shown in Figure 1, the DME prognosis information forecasting system based on ensemble machine learning, including preprocessing module 1, feature
Extraction module 2, network struction module 3, Fusion Features module 4, data processing module 5 and prediction module 6；Wherein:
The preprocessing module 1 pre-processes to OCT image, clinical variable text data, and processing result is sent
To the characteristic extracting module 2；
The characteristic extracting module 2 carries out characteristics of image to pretreated OCT image using three kinds of deep learning models
It extracts；
The network struction module 3 carries out the building of deep learning network according to the characteristic extracting module 2；
The Fusion Features module 4 merges the characteristics of image that the characteristic extracting module 2 obtains；
The text data that the data processing module 5 generates image co-registration feature and deep learning network is handled,
Generating probability distribution map；
The prediction module 6 generates the predicted value of CFT and BCVA according to probability distribution graph, completes the pre- of DME prognosis information
It surveys.
More specifically, the OCT image is tiff format, and the clinical variable text data includes but are not limited to:
Gender, age, BCVA, the research ETDRS scoring of early treatment diabetic retinopathy, intraocular pressure, diabetes type, diabetes
The course of disease, before whether Zeng Hangquan endolaser photocoagulation, serum casual glucose, serum baseline glycosylated hemoglobin value
HbA1c, if with hypertension, the anti vegf agents type of injection.
More specifically, described that OCT image, clinical variable text data are pre-processed specifically: the OCT image is pre-
Processing, including for the removal of image saturated pixel value, denoising and edge deletion；The clinical variable text data, including from setting
Standby upper reading clinical data is simultaneously screened, and it is spare to filter out valuable clinical data.
More specifically, it includes AlexNet deep learning mould that the characteristic extracting module 2, which includes three kinds of deep learning models,
Type, Vgg16 deep learning model and ResNet18 deep learning model.
Embodiment 2
More specifically, on the basis of embodiment 1, as shown in Figure 2 and Figure 3, a kind of DME based on ensemble machine learning is pre-
Information prediction systematic difference method afterwards, comprising the following steps:
S1: collecting the OCT image and clinical variable text data for being diagnosed as DME patient, to OCT image, clinical variable text
Notebook data is pre-processed；
S2: carrying out image characteristics extraction to pretreated OCT image using three kinds of deep learning models, each by merging
Characteristics of image constructs deep learning network；
S3: building ensemble machine learning model, the blending image feature obtained to deep learning network and pretreated
Clinical variable text data is handled, generating probability distribution map；
S4: generating the predicted value of CFT and BCVA according to probability distribution graph, completes the prediction of DME prognosis information.
Wherein, the detailed process of the step S2 are as follows:
S21: creating and finely tunes three kinds of deep learning models of the pre-training on ImageNet data set, including freezes net
Network shallow-layer weight and alternative networks task layer；
S22: three kinds of deep learning models carry out image characteristics extraction to pretreated OCT image respectively；
In the specific implementation process, AlexNet deep learning model, Vgg16 deep learning model and ResNet18 depth
The depth characteristic characteristic dimension that learning model extracts is 80；
S23: the image characteristics extraction extracted merge simultaneously dimensionality reduction, forms deep learning network.
In the specific implementation process, made by the depth characteristic that feature series connection strategy integrates three kinds of deep learning model extractions
To merge depth characteristic, the fusion depth characteristic dimension is 240, and PCA dimensionality reduction technology is recycled to carry out dimensionality reduction, the spy after dimensionality reduction
Levying dimension is 20, forms deep learning network.
Wherein, the step S3 specifically:
S31: the blending image feature that deep learning network is obtained and pretreated clinical variable text data carry out
Fusion, obtains fusion feature；
In the specific implementation process, the characteristic dimension of blending image feature is 20, the feature dimensions of clinical variable text data
Degree is 20, and the characteristic dimension of fusion feature is 40；
S32: ensemble machine learning model is constructed using four machine learning models, fusion feature is handled, is obtained
CFT predicted value and BCVA predicted value；
S33: CFT probability distribution graph and BCVA probability distribution graph are generated according to CFT predicted value and BCVA predicted value.
Wherein, four machine learning models include SVM machine learning model, lasso machine learning model,
Decision Tree machine learning model and Random Forest machine learning model.
In the specific implementation process, multiple subsets are randomly selected and created from training set using random sampling methods, this
It is randomly selected in embodiment 20 times, creates 20 training subsets；Respectively SVM machine learning model, lasso machine learning mould
Type, Decision Tree machine learning model and Random Forest machine learning model create 20 training subsets, pass through
20 SVM machine learning models, 20 lasso machine learning models, 20 Decision Tree machines are respectively trained in training subset
Device learning model and 20 Random Forest machine learning models construct ensemble machine learning by 80 machine learning models
Model symbiosis is at 80 predicted values, to generate the probability distribution graph of 80 dlinial prediction values.
Wherein, in the step S4, by choosing three regions that numeric distribution is most concentrated in CFT probability distribution graph,
The average value for falling in region interior prediction value is calculated, the predicted value of CFT is obtained；By choosing numeric distribution in BCVA probability distribution graph
Two regions most concentrated calculate the average value for falling in region interior prediction value, obtain the predicted value of BCVA.
In the specific implementation process, two primary treatments of DME are accurately predicted as a result, i.e. structure C FT, function are replied in dissection
Result BCVA can be replied, the information about DME patient to anti-vegf drug medical can be provided for oculist, which can be with
Helping doctor is that patient formulates better medical scheme.For estimated to the good patient of anti-vegf therapeutic response, for emphasizing
Subject anti-VEGF treatment's is promising as a result, and patient is encouraged to adhere to standard regimens.Anti-vegf is treated as prediction
The DME patient of poor response can recommend other treatment mode to it, such as vitreous chamber injection of hormone, full retinal laser light
Solidifying art, vitrectomy etc..It is significantly for the treatment time and financial burden angle for saving DME patient.
In the specific implementation process, scheme of the present invention is all based on the common clinical information of DME patient, such as OCT
Image, and common clinical variable text data, this will allow more doctors using the system, without to new ophthalmology
Inspection machine and intelligent algorithm carry out additional investment.
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description
To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this
Made any modifications, equivalent replacements, and improvements etc., should be included in the claims in the present invention within the spirit and principle of invention
Protection scope within.
Claims (9)
1. the DME prognosis information forecasting system based on ensemble machine learning, it is characterised in that: including preprocessing module (1), feature
Extraction module (2), network struction module (3), Fusion Features module (4), data processing module (5) and prediction module (6)；Its
In:
The preprocessing module (1) pre-processes to OCT image, clinical variable text data, and processing result is sent to
The characteristic extracting module (2)；
The characteristic extracting module (2) carries out characteristics of image to pretreated OCT image using three kinds of deep learning models and mentions
It takes；
The network struction module (3) carries out the building of deep learning network according to the characteristic extracting module (2)；
The Fusion Features module (4) merges the characteristics of image that the characteristic extracting module (2) obtains；
The text data that the data processing module (5) generates image co-registration feature and deep learning network is handled, raw
At probability distribution graph；
The prediction module (6) generates the predicted value of CFT and BCVA according to probability distribution graph, completes the prediction of DME prognosis information.
2. the DME prognosis information forecasting system according to claim 1 based on ensemble machine learning, it is characterised in that: institute
The OCT image stated is tiff format, and the clinical variable text data includes but are not limited to: gender, age, BCVA, early stage
Treat the research ETDRS scoring of diabetic retinopathy, intraocular pressure, diabetes type, diabetic duration, before whether Zeng Hangquan
Endolaser photocoagulation, serum casual glucose, serum baseline glycosylated hemoglobin value HbA1c, if with hypertension, note
The anti vegf agents type penetrated.
3. the DME prognosis information forecasting system according to claim 2 based on ensemble machine learning, it is characterised in that: institute
It states and OCT image, clinical variable text data is pre-processed specifically: the OCT image pretreatment, including it is full for image
With pixel value removal, denoising and edge deletion；The clinical variable text data, including read clinical data from equipment and go forward side by side
Row screening, it is spare to filter out valuable clinical data.
4. the DME prognosis information forecasting system according to claim 3 based on ensemble machine learning, it is characterised in that: institute
It includes AlexNet deep learning model, Vgg16 deep learning mould that characteristic extracting module (2), which are stated, including three kinds of deep learning models
Type and ResNet18 deep learning model.
5. a kind of application method of the DME prognosis information forecasting system based on ensemble machine learning as claimed in claim 4,
It is characterized in that, comprising the following steps:
S1: the OCT image and clinical variable text data for being diagnosed as DME patient are collected, to OCT image, clinical variable textual data
According to being pre-processed；
S2: image characteristics extraction is carried out to pretreated OCT image using three kinds of deep learning models, by merging each image
Feature constructs deep learning network；
S3: building ensemble machine learning model, the blending image feature and pretreated clinic that deep learning network is obtained
Variable text data is handled, generating probability distribution map；
S4: generating the predicted value of CFT and BCVA according to probability distribution graph, completes the prediction of DME prognosis information.
6. the application method of the DME prognosis information forecasting system according to claim 5 based on ensemble machine learning, special
Sign is, the detailed process of the step S2 are as follows:
S21: creating and finely tunes three kinds of deep learning models of the pre-training on ImageNet data set, including to freeze network shallow
Layer weight and alternative networks task layer；
S22: three kinds of deep learning models carry out image characteristics extraction to pretreated OCT image respectively；
S23: the image characteristics extraction extracted merge simultaneously dimensionality reduction, forms deep learning network.
7. the application method of the DME prognosis information forecasting system according to claim 6 based on ensemble machine learning, special
Sign is, the step S3 specifically:
S31: blending image feature that deep learning network obtains and pretreated clinical variable text data are merged,
Obtain fusion feature；
S32: ensemble machine learning model is constructed using four machine learning models, fusion feature is handled, it is pre- to obtain CFT
Measured value and BCVA predicted value；
S33: CFT probability distribution graph and BCVA probability distribution graph are generated according to CFT predicted value and BCVA predicted value.
8. the application method of the DME prognosis information forecasting system according to claim 7 based on ensemble machine learning, special
Sign is that four machine learning models include SVM machine learning model, lasso machine learning model, Decision
Tree machine learning model and Random Forest machine learning model.
9. the application method of the DME prognosis information forecasting system according to claim 8 based on ensemble machine learning, special
Sign is, in the step S4, by choosing three regions that numeric distribution is most concentrated in CFT probability distribution graph, calculating is fallen
In the average value of region interior prediction value, the predicted value of CFT is obtained；It is most concentrated by choosing numeric distribution in BCVA probability distribution graph
Two regions, calculate and fall in the average value of region interior prediction value, obtain the predicted value of BCVA.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910502749.9A CN110309862A (en) | 2019-06-11 | 2019-06-11 | DME prognosis information forecasting system and its application method based on ensemble machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910502749.9A CN110309862A (en) | 2019-06-11 | 2019-06-11 | DME prognosis information forecasting system and its application method based on ensemble machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110309862A true CN110309862A (en) | 2019-10-08 |
Family
ID=68077173
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910502749.9A Pending CN110309862A (en) | 2019-06-11 | 2019-06-11 | DME prognosis information forecasting system and its application method based on ensemble machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110309862A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112132782A (en) * | 2020-08-21 | 2020-12-25 | 广东省人民医院 | Method and terminal for processing DME typing based on deep neural network |
| CN112133441A (en) * | 2020-08-21 | 2020-12-25 | 广东省人民医院 | Establishment method and terminal of MH post-operation fissure hole state prediction model |
| WO2021139446A1 (en) * | 2020-09-30 | 2021-07-15 | 平安科技（深圳）有限公司 | Anti-vascular endothelial growth factor (vegf) curative effect prediction apparatus and method |
| CN115083005A (en) * | 2022-06-13 | 2022-09-20 | 广东省人民医院 | ROP image classification system and method based on deep learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015189264A1 (en) * | 2014-06-10 | 2015-12-17 | Ventana Medical Systems, Inc. | Predicting breast cancer recurrence directly from image features computed from digitized immunohistopathology tissue slides |
| CN109119167A (en) * | 2018-07-11 | 2019-01-01 | 山东师范大学 | Pyemia anticipated mortality system based on integrated model |
| CN109658411A (en) * | 2019-01-21 | 2019-04-19 | 杭州英库医疗科技有限公司 | A kind of correlation analysis based on CT images feature Yu Patients with Non-small-cell Lung prognosis situation |
| CN109785430A (en) * | 2019-01-28 | 2019-05-21 | 上海交通大学医学院附属上海儿童医学中心 | The construction method and device of a kind of TAPVC risk forecast model of postoperative restenosis in the heart |
| CN109829882A (en) * | 2018-12-18 | 2019-05-31 | 苏州比格威医疗科技有限公司 | A kind of stages of DR prediction technique |
-
2019
- 2019-06-11 CN CN201910502749.9A patent/CN110309862A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015189264A1 (en) * | 2014-06-10 | 2015-12-17 | Ventana Medical Systems, Inc. | Predicting breast cancer recurrence directly from image features computed from digitized immunohistopathology tissue slides |
| CN109119167A (en) * | 2018-07-11 | 2019-01-01 | 山东师范大学 | Pyemia anticipated mortality system based on integrated model |
| CN109829882A (en) * | 2018-12-18 | 2019-05-31 | 苏州比格威医疗科技有限公司 | A kind of stages of DR prediction technique |
| CN109658411A (en) * | 2019-01-21 | 2019-04-19 | 杭州英库医疗科技有限公司 | A kind of correlation analysis based on CT images feature Yu Patients with Non-small-cell Lung prognosis situation |
| CN109785430A (en) * | 2019-01-28 | 2019-05-21 | 上海交通大学医学院附属上海儿童医学中心 | The construction method and device of a kind of TAPVC risk forecast model of postoperative restenosis in the heart |
Non-Patent Citations (3)
| Title |
|---|
| 潘浩 等: "深度学习在肺癌患者生存预测中的应用研究", 《计算机工程与应用》 * |
| 胡忆群 等: "Avastin 联合黄斑格栅样光凝治疗 DME 的视力预后的多因素分析", 《国际眼科杂志》 * |
| 贾建华: "《谱聚类集成算法研究》", 31 August 2011, 天津大学出版社 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112132782A (en) * | 2020-08-21 | 2020-12-25 | 广东省人民医院 | Method and terminal for processing DME typing based on deep neural network |
| CN112133441A (en) * | 2020-08-21 | 2020-12-25 | 广东省人民医院 | Establishment method and terminal of MH post-operation fissure hole state prediction model |
| CN112132782B (en) * | 2020-08-21 | 2023-09-05 | 广东省人民医院 | Method and terminal for processing DME (DME) typing based on deep neural network |
| WO2021139446A1 (en) * | 2020-09-30 | 2021-07-15 | 平安科技（深圳）有限公司 | Anti-vascular endothelial growth factor (vegf) curative effect prediction apparatus and method |
| CN115083005A (en) * | 2022-06-13 | 2022-09-20 | 广东省人民医院 | ROP image classification system and method based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110309862A (en) | DME prognosis information forecasting system and its application method based on ensemble machine learning | |
| Zhang et al. | A survey on computer aided diagnosis for ocular diseases | |
| Batista et al. | Rim-one dl: A unified retinal image database for assessing glaucoma using deep learning | |
| Patel et al. | Pars plana vitrectomy with and without peeling of the inner limiting membrane for diabetic macular edema | |
| CN109528155B (en) | Intelligent screening system suitable for high myopia complicated with open angle glaucoma and establishment method thereof | |
| CN113243887B (en) | Intelligent diagnosis and treatment instrument for macular degeneration of old people | |
| Klyce et al. | Screening patients with the corneal navigator | |
| Lalithadevi et al. | Detection of diabetic retinopathy and related retinal disorders using fundus images based on deep learning and image processing techniques: A comprehensive review | |
| Xu et al. | Topical medication instillation techniques for glaucoma | |
| Hu et al. | Comparison of Hydrus and iStent microinvasive glaucoma surgery implants in combination with phacoemulsification for treatment of open-angle glaucoma: systematic review and network meta-analysis | |
| Phu et al. | Application of pattern recognition analysis to optimize hemifield asymmetry patterns for early detection of glaucoma | |
| AU2021100492A4 (en) | DME Prognostic Information Prediction System Based on Integrated Machine Learning and Application Method Thereof | |
| Coleman et al. | A new smartphone-based optic nerve head biometric for verification and change detection | |
| Amiri et al. | Detection of topographic images of keratoconus disease using machine vision | |
| Kurysheva et al. | Comparative study of predictors of hypotensive efficacy of laser peripheral iridotomy and lensectomy in patients with primary anterior chamber angle closure based on machine learning methods | |
| Smits et al. | Machine learning in the detection of the glaucomatous disc and visual field | |
| Tekouabou et al. | Machine learning aprroach for early detection of glaucoma from visual fields | |
| Yue et al. | CCNET: Cross coordinate network for joint diabetic retinopathy and diabetic macular edema grading | |
| Demirel et al. | Predicting progressive glaucomatous optic neuropathy using baseline standard automated perimetry data | |
| Wang et al. | Advances in artificial intelligence models and algorithms in the field of optometry | |
| Prasher et al. | Automated Eye Disease Classification using MobileNetV3 and EfficientNetB0 Models using Transfer Learning | |
| Chen et al. | Artificial intelligence assisted pterygium diagnosis: current status and perspectives | |
| Du et al. | Artificial intelligence-aided diagnosis and treatment in the field of optometry | |
| Jipeng et al. | An optimized method using CNN, RF, cuckoo search and HOG for early detection of eye disease in humans | |
| Ranadive et al. | Predicting Glaucoma Diagnosis Using AI |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191008 |
|
| RJ01 | Rejection of invention patent application after publication |