CN110504027A - A kind of X-Ray rabat pneumonia intelligent diagnosis system and method based on deep learning - Google Patents
A kind of X-Ray rabat pneumonia intelligent diagnosis system and method based on deep learning Download PDFInfo
- Publication number
- CN110504027A CN110504027A CN201910767278.4A CN201910767278A CN110504027A CN 110504027 A CN110504027 A CN 110504027A CN 201910767278 A CN201910767278 A CN 201910767278A CN 110504027 A CN110504027 A CN 110504027A
- Authority
- CN
- China
- Prior art keywords
- user
- pneumonia
- picture
- prediction
- model
- 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
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/25—Integrating or interfacing systems involving database management systems
- G06F16/252—Integrating or interfacing systems involving database management systems between a Database Management System and a front-end application
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/31—User authentication
-
- 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
- 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
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Public Health (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Security & Cryptography (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Pathology (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The present invention discloses a kind of X-Ray rabat pneumonia intelligent diagnosis system and method based on deep learning, belong to Medical image classification technology field, the system includes that user logs in and registration module, picture format conversion module and pneumonia prediction module, with B/S framework, divide the system into four layers, respectively podium level, supporting layer, service layer and application layer.User installation process is simple, mitigates the dependence to home environment.By friendly interface and the system module designed around actual demand, the fusion of this system and other systems progress functionally can be made.X-Ray rabat pneumonia intelligent diagnosing method based on deep learning enhances transfer learning by data and the convergence rate that means accelerate model training such as improves to network structure, improves the accuracy rate of pneumonia identification and the generalization ability of model.System and method for of the invention can significantly reduce artificial diagosis and perplex to doctor's bring.
Description
Technical field
The present invention relates to Medical image classification technology field more particularly to a kind of X-Ray rabat lungs based on deep learning
Scorching intelligent diagnosis system and method.
Background technique
Medical image is one of information important in medical diagnosis on disease, and wherein imaging technique includes: computer tomography skill
Art (Computed Tomography, CT), positron emission tomography (Positron Emission Tomography,
PET), single photon emission tomographic imaging, Magnetic resonance imaging, ultrasonic imaging etc., traditional radiodiagnosis, which is still important, to be put
Penetrate one of diagnosis basis.Thoracic cavity is called the mirror of human health and disease, because it contains many kinds of important groups of human body
Knit structure, can provide human body various information, as the diagnosis of pulmonary disease, fracture of rib and damage, Heart enlargement symptom,
Cardiopulmonary coefficient etc. all can be identified and be confirmed by X-Ray rabat.Chest x-ray is still the routine inspection project of hospital image diagnosis,
Even if the case for needing CT, MRI to check, also often needs to refer to the performance of chest X-ray.X-Ray rabat is with its cheap valence
Lattice and faint radiological dose account for 40% or more of all irradiation image diagnosis.This embody X-Ray rabat in medical domain
Significant application value.
But how to analyze rabat is an extremely challenging task, even veteran expert also often feels
To intractable.As depth learning technology in recent years is in the huge hair in the fields such as computer vision and image classification, segmentation, identification
Exhibition, scientific research personnel propose many aided diagnosis methods in terms of the computer diagnosis of X-Ray rabat, and Aramato etc. is proposed
It is a kind of that area-of-interest is extracted based on the method for the wealthy value of more gray values, then letter is carried out using nine features of area-of-interest
Tubercle is extracted in single classification.Awai et al. positions Lung neoplasm using morphological method.Li et al. people continues the research achievement of sato,
And devise can enhance it is specific move back the performance being woven on image and weaken the Selective long-range DEPT function that remaining tissue shows, this
Sample makes area-of-interest more obvious.Pranav et al. proposes a kind of network C heXNet based on deep learning can be right
Lung X-ray-Ray image carries out auxiliary diagnosis.Guan et al. is in such a way that deep learning merges attention model come to lung X-ray-
Ray image carries out auxiliary diagnosis.Above-mentioned is the correlative study for pulmonary nodule mostly, even if having for pneumonia automatic identification
Lung X-ray-the Ray that correlative study is also based on deep learning classifies more, lacks specific aim to the intelligent diagnostics of pneumonia, additionally due to
Pneumonia focal area is easy to be blocked by normal institutional framework, so that subtle pneumonia lesion and institutional framework are difficult to differentiate between,
Lesion testing result is also easy to omit pneumonia lesion.It tends not to reach higher using traditional computer-aided diagnosis means
Accuracy rate.
Summary of the invention
In view of the above shortcomings of the prior art, the present invention provides a kind of X-Ray rabat pneumonia intelligence based on deep learning
Diagnostic system and method.
In order to solve the above technical problems, the technical solution used in the present invention is: a kind of X-Ray chest based on deep learning
Piece pneumonia intelligent diagnosis system, system include that three modules are respectively as follows: user's login and registration module, picture format conversion module
It is as shown in Figure 2 with the brief operation flow of pneumonia prediction module, system;
Wherein, user, which logs in, provides login function, registering functional and resetting cryptographic function with registration module, for user
The entrance for entering system is provided, process is as shown in Figure 3;
The login function needs user to input account and password in login interface and submits to system, System Back-end according to
The account of input inquires corresponding password into the user message table of database, if returned the result as sky, illustrates that user is defeated
The account entered is not present, if entered passwords does not match with user for the result returned, illustrates user password input error, only
Have when the password of user's input and the password inquired from database match, system can just be shown jumps interface accordingly;
The registering functional is that user can input account, password, phone, email address and register, and can be passed through from the background
JS script carries out validity judgement to these information, increases one newly in the user message table of database if information is all legal
Data;
The resetting cryptographic function is can to give password page input account and mailbox for change when user forgets Password
Verification information, when mailbox verification information is correct, system can allow user to carry out password resetting, and modify phase in user message table
The encrypted message answered.
Wherein, picture format conversion module, which provides, selects local picture and format conversion function, and format conversion function is used
In the dicom formatted data of image center is converted to jpeg format needed for this system, process is as shown in Figure 4.
Wherein, pneumonia prediction module provides patient information addition and picture prediction function, for carrying out patient information
Corresponding image data is simultaneously added in patients database by typing, then using trained prediction model in this system to this
The data of patient carry out forecast analysis, and process is as shown in Figure 5;
Patient information addition is user by by patient's name, gender, age, medical date input system, finally
Picture is added in patient information, JS verification can be carried out to these data from the background, saved later into database.
The picture prediction function is one anticipatory remark of selection ground picture, is completed by prediction model trained in system pair
Whether patient suffers from the prediction of pneumonia.
The system uses B/S framework, divides the system into four layers, respectively podium level, supporting layer, service layer and application layer,
Its framework is as shown in Figure 1;
Wherein, application layer includes system call interfaces, web access interface and result visualization interface, is connected with user terminal
It connects;
Service layer includes user's registration, user authentication, patient imports, picture imports, user logs in, format conversion, model
The user operable interface of load and picture prediction;Wherein, user's registration, user's login, user authentication belong to login and registration
The service that module provides;Format conversion belongs to the service of picture format conversion module offer;Patient imports, picture imports, model
Load, picture prediction belong to the service of pneumonia prediction module offer;
Supporting layer includes the classification method based on depth convolutional neural networks, the data base administration based on relevant database
System, traditional images processing method, medical image processing method;Classification method, biography wherein based on depth convolutional neural networks
System image processing method provides service for picture prediction;Data base management system based on relevant database be user's registration,
User logs in, user authentication provides service;Medical image processing method provides service for format conversion;
Podium level use Keras frame, according to convolutional layer, BN layers, advanced activation primitive layer, pond layer, full articulamentum with
And softmax layers be designed depth convolutional neural networks, selects loss function binary cross-entropy and optimization
Function RMSProp optimizes convolutional neural networks;Data depositary management using sqlite3 relevant database as this system
Reason system;Platform using ITK software library as medical image processing method;Using opencv computer vision library as tradition
The platform of image processing method.
In order to solve the above technical problems, the technical solution used in the present invention is: a kind of use the X- based on deep learning
The method that Ray rabat pneumonia intelligent diagnosis system is diagnosed, the process of this method is as shown in fig. 6, include the following steps:
Step 1: obtaining the X-Ray rabat image of a dicom format, the ReadImage method in the library ITK is called to read
Dicom format picture, later by calling the pixel square in ITK in GetArrayFromImage method extraction dicom image
Battle array, finally by the imwrite method in opencv by extracted picture element matrix preservation at the picture of jpeg format;
Step 2: selection data set Chest X-Ray Images (Pneumonia) generates training set and test set;
Step 3: depth convolutional neural networks VGG prediction model is established, as shown in fig. 7, wherein VGG model includes six volumes
Lamination, BN layers, advanced activation primitive, two full articulamentums and last softmax layer, and model training the number of iterations is set
The value of epoch;
Transfer learning is carried out to model using the trained weight of ImageNet for the weight load of model, in convolutional layer
BatchNormalization layers are added between activation primitive to accelerate network convergence rate, advanced activation primitive
LeakyRelu replaces Relu, excellent using cross entropy loss function binary cross-entropy as model optimization index
Change function and model convergence is accelerated using RMSProp method;
Step 4: VGG prediction model being trained with training set;
Step 5: using the method for overturning, rotation, the enhancing of these data of affine transformation, the image data of training set being carried out
Enhancing processing, obtains new training set;
Step 6: input test collection tests trained VGG prediction model, obtains predictablity rate;
Step 7: step 4 is repeated to step 7, and training is iterated to depth convolutional neural networks VGG prediction model,
Until the number of iterations reaches the value of preset epoch, stopping iteration;
Step 8: the highest VGG model of accuracy rate on test set is saved;
Step 9: the picture for the jpeg format that step 1 obtains is input to the highest VGG model of accuracy rate that step 8 saves
In, obtain the classification prediction result of picture.
The beneficial effects of adopting the technical scheme are that
1, the present invention is compared to traditional deep learning algorithm, by data enhancing, transfer learning and to network structure
It the convergence rate that means accelerate model training such as improves, improves the accuracy rate of pneumonia identification and the extensive energy of model
Power;
2, the BS application based on Django that present invention provides one, simplifies user installation process, mitigates to local ring
The dependence in border.By friendly interface and the system module designed around actual demand, can this system be easy to
Ground is received by each hospital, can also be easy to and the other systems of hospital carry out fusion functionally.Based on MVC framework point
Application also improve the efficiency of later maintenance and redevelopment;
3, by intelligent auxiliary diagnosis system, artificial diagosis can also be significantly reduced and perplexed to doctor's bring.
Detailed description of the invention
Fig. 1 is a kind of X-Ray rabat pneumonia intelligent diagnosis system architecture diagram based on deep learning of the present invention;
Fig. 2 is a kind of brief business process map of X-Ray rabat pneumonia intelligent diagnosis system based on deep learning of the present invention;
Fig. 3 is that user of the present invention logs in and registration module flow chart;
Fig. 4 is picture format conversion module flow chart of the present invention;
The position Fig. 5 pneumonia prediction module flow chart of the present invention;
Fig. 6 is a kind of X-Ray rabat pneumonia intelligent diagnosing method flow chart based on deep learning of the present invention;
Fig. 7 is depth convolutional neural networks VGG prediction model structure chart of the present invention;
Fig. 8 is user's login interface figure in the embodiment of the present invention;
Fig. 9 is user's registration surface chart in the embodiment of the present invention;
Figure 10 is picture format transition interface figure in the embodiment of the present invention;
Figure 11 is to add patient information in the embodiment of the present invention and upload patient X-Ray rabat and carry out prediction interface figure;
Figure 12 is patient's X-Ray rabat pneumonia prediction result output interface in the embodiment of the present invention.
Specific embodiment
With reference to the accompanying drawings and examples, specific embodiments of the present invention will be described in further detail.Implement below
Example is not intended to limit the scope of the invention for illustrating the present invention.
A kind of X-Ray rabat pneumonia intelligent diagnosis system based on deep learning, system include that three modules are respectively as follows: use
Family logs in and registration module, picture format conversion module and pneumonia prediction module, and the brief operation flow of system is as shown in Figure 2;
Wherein, user, which logs in, provides login function, registering functional and resetting cryptographic function with registration module, for user
The entrance for entering system is provided, process is as shown in Figure 3;
The login function needs user to input account and password in login interface and submits to system, user's login interface
As shown in figure 8, account of the System Back-end according to input, corresponding password is inquired into the user message table of database, if returned
Returning result is sky, then illustrates that the account of user's input is not present, if returning the result of goods, entered passwords does not match with user,
Illustrate user password input error, only when the password of user's input and the password inquired from database match, system
It can just show and jump interface accordingly;
The registering functional is that user can input account, password, phone, email address and register, user's registration circle
Face is as shown in figure 9, backstage can carry out validity judgement to these information by JS script, in data if information is all legal
Increase a data in the user message table in library newly;
The resetting cryptographic function is can to give password page input account and mailbox for change when user forgets Password
Verification information, when mailbox verification information is correct, system can allow user to carry out password resetting, and modify phase in user message table
The encrypted message answered.
Wherein, picture format conversion module, which provides, selects local picture and format conversion function, and picture format converts boundary
Face is as shown in Figure 10, and format conversion function is used to the dicom formatted data of image center being converted to jpeg lattice needed for this system
Formula.
Wherein, pneumonia prediction module provides patient information addition and picture prediction function, for carrying out patient information
Corresponding image data is simultaneously added in patients database by typing, is added patient information and is uploaded patient's X-Ray rabat and goes forward side by side
Row prediction interface is as shown in figure 11, then carries out prediction point using data of the prediction model trained in this system to the patient
Analysis;
Patient information addition is user by by patient's name, gender, age, medical date input system, finally
It is added to picture in patient information, JS verification can be carried out to these data from the background, saved later into database.
The picture prediction function is one anticipatory remark of selection ground picture, is completed by prediction model trained in system pair
Whether patient suffers from the prediction of pneumonia.
The system uses B/S framework, divides the system into four layers, respectively podium level, supporting layer, service layer and application layer,
Its framework is as shown in Figure 1;
Wherein, application layer includes system call interfaces, web access interface and result visualization interface, is connected with user terminal
It connects;
Service layer includes user's registration, user authentication, patient imports, picture imports, user logs in, format conversion, model
The user operable interface of load and picture prediction;Wherein, user's registration, user's login, user authentication belong to login and registration
The service that module provides;Format conversion belongs to the service of picture format conversion module offer;Patient imports, picture imports, model
Load, picture prediction belong to the service of pneumonia prediction module offer;
Supporting layer includes the classification method based on depth convolutional neural networks, the data base administration based on relevant database
System, traditional images processing method, medical image processing method;Classification method, biography wherein based on depth convolutional neural networks
System image processing method provides service for picture prediction;Data base management system based on relevant database be user's registration,
User logs in, user authentication provides service;Medical image processing method provides service for format conversion;
Podium level use Keras frame, according to convolutional layer, BN layers, advanced activation primitive layer, pond layer, full articulamentum with
And softmax layers be designed depth convolutional neural networks, selects loss function binary cross-entropy and optimization
Function RMSProp optimizes convolutional neural networks;Data depositary management using sqlite3 relevant database as this system
Reason system;Platform using ITK software library as medical image processing method;Using opencv computer vision library as tradition
The platform of image processing method.
In order to solve the above technical problems, the technical solution used in the present invention is: a kind of use the X- based on deep learning
The method that Ray rabat pneumonia intelligent diagnosis system is diagnosed, the process of this method is as shown in fig. 6, include the following steps:
Step 1: obtaining the X-Ray rabat image of a dicom format, the ReadImage method in the library ITK is called to read
Dicom format picture, later by calling the pixel square in ITK in GetArrayFromImage method extraction dicom image
Battle array, finally by the imwrite method in opencv by extracted picture element matrix preservation at the picture of jpeg format;
Step 2: selection data set Chest X-Ray Images (Pneumonia) generates training set and test set;
Step 3: depth convolutional neural networks VGG prediction model is established, as shown in fig. 7, wherein VGG model includes six volumes
Model training the number of iterations is arranged in lamination, BN layers, advanced activation primitive, two full articulamentums and last softmax layer
Epoch=50;
Transfer learning is carried out to model using the trained weight of ImageNet for the weight load of model, in convolutional layer
BatchNormalization layers are added between activation primitive to accelerate network convergence rate, advanced activation primitive
LeakyRelu replaces Relu, excellent using cross entropy loss function binary cross-entropy as model optimization index
Change function and model convergence is accelerated using RMSProp method;
Step 4: VGG prediction model being trained with training set;
Step 5: using the method for overturning, rotation, the enhancing of these data of affine transformation, the image data of training set being carried out
Enhancing processing, obtains new training set;
Step 6: input test collection tests trained VGG prediction model, obtains predictablity rate;
Step 7: step 4 is repeated to step 7, and training is iterated to depth convolutional neural networks VGG prediction model,
Until the number of iterations reaches the value of preset epoch, stopping iteration;
Step 8: the highest VGG model of accuracy rate on test set is saved;
Step 9: the picture for the jpeg format that step 1 obtains is input to the highest VGG model of accuracy rate that step 8 saves
In, obtain the classification prediction result of picture.
Prediction result is arranged in the present embodiment and exports rule: 1 is pneumonia, and 0 is normal;
The X-Ray rabat image prediction result output of patient is 0 in the present embodiment, can determine whether that the patient does not suffer from an inflammation of the lungs,
As a result output interface is as shown in figure 12.
Claims (6)
1. a kind of X-Ray rabat pneumonia intelligent diagnosis system based on deep learning, it is characterised in that log in and infuse including user
Volume module, picture format conversion module and pneumonia prediction module, the system use B/S framework, divide the system into four layers, respectively
Podium level, supporting layer, service layer and application layer;
Wherein, application layer includes system call interfaces, web access interface and result visualization interface, is connected with user terminal;
Service layer includes user's registration, user authentication, patient imports, picture imports, user logs in, format conversion, model load
With the user operable interface of picture prediction;Wherein, user's registration, user's login, user authentication belong to login and registration module
The service of offer;Format conversion belongs to the service of picture format conversion module offer;Patient imports, picture imports, model loads,
Picture prediction belongs to the service of pneumonia prediction module offer;
Supporting layer includes the classification method based on depth convolutional neural networks, the data base administration system based on relevant database
System, traditional images processing method, medical image processing method;Classification method, tradition wherein based on depth convolutional neural networks
Image processing method provides service for picture prediction;Data base management system based on relevant database is user's registration, uses
Family logs in, user authentication provides service;Medical image processing method provides service for format conversion;
Podium level use Keras frame, according to convolutional layer, BN layers, advanced activation primitive layer, pond layer, full articulamentum and
Softmax layers are designed depth convolutional neural networks, select loss function binary cross-entropy and optimization letter
Number RMSProp optimizes convolutional neural networks;Data base administration using sqlite3 relevant database as this system
System;Platform using ITK software library as medical image processing method;Using opencv computer vision library as tradition figure
As the platform of processing method.
2. a kind of X-Ray rabat pneumonia intelligent diagnosis system based on deep learning according to claim 1, feature exist
It is logged in user and provides login function, registering functional and resetting cryptographic function with registration module, enter system for providing user
The entrance of system;
The login function needs user to input account and password in login interface and submits to system, and System Back-end is according to input
Account, corresponding password is inquired into the user message table of database, if returned the result as sky, illustrates user's input
Account is not present, if entered passwords does not match with user for the result returned, illustrates user password input error, only when
When the password of user's input and the password inquired from database match, system can just be shown jumps interface accordingly;
The registering functional is that user can input account, password, phone, email address and register, and can pass through JS foot from the background
This carries out validity judgement to these information, increases a number newly in the user message table of database if information is all legal
According to;
The resetting cryptographic function be when user forgets Password, can give for change the password page input account and mailbox verifying
Information, when mailbox verification information is correct, system can allow user to carry out password resetting, and modify corresponding in user message table
Encrypted message.
3. a kind of X-Ray rabat pneumonia intelligent diagnosis system based on deep learning according to claim 1, feature exist
It is provided in picture format conversion module and selects local picture and format conversion function, format conversion function is used for image center
Dicom formatted data be converted to jpeg format needed for this system.
4. a kind of X-Ray rabat pneumonia intelligent diagnosis system based on deep learning according to claim 1, feature exist
Patient information addition and picture prediction function are provided in pneumonia prediction module, for patient information to be carried out typing and will be corresponded to
Image data be added in patients database, then using trained prediction model in this system to the data of the patient into
Row forecast analysis;
The patient information addition is user by will finally scheme patient's name, gender, age, medical date input system
Piece is added in patient information, can be carried out JS verification to these data from the background, be saved later into database;
The picture prediction function is one anticipatory remark of selection ground picture, is completed by prediction model trained in system to patient
Whether the prediction of pneumonia is suffered from.
5. a kind of X-Ray rabat pneumonia intelligent diagnosis system using described in claim 1 based on deep learning is diagnosed
Method, it is characterised in that include the following steps:
Step 1: obtaining the X-Ray rabat image of a dicom format, the ReadImage method in the library ITK is called to read
Dicom format picture, later by calling the pixel square in ITK in GetArrayFromImage method extraction dicom image
Battle array, finally by the imwrite method in opencv by extracted picture element matrix preservation at the picture of jpeg format;
Step 2: selection data set Chest X-Ray Images (Pneumonia) generates training set and test set;
Step 3: establish depth convolutional neural networks VGG prediction model, wherein VGG model include six convolutional layers, BN layers, it is advanced
Activation primitive, two full articulamentums and last softmax layer, and the value of model training the number of iterations epoch is set;
Step 4: VGG prediction model being trained with training set;
Step 5: using the method for overturning, rotation, the enhancing of these data of affine transformation, the image data of training set being enhanced
Processing, obtains new training set;
Step 6: input test collection tests trained VGG prediction model, obtains predictablity rate;
Step 7: step 4 is repeated to step 7, and training is iterated to depth convolutional neural networks VGG prediction model, until
The number of iterations reaches the value of preset epoch, stops iteration;
Step 8: the highest VGG model of accuracy rate on test set is saved;
Step 9: the picture for the jpeg format that step 1 obtains is input in the highest VGG model of accuracy rate that step 8 saves,
Obtain the classification prediction result of picture.
6. a kind of X-Ray rabat pneumonia intelligence using described in claim 1 based on deep learning according to claim 5
The method that energy diagnostic system is diagnosed, it is characterised in that depth convolutional neural networks VGG prediction mould is established in the step 3
Type carries out transfer learning to model using the trained weight of ImageNet for the weight load of model, in convolutional layer and swashs
BatchNormalization layers are added between function living to accelerate network convergence rate, advanced activation primitive LeakyRelu generation
For Relu, using cross entropy loss function binary cross-entropy as model optimization index, majorized function is used
RMSProp method accelerates model convergence.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910767278.4A CN110504027A (en) | 2019-08-20 | 2019-08-20 | A kind of X-Ray rabat pneumonia intelligent diagnosis system and method based on deep learning |
PCT/CN2019/107225 WO2021031279A1 (en) | 2019-08-20 | 2019-09-23 | Deep-learning-based intelligent pneumonia diagnosis system and method for x-ray chest radiograph |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910767278.4A CN110504027A (en) | 2019-08-20 | 2019-08-20 | A kind of X-Ray rabat pneumonia intelligent diagnosis system and method based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110504027A true CN110504027A (en) | 2019-11-26 |
Family
ID=68588848
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910767278.4A Pending CN110504027A (en) | 2019-08-20 | 2019-08-20 | A kind of X-Ray rabat pneumonia intelligent diagnosis system and method based on deep learning |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN110504027A (en) |
WO (1) | WO2021031279A1 (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112349425A (en) * | 2020-02-10 | 2021-02-09 | 胡秋明 | Novel artificial intelligent rapid screening system for coronavirus infection pneumonia |
CN112631898A (en) * | 2020-12-09 | 2021-04-09 | 南京理工大学 | Software defect prediction method based on CNN-SVM |
CN113205126A (en) * | 2021-04-27 | 2021-08-03 | 武汉工程大学 | Chest X-ray film identification and detection system and method based on neural network |
CN113476065A (en) * | 2021-09-06 | 2021-10-08 | 华南理工大学 | Multiclass pneumonia diagnostic system |
CN114821826A (en) * | 2022-02-03 | 2022-07-29 | 北京泽桥医疗科技股份有限公司 | Visual system for interactive three-dimensional human body medical model |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113569866A (en) * | 2021-07-15 | 2021-10-29 | 桂林电子科技大学 | Method for identifying HPV test paper based on deep learning |
CN113610807B (en) * | 2021-08-09 | 2024-02-09 | 西安电子科技大学 | New coronaries pneumonia segmentation method based on weak supervision multitask learning |
CN114461516B (en) * | 2021-12-23 | 2024-03-29 | 连邦网络科技服务南通有限公司 | Software graphic interface test method capable of automatically deleting data and server |
CN116468959B (en) * | 2023-06-15 | 2023-09-08 | 清软微视(杭州)科技有限公司 | Industrial defect classification method, device, electronic equipment and storage medium |
CN116525077B (en) * | 2023-06-29 | 2023-09-08 | 安翰科技(武汉)股份有限公司 | Processing method and system for artificial intelligent medical instrument test data |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106372390A (en) * | 2016-08-25 | 2017-02-01 | 姹ゅ钩 | Deep convolutional neural network-based lung cancer preventing self-service health cloud service system |
CN107749308A (en) * | 2017-10-18 | 2018-03-02 | 北京航空航天大学 | A kind of background management system that support is provided for Thyroid ultrasound diagnosis APP |
CN109411036A (en) * | 2017-05-09 | 2019-03-01 | 北京华信诚达科技有限公司 | A kind of cloud platform system and method monitoring various disease surgery complication |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
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 |
CN104732468B (en) * | 2015-04-08 | 2018-06-08 | 南京吉星兆健康信息咨询有限公司 | A kind of Telemedicine System based on mobile interchange |
CN108281158A (en) * | 2018-01-12 | 2018-07-13 | 平安科技(深圳)有限公司 | Voice biopsy method, server and storage medium based on deep learning |
CN108776962A (en) * | 2018-04-11 | 2018-11-09 | 浙江师范大学 | A method of the structure good pernicious prediction model of lung neoplasm |
CN108876775B (en) * | 2018-06-12 | 2022-10-18 | 湖南大学 | Method for rapidly detecting diabetic retinopathy |
CN109379329B (en) * | 2018-09-05 | 2021-12-21 | 中国人民解放军战略支援部队信息工程大学 | Network security protocol fuzzy test method and system based on LSTM |
CN109712693A (en) * | 2018-12-14 | 2019-05-03 | 贵州大学 | A kind of intelligence pathological diagnosis method and system |
CN109686444A (en) * | 2018-12-27 | 2019-04-26 | 上海联影智能医疗科技有限公司 | System and method for medical image classification |
-
2019
- 2019-08-20 CN CN201910767278.4A patent/CN110504027A/en active Pending
- 2019-09-23 WO PCT/CN2019/107225 patent/WO2021031279A1/en active Application Filing
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106372390A (en) * | 2016-08-25 | 2017-02-01 | 姹ゅ钩 | Deep convolutional neural network-based lung cancer preventing self-service health cloud service system |
CN109411036A (en) * | 2017-05-09 | 2019-03-01 | 北京华信诚达科技有限公司 | A kind of cloud platform system and method monitoring various disease surgery complication |
CN107749308A (en) * | 2017-10-18 | 2018-03-02 | 北京航空航天大学 | A kind of background management system that support is provided for Thyroid ultrasound diagnosis APP |
Non-Patent Citations (1)
Title |
---|
钟文雅: ""基于深度卷积神经网络的乳腺X线图像的分类方法研究"", 《中国优秀硕士论文 信息科技》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112349425A (en) * | 2020-02-10 | 2021-02-09 | 胡秋明 | Novel artificial intelligent rapid screening system for coronavirus infection pneumonia |
CN112631898A (en) * | 2020-12-09 | 2021-04-09 | 南京理工大学 | Software defect prediction method based on CNN-SVM |
CN113205126A (en) * | 2021-04-27 | 2021-08-03 | 武汉工程大学 | Chest X-ray film identification and detection system and method based on neural network |
CN113476065A (en) * | 2021-09-06 | 2021-10-08 | 华南理工大学 | Multiclass pneumonia diagnostic system |
CN113476065B (en) * | 2021-09-06 | 2022-04-22 | 华南理工大学 | Multiclass pneumonia diagnostic system |
CN114821826A (en) * | 2022-02-03 | 2022-07-29 | 北京泽桥医疗科技股份有限公司 | Visual system for interactive three-dimensional human body medical model |
Also Published As
Publication number | Publication date |
---|---|
WO2021031279A1 (en) | 2021-02-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110504027A (en) | A kind of X-Ray rabat pneumonia intelligent diagnosis system and method based on deep learning | |
Khan et al. | Deep neural architectures for medical image semantic segmentation | |
Zhong et al. | 3D fully convolutional networks for co-segmentation of tumors on PET-CT images | |
Chen et al. | Fully automated multiorgan segmentation in abdominal magnetic resonance imaging with deep neural networks | |
Kurgan et al. | Knowledge discovery approach to automated cardiac SPECT diagnosis | |
Young et al. | Computational cardiac atlases: from patient to population and back | |
Li et al. | When medical images meet generative adversarial network: recent development and research opportunities | |
CN110288609B (en) | Multi-modal whole-heart image segmentation method guided by attention mechanism | |
CN106456098B (en) | The generation method and system of decay pattern | |
Banerjee et al. | A completely automated pipeline for 3D reconstruction of human heart from 2D cine magnetic resonance slices | |
CN106170246A (en) | For four-dimensional (4D) stream equipment of nuclear magnetic resonance, method and product | |
CN109035355A (en) | System and method for PET image reconstruction | |
CN107638188A (en) | Image attenuation bearing calibration and device | |
CN109378054A (en) | A kind of multi-modality images assistant diagnosis system and its building method | |
Liu et al. | Deep learning-based automated left ventricular ejection fraction assessment using 2-D echocardiography | |
CN105184794B (en) | A kind of CSM Computer Aided Analysis Systems and method based on tensor image | |
Lei et al. | Echocardiographic image multi‐structure segmentation using Cardiac‐SegNet | |
Li et al. | Automatic quantification of epicardial adipose tissue volume | |
Shekhar et al. | High-speed registration of three-and four-dimensional medical images by using voxel similarity | |
WO2023134030A1 (en) | Pet system attenuation correction method based on flow model | |
CN108038904A (en) | Medical image three-dimensional reconstruction system | |
Wang et al. | IGNFusion: an unsupervised information gate network for multimodal medical image fusion | |
CN110458779A (en) | The correction information acquiring method of correction for attenuation is carried out to the PET image of breathing or heart | |
CN105868537A (en) | PET-CT (Positron emission tomography-computed tomography) dynamic medical image intelligent quantitative analysis system and analysis method | |
CN103377369B (en) | Generate the method and system of the cardiovascular measurement identified for individuality |
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 |