CN108985366A - B ultrasound image recognition algorithm and system based on convolution depth network - Google Patents
B ultrasound image recognition algorithm and system based on convolution depth network Download PDFInfo
- Publication number
- CN108985366A CN108985366A CN201810733678.9A CN201810733678A CN108985366A CN 108985366 A CN108985366 A CN 108985366A CN 201810733678 A CN201810733678 A CN 201810733678A CN 108985366 A CN108985366 A CN 108985366A
- Authority
- CN
- China
- Prior art keywords
- ultrasound image
- convolution
- neural network
- input
- network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Abstract
The invention proposes a kind of B ultrasound image recognition algorithm and system based on convolution depth network comprising following steps, S1 obtain B ultrasound image, and by diagnosis result as image tag；The B ultrasound image that step S1 is obtained is trained neural network as sample set, trained neural network is saved by S2；S3 obtains recognition result according to output vector using B ultrasound image to be identified as input.By the way that directly using digitized image pixel as input, training obtains the supplementary model based on convolution, effectively realizes the identification and reconstruct of image, there is the advantages of intelligence degree is high, normalizing operation.
Description
Technical field
The present invention relates to field of image recognition more particularly to a kind of B ultrasound image recognition algorithms based on convolution depth network
And system.
Background technique
The range of B-mode ultrasonography is very wide, specifically includes that (1) abdomen examination: including liver, gallbladder, pancreas, spleen and abdominal cavity etc.；
(2) gynecologial examination；(3) urinary system inspection；(4) superficial tumor and lesion；(5) heart and limb vessel inspection.
For ultrasound diagnosis as a result, it is desirable to medical practitioner diagnoses, and examining report is provided, strong depend-ence doctor
People's experience and level be easy to cause mistaken diagnosis, fail to pinpoint a disease in diagnosis；And each doctors experience is not identical, therefore identical B ultrasound image, it is different
Doctor can also provide different diagnosis, cause to perplex to patient.
Summary of the invention
In view of this, the invention proposes a kind of intelligent, standardized B ultrasound image recognitions based on convolution depth network
Algorithm and system.
The technical scheme of the present invention is realized as follows:
On the one hand, the present invention provides a kind of B ultrasound image recognition algorithms based on convolution depth network comprising following step
Suddenly,
S1 obtains B ultrasound image, and by diagnosis result as image tag；
The B ultrasound image that step S1 is obtained is trained neural network as sample set by S2, by trained nerve net
Network saves；
S3 obtains recognition result according to output vector using B ultrasound image to be identified as input.
On the basis of above technical scheme, it is preferred that the deep neural network is the convolution depth nerve of multilayer
Network, including input layer, 3 convolutional layers, 3 pond layers and 1 output layer, in which: input layer is the pixel of two dimensional image,
First convolutional layer has 24 convolution characteristic patterns, and second convolutional layer has 48 convolution characteristic patterns, and output layer is set as 10 sections
Point.
On the basis of above technical scheme, it is preferred that in step S2, sample set is input to the nerve net set
Network carries out pre-training using restricted Boltzmann innovatory algorithm, then to obtained models coupling backpropagation BP algorithm
The complete training process to neural network is completed in adjusting parameter and biasing.
On the basis of above technical scheme, it is preferred that in step S2 training include,
S2.1, to netinit；Random initializtion is carried out to convolution kernel, full articulamentum weight and biasing；
Training sample and tally set are imported the network initialized and carry out pre-training by S2.2；Boltzmann is improved first
Algorithm model, introduces conditional Gaussian distribution, and building is restricted Boltzmann innovatory algorithm model based on recruitment factor, then draws
Enter convolution operation；Next carry out training pattern using weight uncertainty method to alleviate overfitting problem；Finally, base
In the model construction one supplement depth confidence net based on convolution；Pre-training is carried out using image data, after pre-training, is made
For deep neural network model, input picture obtains result；
S2.3 compares reality output and label, obtains error, using model as neural network, utilizes weight
Uncertainty BP algorithm is finely adjusted, and obtains trained neural network model.
Second aspect, the present invention provides a kind of B ultrasound image identification systems based on convolution depth network, based on the present invention
B ultrasound image recognition algorithm based on convolution depth network described in first aspect, including client input, cloud server end and multiple
Core end, client input and review end are connect with Cloud Server end signal respectively, wherein
End is checked, inputs B ultrasound image and diagnostic result for doctor；
Cloud server end stores neural network；And the B ultrasound image to be identified of client input input is inputted into neural network
It is calculated, obtain recognition result and is sent to client input；
Client input inputs B ultrasound image to be identified for client, receives and show recognition result.
Of the invention B ultrasound image recognition algorithm and system based on convolution depth network has following compared with the existing technology
The utility model has the advantages that
By the way that directly using digitized image pixel as input, training obtains the supplementary model based on convolution, effectively
The identification and reconstruct of image are realized, there is the advantages of intelligence degree is high, normalizing operation.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the flow chart of the B ultrasound image recognition algorithm of the invention based on convolution depth network；
Fig. 2 is the frame diagram of the B ultrasound image identification system of the invention based on convolution depth network.
Specific embodiment
Below in conjunction with embodiment of the present invention, the technical solution in embodiment of the present invention is carried out clearly and completely
Description, it is clear that described embodiment is only some embodiments of the invention, rather than whole embodiments.Base
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts all
Other embodiments shall fall within the protection scope of the present invention.
As shown in Figure 1, the B ultrasound image recognition algorithm of the invention based on convolution depth network comprising following steps,
S1 obtains B ultrasound image, and by diagnosis result as image tag.
Specifically, the deep neural network is the convolution deep neural network of multilayer, including input layer, 3 convolution
Layer, 3 pond layers and 1 output layer, in which: input layer is the pixel of two dimensional image, and first convolutional layer has 24 convolution
Characteristic pattern, second convolutional layer have 48 convolution characteristic patterns, and output layer is set as 10 nodes.
The B ultrasound image that step S1 is obtained is trained neural network as sample set by S2, by trained nerve net
Network saves.
Specifically, sample set to be input to the neural network set in step S2, changed using restricted Boltzmann
Pre-training is carried out into algorithm, then to obtained models coupling backpropagation BP algorithm adjusting parameter and biasing, is completed to nerve
The complete training process of network.
Specifically, training includes in step S2,
S2.1, to netinit；Random initializtion is carried out to convolution kernel, full articulamentum weight and biasing；
Training sample and tally set are imported the network initialized and carry out pre-training by S2.2；Boltzmann is improved first
Algorithm model, introduces conditional Gaussian distribution, and building is restricted Boltzmann innovatory algorithm model based on recruitment factor, then draws
Enter convolution operation；Next carry out training pattern using weight uncertainty method to alleviate overfitting problem；Finally, base
In the model construction one supplement depth confidence net based on convolution；Pre-training is carried out using image data, after pre-training, is made
For deep neural network model, input picture obtains result.
S2.3 compares reality output and label, obtains error, using model as neural network, utilizes weight
Uncertainty BP algorithm is finely adjusted, and obtains trained neural network model.
S3 obtains recognition result according to output vector using B ultrasound image to be identified as input.
As shown in Fig. 2, the B ultrasound image identification system of the invention based on convolution depth network, is based on first party of the present invention
The B ultrasound image recognition algorithm based on convolution depth network in face, including client input 1, cloud server end 2 and review end 3, visitor
Family input terminal 1 and review end 3 are connect with 2 signal of cloud server end respectively, wherein
End 3 is checked, inputs B ultrasound image and diagnostic result for doctor；
Cloud server end 2 stores neural network；And the B ultrasound image to be identified for inputting client input 1 inputs nerve net
Network is calculated, and is obtained recognition result and is sent to client input 1；
Client input 1 inputs B ultrasound image to be identified for client, receives and show recognition result.
The foregoing is merely better embodiments of the invention, are not intended to limit the invention, all of the invention
Within spirit and principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (5)
1. a kind of B ultrasound image recognition algorithm based on convolution depth network, it is characterised in that: it includes the following steps,
S1 obtains B ultrasound image, and by diagnosis result as image tag；
The B ultrasound image that step S1 is obtained is trained neural network as sample set, trained neural network is protected by S2
It deposits；
S3 obtains recognition result according to output vector using B ultrasound image to be identified as input.
2. the B ultrasound image recognition algorithm as described in claim 1 based on convolution depth network, it is characterised in that: the depth
The convolution deep neural network that neural network is multilayer, including input layer, 3 convolutional layers, 3 pond layers and 1 output layer are spent,
Wherein: input layer is the pixel of two dimensional image, and first convolutional layer has 24 convolution characteristic patterns, and second convolutional layer has 48
Convolution characteristic pattern, output layer are set as 10 nodes.
3. the B ultrasound image recognition algorithm as described in claim 1 based on convolution depth network, it is characterised in that: in step S2,
Sample set is input to the neural network set, carries out pre-training using restricted Boltzmann innovatory algorithm, it is then right
The complete training process to neural network is completed in obtained models coupling backpropagation BP algorithm adjusting parameter and biasing.
4. the B ultrasound image recognition algorithm as described in claim 1 based on convolution depth network, it is characterised in that: in step S2
Training includes,
S2.1, to netinit；Random initializtion is carried out to convolution kernel, full articulamentum weight and biasing；
Training sample and tally set are imported the network initialized and carry out pre-training by S2.2；Boltzmann algorithm is improved first
Model introduces conditional Gaussian distribution, and building is restricted Boltzmann innovatory algorithm model based on recruitment factor, then introduces volume
Product operation；Next carry out training pattern using weight uncertainty method to alleviate overfitting problem；Finally, being based on mould
Type constructs a supplement depth confidence net based on convolution；Pre-training is carried out using image data, after pre-training, as depth
Spend neural network model, input picture, obtain as a result,
S2.3 compares reality output and label, obtains error, using model as neural network, utilizes weight
Uncertainty BP algorithm is finely adjusted, and obtains trained neural network model.
5. a kind of B ultrasound image identification system based on convolution depth network is based on convolution depth net based on described in claim 1
The B ultrasound image recognition algorithm of network, it is characterised in that: including client input (1), cloud server end (2) and review end (3), visitor
Family input terminal (1) and review end (3) are connect with cloud server end (2) signal respectively, wherein
It checks end (3), inputs B ultrasound image and diagnostic result for doctor；
Cloud server end (2) stores neural network；And the B ultrasound image to be identified of client input (1) input is inputted into nerve net
Network is calculated, and is obtained recognition result and is sent to client input (1)；
Client input (1) inputs B ultrasound image to be identified for client, receives and show recognition result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810733678.9A CN108985366A (en) | 2018-07-06 | 2018-07-06 | B ultrasound image recognition algorithm and system based on convolution depth network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810733678.9A CN108985366A (en) | 2018-07-06 | 2018-07-06 | B ultrasound image recognition algorithm and system based on convolution depth network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108985366A true CN108985366A (en) | 2018-12-11 |
Family
ID=64536327
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810733678.9A Pending CN108985366A (en) | 2018-07-06 | 2018-07-06 | B ultrasound image recognition algorithm and system based on convolution depth network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108985366A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109658399A (en) * | 2018-12-13 | 2019-04-19 | 深圳先进技术研究院 | A kind of neck patch image-recognizing method and device |
| CN109829512A (en) * | 2019-03-01 | 2019-05-31 | 华东师范大学 | A kind of image recognition mould group based on deep neural network |
| CN110246135A (en) * | 2019-07-22 | 2019-09-17 | 新名医(北京)科技有限公司 | Monitor Follicles method, apparatus, system and storage medium |
| CN110378888A (en) * | 2019-07-22 | 2019-10-25 | 新名医(北京)科技有限公司 | A kind of physiology phase monitoring method, device, ultrasonic device and storage medium |
| CN110378372A (en) * | 2019-06-11 | 2019-10-25 | 中国科学院自动化研究所南京人工智能芯片创新研究院 | Diagram data recognition methods, device, computer equipment and storage medium |
| CN112861876A (en) * | 2021-01-25 | 2021-05-28 | 北京小白世纪网络科技有限公司 | Automatic liver cancer ultrasonic image identification method and device based on convolutional neural network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101161204A (en) * | 2007-11-20 | 2008-04-16 | 深圳市迈科龙影像技术有限公司 | Method and apparatus for recognizing intelligently ultrasound womb contraceptive ring image |
| US20090318779A1 (en) * | 2006-05-24 | 2009-12-24 | Bao Tran | Mesh network stroke monitoring appliance |
| CN105574820A (en) * | 2015-12-04 | 2016-05-11 | 南京云石医疗科技有限公司 | Deep learning-based adaptive ultrasound image enhancement method |
| CN205665697U (en) * | 2016-04-05 | 2016-10-26 | 陈进民 | Medical science video identification diagnostic system based on cell neural network or convolution neural network |
| CN107145893A (en) * | 2017-03-13 | 2017-09-08 | 中国矿业大学 | A kind of image recognition algorithm and system based on convolution depth network |
-
2018
- 2018-07-06 CN CN201810733678.9A patent/CN108985366A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090318779A1 (en) * | 2006-05-24 | 2009-12-24 | Bao Tran | Mesh network stroke monitoring appliance |
| CN101161204A (en) * | 2007-11-20 | 2008-04-16 | 深圳市迈科龙影像技术有限公司 | Method and apparatus for recognizing intelligently ultrasound womb contraceptive ring image |
| CN105574820A (en) * | 2015-12-04 | 2016-05-11 | 南京云石医疗科技有限公司 | Deep learning-based adaptive ultrasound image enhancement method |
| CN205665697U (en) * | 2016-04-05 | 2016-10-26 | 陈进民 | Medical science video identification diagnostic system based on cell neural network or convolution neural network |
| CN107145893A (en) * | 2017-03-13 | 2017-09-08 | 中国矿业大学 | A kind of image recognition algorithm and system based on convolution depth network |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109658399A (en) * | 2018-12-13 | 2019-04-19 | 深圳先进技术研究院 | A kind of neck patch image-recognizing method and device |
| CN109829512A (en) * | 2019-03-01 | 2019-05-31 | 华东师范大学 | A kind of image recognition mould group based on deep neural network |
| CN110378372A (en) * | 2019-06-11 | 2019-10-25 | 中国科学院自动化研究所南京人工智能芯片创新研究院 | Diagram data recognition methods, device, computer equipment and storage medium |
| CN110246135A (en) * | 2019-07-22 | 2019-09-17 | 新名医(北京)科技有限公司 | Monitor Follicles method, apparatus, system and storage medium |
| CN110378888A (en) * | 2019-07-22 | 2019-10-25 | 新名医(北京)科技有限公司 | A kind of physiology phase monitoring method, device, ultrasonic device and storage medium |
| CN110246135B (en) * | 2019-07-22 | 2022-01-18 | 新名医(北京)科技有限公司 | Follicle monitoring method, device, system and storage medium |
| CN110378888B (en) * | 2019-07-22 | 2022-01-25 | 新名医(北京)科技有限公司 | Physiological period monitoring method and device, ultrasonic equipment and storage medium |
| CN112861876A (en) * | 2021-01-25 | 2021-05-28 | 北京小白世纪网络科技有限公司 | Automatic liver cancer ultrasonic image identification method and device based on convolutional neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108985366A (en) | B ultrasound image recognition algorithm and system based on convolution depth network | |
| Abdi et al. | Automatic quality assessment of echocardiograms using convolutional neural networks: feasibility on the apical four-chamber view | |
| Brehar et al. | Comparison of deep-learning and conventional machine-learning methods for the automatic recognition of the hepatocellular carcinoma areas from ultrasound images | |
| CN109978778A (en) | Convolutional neural networks medicine CT image denoising method based on residual error study | |
| CN108537794A (en) | Medical image processing method, device and computer readable storage medium | |
| Han et al. | Automatic segmentation of human placenta images with U-Net | |
| Jin et al. | DC2Anet: generating lumbar spine MR images from CT scan data based on semi-supervised learning | |
| CN110223289A (en) | A kind of image processing method and system | |
| CN110163877A (en) | A kind of method and system of MRI ventricular structure segmentation | |
| Ma et al. | A novel deep learning framework for automatic recognition of thyroid gland and tissues of neck in ultrasound image | |
| Khorami et al. | Optimal diagnosis of COVID-19 based on convolutional neural network and red Fox optimization algorithm | |
| Xie et al. | Vessel lumen segmentation in carotid artery ultrasounds with the U-Net convolutional neural network | |
| Wang et al. | Deep transfer learning-based multi-modal digital twins for enhancement and diagnostic analysis of brain mri image | |
| Indhumathi et al. | Hybrid pixel based method for multimodal image fusion based on Integration of Pulse Coupled Neural Network (PCNN) and Genetic Algorithm (GA) using Empirical Mode Decomposition (EMD) | |
| Shankarlal et al. | Computer-aided detection and diagnosis of thyroid nodules using machine and deep learning classification algorithms | |
| Lu et al. | CDA-net for automatic prostate segmentation in MR images | |
| Ruan et al. | An efficient tongue segmentation model based on u-net framework | |
| CN115690556B (en) | Image recognition method and system based on multi-mode imaging features | |
| Sengan et al. | Echocardiographic image segmentation for diagnosing fetal cardiac rhabdomyoma during pregnancy using deep learning | |
| Xia et al. | AwCPM-Net: a collaborative constraint GAN for 3D coronary artery reconstruction in intravascular ultrasound sequences | |
| CN115239740A (en) | GT-UNet-based full-center segmentation algorithm | |
| CN114494482A (en) | Method for generating CT blood vessel imaging based on flat scanning CT | |
| Farhad et al. | A review of medical diagnostic video analysis using deep learning techniques | |
| Wang et al. | Enhanced three‐dimensional U‐Net with graph‐based refining for segmentation of gastrointestinal stromal tumours | |
| CN116402906B (en) | Signal grade coding method and system based on kidney echo |
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: 20181211 |
|
| RJ01 | Rejection of invention patent application after publication |