CN108230277A - A kind of dual intensity CT picture breakdown methods based on convolutional neural networks - Google Patents
A kind of dual intensity CT picture breakdown methods based on convolutional neural networks Download PDFInfo
- Publication number
- CN108230277A CN108230277A CN201810136595.1A CN201810136595A CN108230277A CN 108230277 A CN108230277 A CN 108230277A CN 201810136595 A CN201810136595 A CN 201810136595A CN 108230277 A CN108230277 A CN 108230277A
- Authority
- CN
- China
- Prior art keywords
- convolutional neural
- neural networks
- dual intensity
- dual
- images
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000009977 dual effect Effects 0.000 title claims abstract description 90
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 72
- 238000000034 method Methods 0.000 title claims abstract description 39
- 230000015556 catabolic process Effects 0.000 title claims abstract description 33
- 239000000463 material Substances 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 12
- 238000013507 mapping Methods 0.000 claims abstract description 4
- 230000001537 neural effect Effects 0.000 claims description 2
- 239000000758 substrate Substances 0.000 abstract description 5
- 238000012545 processing Methods 0.000 abstract description 3
- 239000003814 drug Substances 0.000 abstract description 2
- 210000000988 bone and bone Anatomy 0.000 description 6
- 230000003595 spectral effect Effects 0.000 description 6
- 238000013170 computed tomography imaging Methods 0.000 description 5
- 230000008520 organization Effects 0.000 description 5
- 238000012805 post-processing Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- PLXMOAALOJOTIY-FPTXNFDTSA-N Aesculin Natural products OC[C@@H]1[C@@H](O)[C@H](O)[C@@H](O)[C@H](O)[C@H]1Oc2cc3C=CC(=O)Oc3cc2O PLXMOAALOJOTIY-FPTXNFDTSA-N 0.000 description 1
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000007630 basic procedure Methods 0.000 description 1
- 238000012730 cephalometry Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000009659 non-destructive testing Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
Classifications
-
- G06T5/80—
-
- 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
-
- 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
- G06N3/08—Learning methods
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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]
Abstract
The present invention relates to the picture breakdown of medicine dual intensity and technical field of image processing more particularly to dual intensity CT picture breakdown methods, more particularly to a kind of dual intensity CT picture breakdown methods based on convolutional neural networks.A kind of dual intensity CT picture breakdown methods based on convolutional neural networks, include the following steps：Convolutional neural networks model is designed as the mapping function D (μ in dual intensity decomposition modelH,L；Θ)；It is trained by convolutional neural networks model and training data set pair convolutional neural networks, convolutional neural networks parameter Θ is effectively estimated；Sill efficient-decomposition is carried out to dual intensity CT images using the convolutional neural networks parameter Θ that the convolutional neural networks after training and step 2 obtain.The present invention realizes the reasonable diffluence of different substrate materials material in high energy CT images, low energy CT images, so as to effectively promote the quality that dual intensity CT image substrates material decomposes by the foundation of dual input, the convolutional neural networks model of dual output and intersection convolution.
Description
Technical field
The present invention relates to the picture breakdown of medicine dual intensity and technical field of image processing more particularly to dual intensity CT picture breakdown sides
Method, more particularly to a kind of dual intensity CT picture breakdown methods based on convolutional neural networks.
Background technology
Dual intensity CT image reconstructions have been increasingly being applied to the fields such as medical imaging, safety inspection, non-destructive testing, phase
Than in traditional monoenergetic spectrum CT imaging techniques, dual intensity CT can utilize different power spectrum hypograph dampening informations to realize to different material
The identification of material.Dual intensity CT technologies have broken the physical constraints of traditional monoenergetic spectrum CT, become CT imaging field researchs hot spot and
Difficulties.
The core theory of dual intensity CT imaging techniques is dual intensity CT image reconstruction algorithms, wherein material and energy intersection information
Decoupling is one of critical issue.Compared with conventional CT image reconstruction theory, the difficulty of dual intensity CT image reconstruction algorithms
The features such as being the non-linear of problem, multi-solution and ill-posedness, high dimension.Due to dual intensity CT imaging applications demand difference, institute
The imaging pattern difference of use and the data information collected are different, therefore corresponding dual intensity CT image reconstruction algorithms
It is different.At present, dual intensity CT image reconstruction algorithms can be mainly divided into three classes：Class of Iterative directly reconstructs algorithm, based on projection
The image reconstruction algorithm of domain pretreatment and the image reconstruction algorithm based on image area post processing.
Iteration dual intensity CT image reconstruction algorithms are suitable for geometrically consistent or inconsistent high and low energy set of projections, weight
It is higher to build signal noise ratio (snr) of image.But these methods usually require huge computing cost, calculating speed is slow, seriously reduces algorithm
Practicability.Image reconstruction algorithm based on projection domain pretreatment makes full use of high and low spectral information and polychrome to project generation mould
Non-linear Solve problems are converted into linear solution problem by type, and dual intensity CT figures are realized thereby using conventional CT image reconstruction algorithms
As rebuilding, the influence of hardening artifact can be theoretically effectively eliminated, obtains accurate physical parameter distribution information, calculates letter
Just, it is efficient, it is the mainstream method for reconstructing of current dual intensity CT Image Reconstruction Technologies.But this kind of method is highly dependent on and calibrated
Journey, and high and low energy data for projection collection is required to be consistent on space geometry, i.e., per a pair of high and low energy Cephalometry all
It needs along identical ray path.
Image reconstruction algorithm based on image area post processing is utilized respectively conventional CT image to high and low energy data for projection first
Algorithm for reconstructing reconstructs high and low energy CT images, carries out material decomposition in image area to high and low energy CT reconstruction images, obtains object
The physical parameter distribution image of tomography.The decomposition for realizing the different materials under high low energy image is the image based on post processing of image
The key of algorithm for reconstructing.Dual intensity CT image reconstruction algorithms based on image area post processing are for the space of high and low energy data for projection
Geometrical consistency is of less demanding, calculates simplicity, can inhibit hardening artifact to a certain extent.Also, this method can be answered directly
For in existing imaging system, not needing to additionally increase hardware device, cost is saved.Therefore, this method is widely used in
In existing dual intensity CT imaging systems.Based on this, this patent devises a kind of dual intensity CT picture breakdowns based on convolutional neural networks
Method.
Invention content
The sill image obtained for existing dual intensity CT picture breakdown methods contains that much noise, signal-to-noise ratio are relatively low to ask
Topic, the present invention provides a kind of dual intensity CT picture breakdown methods based on convolutional neural networks, pass through dual input, the volume of dual output
Product neural network model and the foundation for intersecting convolution, realize different substrate materials material in high energy CT images, low energy CT images reasonable point
Stream, so as to effectively promote the quality that dual intensity CT image substrates material decomposes.
To achieve these goals, the present invention uses following technical scheme：
A kind of dual intensity CT picture breakdown methods based on convolutional neural networks, include the following steps：
Step 1：Convolutional neural networks model is designed as the mapping function D (μ in dual intensity decomposition modelH,L；Θ)；
Step 2：It is trained by convolutional neural networks model and training data set pair convolutional neural networks, to convolutional Neural net
Network parameter Θ is effectively estimated；
Step 3：Using the convolutional neural networks parameter Θ that the convolutional neural networks after training and step 2 obtain to dual intensity CT images
Carry out sill efficient-decomposition.
Preferably, the convolutional neural networks modelling is dual input, the network structure model of dual output is to realize height
The direct output directly inputted with different materials image of energy CT image, low energy CT images.
Preferably, the dual input, dual output network structure model establish intersect convolution with realize high energy CT images,
The reasonable diffluence of different substrate materials material information in low energy CT images.
Preferably, short chain is established in the convolutional neural networks model to connect.
Preferably, the training dataset includes the input data and output data of convolutional neural networks model, described defeated
Go out data for sill image, the input data is that the dual intensity CT obtained according to sill image and corresponding energy information schemes
Picture, the dual intensity CT images include high energy CT images and low energy CT images；Using output data as label data.
Compared with prior art, the device have the advantages that：
1st, the convolutional neural networks model the present invention is based on dual input, dual output decomposes dual intensity CT images, can effectively keep away
Exempt from different-energy, different materials information in input, the information crosstalk of output terminal.
2nd, the present invention is based on the convolutional neural networks models for intersecting convolution can make in high energy CT images, low energy CT images not
With material information reasonable diffluence, different output terminals are reached along crossover network structure.
3rd, the present invention is based on network residual error design philosophys, and short chain is established in convolutional neural networks model and connects and can effectively be promoted
The training effectiveness of convolutional neural networks, conducive to the design of subsequently more profound network.
Description of the drawings
Fig. 1 is a kind of basic procedure schematic diagram of the dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention.
Fig. 2 is a kind of dual input of the dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention, dual output net
Network structural model.
Fig. 3 is a kind of dual input of the dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention, dual output friendship
Pitch network structure model.
Fig. 4 is a kind of dual intensity CT image convolutions god of dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention
Through network model；
Fig. 5 is a kind of being filled by bone and organization material for dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention
Digital Simulation test body mould.
Fig. 6 is a kind of SpekCalc software emulations of the dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention
X ray it is high and low can spectral information figure.
Fig. 7 is a kind of utilization Fig. 6 test body moulds of dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention
With the high energy CT images of Fig. 7 spectral informations generation.
Fig. 8 is a kind of utilization Fig. 6 test body moulds of dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention
With the low energy CT images of Fig. 7 spectral informations generation.
Fig. 9 is that a kind of dual intensity CT picture breakdowns method based on convolutional neural networks of the present invention is obtained using emulation data
Bone image.
Figure 10 is that a kind of dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention are obtained by emulating data
Organization chart picture.
Figure 11 is a kind of practical high energy CT images of the dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention.
Figure 12 is a kind of practical low energy CT images of the dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention.
Figure 13 is that a kind of dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention are obtained by real data
Bone image.
Figure 14 is that a kind of dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention are obtained by real data
Organization chart picture.
Specific embodiment
In order to make it easy to understand, explanation explained below is made to the part noun occurred in the specific embodiment of the present invention：
BP algorithm：Error backpropagation algorithm.
Below in conjunction with the accompanying drawings with specific embodiment the present invention will be further explained explanation：
Embodiment one：
As shown in Figure 1, a kind of dual intensity CT picture breakdown methods based on convolutional neural networks of the present invention, include the following steps：
Step S101：Design dual input, dual output convolutional neural networks model as the mapping function D in dual intensity decomposition model
(μH, L；Θ), wherein μH,LFor dual intensity CT images, the convolutional neural networks modelling is dual input, the network knot of dual output
Structure model, as shown in Fig. 2, A, B are the high energy CT images of input, low energy CT images in figure, M1, M2 are the sill 1 of output, base
Material 2, " → " represent two heterogeneous networks structures；And dual input, dual output network structure model in establish intersection volume
Product, as shown in Figure 3；Short chain is established in the convolutional neural networks model to connect.Specific convolutional neural networks model such as Fig. 4 institutes
Show：Wherein, convolutional neural networks model includes 12 convolutional layers, and except last output convolution, (totally two, correspond to two number of tags
According to) outside, each convolution is followed by a mass (Batch Normalization, BN) and linear amending unit
(rectified linear unit, ReLU), each square represents the combination of convolution, BN and ReLU in Fig. 4, Data_A,
Data_B, Label_A, Label_B represent successively input low energy CT images, input high energy CT images, 1 image of output sill,
Export 2 image of sill.Entire convolutional neural networks model is divided into three levels, and wherein first layer (Stage 1), which is characterized, carries
Layer is taken, the second layer (Stage 2) is shunting layer, and third layer (Stage 3) is output layer.Feature extraction layer is by 1 × 7 × 7 × 64
Convolution composition, output 64 dimension characteristic images；Shunting layer intersects convolution by 4 and 2 residual blocks form, and convolution dimension is
64×3×3×64；Output layer is made of the convolution that dimension is 64 × 5 × 5 × 1, and 64 dimensional features are combined as output image.
Step S102：It is trained by convolutional neural networks model and training data set pair convolutional neural networks, to volume
Product neural network parameter Θ is effectively estimated；As a kind of embodiment, by BP algorithm combined training data set and set
The convolutional neural networks model of meter is trained convolutional neural networks, so as to effectively be estimated to convolutional neural networks parameter Θ
Meter, the convolutional neural networks after being trained, and during supervised training, initial learning rate, step-length, weighted value are set respectively
For 10e-6,0.95,0.0005.The training dataset includes the input data and output data of convolutional neural networks model, institute
Output data is stated as sill image, the input data is double to be obtained according to decomposition sill image and corresponding energy information
Energy CT images, the dual intensity CT images include high energy CT images and low energy CT images；Using output data as label data.As
A kind of embodiment obtains the high and low energy CT images of human body using dual intensity CT equipment, according to doctor's clinical experience, high to human body,
Low energy CT images are split, and bone image (sill 1) and organization chart picture (sill 2) under different-energy are obtained, by it
As output data, i.e. label data, the high and low energy spectral information of X ray is obtained using SpekCalc softwares, according to high and low energy
Spectral information generation projects and rebuilds to obtain high and low energy CT images, using the high and low energy CT images of acquisition as input data, with
This establishes training dataset.As a kind of embodiment, the bone image and tissue of 1300 pair of 512 × 512 pixel are obtained altogether
Image, and generate the high and low energy CT images of 1300 pair of 512 × 512 pixel.By the bone image of 1300 pair of 512 × 512 pixel
It is split with organization chart picture, stride 48, it is refreshing as convolution to obtain the image block that 105300 pairs of sizes are 128 × 128 pixels
Output data through network model, i.e. label data；The high and low energy CT images of 1300 pair of 512 × 512 pixel are split,
Stride is 48, obtains input number of image block of 105300 pairs of sizes for 128 × 128 pixels as convolutional neural networks model
According to.
Step S103：Θ pairs of the convolutional neural networks parameter obtained using the convolutional neural networks after training and step S102
Dual intensity CT images carry out sill efficient-decomposition.
As a kind of embodiment, generation Fig. 7, high energy CT images shown in Fig. 8 and low energy CT are emulated using Fig. 5, Fig. 6
Image, using the convolutional neural networks after training to Fig. 7, Fig. 8 processing, image result such as Fig. 9, figure are decomposed in obtained output
Shown in 10.
As a kind of embodiment, high energy CT images and low energy CT images as shown in Figure 11, Figure 12 utilize this implementation
The network decomposition result that example method obtains is respectively as shown in Figure 13, Figure 14.
Illustrated above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications also should
It is considered as protection scope of the present invention.
Claims (5)
- A kind of 1. dual intensity CT picture breakdown methods based on convolutional neural networks, which is characterized in that include the following steps：Step 1：Convolutional neural networks model is designed as the mapping function D (μ in dual intensity decomposition modelH,L；Θ)；Step 2：It is trained by convolutional neural networks model and training data set pair convolutional neural networks, to convolutional Neural net Network parameter Θ is effectively estimated；Step 3：Using the convolutional neural networks parameter Θ that the convolutional neural networks after training and step 2 obtain to dual intensity CT images Carry out sill efficient-decomposition.
- 2. a kind of dual intensity CT picture breakdown methods based on convolutional neural networks according to claim 1, which is characterized in that The convolutional neural networks modelling is dual input, the network structure model of dual output is to realize high energy CT images, low energy CT The direct output directly inputted with different materials image of image.
- 3. a kind of dual intensity CT picture breakdown methods based on convolutional neural networks according to claim 2, which is characterized in that The dual input, dual output network structure model establish and intersect convolution to realize in high energy CT images, low energy CT images not With the reasonable diffluence of sill information.
- 4. a kind of dual intensity CT picture breakdown methods based on convolutional neural networks according to claim 1, which is characterized in that Short chain is established in the convolutional neural networks model to connect.
- 5. a kind of dual intensity CT picture breakdown methods based on convolutional neural networks according to claim 1, which is characterized in that The training dataset includes the input data and output data of convolutional neural networks model, and the output data is sill figure Picture, the input data are the dual intensity CT images obtained according to sill image and corresponding energy information, the dual intensity CT images Including high energy CT images and low energy CT images；Using output data as label data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810136595.1A CN108230277B (en) | 2018-02-09 | 2018-02-09 | Dual-energy CT image decomposition method based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810136595.1A CN108230277B (en) | 2018-02-09 | 2018-02-09 | Dual-energy CT image decomposition method based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108230277A true CN108230277A (en) | 2018-06-29 |
| CN108230277B CN108230277B (en) | 2020-09-11 |
Family
ID=62661486
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810136595.1A Active CN108230277B (en) | 2018-02-09 | 2018-02-09 | Dual-energy CT image decomposition method based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108230277B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109523024A (en) * | 2018-11-22 | 2019-03-26 | 天津大学 | Energy spectrum correction method towards medical X-ray detector |
| CN109727238A (en) * | 2018-12-27 | 2019-05-07 | 贵阳朗玛信息技术股份有限公司 | The recognition methods of x-ray chest radiograph and device |
| CN109916933A (en) * | 2019-01-04 | 2019-06-21 | 中国人民解放军战略支援部队信息工程大学 | X ray computer tomographic imaging spectra estimation method based on convolutional neural networks |
| CN110189389A (en) * | 2019-05-31 | 2019-08-30 | 首都师范大学 | Dual intensity spectrum CT projection domain sill decomposition method and device based on deep learning |
| CN110428395A (en) * | 2019-06-20 | 2019-11-08 | 浙江大学 | More material decomposition methods of monoenergetic spectrum CT image |
| CN110675467A (en) * | 2019-09-30 | 2020-01-10 | 中国科学院深圳先进技术研究院 | Material decomposition method and device for energy spectrum CT, storage medium and electronic equipment |
| CN110706299A (en) * | 2019-09-16 | 2020-01-17 | 天津大学 | Substance decomposition imaging method for dual-energy CT |
| CN111179228A (en) * | 2019-12-16 | 2020-05-19 | 浙江大学 | Single-energy CT energy spectrum imaging method based on deep learning |
| CN111192224A (en) * | 2020-01-13 | 2020-05-22 | 北京联合大学 | Image enhancement method and device, electronic equipment and computer readable storage medium |
| CN112930543A (en) * | 2018-10-10 | 2021-06-08 | 利普麦德股份有限公司 | Neural network processing device, neural network processing method, and neural network processing program |
| CN113222137A (en) * | 2020-02-06 | 2021-08-06 | 苹果公司 | Neural rendering |
| US11967015B2 (en) | 2020-02-06 | 2024-04-23 | Apple Inc. | Neural rendering |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101647706A (en) * | 2008-08-13 | 2010-02-17 | 清华大学 | Image reconstruction method for high-energy double-energy CT system |
-
2018
- 2018-02-09 CN CN201810136595.1A patent/CN108230277B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101647706A (en) * | 2008-08-13 | 2010-02-17 | 清华大学 | Image reconstruction method for high-energy double-energy CT system |
Non-Patent Citations (3)
| Title |
|---|
| WOO-JIN LEE ET AL.: "Material depth reconstruction method of multi-energy X-ray images using neural network", 《2012 ANNUAL INTERNATIONAL CONFERENCE OF THE IEEE ENGINEERING IN MEDICINE AND BIOLOGY SOCIETY》 * |
| 李彦冬等: "卷积神经网络研究综述", 《计算机应用》 * |
| 汤鹏杰等: "并行交叉的深度卷积神经网络模型", 《中国图象图形学报》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112930543A (en) * | 2018-10-10 | 2021-06-08 | 利普麦德股份有限公司 | Neural network processing device, neural network processing method, and neural network processing program |
| CN109523024A (en) * | 2018-11-22 | 2019-03-26 | 天津大学 | Energy spectrum correction method towards medical X-ray detector |
| CN109727238A (en) * | 2018-12-27 | 2019-05-07 | 贵阳朗玛信息技术股份有限公司 | The recognition methods of x-ray chest radiograph and device |
| CN109916933A (en) * | 2019-01-04 | 2019-06-21 | 中国人民解放军战略支援部队信息工程大学 | X ray computer tomographic imaging spectra estimation method based on convolutional neural networks |
| CN109916933B (en) * | 2019-01-04 | 2021-10-01 | 中国人民解放军战略支援部队信息工程大学 | X-ray computed tomography energy spectrum estimation method based on convolutional neural network |
| CN110189389A (en) * | 2019-05-31 | 2019-08-30 | 首都师范大学 | Dual intensity spectrum CT projection domain sill decomposition method and device based on deep learning |
| CN110189389B (en) * | 2019-05-31 | 2022-08-05 | 首都师范大学 | Deep learning-based decomposition method and device for base material of dual-energy-spectrum CT projection domain |
| CN110428395A (en) * | 2019-06-20 | 2019-11-08 | 浙江大学 | More material decomposition methods of monoenergetic spectrum CT image |
| CN110706299A (en) * | 2019-09-16 | 2020-01-17 | 天津大学 | Substance decomposition imaging method for dual-energy CT |
| CN110706299B (en) * | 2019-09-16 | 2023-04-07 | 天津大学 | Substance decomposition imaging method for dual-energy CT |
| CN110675467A (en) * | 2019-09-30 | 2020-01-10 | 中国科学院深圳先进技术研究院 | Material decomposition method and device for energy spectrum CT, storage medium and electronic equipment |
| CN111179228A (en) * | 2019-12-16 | 2020-05-19 | 浙江大学 | Single-energy CT energy spectrum imaging method based on deep learning |
| CN111192224A (en) * | 2020-01-13 | 2020-05-22 | 北京联合大学 | Image enhancement method and device, electronic equipment and computer readable storage medium |
| CN111192224B (en) * | 2020-01-13 | 2024-03-19 | 北京联合大学 | Image enhancement method, image enhancement device, electronic equipment and computer readable storage medium |
| CN113222137A (en) * | 2020-02-06 | 2021-08-06 | 苹果公司 | Neural rendering |
| US11967015B2 (en) | 2020-02-06 | 2024-04-23 | Apple Inc. | Neural rendering |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108230277B (en) | 2020-09-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108230277A (en) | A kind of dual intensity CT picture breakdown methods based on convolutional neural networks | |
| Zhang et al. | Improving CBCT quality to CT level using deep learning with generative adversarial network | |
| Britten et al. | The addition of computer simulated noise to investigate radiation dose and image quality in images with spatial correlation of statistical noise: an example application to X-ray CT of the brain | |
| CN108564553A (en) | Low-dose CT image noise suppression method based on convolutional neural networks | |
| CN110675461A (en) | CT image recovery method based on unsupervised learning | |
| CN108175402A (en) | The intelligent identification Method of electrocardiogram (ECG) data based on residual error network | |
| CN108122265A (en) | A kind of CT reconstruction images optimization method and system | |
| CN107356615A (en) | A kind of method and system for dual-energy x-ray CT | |
| CN108784697A (en) | A kind of tissue distribution of conductivity method for reconstructing and system based on deep learning neural network | |
| Liu et al. | Deep learning with noise‐to‐noise training for denoising in SPECT myocardial perfusion imaging | |
| CN106530366A (en) | Energy spectrum CT image reconstruction method and energy spectrum CT imaging system | |
| CN106780641A (en) | A kind of Low-dose X-ray CT image reconstruction method | |
| CN110211139A (en) | Automatic segmentation Radiotherapy of Esophageal Cancer target area and the method and system for jeopardizing organ | |
| CN110189389A (en) | Dual intensity spectrum CT projection domain sill decomposition method and device based on deep learning | |
| CN111861910A (en) | CT image noise reduction system and method | |
| CN109615674A (en) | The double tracer PET method for reconstructing of dynamic based on losses by mixture function 3D CNN | |
| CN110009613A (en) | Low-dose CT imaging method, apparatus and system based on the dense network of depth | |
| CN109191462A (en) | A kind of CT anthropomorphic phantom generation method | |
| CN108053456A (en) | A kind of PET reconstruction images optimization method and system | |
| CN108133755A (en) | A kind of atlas pivot surgery simulation system and its analogy method based on three-dimensional visualization | |
| Zhang et al. | Hierarchical organ-aware total-body standard-dose pet reconstruction from low-dose PET and CT images | |
| CN109903225A (en) | A kind of medical image Enhancement Method based on deep learning | |
| Luo et al. | Ultra-dense denoising network: application to cardiac catheter-based x-ray procedures | |
| Xie et al. | Contextual loss based artifact removal method on CBCT image | |
| CN111080736A (en) | Low-dose CT image reconstruction method based on sparse transformation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |