CN110070612A - A kind of CT image layer interpolation method based on generation confrontation network - Google Patents
A kind of CT image layer interpolation method based on generation confrontation network Download PDFInfo
- Publication number
- CN110070612A CN110070612A CN201910340691.2A CN201910340691A CN110070612A CN 110070612 A CN110070612 A CN 110070612A CN 201910340691 A CN201910340691 A CN 201910340691A CN 110070612 A CN110070612 A CN 110070612A
- Authority
- CN
- China
- Prior art keywords
- image
- layer
- confrontation network
- network
- generator
- 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
Classifications
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Graphics (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Geometry (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Processing (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The present invention relates to a kind of based on the CT image layer interpolation method for generating confrontation network；Include: S1, for CT image to be processed, obtains thick-layer CT image, and carry out linear normalization processing for the thick-layer CT image；S2, for thick-layer CT image after normalized adjacent two layers be combined be input in advance training generation confrontation network generator；S3, the generation of the training in advance is fought into the output of the generator of network as CT interlayer interpolation image；The method of the present invention can obtain CT interlayer interpolation image using confrontation network is generated automatically, and the model structure is simple, fast convergence rate and precision are high, calculation amount is small, and image is accurately convenient for the foundation of subsequent 3-D image.
Description
Technical field
The invention belongs to technical field of medical image processing more particularly to a kind of CT image layers based on generation confrontation network
Interpolation method.
Background technique
With the continuous development of medical imaging technology, medical image develops into sequence two now from the X-ray of early stage
Dimension word faultage image.Image interpolation has special status in medical image processing, from being generated to of medical image after
Processing is typically necessary carry out interpolation processing.Lehmann etc. thinks since having computer graphics and Digital Image Processing, just
Have an image interpolation, so-called image interpolation be exactly an image data reproduction at process.MRI, CT imaging etc. tomoscans at
Seem that coplanar sampling is carried out by adjustable distance and thickness, obtains cross-section image object under examination.It is ground in digital medical image
In studying carefully, it is often desirable that be sliced according to two-dimensional sequence and restore its 3D shape, rebuild virtual organ or tissue, with adjuvant clinical point
Analysis and diagnosis.Clinically in the operation of some complexity, doctor needs to know before surgery the three-dimensional spatial information of tissue, because
This, the three-dimensional reconstruction of medical image comes into being.And the three-dimensional reconstruction of image is just needed the sequence section heap clinically obtained
It stacks.
In terms of time and hardware costs, the domestic spiral CT for clinic is 64 rows, interlayer mostly at present
Away from usually 3mm-7mm, the sequence section interlamellar spacing of acquisition is much larger than the distance between adjacent pixel in lamella.Therefore, it is building
When vertical three-dimensional visualization model, faultage image interlamellar spacing is too big, and interlayer resolution ratio is lower, needs to generate newly by interpolation method
Image level.The target of faulted images interpolation is by interpolation method between adjacent two width faultage image to generate new interpolation
Image reduces the spacing between image which adds the quantity of the sequence faultage image, is convenient for subsequent image three-dimensional
It rebuilds.Faulted images interpolation increases image data, and interpolation itself also consumes the regular hour, therefore, in faultage image
The applicable new algorithm of comparison must be developed in interpolation technique, can really improve the precision and efficiency of three-dimensional reconstruction.Tomography
The interpolation method of image can be divided into interpolation method, shape based interpolation method and the interpolation side based on small echo based on gray scale
Method.Though these methods improve interpolation precision to a certain extent, algorithm is complicated, and it is therefore necessary to further increase algorithm essence
The complexity calculated is reduced while spending.
Summary of the invention
(1) technical problems to be solved
In order to solve the technical problem of the interpolation method convergence rate slow-motion of conventional CT image interlayer and effect difference, the present invention is mentioned
For a kind of based on the CT image layer interpolation method for generating confrontation network.
(2) technical solution
In order to achieve the above object, the main technical schemes that the present invention uses include:
S1, for CT image to be processed, obtain thick-layer CT image, and carry out linear normalizing for the thick-layer CT image
Change processing；
S2, it is combined for the adjacent two layers of thick-layer CT image after normalized and is input in advance trained generation pair
The generator of anti-network；
S3, the generation of the training in advance is fought into the output of the generator of network as CT interlayer interpolation image.
Optionally, further includes:
S4, three-dimensional reconstruction acquisition 3-D image is carried out using the thick-layer CT image and the CT interlayer interpolation image.
Optionally, further include before step S2
A1, building include that the generation of generator and arbiter fights network；
A2, linear normalization operation is carried out to multiple thin-layer CT sample images, by thin-layer CT sample graph each after normalization
K-1 layers of picture and K+1 tomographic image are combined acquisition training sample；
A3, all training samples are input to the generation for generating confrontation network model and generating confrontation network
The confrontation network model that generates is generated the image for the generator output for fighting network as K layers of new interpolation image by device；
A4, using the K tomographic image of new the K layers of interpolation image and corresponding thin-layer CT sample image as the life
The input of the arbiter of confrontation network is generated at confrontation network model；
The knot that the loss function constructed in advance is optimized using the method for backpropagation, and is optimized according to loss function
Fruit updates the network weight of generator and arbiter, carries out structural similarity assessment, weight to K layers new of the interpolation image
Multiple above-mentioned steps A1 to A5, and save the power that generation confrontation network when structural similarity assesses optimal in iterative process is used as
Weight obtains the generation trained in advance and fights network.
Optionally, in step A1, the generator for generating confrontation network includes the first convolutional coding structure, generates confrontation network
Arbiter includes the second convolutional coding structure；
First convolutional coding structure is used for the input picture of generator into down-sampling, and uses closest interpolation algorithm pair
The characteristic pattern of down-sampling is up-sampled to restore the size of image；
The image that second convolutional coding structure is used to input arbiter carries out feature extraction, and using big with characteristic pattern etc.
Small convolution sum makes arbiter output be scalar.
Optionally, first convolutional coding structure include: the first convolutional layer, using Relu as the first active coating of activation primitive and
First regularization layer Instance Normalization；
Second convolutional coding structure include: the second convolutional layer, using Leaky Relu as the second active coating of activation primitive and
Second regularization layer Instance Normalization.
Optionally, the 3-D image includes: three-dimensional reconstruction sagittal plane result images and/or three-dimensional reconstruction coronal-plane knot
Fruit image.
Optionally, the thick-layer CT image with a thickness of 1mm.
Optionally, the thin-layer CT image with a thickness of 5mm.
(3) beneficial effect
The beneficial effects of the present invention are: on the one hand, the method for the present invention utilizes the CT interlayer interpolation image that can be automatically generated,
The structure for generating confrontation net is simple, using convenient, and fast convergence rate and calculation amount are small；On the other hand, the method for the present invention generates
CT interlayer interpolation image precise effect it is preferable, provide the foundation for subsequent 3-dimensional reconstruction.
Detailed description of the invention
Fig. 1 is a kind of CT image layer interpolation method flow based on generation confrontation network that one embodiment of the invention provides
Figure；
Fig. 2 implements the process signal that generation confrontation network is obtained using training set and test set provided for the present invention one
Figure；
Fig. 3 is the generator network structure for the generation confrontation network that one embodiment of the invention provides；
Fig. 4 is the arbiter network structure for the generation confrontation network that one embodiment of the invention provides；
Fig. 5 a is the K layers of new interpolation image generated using sample set training that one embodiment of the invention provides；
Fig. 5 b provides the CT interlayer interpolation image that CT image to be processed generates for one embodiment of the invention；
Fig. 6 a is the three-dimensional reconstruction sagittal plane result figure that one embodiment of the invention provides；
Fig. 6 b is the three-dimensional reconstruction coronal-plane result figure that one embodiment of the invention provides.
Specific embodiment
In order to preferably explain the present invention, in order to understand, with reference to the accompanying drawing, by specific embodiment, to this hair
It is bright to be described in detail.
Embodiment one
A kind of CT image layer interpolation method based on generation confrontation network is present embodiments provided, following step is specifically included
It is rapid:
As shown in Fig. 2, the present invention obtains CT interlayer interpolation image using confrontation network is generated, generating confrontation network has knot
The convenient feature of structure simple application, and generate confrontation network and needed during use using a large amount of data at confrontation net
Network is trained, such as establishes sample set and test set verifies network model, wherein obtaining generation pair trained in advance
Anti- network the following steps are included:
A1, building include that the generation of generator and arbiter fights network；For example, as shown in Figure 3 in step A1,
The generator for generating confrontation network includes the first convolutional coding structure, as shown in figure 4, the arbiter for generating confrontation network includes volume Two
Product structure；First convolutional coding structure is used for the input picture of generator into down-sampling, and is adopted under using closest interpolation algorithm
The characteristic pattern of sample is up-sampled to restore the size of image；
The image that second convolutional coding structure is used to input arbiter carries out feature extraction, and uses and the sizes such as characteristic pattern
Convolution sum makes arbiter output be scalar.
Preferably, the first convolutional coding structure includes: the first convolutional layer, using Relu as the first active coating of activation primitive and first
Regularization layer Instance Normalization；
Second convolutional coding structure includes: the second convolutional layer, using Leaky Relu as the second active coating of activation primitive and second
Regularization layer Instance Normalization.
For example, in the present embodiment generator, arbiter network depth, the size of convolution nuclear volume, convolution sum can
It is modified according to the size of actual input picture, the present embodiment is simultaneously not limited thereof, and whole network framework does not make
With full articulamentum；
A2, linear normalization operation is carried out to multiple thin-layer CT sample images, by thin-layer CT sample graph each after normalization
K-1 layers of picture and K+1 tomographic image are combined acquisition training sample；Wherein, K is the positive integer more than or equal to 1；Citing comes
It says, CT image includes thin-layer CT image and thick-layer CT image, and the present invention is trained using thin-layer CT image and is desirable to network science
The information of thin layer is practised, and last whole network will be applied on thick-layer CT image, so that the effect of three-dimensional reconstruction more preferably exists.
The present embodiment medium bed CT image is being answered with a thickness of 5mm for trained thin-layer CT image with a thickness of 1mm
With the image that other thickness also may be selected in the process, the present embodiment is not defined specific thickness and is only used for illustrating.
A3, training sample is input to the generator for generating confrontation network, the generator output of confrontation network will be generated
Image is as K layers of new interpolation image；Specific as shown in Figure 5 a, training sample set is input to generation confrontation by the present embodiment
The generator of network obtains K layers of new interpolation image；
A4, using the K tomographic image of new the K layers of interpolation image and corresponding thin-layer CT sample image as the life
The input of the arbiter of confrontation network is generated at confrontation network model；
The knot that the loss function constructed in advance is optimized using the method for backpropagation, and is optimized according to loss function
Fruit updates the network weight of generator and arbiter, and ties in the training process to K layers new of the interpolation image
Structure similarity assessment, repeat the above steps A1 to A5, and saves the generation pair when structural similarity assesses optimal in iterative process
The weight that anti-network is used as obtains the generation trained in advance and fights network.
Specifically for example, during generating confrontation network first with sample image training, building loss in advance
Function, the method optimization loss function of backpropagation simultaneously carry out more the weight of the generator and arbiter that generate confrontation network
Newly；Secondly, this method also carries out structural similarity assessment to K layers of new interpolation image, K layers new in iterative process are inserted
The structural similarity of value image when assessing optimal the weight of corresponding generator and arbiter as generation confrontation trained in advance
The generator of network and the weight of arbiter.
As shown in Figure 1, obtaining CT interlayer for any CT image to be processed after obtaining generation confrontation network trained in advance
Interpolation image the following steps are included:
S1, for CT image to be processed, obtain thick-layer CT image, and carry out at linear normalization for thick-layer CT image
Reason；
S2, it is combined for the adjacent two layers of thick-layer CT image after normalized and is input in advance trained generation pair
The generator of anti-network；
S3, the generation of training in advance is fought into the output of the generator of network as CT interlayer interpolation image；Specifically, this
The CT interlayer interpolation image of the generator output of the generation confrontation network of embodiment training in advance is as shown in Figure 5 b；
Preferably, further include in actual application step S4 after obtaining CT interlayer interpolation image, utilize thick-layer
CT image and CT interlayer interpolation image carry out three-dimensional reconstruction and obtain 3-D image；
For example, using CT image to be processed thick-layer CT image and step S3 obtain CT interlayer interpolation image into
Row three-dimensional reconstruction, obtained three-dimensional reconstruction sagittal plane result is as shown in Figure 6 a, obtained three-dimensional reconstruction coronal-plane result such as Fig. 6 b
It is shown.
Finally, it should be noted that above-described embodiments are merely to illustrate the technical scheme, rather than to it
Limitation；Although the present invention is described in detail referring to the foregoing embodiments, those skilled in the art should understand that:
It can still modify to technical solution documented by previous embodiment, or to part of or all technical features into
Row equivalent replacement；And these modifications or substitutions, it does not separate the essence of the corresponding technical solution various embodiments of the present invention technical side
The range of case.
Claims (8)
1. a kind of based on the CT image layer interpolation method for generating confrontation network, generating confrontation network includes generator and differentiation
Device, which comprises the following steps:
S1, for CT image to be processed, obtain thick-layer CT image, and carry out at linear normalization for the thick-layer CT image
Reason；
S2, it is combined for the adjacent two layers of thick-layer CT image after normalized and is input in advance trained generation and fights net
The generator of network；
S3, the generation of the training in advance is fought into the output of the generator of network as CT interlayer interpolation image.
2. the method as described in claim 1, which is characterized in that further include:
S4, three-dimensional reconstruction acquisition 3-D image is carried out using the thick-layer CT image and the CT interlayer interpolation image.
3. the method as described in claim 1, which is characterized in that further include before step S2
A1, building include that the generation of generator and arbiter fights network；
A2, linear normalization operation is carried out to multiple thin-layer CT sample images, by thin-layer CT sample image each after normalization
K-1 layers are combined acquisition training sample with K+1 tomographic image；
A3, all training samples are input to the generator for generating confrontation network model and generating confrontation network, it will
The confrontation network model that generates generates the image for the generator output for fighting network as K layers of new interpolation image；
A4, using the K tomographic image of new the K layer interpolation image and corresponding thin-layer CT sample image as the generation pair
Anti- network model generates the input of the arbiter of confrontation network；
The loss function constructed in advance is optimized using the method for backpropagation, and more according to the result of loss function optimization
The network weight that new life grows up to be a useful person with arbiter, to new K layer of the interpolation image progress structural similarity assessment, in repetition
Step A1 to A5 is stated, and saves the weight that generation confrontation network when structural similarity assesses optimal in iterative process is used as, is obtained
Take the generation confrontation network trained in advance.
4. method as claimed in claim 3, which is characterized in that in step A1, the generator for generating confrontation network includes the
One convolutional coding structure, the arbiter for generating confrontation network includes the second convolutional coding structure；
First convolutional coding structure is used for the input picture of generator into down-sampling, and is adopted under using closest interpolation algorithm
The characteristic pattern of sample is up-sampled to restore the size of image；
The image that second convolutional coding structure is used to input arbiter carries out feature extraction, and uses and the sizes such as characteristic pattern
Convolution sum makes arbiter output be scalar.
5. method as claimed in claim 4, which is characterized in that
First convolutional coding structure includes: the first convolutional layer, using Relu as the first active coating of activation primitive and the first regularization
Layer Instance Normalization；
Second convolutional coding structure includes: the second convolutional layer, using Leaky Relu as the second active coating of activation primitive and second
Regularization layer Instance Normalization.
6. the method as shown in claim 2, which is characterized in that the 3-D image includes: three-dimensional reconstruction sagittal plane result figure
Picture and/or three-dimensional reconstruction coronal-plane result images.
7. the method as described in claim 1, which is characterized in that the thick-layer CT image with a thickness of 1mm.
8. method as claimed in claim 3, which is characterized in that the thin-layer CT image with a thickness of 5mm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910340691.2A CN110070612B (en) | 2019-04-25 | 2019-04-25 | CT image interlayer interpolation method based on generation countermeasure network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910340691.2A CN110070612B (en) | 2019-04-25 | 2019-04-25 | CT image interlayer interpolation method based on generation countermeasure network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110070612A true CN110070612A (en) | 2019-07-30 |
| CN110070612B CN110070612B (en) | 2023-09-22 |
Family
ID=67369003
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910340691.2A Active CN110070612B (en) | 2019-04-25 | 2019-04-25 | CT image interlayer interpolation method based on generation countermeasure network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110070612B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110796613A (en) * | 2019-10-10 | 2020-02-14 | 东软医疗系统股份有限公司 | Automatic image artifact identification method and device |
| CN112435341A (en) * | 2020-11-23 | 2021-03-02 | 推想医疗科技股份有限公司 | Training method and device for three-dimensional reconstruction network, and three-dimensional reconstruction method and device |
| CN112509091A (en) * | 2020-12-10 | 2021-03-16 | 上海联影医疗科技股份有限公司 | Medical image reconstruction method, apparatus, device and medium |
| CN112633306A (en) * | 2019-09-24 | 2021-04-09 | 杭州海康威视数字技术股份有限公司 | Method and device for generating confrontation image |
| CN113706358A (en) * | 2021-07-09 | 2021-11-26 | 清华大学 | Method and device for encrypting tomographic image interlamellar spacing |
| US20220114259A1 (en) * | 2020-10-13 | 2022-04-14 | International Business Machines Corporation | Adversarial interpolation backdoor detection |
| CN113706358B (en) * | 2021-07-09 | 2024-07-12 | 清华大学 | Encryption method and encryption device for layer spacing of tomographic image |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107968962A (en) * | 2017-12-12 | 2018-04-27 | 华中科技大学 | A kind of video generation method of the non-conterminous image of two frames based on deep learning |
| GB201809604D0 (en) * | 2018-06-12 | 2018-07-25 | Tom Tom Global Content B V | Generative adversarial networks for image segmentation |
| CN108629816A (en) * | 2018-05-09 | 2018-10-09 | 复旦大学 | The method for carrying out thin layer MR image reconstruction based on deep learning |
| GB201902459D0 (en) * | 2019-02-22 | 2019-04-10 | Facesoft Ltd | Facial shape representation and generation system and method |
-
2019
- 2019-04-25 CN CN201910340691.2A patent/CN110070612B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107968962A (en) * | 2017-12-12 | 2018-04-27 | 华中科技大学 | A kind of video generation method of the non-conterminous image of two frames based on deep learning |
| CN108629816A (en) * | 2018-05-09 | 2018-10-09 | 复旦大学 | The method for carrying out thin layer MR image reconstruction based on deep learning |
| GB201809604D0 (en) * | 2018-06-12 | 2018-07-25 | Tom Tom Global Content B V | Generative adversarial networks for image segmentation |
| GB201902459D0 (en) * | 2019-02-22 | 2019-04-10 | Facesoft Ltd | Facial shape representation and generation system and method |
Non-Patent Citations (8)
| Title |
|---|
| MARK KOREN ET AL: "Frame Interpolation Using Generative Adversarial Networks", 《HTTP://CS231N.STANFORD.EDU/REPORTS/2017/PDFS/317.PDF》 * |
| MARK KOREN ET AL: "Frame Interpolation Using Generative Adversarial Networks", 《HTTP://CS231N.STANFORD.EDU/REPORTS/2017/PDFS/317.PDF》, 2 July 2017 (2017-07-02), pages 1 - 8 * |
| VIREN JAIN: "Adversarial Image Alignment and Interpolation", 《HTTPS://ARXIV.ORG/ABS/1707.00067》 * |
| VIREN JAIN: "Adversarial Image Alignment and Interpolation", 《HTTPS://ARXIV.ORG/ABS/1707.00067》, 30 June 2017 (2017-06-30), pages 1 - 10 * |
| ZEJU LI ET AL: "Reconstruction of Thin-Slice Medical Image Using Generative Adversarial Network", 《MACHINE LEARNING IN MEDICAL IMAGING》 * |
| ZEJU LI ET AL: "Reconstruction of Thin-Slice Medical Image Using Generative Adversarial Network", 《MACHINE LEARNING IN MEDICAL IMAGING》, 7 September 2017 (2017-09-07), pages 325 - 333, XP047437117, DOI: 10.1007/978-3-319-67389-9_38 * |
| 孟博: "基于低剂量CT的层间插值技术研究", 中国优秀硕士学位论文全文数据库 信息科技辑, no. 11, pages 138 - 865 * |
| 林毓秀;刘慧;刘绍玲;张彩明;: "三维CT层间图像超分辨重建与修复", 计算机辅助设计与图形学学报, no. 06, pages 66 - 76 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112633306A (en) * | 2019-09-24 | 2021-04-09 | 杭州海康威视数字技术股份有限公司 | Method and device for generating confrontation image |
| CN112633306B (en) * | 2019-09-24 | 2023-09-22 | 杭州海康威视数字技术股份有限公司 | Method and device for generating countermeasure image |
| CN110796613A (en) * | 2019-10-10 | 2020-02-14 | 东软医疗系统股份有限公司 | Automatic image artifact identification method and device |
| CN110796613B (en) * | 2019-10-10 | 2023-09-26 | 东软医疗系统股份有限公司 | Automatic identification method and device for image artifacts |
| US20220114259A1 (en) * | 2020-10-13 | 2022-04-14 | International Business Machines Corporation | Adversarial interpolation backdoor detection |
| CN112435341A (en) * | 2020-11-23 | 2021-03-02 | 推想医疗科技股份有限公司 | Training method and device for three-dimensional reconstruction network, and three-dimensional reconstruction method and device |
| CN112509091A (en) * | 2020-12-10 | 2021-03-16 | 上海联影医疗科技股份有限公司 | Medical image reconstruction method, apparatus, device and medium |
| CN112509091B (en) * | 2020-12-10 | 2023-11-14 | 上海联影医疗科技股份有限公司 | Medical image reconstruction method, device, equipment and medium |
| CN113706358A (en) * | 2021-07-09 | 2021-11-26 | 清华大学 | Method and device for encrypting tomographic image interlamellar spacing |
| CN113706358B (en) * | 2021-07-09 | 2024-07-12 | 清华大学 | Encryption method and encryption device for layer spacing of tomographic image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110070612B (en) | 2023-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110070612A (en) | A kind of CT image layer interpolation method based on generation confrontation network | |
| US11398029B2 (en) | Systems and methods for medical acquisition processing and machine learning for anatomical assessment | |
| JP5484444B2 (en) | Medical image diagnostic apparatus and volume calculation method | |
| CN103514615B (en) | Method and apparatus for iterative approximation | |
| JP2021521993A (en) | Image enhancement using a hostile generation network | |
| US8103070B2 (en) | Volume rendering apparatus and method | |
| US8873819B2 (en) | Method for sorting CT image slices and method for constructing 3D CT image | |
| CN111951344B (en) | Magnetic resonance image reconstruction method based on cascade parallel convolution network | |
| JPH02178786A (en) | Method and apparatus for determining three-dimensional image from tomograph data using variable threshold | |
| WO2020206755A1 (en) | Ray theory-based method and system for ultrasound ct image reconstruction | |
| CN113436290A (en) | Method and system for selectively removing streak artifacts and noise from images using a deep neural network | |
| CN109598697A (en) | The determination of two-dimentional mammary gland radiography data group | |
| US11605162B2 (en) | Systems and methods for determining a fluid and tissue volume estimations using electrical property tomography | |
| WO2018155046A1 (en) | Medical imaging device, image processing method, and program | |
| EP3987445A1 (en) | Method and system for generating a synthetic elastrography image | |
| JP2021087629A (en) | Medical data processing device | |
| US11748921B2 (en) | Learnable extrapolation for deep learning-based medical image reconstruction | |
| GB2576945A (en) | Image processing methods | |
| Gunduzalp et al. | 3d u-netr: Low dose computed tomography reconstruction via deep learning and 3 dimensional convolutions | |
| CN112669450B (en) | Human body model construction method and personalized human body model construction method | |
| CN115908610A (en) | Method for obtaining attenuation correction coefficient image based on single-mode PET image | |
| CN113192155B (en) | Spiral CT cone beam scanning image reconstruction method, scanning system and storage medium | |
| JP2023067357A (en) | Inference device, medical image diagnostic apparatus, inference method, and trained neural network generation method | |
| CN112819789A (en) | Apparatus and method for cardiac assessment | |
| Jeong et al. | Investigating the use of traveltime and reflection tomography for deep learning-based sound-speed estimation in ultrasound computed tomography |
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 |