CN109191462A - A kind of CT anthropomorphic phantom generation method - Google Patents
A kind of CT anthropomorphic phantom generation method Download PDFInfo
- Publication number
- CN109191462A CN109191462A CN201811241802.6A CN201811241802A CN109191462A CN 109191462 A CN109191462 A CN 109191462A CN 201811241802 A CN201811241802 A CN 201811241802A CN 109191462 A CN109191462 A CN 109191462A
- Authority
- CN
- China
- Prior art keywords
- image
- clinical
- training
- ray
- neural 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
- 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
- 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
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/003—Reconstruction from projections, e.g. tomography
- G06T11/005—Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating
-
- 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]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a kind of CT anthropomorphic phantom generation methods, comprising the following steps: (1) establishes Neural Network Data collection;(2) training convolutional neural networks;(3) divide Clinical CT image;(4) physical parameter is demarcated;(5) CT scan process is simulated.This method is split existing medical image using convolutional neural networks, the importing and calibration of physical message are artificially carried out to segmentation result, absorption process of the X-ray in tissue is simulated, so that obtaining data for projection carries out subsequent detection and image reconstruction process.The method easily and fast simulation real human body structure, realization computed tomography emulation experiment can push the further development of accurate medical treatment.
Description
Technical field
The present invention relates to the fields computed tomography (CT), and in particular to a kind of CT anthropomorphic phantom generation method utilizes
The method of medical image segmentation generates body mould database simulation computer tomographic scanning procedure, can promote the scientific research of the field CT
Fast development.
Background technique
Computed tomography (CT) is one of common disease detection means of current medical field, high-precision CT imaging
Technology plays a crucial role the development of China's medical field.But clinical diagnosis information be related to individual privacy and it is long when
Indirect raying can cause irreversible injury to human body, so that disclosed scan data scarcity of resources, is the scientific research in the field CT
Personnel bring heavy burden, significantly limit the development in the field.Wherein the acquisition and use of body mould database are emulation doctors
With one of the key problem in CT scan and detection process.
In fact, CT image is being rebuild by the data for projection of X-ray scanning human body, in the practical life for carrying out equipment
Before production, when scientific research, needs to carry out a large amount of emulation experiment to the design architecture for system of making rational planning for, manikin number
The non-medical staff matter of utmost importance to be faced is configured to according to library.The people of part arbitrary elliptical composition has been formd at present
Body Model, but it is very remote with real human body institutional framework gap, it is unable to satisfy the demand of high-precision medical imaging, however is drawn big
Amount body mould similar with tissue is excessively complicated, takes a long time, and more demanding to basic medical, it is difficult to realize.In recent years
With the development in the field CT, the medical image largely increased income is easily obtained, but body mould database is still extremely rare.With machine
The advantage of the progress of device learning art, convolutional neural networks is increasingly apparent, using neural network to existing medical image into
Row segmentation, and subsequent physical parameter calibration is artificially carried out, it can easily and efficiently construct a large amount of high quality human body model data
Library has strategic meaning to the progress of accurate medical field, plays great promotion to the development of current medical field
Effect.
Summary of the invention
For the rare problem in human body model data library during medical CT scan emulation experiment, the invention proposes one kind
CT anthropomorphic phantom generation method.Existing medical image is split using convolutional neural networks, artificially to segmentation result into
Absorption process of the X-ray in tissue is simulated in the importing and calibration of row physical message, thus after obtaining data for projection progress
Continuous detection and image reconstruction process.The method can easily and fast simulation real human body structure, realization computerized tomography be swept
Emulation experiment is retouched, the further development of accurate medical treatment has been pushed.
The purpose of the present invention is what is be achieved through the following technical solutions:
A kind of CT anthropomorphic phantom generation method, comprising the following steps:
(1) it establishes Neural Network Data collection: collecting Clinical CT image and segmentation result database as data set, from number
The n group Clinical CT image for accounting for about sum 70% and segmentation result are randomly selected as training set, by residue about 30% according to concentrating
Data as test set;
(2) training convolutional neural networks: establishing neural network model, using the n group training set divided in step (1) to mind
It is trained through network, makes loss function converge to minimum by back-propagation algorithm, terminate the training of network, and utilize survey
Examination collection test network training effect;
(3) divide Clinical CT image: further collecting 100 or more Clinical CT images, certain pretreatment is done to it,
Image segmentation is carried out using trained neural network, extracts the profile information of different tissues structural integrity;
(4) it demarcates physical parameter: using X-ray in the intracorporal attenuation law of people, being arranged in image according to the actual situation not
With the absorption coefficient of institutional framework, software is analyzed by MATLAB, CT image segmentation result is demarcated, construct real human body
Interior physical effect model, that is, body mould;
(5) it simulates CT scan process: the body mould demarcated is used for the simulated experiment of CT scan, to obtain across specified
The X-ray projection data of body mould carries out subsequent X-ray detection X and image reconstruction process.
Compared with prior art, the beneficial effects brought by the technical solution of the present invention are as follows:
1. the present invention is split existing clinical medicine image by convolutional neural networks, different human body organizer is obtained
The profile of official, and think to carry out calibration setting to physical process therein, to simulate real human body histoorgan, emulation is clinical
CT scan process.
2. this method is suitable for any kind of CT detector and image reconstruction algorithm, can generate in a short time a large amount of
Body mould database, has the characteristics that convenient, fast, high-precision, solves the problems, such as that the field CT body mould database is rare, effectively mentions
The efficiency of the high field scientific research.
3. this method body mould database obtained can be used for the CT prototype emulation experiment of different model different scanning mode,
The exploitation duration that non-medical staff carries out CT technology can greatly be shortened, established for the acquisition of high-precision medical image good
Basis.
Detailed description of the invention
Fig. 1 is the simulation process block schematic illustration of CT scan.
Fig. 2 is the flow diagram of the method for the present invention.
Specific embodiment
The invention will be further described with reference to the accompanying drawing.
The anthropomorphic phantom generation method towards Medical CT that the invention proposes a kind of, first by neural network to existing
It discloses clinical medicine image to be split, obtains the profile of different human body institutional framework, the softwares such as subsequent artificial utilization Matlab
The matching and calibration that physical message is carried out to segmentation result, by absorption coefficient of the X-ray in body mould and different institutional frameworks
It is corresponding, to simulate the process of entire CT scan, the projection of a large amount of different human body institutional framework can be obtained in a short time
Data are had laid a good foundation for further system optimization with image reconstruction process.Medical CT scan simulation process frame
Scheme as shown in Figure 1, the preparation method flow chart of body mould is as shown in Fig. 2, specific embodiment is as follows:
Step1: Neural Network Data collection is established.Collect the CT image and segmentation result of clinically each histoorgan of human body
Database carries out the training and test of neural network as data set.The n group for accounting for about sum 70% is randomly selected from data set
Clinical image and segmentation result are as training set, it is noted that as far as possible including the different tissues organic image of human body when selection,
The rest part of data set is as test set.
Step2: training convolutional neural networks.Neural network model is established, the n group training set pair divided in Step1 is utilized
Neural network is trained, and makes loss function converge to minimum by back-propagation algorithm, terminates the training of network, and is utilized
Test set test network training effect.
Step3: segmentation Clinical CT image.The CT image for collecting a large amount of disclosed human body different tissues organs, does one to it
Fixed pretreatment, pretreatment generally comprise image denoising, enhancing etc., carry out image segmentation using trained neural network, mention
Take out the profile information of different tissues structural integrity.
Step4: calibration physical parameter.Using X-ray in the intracorporal attenuation law of people, it is arranged in image according to the actual situation
The absorption coefficient of different tissues structure demarcates CT image segmentation result by softwares such as Matlab, constructs real human body
Interior physical effect model.
Step5: simulation CT scan process.The body mould demarcated is used for the simulated experiment of CT scan, thus obtain across
The X-ray projection data of specified body mould, carries out subsequent X-ray detection X and image reconstruction process.
Above-mentioned steps generate the body mould database that can simulate real human body structure, can be directly used for different scanning side
The emulation experiment of X-ray detector and image reconstruction algorithm under formula, the body structure of the convenient and efficient a large amount of clinical patients of simulation,
It haves laid a good foundation for subsequent CT system structure optimization.
When concrete application, used data set is answered widely distributed when training and testing convolutional neural networks model, as far as possible
Information comprising different patient's different tissues structures, such as lung, brain, abdomen.Terminate network instruction when loss function convergence
Practice, and test network training effect.The result of CT image segmentation requires clear-cut, boundary line point between each human tissue structure
Bright, edge is complete.Root is wanted when demarcating followed by softwares such as Matlab to absorption coefficient of the different tissues ingredient to X-ray
According to the parameter that specific requirements selection needs to demarcate, such as photoelectric effect, Compton scattering, to meet the practical feelings of human body as far as possible
Condition.The emulation experiment for the body mould database progress CT scan finally established using this method and subsequent X-ray detection and figure
As reconstruction process.
The present invention is not limited to embodiments described above.Above the description of specific embodiment is intended to describe and say
Bright technical solution of the present invention, the above mentioned embodiment is only schematical, is not restrictive.This is not being departed from
In the case of invention objective and scope of the claimed protection, those skilled in the art may be used also under the inspiration of the present invention
The specific transformation of many forms is made, within these are all belonged to the scope of protection of the present invention.
Claims (2)
1. a kind of CT anthropomorphic phantom generation method, which comprises the following steps:
(1) it establishes Neural Network Data collection: collecting Clinical CT image and segmentation result database as data set, from data set
In randomly select the n group Clinical CT image for accounting for sum 70% and segmentation result as training set, by remaining about 30% data
As test set;
(2) training convolutional neural networks: establishing neural network model, using the n group training set divided in step (1) to nerve net
Network is trained, and makes loss function converge to minimum by back-propagation algorithm, terminates the training of network, and utilize test set
Test network training effect;
(3) divide Clinical CT image: further collecting 100 or more Clinical CT images, certain pretreatment is done to it, utilize
Trained neural network carries out image segmentation, extracts the profile information of different tissues structural integrity;
(4) it demarcates physical parameter: using X-ray in the intracorporal attenuation law of people, different groups in image is set according to the actual situation
The absorption coefficient for knitting structure demarcates CT image segmentation result by analyzing software, constructs the physics effect in real human body
Answer model i.e. body mould;
(5) it simulates CT scan process: the body mould demarcated being used for the simulated experiment of CT scan, to obtain across specified body mould
X-ray projection data, carry out subsequent X-ray detection X and image reconstruction process.
2. a kind of CT anthropomorphic phantom generation method according to claim 1, which is characterized in that analysis software is in step (4)
MATLAB。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811241802.6A CN109191462A (en) | 2018-10-18 | 2018-10-18 | A kind of CT anthropomorphic phantom generation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811241802.6A CN109191462A (en) | 2018-10-18 | 2018-10-18 | A kind of CT anthropomorphic phantom generation method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109191462A true CN109191462A (en) | 2019-01-11 |
Family
ID=64942979
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811241802.6A Pending CN109191462A (en) | 2018-10-18 | 2018-10-18 | A kind of CT anthropomorphic phantom generation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109191462A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110702706A (en) * | 2019-09-20 | 2020-01-17 | 天津大学 | Method for simulating output data of energy spectrum CT system |
CN112561825A (en) * | 2020-12-22 | 2021-03-26 | 清华大学 | Image processing method and device based on X-ray imaging |
CN112951384A (en) * | 2021-02-04 | 2021-06-11 | 慧影医疗科技(北京)有限公司 | Data simulation generation method and system for medical imaging equipment |
CN113379860A (en) * | 2021-04-25 | 2021-09-10 | 上海索骥信息科技有限公司 | Medical image data generation method and system with marker based on digital object |
CN114707416A (en) * | 2022-04-18 | 2022-07-05 | 成都理工大学 | Method, device and system for detecting irradiation dose in human body and computer equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101458826A (en) * | 2008-11-25 | 2009-06-17 | 中国科学院等离子体物理研究所 | Digital human body modeling method for assigning density, constituent by CT value |
CN101706844A (en) * | 2009-11-18 | 2010-05-12 | 中国人民解放军第三军医大学第二附属医院 | Mandible firearm wound simulation method |
CN107169974A (en) * | 2017-05-26 | 2017-09-15 | 中国科学技术大学 | It is a kind of based on the image partition method for supervising full convolutional neural networks more |
WO2018156803A1 (en) * | 2017-02-24 | 2018-08-30 | Bayer Healthcare Llc | Systems and methods for generating simulated computed tomography (ct) images |
CN108615237A (en) * | 2018-05-08 | 2018-10-02 | 上海商汤智能科技有限公司 | A kind of method for processing lung images and image processing equipment |
-
2018
- 2018-10-18 CN CN201811241802.6A patent/CN109191462A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101458826A (en) * | 2008-11-25 | 2009-06-17 | 中国科学院等离子体物理研究所 | Digital human body modeling method for assigning density, constituent by CT value |
CN101706844A (en) * | 2009-11-18 | 2010-05-12 | 中国人民解放军第三军医大学第二附属医院 | Mandible firearm wound simulation method |
WO2018156803A1 (en) * | 2017-02-24 | 2018-08-30 | Bayer Healthcare Llc | Systems and methods for generating simulated computed tomography (ct) images |
CN107169974A (en) * | 2017-05-26 | 2017-09-15 | 中国科学技术大学 | It is a kind of based on the image partition method for supervising full convolutional neural networks more |
CN108615237A (en) * | 2018-05-08 | 2018-10-02 | 上海商汤智能科技有限公司 | A kind of method for processing lung images and image processing equipment |
Non-Patent Citations (1)
Title |
---|
邓金城等: "深度卷积神经网络在放射治疗计划图像分割中的应用", 《中国医学物理学杂志》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110702706A (en) * | 2019-09-20 | 2020-01-17 | 天津大学 | Method for simulating output data of energy spectrum CT system |
CN112561825A (en) * | 2020-12-22 | 2021-03-26 | 清华大学 | Image processing method and device based on X-ray imaging |
CN112951384A (en) * | 2021-02-04 | 2021-06-11 | 慧影医疗科技(北京)有限公司 | Data simulation generation method and system for medical imaging equipment |
CN113379860A (en) * | 2021-04-25 | 2021-09-10 | 上海索骥信息科技有限公司 | Medical image data generation method and system with marker based on digital object |
CN114707416A (en) * | 2022-04-18 | 2022-07-05 | 成都理工大学 | Method, device and system for detecting irradiation dose in human body and computer equipment |
CN114707416B (en) * | 2022-04-18 | 2023-04-07 | 成都理工大学 | Method, device and system for detecting irradiation dose in human body and computer equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109191462A (en) | A kind of CT anthropomorphic phantom generation method | |
CN107330949A (en) | A kind of artifact correction method and system | |
CN1663530B (en) | Methods and apparatus for processing image data to aid in detecting disease | |
CN109381212A (en) | A kind of image formation control method and system | |
CN104700438B (en) | Image rebuilding method and device | |
CN109272472B (en) | Noise and artifact eliminating method for medical energy spectrum CT image | |
CN109961834A (en) | The generation method and equipment of diagnostic imaging report | |
CN107133549A (en) | ECT motion gates signal acquiring method and ECT image rebuilding methods | |
CN108257132A (en) | A kind of method of the CT image quality measures based on machine learning | |
CN108230277A (en) | A kind of dual intensity CT picture breakdown methods based on convolutional neural networks | |
CN109255354A (en) | medical CT-oriented computer image processing method and device | |
CN105147312A (en) | PET image acquiring method and system | |
CN109215040B (en) | Breast tumor segmentation method based on multi-scale weighted learning | |
CN105678821B (en) | A kind of dynamic PET images method for reconstructing based on self-encoding encoder image co-registration | |
US20230121358A1 (en) | Ct image generation method for attenuation correction of pet images | |
CN102496156A (en) | Medical image segmentation method based on quantum-behaved particle swarm cooperative optimization | |
CN106999138A (en) | Diagnosing image method and apparatus and its recording medium | |
Nordstrom | The quantitative imaging network in precision medicine | |
CN110459300A (en) | A kind of lung cancer pathology type diagnostic method based on computer vision and CT images | |
CN109961419B (en) | Correction information acquisition method for attenuation correction of PET activity distribution image | |
CN107126257A (en) | A kind of method that cardiac module is simulated and assessed to structural heart disease operation | |
CN110009007A (en) | A kind of artificial intelligence surgical assistant system towards polymorphic type disease | |
Banjšak et al. | Implementation of artificial intelligence in chronological age estimation from orthopantomographic X-ray images of archaeological skull remains | |
CN109903225A (en) | A kind of medical image Enhancement Method based on deep learning | |
Gargiulo et al. | Quantitative color three-dimensional computer tomography imaging of human long-term denervated muscle |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190111 |
|
WD01 | Invention patent application deemed withdrawn after publication |