CN108932723A - A kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape - Google Patents
A kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape Download PDFInfo
- Publication number
- CN108932723A CN108932723A CN201810275423.2A CN201810275423A CN108932723A CN 108932723 A CN108932723 A CN 108932723A CN 201810275423 A CN201810275423 A CN 201810275423A CN 108932723 A CN108932723 A CN 108932723A
- Authority
- CN
- China
- Prior art keywords
- aorta
- dissection
- dimensional
- nake
- image
- 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
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- 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
-
- 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
- G06T2207/10004—Still image; Photographic image
- G06T2207/10012—Stereo images
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape, this method has fully considered the three-D space structure information of dissection of aorta, by minimizing energy functional on the basis of medical volume data, directly aorta clamp tunic is extracted in a manner of 3 D stereo, greatly enhance efficiency, and to a very small extent by the control of the offset error of target position, the three-dimensional space serial correlation between CT image sequence is considered simultaneously, so that the three-dimensional sandwich mould of building is more nearly true value；Its realization process is：(1) dissection of aorta image background information is converted into white as target image using Threshold segmentation；(2) building is suitable for the three-dimensional S nake parted pattern gross energy of dissection of aorta Image Segmentation Methods Based on Features extraction；(3) triangulation is carried out to construct initial interlayer curved surface profile model needed for three-dimensional S nake model evolution using interlayer sampling point set；(4) initial surface profile is driven to realize that final dissection of aorta three-dimensional segmentation is extracted to final real goal interlayer profile iteration convergence by three-dimensional S nake parted pattern gross energy；The present invention precisely divides field in dissection of aorta and has important application value.
Description
Technical field
The invention belongs to technical field of image processing；It is related to a kind of three-dimensional S nake active based on aorta curved-surface shape
Arteries and veins interlayer division and extracting method；It can be used for carrying out dissection of aorta whole segmentation to extract.
Background technique
Dissection of aorta is presently the most dangerous a kind of cardiovascular disease, and main cause is active caused by hypertension
The rupture of astillen inner membrance, blood flow vertical peeling and intramural hematoma for being formed between inner membrance and middle layer, once rupture haemorrhage, rescue ten
Divide difficulty, dangerous degree is significantly larger than the diseases such as cerebral infarction, myocardial infarction and malignant tumour；Electronic computer tomography is swept at present
Retouching (Computed Tomography, CT) is Main Diagnosis mode in aortic dissection, passes through the dissection of aorta of acquisition
Patient's splanchnocoel CT image sequence obtains the position of interlayer and break size in aorta, provides auxiliary for operative treatment；In early days
The segmentation extraction of dissection of aorta part is successively extracted by professional image doctor Freehandhand-drawing mode in CT image；With CT scan
The diversification of imaging mode and the raising of scanning accuracy, it is very low that Freehandhand-drawing segmentation extracting mode becomes very cumbersome, time-consuming and efficiency,
Very strong subjectivity and uncertainty are had also with hand drawing interlayer data；And with computer image processing technology
Fast development, the dividing method based on image procossing are widely applied in medicine CT image, but this method is mainly for individual
Sandwiched area pixel extracts in CT image, interferes vulnerable to other noise pixel points, while not considering that interlayer three-dimensional space is continuous
Correlation, so that there are large errors for the three-dimensional sandwich mould finally constructed, therefore it is three-dimensional automatic to find a kind of dissection of aorta
Dividing extraction algorithm is difficult point.
Summary of the invention
This method outstanding advantages are the three-D space structure information for having fully considered dissection of aorta, by medicine body number
Energy functional is minimized on the basis of, is directly extracted aorta clamp tunic in a manner of 3 D stereo, largely
Efficiency is improved, and to a very small extent by the control of the offset error of target position, while considered between CT image sequence
Three-dimensional space serial correlation so that building three-dimensional sandwich mould be more nearly true value；The technical side that the present invention uses
Case is a kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape, is included the following steps：
(1) by one group of aorta regions CT image using Threshold segmentation by image background information be converted to after white as
Target image；
(2) building is suitable for the three-dimensional S nake parted pattern gross energy of dissection of aorta Image Segmentation Methods Based on Features extraction, including inside
Coupling energy between energy, external energy and CT image sequence；
(3) Delaunay Triangulation is carried out using the sampled point set of aorta clamp layer region to be split, construction is three-dimensional
Dissection of aorta initial surface skeleton pattern needed for Snake model evolution；
(4) it is obtained by the driving of three-dimensional S nake parted pattern gross energy obtained in step (2) by step (3) initial
Interlayer curved surface profile is to final real goal interlayer profile iteration convergence, when converging at energy minimum, develops and stops, realizing
Final dissection of aorta three-dimensional segmentation extracts result；
In step (1), the characteristics of according to aorta regions CT image sequence, one group of aorta regions CT image is utilized into threshold
Image background information is all converted to white and as target image by value partitioning algorithm, is constructed subsequent dissection of aorta and is extracted point
Cut required pictures；
In step (2), building is suitable for extracting the three-dimensional S nake parted pattern gross energy E of dissection of aorta featuretotal,
Including internal energy Eintern, external energy EexternAnd the coupling ENERGY E between CT image sequencecouple, gross energy expression
For：
In step (3), Delaunay Triangulation is carried out using dissection of aorta area sampling point set, to construct
Initial dissection of aorta curved surface profile model needed for three-dimensional S nake model evolution；One group of aorta CT image sequence is constructed first
Interlayer contour curve collection on every image is combined into one group of closed curve { B by the profile of target interlayer in columnk, k=1,
2 ..., n }, n is aorta CT amount of images, wherein Bk=(x, y) | s (x, y)=zk, it is sampled obtain curve discrete and be adopted
Sampling point set P, building model obtain target dissection of aorta curved surface S (x, y)；Then to the CT of dissection of aorta sufferer to be split
Interlayer pixel in image sequence carries out profile point sampling, is carried out by Delaunay Triangulation to dissection of aorta profile
Curved surface modeling, surface model are multiple triangular elements by subdivision, which will be used as three-dimensional S nake parted pattern iteration
The initial interlayer curved surface profile to develop；
In step (4), by the three-dimensional S nake parted pattern gross energy E of the dissection of aorta feature of buildingtotalAs driving
Power drives dissection of aorta initial surface profile to be split towards real goal interlayer profile iteration convergence, when converging to energy
When at amount minimum, develop and stop, realizing that final dissection of aorta three-dimensional segmentation extracts result；The present invention is compared with prior art
It has the following advantages that：
1. the present invention solves existing be based in two-dimensional CT image segmentation extraction algorithm vulnerable to the interference of other pixels, precision
It is not high, and whole three-dimensional spatial information is not considered, lack the space serial correlation problem between CT image sequence；
2. the present invention is using three-dimensional S nake surface model dividing method is based on, according to the morphological feature structure of dissection of aorta
It builds and is extracted with objective contour surface model similar in true interlayer curved surface, the whole three-dimensional segmentation for realizing dissection of aorta, it should
Inventive method precision is higher and ensure that the spatial continuity for the dissection of aorta curved surface that segmentation is extracted.
Detailed description of the invention
Fig. 1 is algorithm flow chart of the invention；
Fig. 2 is that aorta CT image background is converted to white result schematic diagram；
Fig. 3 is the configuration sampling point schematic diagram of target interlayer；
Fig. 4 is part curved surface schematic diagram after sufferer dissection of aorta triangulation；
Fig. 5 is dissection of aorta image external force field pattern；(a), (b), (c), (d) are different shape dissection of aorta
Force distribution figure；
Fig. 6 is that dissection of aorta three-dimensional segmentation extracts stereoscopic display result in result and aorta lumen；It (a) is case 1
Interlayer film；It (b) is display in the interlayer aorta lumen of case 1；It (c) is the interlayer film of case 2；(d) actively for the interlayer of case 2
The intracavitary display of arteries and veins；
Specific embodiment
Algorithm flow chart of the invention will be as shown in Figure 1, first will using Threshold segmentation by one group of aorta regions CT image
Image background information is used as target image after being converted to white；Then building is suitable for the extraction of dissection of aorta Image Segmentation Methods Based on Features
Three-dimensional S nake parted pattern gross energy；Then Delaunay is carried out using the sampled point set of aorta clamp layer region to be split
Triangulation, dissection of aorta initial surface skeleton pattern needed for constructing three-dimensional S nake model evolution；Finally by three-dimensional
Snake parted pattern gross energy drives initial interlayer curved surface profile to final real goal interlayer profile iteration convergence, realizes most
Whole dissection of aorta three-dimensional segmentation is extracted；With reference to the accompanying drawing, the specific implementation process of technical solution of the present invention is carried out detailed
Description.
1. dissection of aorta image background information is converted to white and as target image using Threshold segmentation
One group of aorta CT image sequence is utilized Threshold Segmentation Algorithm first by the characteristics of according to dissection of aorta CT image
It converts background parts to after white as target image, to obtain subsequent to carry out required for dissection of aorta segmentation extraction
CT image collection, as shown in Figure 2；Since, there are noise, the present invention carries out pretreatment operation using median filtering in CT image
To eliminate noise jamming, high quality graphic collective data is obtained, to be conducive to accurately construct active interlayer curved surface profile model.
2. the three-dimensional S nake parted pattern gross energy that building is suitable for the extraction of dissection of aorta Image Segmentation Methods Based on Features
Three-dimensional S nake segmentation is under the collective effect of the coupling energy between external energy, internal energy and slice
The curved surface profile of progress develops, from energy outside power drive curved surface profile to be split develop to objective contour edge, energy internal force with
Keep the smooth and continuous of curved surface profile；Medicine CT image sequence has one between that is, adjacent scanned picture there are spatial coherence
Fixed coupling energy, therefore profile cannot be driven to carry out three-dimensional S nake iteration only in conjunction with energy internal force and energy external force, it is no
Then curved surface profile is interfered vulnerable to other marginal informations, causes to divide error；Limited model row is generated according to power inside Snake model
For contractility and rigidity, guarantee that curved surface is smooth continuous；Therefore the surface model internal energy in the present invention is defined as：
Eintern=α (s) | vs(s)|2+β(s)|vss(s)|2
vs(s) it is Elastic Term, is v first derivative, it acts as curve elasticity is guaranteed, reduces the effect of curve inner area；?
It is presented as the coordinate difference of adjacent pixel in image, it is smaller that difference gets over mini Mod internal energy, and it is better to represent curve smoothing；vss
(s) it is rigid item, is v second dervative, it acts as line smoothing is ensured, hinders curved, a certain volume data can be converted into
The growth in the normal vector direction between its adjacent volume data is measured；α (s), β (s) be respectively curve coefficient of elasticity and just
Property coefficient, α (s) guarantee curved spring, β (s) guarantee curved surface resistance bending.
Surface model external energy is defined as according to characteristics of image：
I (x) indicates the strength information in target image；Make energy curve convergence to entire figure by setting external energy
As upper intensity most dark areas, and intensity most dark areas is the position where dissection of aorta on target image, to reach final
Converge to the segmentation object of dissection of aorta；
Coupling energy definition between CT scan picture is：
xi(s)、xj(s) it is respectively aorta contour curve on adjacent two width CT image, coupling adjacent profile joined CT
The space correlation continuity of image sequence is 0 for the profile coupling energy not in adjacent C T image；Non-negative parameter ξ is determined
Bonding force acts on the intensity on profile；Then three-dimensional segmentation model gross energy is defined as in the present invention：
Image force driving model is generated using these energy terms to decline by iterative gradient to the direction for minimizing energy
Develop, converges to real goal profile.
3. it is bent to construct initial interlayer needed for three-dimensional S nake model evolution to carry out triangulation using interlayer sampling point set
Facial contour model
Profile point sampling is carried out according to the dissection of aorta pixel in the CT image of extraction to be split, with Delaunay tri-
Angle partitioning techniques carry out curved surface modeling to dissection of aorta profile, obtain the curved surface profile for being split into numerous triangular elements,
It is first when will be close to the part constructing curve model of true dissection of aorta profile as the evolution of three-dimensional S nake parted pattern iteration
Beginning profile；Method is sampled using ε and realizes uniform sampling, point p each of on dissection of aorta contour surface to interlayer in the present invention
There is corresponding nearest sampled point q, and distance is less than ε f (p) between p, q, ε indicates p to the nearest of picture centre less than 1, f (p)
Distance, as shown in Figure 3.Using three-dimensional S nake model to the interlayer film three-dimensional segmentation in human body splanchnocoel CT scan image sequence
When, dissection of aorta initial surface skeleton pattern need to be defined, and triangulation, sampled point are carried out to sampled point each in the model
Triangular element is formed after triangulation, finally by Snake iterative algorithm to establishing dissection of aorta curved surface profile mould
Type is smoothed, and smoothed out partial model is as shown in Figure 4.
4. realizing final dissection of aorta three to final real goal interlayer profile iteration convergence by initial surface profile
Dimension segmentation is extracted
Drive initial surface profile towards target interlayer profile iteration convergence by energy, when converging to minimum energy term
When, it obtains final dissection of aorta three-dimensional segmentation and extracts result；It minimizes energy term and uses Conjugate gradient descent method, pass through
Gradient descent algorithm energy minimizes power；Later newly the local linear expression formula of energy is each iteration：
According to the energy term of the suitable dissection of aorta of construction, external force field is as shown in Figure 5；It can see external force field driving wheel
Exterior feature, until converging to interlayer reaches energy minimum, obtains finally extracting segmentation result towards movement at interlayer.
Fig. 6 is that dissection of aorta three-dimensional segmentation extracts stereoscopic display result in result and aorta lumen；RED sector is base
In three-dimensional S nake model segmentation extract dissection of aorta three-dimensional configuration, and by after aorta cavity transparency process by its
It is shown in aorta lumen, dissection of aorta initial position and final position show clear and definite；And it is based on three-dimensional S nake mould
The dissection of aorta that type extracts is smoothly continuous, intermediate free of discontinuities, can show that dissection of aorta originates cut position well
Set with interlayer expansion range, there is greater significance to success rate of operation and postoperative rehabilitation is improved.
Claims (1)
1. a kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape, includes the following steps：
(1) image background information is converted into white later as target using Threshold segmentation by one group of aorta regions CT image
Image；
(2) building is suitable for the three-dimensional S nake parted pattern gross energy of dissection of aorta Image Segmentation Methods Based on Features extraction, including internal energy
Coupling energy between amount, external energy and CT image sequence；
(3) Delaunay Triangulation is carried out using the sampled point set of aorta clamp layer region to be split, constructs three-dimensional S nake
Dissection of aorta initial surface skeleton pattern needed for model evolution；
(4) the initial interlayer obtained by step (3) is driven by three-dimensional S nake parted pattern gross energy obtained in step (2)
Curved surface profile is to final real goal interlayer profile iteration convergence, when converging at energy minimum, develops and stops, and realizes final
Dissection of aorta three-dimensional segmentation extracts result；
In step (1), the characteristics of according to aorta regions CT image sequence, one group of aorta regions CT image is utilized into threshold value point
It cuts algorithm and image background information is all converted into white and as target image, construct subsequent dissection of aorta and extract segmentation institute
Need pictures；
In step (2), building is suitable for extracting the three-dimensional S nake parted pattern gross energy E of dissection of aorta featuretotal, including
Internal energy Eintern, external energy EexternAnd the coupling ENERGY E between CT image sequencecouple, gross energy is expressed as：In step (3), Delaunay tri- is carried out using dissection of aorta area sampling point set
Angle subdivision, to construct initial dissection of aorta curved surface profile model needed for three-dimensional S nake model evolution；Building one first
It is bent to be combined into one group of closure by the profile of target interlayer in group aorta CT image sequence for interlayer contour curve collection on every image
Line { Bk, k=1,2 ..., n }, n is aorta CT amount of images, wherein Bk=(x, y) | s (x, y)=zk, it is sampled
To curve discrete sampled point set P, constructs model and obtain target dissection of aorta curved surface S (x, y)；Then to aorta to be split
Interlayer pixel in the CT image sequence of interlayer sufferer carries out profile point sampling, by Delaunay Triangulation to aorta
Interlayer profile carries out curved surface modeling, and surface model is multiple triangular elements by subdivision, which will be used as three-dimensional S nake
The initial interlayer curved surface profile that parted pattern iteration develops；
In step (4), by the three-dimensional S nake parted pattern gross energy E of the dissection of aorta feature of buildingtotalAs driving force,
Drive dissection of aorta initial surface profile to be split towards real goal interlayer profile iteration convergence, when converging to energy most
When small place, develop and stop, realizing that final dissection of aorta three-dimensional segmentation extracts result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810275423.2A CN108932723A (en) | 2018-03-26 | 2018-03-26 | A kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810275423.2A CN108932723A (en) | 2018-03-26 | 2018-03-26 | A kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108932723A true CN108932723A (en) | 2018-12-04 |
Family
ID=64448992
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810275423.2A Pending CN108932723A (en) | 2018-03-26 | 2018-03-26 | A kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108932723A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111932552A (en) * | 2020-07-21 | 2020-11-13 | 深圳睿心智能医疗科技有限公司 | Aorta modeling method and device |
| CN112330708A (en) * | 2020-11-24 | 2021-02-05 | 沈阳东软智能医疗科技研究院有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
| CN112561871A (en) * | 2020-12-08 | 2021-03-26 | 中国医学科学院北京协和医院 | Aortic dissection method and device based on flat scanning CT image |
| CN112951056A (en) * | 2021-02-19 | 2021-06-11 | 武汉市中心医院 | Virtual model three-dimensional mapping device and method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130094750A1 (en) * | 2011-10-12 | 2013-04-18 | Tolga Tasdizen | Methods and systems for segmentation of cells for an automated differential counting system |
| CN105741355A (en) * | 2016-02-01 | 2016-07-06 | 华侨大学 | Block segmentation method for triangular grid model |
| CN107330281A (en) * | 2017-07-05 | 2017-11-07 | 大连理工大学 | The full-automatic personalized reconstructing method of fracture steel plate model |
-
2018
- 2018-03-26 CN CN201810275423.2A patent/CN108932723A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130094750A1 (en) * | 2011-10-12 | 2013-04-18 | Tolga Tasdizen | Methods and systems for segmentation of cells for an automated differential counting system |
| CN105741355A (en) * | 2016-02-01 | 2016-07-06 | 华侨大学 | Block segmentation method for triangular grid model |
| CN107330281A (en) * | 2017-07-05 | 2017-11-07 | 大连理工大学 | The full-automatic personalized reconstructing method of fracture steel plate model |
Non-Patent Citations (2)
| Title |
|---|
| DUAN XIAOJIE ET AL.: ""Segmentation of the Aortic Dissection from CT Images Based on Spatial Continuity Prior Model"", 《2016 8TH INTERNATIONAL CONFERENCE ON INFORMATION TECHNOLOGY IN MEDICINE AND EDUCATION》 * |
| 刘星: ""基于三维Snake模型的肝脏CT图像交互式分割系统的设计与实现"", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111932552A (en) * | 2020-07-21 | 2020-11-13 | 深圳睿心智能医疗科技有限公司 | Aorta modeling method and device |
| CN111932552B (en) * | 2020-07-21 | 2023-12-01 | 深圳睿心智能医疗科技有限公司 | Aorta modeling method and device |
| CN112330708A (en) * | 2020-11-24 | 2021-02-05 | 沈阳东软智能医疗科技研究院有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
| CN112330708B (en) * | 2020-11-24 | 2024-04-23 | 沈阳东软智能医疗科技研究院有限公司 | Image processing method, device, storage medium and electronic equipment |
| CN112561871A (en) * | 2020-12-08 | 2021-03-26 | 中国医学科学院北京协和医院 | Aortic dissection method and device based on flat scanning CT image |
| CN112951056A (en) * | 2021-02-19 | 2021-06-11 | 武汉市中心医院 | Virtual model three-dimensional mapping device and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11534136B2 (en) | Three-dimensional segmentation from two-dimensional intracardiac echocardiography imaging | |
| CN108932723A (en) | A kind of three-dimensional S nake dissection of aorta dividing method based on curved-surface shape | |
| JP5108905B2 (en) | Method and apparatus for automatically identifying image views in a 3D dataset | |
| US20170337686A1 (en) | Kind of x-ray chest image rib suppression method based on poisson model | |
| Leung et al. | Automated border detection in three-dimensional echocardiography: principles and promises | |
| CN109598722B (en) | Image analysis method based on recurrent neural network | |
| CN103077550B (en) | A kind of four dimensional rebuilding method of non-gate ICUS image sequence medium vessels | |
| Zagrodsky et al. | Registration-assisted segmentation of real-time 3-D echocardiographic data using deformable models | |
| CN108053433A (en) | A kind of multi-modal arteria carotis MRI method for registering based on physical alignment and outline | |
| JP6243535B2 (en) | Model-based segmentation of anatomical structures | |
| Banerjee et al. | A completely automated pipeline for 3D reconstruction of human heart from 2D cine magnetic resonance slices | |
| Girum et al. | Learning with context feedback loop for robust medical image segmentation | |
| CN104200465B (en) | The dividing method and device of cardiac three-dimensional image | |
| Hernandez et al. | Deep learning in spatiotemporal cardiac imaging: A review of methodologies and clinical usability | |
| WO2022213654A1 (en) | Ultrasonic image segmentation method and apparatus, terminal device, and storage medium | |
| Chartrand et al. | Semi-automated liver CT segmentation using Laplacian meshes | |
| CN105139377A (en) | Rapid robustness auto-partitioning method for abdomen computed tomography (CT) sequence image of liver | |
| Zeng et al. | Liver segmentation in magnetic resonance imaging via mean shape fitting with fully convolutional neural networks | |
| Heyde et al. | Anatomical image registration using volume conservation to assess cardiac deformation from 3D ultrasound recordings | |
| CN115830016B (en) | Medical image registration model training method and equipment | |
| WO2023040628A1 (en) | Method for automatically turning and segmenting left ventricle in pet three-dimensional image | |
| CN114972362A (en) | Medical image automatic segmentation method and system based on RMAU-Net network | |
| Banerjee et al. | Automated 3D whole-heart mesh reconstruction from 2D cine MR slices using statistical shape model | |
| CN110570416B (en) | Method for visualization and 3D printing of multi-modal cardiac images | |
| CN110033454B (en) | Segmentation method of lung tumor with large-area adhesion lung boundary tissue in CT image |
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: 20181204 |
|
| WD01 | Invention patent application deemed withdrawn after publication |