CN109741359B - Method for segmenting lesion liver of abdominal CT sequence image - Google Patents
Method for segmenting lesion liver of abdominal CT sequence image Download PDFInfo
- Publication number
- CN109741359B CN109741359B CN201910029480.7A CN201910029480A CN109741359B CN 109741359 B CN109741359 B CN 109741359B CN 201910029480 A CN201910029480 A CN 201910029480A CN 109741359 B CN109741359 B CN 109741359B
- Authority
- CN
- China
- Prior art keywords
- liver
- lesion
- shape
- segmentation
- diseased
- 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.)
- Active
Links
Images
Landscapes
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a method for segmenting a lesion liver of an abdominal CT sequence image, which comprises the following steps: constructing a level set energy function by utilizing a gray level offset field and the abdominal CT sequence slice spatial correlation, and performing primary segmentation on the diseased liver; constructing a liver dictionary based on the liver grid, and completing the shape correction of the liver with pathological changes by utilizing the sparse shape combination of dictionary atoms; and constructing a diseased liver shape prior, fusing the diseased liver shape prior into a graph cutting energy function, optimizing a diseased liver cutting result, and finishing final cutting of the diseased liver. The method can effectively segment the lesion liver in the abdominal CT sequence image and can avoid under-segmentation of the lesion region of the liver.
Description
Technical Field
The invention relates to the field of medical image processing, in particular to a method for segmenting a lesion liver of an abdominal CT sequence image.
Background
China is a big liver cancer country, about more than half of newly-increased liver cancer patients and cases of death due to liver cancer occur in China worldwide, and early discovery and early treatment of liver diseases are the main tasks currently facing. At present, the imaging examination of liver diseases is mainly carried out by the modes of ultrasound, CT, MRI imaging and the like, and the treatment of early liver cancer is carried out by surgical treatment, radio frequency ablation, stereotactic radiotherapy and the like. CT has certain advantages for imaging examination of liver diseases, local treatment of liver cancer and curative effect evaluation, and stereotactic radiotherapy of liver cancer patients needs to calculate radiation dose according to CT images and formulate a radiotherapy plan, so the abdominal CT image has an important role in the examination and treatment processes of liver diseases.
In order to provide a basis for the examination, lesion analysis, operation, radiotherapy and the like of liver diseases and guarantee the safety of liver cancer treatment, the diseased liver in an abdominal CT sequence image needs to be accurately segmented, however, the number of slices of a CT sequence is large, the workload of experts for manually segmenting the diseased liver is large, the time consumption is long, and the segmentation result lacks objectivity, so that the computer-assisted segmentation method for researching the diseased liver has great significance. Due to the fact that the shape of the diseased liver is changed, the gray distribution is complex, and the imaging is influenced by noise, the abdominal CT sequence diseased liver segmentation faces huge challenges. Existing segmentation methods mainly include traditional grayscale-based methods and deep learning-based methods. The traditional method is sensitive to image noise and gray distribution, and often causes under-segmentation of the lesion part of the liver; the deep learning-based method requires a large amount of data to perform network training, and is time-consuming.
Disclosure of Invention
The invention aims to provide a method for segmenting a lesion liver of an abdominal CT sequence image. The invention is realized by the following scheme:
a method for segmenting lesion liver of an abdominal CT sequence image comprises the following steps:
(1) the method specifically comprises the following steps of performing initial segmentation on the lesion liver of an input abdominal CT sequence image:
a. selecting an initial slice of a diseased liver, and constructing a level set energy function
Wherein x and y are pixel points in the CT image omega, I (x) is the gray value of x in omega,is a non-negative window function, ciWhere i is 1,2 represents the average gray scale of the object and the background in a circular area with y as the center and r as the radius, b (-) is the gray scale offset field, H (-) is the Heaviside function,represents the inner and outer regions of the zero level set phi (x), H (-) is a Heaviside function,in the term of the perimeter, the length of the week,as a distance regularization term, the parameters u and mu are respectively a term L (phi) for controlling the circumference and a distance regularization term Rp(phi) ofThe weight is used for minimizing the energy function by utilizing a gradient descent method, and the maximum connected domain of the evolution result is reserved as the initial slice liver primary segmentation result; wherein upsilon and mu are constants larger than 0;
b. constructing a non-initial slice level set energy function by taking an initial slice as a starting point and taking a segmentation result of an adjacent slice as position constraint
Wherein the content of the first and second substances,constraint of the liver segmentation position for the jth slice in the abdominal CT sequence, Ij-1And Ij+1Respectively segmenting non-initial slices in the sequence upwards and downwards in an iterative mode for the segmentation result of the liver of the adjacent slice, and reserving the area larger than T in the evolution resultpThe area of each pixel is used as the primary segmentation result of the lesion liver; wherein, TpIs an integer greater than 0.
(2) Finishing the correction of the shape of the lesion liver based on the liver sparse shape combination, which specifically comprises the following steps:
a. respectively carrying out three-dimensional reconstruction on liver CT serial slices of M patients manually segmented by experts, respectively constructing liver triangular grids with the number of vertexes being N, and obtaining liver dictionary atoms d by taking grid vertexes as dictionary elementsi∈R3NI 1,2, …, M, a liver shape dictionary D of size 3N × M is constructed with the first liver atom D1For reference, the remaining liver atom dqAnd d1Aligning to realize one-to-one correspondence of grid vertexes, wherein q is 2,3, …, M; wherein M, N are all integers greater than 0;
b. three-dimensional reconstruction is carried out on the primary segmentation result of the CT sequence of the diseased liver, a diseased liver triangular grid with the number of vertexes N is constructed,using the mesh vertex vector v, v ∈ R3NRepresenting the shape of the lesion liver to be corrected, aligning the lesion liver to a first liver atom, and realizing one-to-one correspondence of grid vertexes;
c. constructing sparse shape combination residual functionWhere T (v, β) represents global transformation of v with β as a parameter, and x ═ x1,x2,…,xM]∈RMFor sparse coefficients, e ∈ R3NRepresenting the under-segmented part of the shape of the lesion liver to be corrected, D is a dictionary of liver shapes, lambda1、λ2Is a constant for controlling the sparsity of x and e; minimizing the residual function by using a simple and fast iterative shrinkage threshold algorithm, solving x, e and beta, and obtaining a corrected lesion liver triangular grid v ' ═ T ' (Dx-e, beta), wherein T ' (. cndot.) represents the inverse transformation of T (.); lambda [ alpha ]1Is a constant greater than 0, λ2Is a constant greater than 0 and less than 1;
d. and (3) carrying out voxelization correction on the three-dimensional grid v' of the lesion liver after voxelization correction, and obtaining a corresponding liver slice shape correction result.
(3) Based on the prior of the shape of the diseased liver, a graph cutting energy function is constructed, the segmentation result of the diseased liver is optimized, and the final segmentation of the diseased liver is completed, and the method specifically comprises the following steps:
a. obtaining the shape prior of the lesion liver according to the initial segmentation result and the shape correction result of the lesion liver section
Wherein f ispRepresenting pixel points p, L in CT slice flevel-setAnd LsscRespectively representing the primary segmentation result and the shape correction result of the liver corresponding to the slice lesion;
b. constructing a graph cut energy function E (f) based on the shape prior of the lesion liver,
wherein, I (f)p) Is a gray term, Sprior(fp) A priori to shape, B (f)p,fq) Is a boundary penalty term, alpha is a weight value for controlling a gray level term and a boundary term, P is a set of all pixels in the image f, N is a weight value for controlling a gray level term and a boundary termpSet of neighborhood pixels, I, of pixel point ppIs the gray scale value of pixel p, d (p, q) is the Euclidean distance between pixels p and q, SPThe number of pixel points in the pixel set P is counted; wherein alpha is a constant greater than 0 and less than 1;
c. and minimizing the graph cut energy function by using a maximum flow-minimum cut algorithm to obtain a lesion liver optimization result, and finishing the final segmentation of the lesion liver.
In the step (1), a CT section with no liver fracture phenomenon and relatively large liver area in the whole sequence is selected as an initial section of the diseased liver, and the section is positioned from top to bottom one third to two fifths of the whole sequence.
In the step (1), preferably, v is 65.025, μ is 1, r is 10, and T ispIs 200.
In the step (2), preferably, M is 54 and N is 1096.
In step (2), λ is preferred1Is 50, λ2Is 0.17.
In the step (3), α is preferably 0.3.
The method of the invention has the following advantages:
an offset field is fused into the level set function, and position constraint is introduced by utilizing the spatial correlation between adjacent slices, so that the influence of noise and other irrelevant organs and tissues on the primary segmentation of the diseased liver can be effectively reduced;
a liver dictionary is constructed by using a standard normal liver, the correction of the primary segmentation result of the lesion liver is realized based on the sparse shape combination of dictionary atoms, and the problem of under-segmentation of the lesion part of the liver can be effectively solved;
a new diseased liver shape prior is constructed and is fused into the graph cut energy function, so that a diseased liver cutting result can be further optimized, an accurate diseased liver final cutting result is obtained, and the cutting precision can be further improved.
Drawings
FIG. 1 is a diagram of the result of primary segmentation of liver with abdominal CT sequence lesion according to an embodiment of the present invention;
FIG. 2 is a graph of lesion liver shape correction results based on sparse shape combination according to an embodiment of the present invention;
FIG. 3 is a prior map of the shape of a diseased liver according to an embodiment of the invention;
fig. 4 is a graph of the optimization result of the lesion liver based on graph segmentation, i.e. a final segmentation result graph of the lesion liver according to the embodiment of the present invention.
Detailed Description
The following describes specific embodiments of the present invention:
example 1
The method comprises the following concrete steps:
(1) and (3) primarily dividing the belly CT serial lesion liver. The method comprises the following concrete steps:
a. selecting an initial slice of a diseased liver, and constructing an energy function of a level set of the initial slice:
wherein x and y are pixel points in the CT image omega, I (x) is the gray value of x in omega,is a non-negative window function, ciWhere i is 1,2 represents the average gray scale of the object and the background in a circular area with y as the center and r as the radius, b (-) is the gray scale offset field, H (-) is the Heaviside function,represents the inner and outer regions of the zero level set phi (x), H (-) is a Heaviside function,in the term of the perimeter, the length of the circle,as a distance regularization term, the parameters upsilon and mu are respectively a term L (phi) for controlling the perimeter and a term R for distance regularizationpAnd (phi) minimizing the energy function based on a gradient descent method, and keeping the maximum connected domain of the evolution result as an initial segmentation result of the CT sequence initial section of the diseased liver. The present embodiment preferably has upsilon of 65.025, preferably has μ of 1, preferably has r of 10, and preferably has the section of one third of the CT sequence from top to bottom as the initial section of the lesion liver.
b. And (4) taking the initial slice of the diseased liver as a starting point, and respectively carrying out upward and downward iterative segmentation on the residual slices of the CT sequence of the diseased liver. During iterative segmentation, according to the spatial correlation between adjacent slices of a CT sequence, introducing a segmentation result of adjacent lesion livers as position constraint, and constructing a level set energy function:
wherein the content of the first and second substances,constraint of the liver segmentation position for the jth slice in the abdominal CT sequence, Ij-1And Ij+1Dividing results of adjacent sliced livers; preserving area greater than T in evolution resultspThe area of each pixel is used as the primary segmentation result of the lesion liver. Preferred T of the present embodimentpIs 200.
Fig. 1 is a graph of the primary segmentation result of the abdomen CT-series lesion liver obtained in this example.
(2) And correcting the shape of the lesion liver. The method comprises the following concrete steps:
a. respectively carrying out liver sequence slices of M patients manually segmented by expertsThree-dimensional reconstruction, respectively constructing liver triangular meshes with the number of vertexes N, and obtaining liver dictionary atoms d by taking the mesh vertexes as dictionary elementsi∈R3NI 1,2, …, M, a liver shape dictionary D of size 3N × M is constructed with the first liver atom D1For reference, the remaining liver atom dqAnd d1And aligning to realize one-to-one correspondence of grid vertexes, wherein q is 2,3, … and M. In this embodiment, M is preferably 54 and N is 1096.
b. Performing three-dimensional reconstruction on the primary segmentation result of the diseased liver completed in the step 1, constructing a diseased liver triangular grid with the number of vertexes N, and utilizing a grid vertex vector v, which belongs to the element R3NAnd representing the shape of the lesion liver to be corrected, aligning the shape with the first atom of the liver dictionary, and realizing the one-to-one correspondence of the grid vertexes. In this embodiment, N is preferably 1096.
c. Constructing sparse shape combination residual functionWhere T (v, β) represents global transformation of v with β as a parameter, and x ═ x1,x2,…,xM]∈RMFor sparse coefficients, e ∈ R3NRepresenting the under-segmented part of the shape of the lesion liver to be corrected, D is a dictionary of liver shapes, lambda1And λ2Is a constant for controlling the sparsity of x and e; and minimizing the residual function by using a simple and fast iterative shrinkage threshold algorithm, solving x, e and beta, and obtaining a corrected lesion liver triangular mesh v ═ T '(Dx-e, beta), wherein T' (. cndot.) represents the inverse transformation of T (. cndot.). Preferred λ for this embodiment1Is 50, λ2Is 0.17.
d. And voxelizing the corrected three-dimensional grid v' of the lesion liver to obtain a corresponding liver slice shape correction result.
Fig. 2 is a diagram of a liver shape correction result of an abdominal CT-series lesion obtained in this embodiment, and it can be seen that compared with fig. 1, the under-segmentation problem of a liver lesion is effectively solved, and the segmentation precision is improved.
(3) And constructing a graph segmentation energy function based on the prior of the shape of the diseased liver, optimizing the segmentation result of the diseased liver, and finishing the final segmentation of the diseased liver. The method comprises the following concrete steps:
a. obtaining shape prior by adopting the pathological liver section initial segmentation result obtained in the step 1 and the step 2 and the corresponding shape correction result
Wherein f ispRepresenting pixel points p, L in CT slice flevel-setAnd LsscRespectively representing the primary segmentation result and the shape correction result of the liver with the corresponding section lesion. Fig. 3 is a diagram of the prior result of the shape of the diseased liver obtained in the embodiment.
b. Constructing a graph cut energy function E (f) based on the shape prior of the lesion liver,
wherein, I (f)p) Is a gray term, Sprior(fp) A priori of the shape of the liver, B (f)p,fq) Is a boundary penalty term, alpha is a weight value for controlling a gray level term and a boundary term, P is a set of all pixels in the image f, N is a weight value for controlling a gray level term and a boundary termpSet of neighborhood pixels, I, of pixel point ppIs the gray scale value of pixel p, d (p, q) is the Euclidean distance between pixels p and q, SPThe number of pixels in the pixel set P. In this embodiment, α is preferably 0.3.
c. And minimizing the graph cut energy function by using a maximum flow-minimum cut algorithm to obtain a lesion liver optimization result and realize final segmentation of the lesion liver.
Fig. 4 is a diagram of an optimized result of the diseased liver obtained in this embodiment, that is, a final segmentation result of the diseased liver, and it can be seen that compared with fig. 2, the segmentation accuracy of the diseased liver is further improved.
Example 2
The method of example 1 was used to perform experiments on abdominal CT sequences with liver lesions, i.e. the 16 th CT sequence in the slice 07 database and the 1 st, 3 rd, 6 th, 14 th and 17 th CT sequences in the 3dircad database. During the experiment, the CT sequences used to construct the liver dictionary were the remaining 54 abdominal CT sequences not used for testing in the XHCSU14, driver 07 and 3Dircadb databases. The above serial slices are all 512 × 512, the planar inter-pixel distance is distributed in the range of 0.53mm to 0.87mm, and the layer thickness is distributed in the range of 1.0mm to 4 mm. The segmentation results of liver lesion in abdominal CT test sequences were evaluated by using Volume Overlay Error (VOE), Relative Volume Difference (RVD), Average Symmetric Surface Distance (ASD), Root Mean Square Symmetric Surface Distance (RMSD), and Maximum Symmetric Surface Distance (MSD), each index being shown in table 1. As can be seen, for the diseased liver, the method has the advantages of small segmentation error and high precision.
TABLE 1
Claims (5)
1. A method for segmenting a lesion liver of an abdominal CT sequence image is characterized by comprising the following steps:
(1) the method specifically comprises the following steps of performing primary liver segmentation on an input abdominal CT sequence image:
a. selecting an initial slice of a diseased liver, and constructing a level set energy function:
wherein x and y are pixel points in the CT image omega, I (x) is the gray value of x in omega,is a non-negative window function, ciI is 1,2 represents the average gray scale of the object and the background in a circular area with y as the center and r as the radius, b (-) is a gray scale offset field,represents the inner and outer regions of the zero level set phi (x), H (-) is a Heaviside function,in the term of the perimeter, the length of the week,as a distance regularization term, the parameters upsilon and mu are respectively a term L (phi) for controlling the perimeter and a term R for distance regularizationp(phi) minimizing the energy function by using a gradient descent method, and keeping the maximum connected domain of the evolution result as the initial slice liver primary segmentation result;
b. and (3) constructing a non-initial slice level set energy function by taking the initial slice as a starting point and the adjacent slice segmentation result as position constraint:
wherein the content of the first and second substances,constraint of the liver segmentation position for the jth slice in the abdominal CT sequence, Ij-1And Ij+1Respectively segmenting non-initial slices in the sequence upwards and downwards in an iterative mode for the segmentation result of the liver of the adjacent slice, and reserving the area larger than T in the evolution resultpThe area of each pixel is used as the primary segmentation result of the lesion liver;
(2) finishing the correction of the shape of the lesion liver based on the liver sparse shape combination, which specifically comprises the following steps:
a. respectively carrying out three-dimensional reconstruction on liver CT serial slices of M patients manually segmented by experts, respectively constructing liver triangular grids with the number of vertexes being N, and obtaining liver dictionary atoms d by taking grid vertexes as dictionary elementsi∈R3NI 1,2, …, M, a liver shape dictionary D of size 3N × M is constructed with the first liver atom D1For reference, the remaining liver atom dqAnd d1Aligning to realize one-to-one correspondence of grid vertexes, wherein q is 2,3, …, M;
b. three-dimensional reconstruction is carried out on the CT sequence primary segmentation result of the diseased liver, a diseased liver triangular grid with the number of vertexes N is constructed, and vertex vectors v and v ∈ R of the grid are utilized3NRepresenting the shape of the diseased liver to be corrected and associating it with the first liver atom d1Aligning to realize one-to-one correspondence of the grid vertexes;
c. constructing sparse shape combination residual function
Where T (v, β) represents global transformation of v with β as a parameter, and x ═ x1,x2,…,xM]∈RMFor sparse coefficients, e ∈ R3NRepresenting the under-segmented part of the shape of the lesion liver to be corrected, D is a dictionary of liver shapes, lambda1、λ2Is a constant for controlling the sparsity of x and e; minimizing the residual function by using a simple and fast iterative shrinkage threshold algorithm, solving x, e and beta, and obtaining a corrected lesion liver triangular mesh v ═ T '(Dx-e, beta), wherein T' (. cndot.) represents the inverse transformation of T (. -);
d. carrying out voxelization correction on the three-dimensional grid v' of the lesion liver after voxelization correction, and obtaining a corresponding liver slice shape correction result;
(3) the method comprises the following steps of constructing a graph segmentation energy function based on shape prior, optimizing a segmentation result of a diseased liver, and realizing final segmentation of the diseased liver, wherein the graph segmentation energy function comprises the following specific steps:
a. obtaining the shape prior of the lesion liver according to the initial segmentation result and the shape correction result of the lesion liver section
Wherein f ispRepresenting pixel points p, L in CT slice flevel-setAnd LsscRespectively representing the primary segmentation result and the shape correction result of the liver corresponding to the slice lesion;
b. constructing a graph cut energy function E (f) based on the shape prior of the lesion liver,
wherein, I (f)p) Is a gray term, Sprior(fp) A priori to shape, B (f)p,fq) Is a boundary penalty term, alpha is a weight value for controlling a gray level term and a boundary term, P is a set of all pixels in the image f, N is a weight value for controlling a gray level term and a boundary termpSet of neighborhood pixels, I, of pixel point ppIs the gray scale value of pixel p, d (p, q) is the Euclidean distance between pixels p and q, SPThe number of pixel points in the pixel set P is counted;
c. and minimizing the graph cut energy function by using a maximum flow-minimum cut algorithm to obtain an optimized segmentation result of the diseased liver, and finishing the final segmentation of the diseased liver.
2. The method for segmenting liver with abdominal CT sequence image lesion of claim 1, wherein in the step (1), the initial slice in the CT sequence is located from top to bottom one third to two fifths of the whole CT sequence without liver rupture phenomenon.
3. The method as claimed in claim 1, wherein in step (1), ν and μ are both constants greater than 0, and r and T are constantpAre integers greater than 0.
4. The method as claimed in claim 1, wherein in step (2), N and M are integers greater than 0, λ1Is a constant greater than 0, λ2Is a constant greater than 0 and less than 1.
5. The method as claimed in claim 1, wherein in the step (3), α is a constant greater than 0 and less than 1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910029480.7A CN109741359B (en) | 2019-01-13 | 2019-01-13 | Method for segmenting lesion liver of abdominal CT sequence image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910029480.7A CN109741359B (en) | 2019-01-13 | 2019-01-13 | Method for segmenting lesion liver of abdominal CT sequence image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109741359A CN109741359A (en) | 2019-05-10 |
CN109741359B true CN109741359B (en) | 2022-05-31 |
Family
ID=66364560
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910029480.7A Active CN109741359B (en) | 2019-01-13 | 2019-01-13 | Method for segmenting lesion liver of abdominal CT sequence image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109741359B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8787658B2 (en) * | 2010-03-05 | 2014-07-22 | Microsoft Corporation | Image segmentation using reduced foreground training data |
CN107424152A (en) * | 2017-08-11 | 2017-12-01 | 联想(北京)有限公司 | The detection method and electronic equipment of organ lesion and the method and electronic equipment for training neuroid |
CN107610121A (en) * | 2017-09-28 | 2018-01-19 | 河北大学 | A kind of initial pose establishing method of liver statistical shape model |
CN108596887A (en) * | 2018-04-17 | 2018-09-28 | 湖南科技大学 | A kind of abdominal CT sequence image liver neoplasm automatic division method |
CN108986114A (en) * | 2018-07-11 | 2018-12-11 | 中南大学 | A kind of abdominal CT sequence image liver automatic division method based on level set and shape description symbols |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10157466B2 (en) * | 2016-01-21 | 2018-12-18 | Riverside Research Institute | Method for automatic tissue segmentation of medical images |
-
2019
- 2019-01-13 CN CN201910029480.7A patent/CN109741359B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8787658B2 (en) * | 2010-03-05 | 2014-07-22 | Microsoft Corporation | Image segmentation using reduced foreground training data |
CN107424152A (en) * | 2017-08-11 | 2017-12-01 | 联想(北京)有限公司 | The detection method and electronic equipment of organ lesion and the method and electronic equipment for training neuroid |
CN107610121A (en) * | 2017-09-28 | 2018-01-19 | 河北大学 | A kind of initial pose establishing method of liver statistical shape model |
CN108596887A (en) * | 2018-04-17 | 2018-09-28 | 湖南科技大学 | A kind of abdominal CT sequence image liver neoplasm automatic division method |
CN108986114A (en) * | 2018-07-11 | 2018-12-11 | 中南大学 | A kind of abdominal CT sequence image liver automatic division method based on level set and shape description symbols |
Also Published As
Publication number | Publication date |
---|---|
CN109741359A (en) | 2019-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11062449B2 (en) | Method and system for extracting vasculature | |
CN110223352B (en) | Medical image scanning automatic positioning method based on deep learning | |
US20220079529A1 (en) | Pseudo-ct generation from mr data using a feature regression model | |
AU2016339009B2 (en) | Pseudo-CT generation from MR data using tissue parameter estimation | |
CN106600609B (en) | Spine segmentation method and system in medical image | |
CN105139377B (en) | A kind of fast robust automatic division method of abdominal CT sequence image liver | |
US7315639B2 (en) | Method of lung lobe segmentation and computer system | |
EP1614069B1 (en) | Volume measurements in 3d datasets | |
CN108830852B (en) | Three-dimensional ultrasonic tumor auxiliary measurement system and method | |
CN108986114B (en) | Automatic abdominal CT sequence image liver segmentation method based on level set and shape descriptor | |
CN107038728A (en) | Profile automation based on iterative approximation is determined | |
CN108629785B (en) | Three-dimensional magnetic resonance pancreas image segmentation method based on self-learning | |
CN106960439B (en) | A kind of vertebrae identification device and method | |
Dewalle-Vignion et al. | Is STAPLE algorithm confident to assess segmentation methods in PET imaging? | |
CN104637056A (en) | Method for segmenting adrenal tumor of medical CT (computed tomography) image based on sparse representation | |
Wang et al. | Deformable torso phantoms of Chinese adults for personalized anatomy modelling | |
EP3631806A1 (en) | Quantified aspects of lesions in medical images | |
Vaassen et al. | Real-world analysis of manual editing of deep learning contouring in the thorax region | |
Lappas et al. | Automatic contouring of normal tissues with deep learning for preclinical radiation studies | |
Gallego et al. | Automatic model-based 3D segmentation of the breast in MRI | |
CN109741359B (en) | Method for segmenting lesion liver of abdominal CT sequence image | |
TW201835800A (en) | Method for determining/correcting defects and associated devices | |
KR101886235B1 (en) | Apparatus and method for providing of body size specific dose and image quality based on ct image | |
Li et al. | A multistep liver segmentation strategy by combining level set based method with texture analysis for CT images | |
CN117427286B (en) | Tumor radiotherapy target area identification method, system and equipment based on energy spectrum CT |
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 |