CN108320277A - Determine the method, apparatus and CT machines on tumour 3 D boundary - Google Patents
Determine the method, apparatus and CT machines on tumour 3 D boundary Download PDFInfo
- Publication number
- CN108320277A CN108320277A CN201710028776.8A CN201710028776A CN108320277A CN 108320277 A CN108320277 A CN 108320277A CN 201710028776 A CN201710028776 A CN 201710028776A CN 108320277 A CN108320277 A CN 108320277A
- Authority
- CN
- China
- Prior art keywords
- feature
- tumour
- faultage
- unit
- boundaries
- 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
- 206010028980 Neoplasm Diseases 0.000 title claims abstract description 81
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000012549 training Methods 0.000 claims abstract description 15
- 238000011156 evaluation Methods 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 7
- 230000009466 transformation Effects 0.000 description 10
- 238000001914 filtration Methods 0.000 description 8
- 230000011218 segmentation Effects 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 6
- 230000003902 lesion Effects 0.000 description 5
- 210000004185 liver Anatomy 0.000 description 5
- 206010019695 Hepatic neoplasm Diseases 0.000 description 4
- 230000006399 behavior Effects 0.000 description 4
- 238000002591 computed tomography Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000000513 principal component analysis Methods 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 210000001367 artery Anatomy 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 208000014018 liver neoplasm Diseases 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 210000003240 portal vein Anatomy 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 241000669244 Unaspis euonymi Species 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000000205 computational method Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 208000019423 liver disease Diseases 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000002271 resection Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 230000001225 therapeutic effect Effects 0.000 description 2
- 210000003462 vein Anatomy 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000003759 clinical diagnosis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000686 essence Substances 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20064—Wavelet transform [DWT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30056—Liver; Hepatic
-
- 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
- G06T2207/30096—Tumor; Lesion
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a kind of method, apparatus and CT machines on determining tumour 3 D boundary.This method includes：Load the body layer data of a specific detected object；Receive a plurality of faultage images selected from the body layer data；Tumor boundaries are determined on each faultage image；The label of the tumor boundaries is received, the label is the classification to tumour；The feature of the faultage image is extracted in voxel range, forms a feature pool；Optimal characteristics are selected from the feature pool；According to the faultage image, the label of the tumor boundaries, the optimal characteristics, one grader of training；Other faultage images are identified with the grader, so that it is determined that corresponding tumor boundaries；The tumor boundaries of each faultage image are connected, so that it is determined that the three-dimensional boundaries of tumour.
Description
Technical field
The present invention relates to medical image processings, the especially determination on tumour 3 D boundary.
Background technology
Liver is the Functional tissue for maintaining human life activity important and complicated, and hepatic disease is multiple, and lesion type is more, hair
Sick rate is high.Computed tomography (Computed Tomography, CT) image has become conventional hand important in clinical diagnosis
One of section, is the important detection methods of liver diseases.Liver tumour treatment means include mainly tumor resection, intervention, radiation at present
Treatment etc., tumor resection is most effective one therapeutic modality.These treatment means are required for accurately understanding tumour in the preoperative
The information such as quantity, position, size and shape contribute to the formulation of liver neoplasm therapeutic scheme.But tumour individual difference is big,
Liver neoplasm and liver parenchyma boundary are fuzzy, and the position of tumour, size, shape, gray scale and texture are different, it is difficult to work out one
The general lesion segmentation algorithm of kind.Manually segmentation is needed with anatomical knowledge and experience, and subjectivity is strong, needs to spend
Take plenty of time and energy.Since tumor boundaries are fuzzy, factors, most of liver segmentation methods such as performance otherness is big can not reach
To clinical requirement precision.
Existing full-automatic dividing liver neoplasm method, main flow be manually to training data carry out feature extraction,
Feature selecting, design grader, obtain prediction model, according to this model to testing number by supervised learning or unsupervised learning
According to being predicted.Characteristic extraction procedure is computationally intensive, it is more to take, the feature that can have been chosen largely by data volume,
Experience and fortune.
Invention content
In view of this, the present invention proposes a kind of method, apparatus and CT machines on determining tumour 3 D boundary.
According to the first aspect of the invention, a kind of method on determining tumour 3 D boundary is provided, including：Load one it is specific by
Examine the body layer data of object；Receive a plurality of faultage images selected from the body layer data；It is determined on each faultage image
Tumor boundaries；The label of the tumor boundaries is received, the label is the classification to tumour；The tomography is extracted in voxel range
The feature of image forms a feature pool；Optimal characteristics are selected from the feature pool；According to the faultage image, the tumour side
The label on boundary, the optimal characteristics, one grader of training；Other faultage images are identified with the grader, so that it is determined that corresponding
Tumor boundaries；The tumor boundaries of each faultage image are connected, so that it is determined that the three-dimensional boundaries of tumour.
In one embodiment, the method further includes：The evaluation to the three-dimensional boundaries is received, if being evaluated as negatively,
Then go to a plurality of faultage images for receiving and being selected from the body layer data.
According to the second aspect of the invention, a kind of device on determining tumour 3 D boundary is provided, including：The load of one data is single
Member loads the body layer data of a specific detected object；One faultage image selecting unit, reception are selected from the body layer data
The a plurality of faultage images selected；One boundary determination unit determines tumor boundaries on each faultage image；One tag unit,
The label of the tumor boundaries is received, the label is the classification to tumour；One feature extraction unit is extracted in voxel range
The feature of the faultage image forms a feature pool；One Feature Selection unit selects optimal characteristics from the feature pool；One
Classifier training unit, according to the faultage image, the label of the tumor boundaries, the optimal characteristics, one classification of training
Device；One boundary recognition unit identifies other faultage images with the grader, so that it is determined that corresponding tumor boundaries；One connects
Order member, connects the tumor boundaries of each faultage image, so that it is determined that the three-dimensional boundaries of tumour.
In one embodiment, the device further includes an evaluation unit, receives the evaluation to the three-dimensional boundaries, if
It is evaluated as negatively, the faultage image selecting unit receives a plurality of faultage images selected from the body layer data again.
According to the third aspect of the invention we, a kind of CT machines are provided, including just like devices described above.
The present invention, which extracts, is characterized in that tumour and the normal structure characteristic to be split for each patient is elected, tradition
Method be to be chosen from the sample set of multiple patients.Therefore compared to the feature selecting of more patient's samples, side of the invention
Method more has specific aim.In addition, being never susceptible to the different scanning parameter even machine of different manufacturers with choosing in sample
Situation needs to do specific processing, and present method avoids this problems.In addition, doctor is every time to some tomograph
After upper lesion segmentation/modification, the feature of final partitioning algorithm is all different.These are characterized in the interaction by doctor and meter
What calculation machine algorithm common choice came out.
Description of the drawings
Below will detailed description of the present invention preferred embodiment by referring to accompanying drawing, make those skilled in the art more
The above and other feature and advantage of the clear present invention, in attached drawing：
Fig. 1 is the flow diagram according to the method on the determination tumour 3 D boundary of the first embodiment of the present invention.
Fig. 2A is the example of a faultage image.
Fig. 2 B are result of the faultage image shown in Fig. 2A through two layers of three-dimensional Stationary Wavelet Transform.
Fig. 3 A to 3C are respectively result of the multiple dimensioned sphericity feature on unenhanced, artery, portal vein image.
Fig. 4 is the structure diagram according to the device on the determination tumour 3 D boundary of the second embodiment of the present invention.
In above-mentioned attached drawing, used reference numeral is as follows：
100 method, 210 feature extraction unit
208 tag unit, 212 Feature Selection unit
200 device, 214 classifier training unit
202 data loading unit, 216 Boundary Recognition unit
204 faultage image selecting unit, 218 connection unit
206 boundary determination unit, 220 evaluation unit
208 tag units
S102、S104、
S106、S108、
S110, S112, step
S114、S116、
S118、S120
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, by the following examples to of the invention further detailed
It describes in detail bright.
Fig. 1 is the flow diagram according to the method 100 on the determination tumour 3 D boundary of the first embodiment of the present invention.Such as
Shown in Fig. 1, method 100 includes step S102, step S104, step S106, step S108, step S110, step S112, step
S114, step S116 and step S118.
In step s 102, the body layer data of a specific detected object is loaded.
In step S104, a plurality of faultage images selected from body layer data are received.
In step s 106, tumor boundaries are determined on each faultage image.
In step S108, the label of tumor boundaries is received, which is the classification to tumour.
In step s 110, the feature of the faultage image is extracted in voxel range, forms a feature pool.Using existing
The algorithm of extraction feature calculate the features such as density, texture, time and shape, and they are cascaded become an attribute to
Amount.For density feature, such as characteristics of mean, Variance feature.Wavelet character may be used in textural characteristics calculating, i.e., first by image
Multi-layer three-dimension wavelet transformation is done, then the filial generation image-region after each transformation extracts feature.Wavelet transformation is separable
, therefore with low frequency and high frequency one-dimensional filtering device successively convolution can be done in three dimensions of image.Fig. 2A is a tomograph
The example of picture.Fig. 2 B are result of the faultage image shown in Fig. 2A through two layers of three-dimensional Stationary Wavelet Transform, wherein the first behavior one
Layer Stationary Wavelet Transform, the second two layers of behavior Stationary Wavelet Transform.For 3-D view, each layer of wavelet transformation can obtain 8
A filial generation.Fig. 2 B list some filial generation images, wherein first is classified as the low frequency filtering in three directions, second is classified as perpendicular to figure
Vertical direction low frequency filtering in the direction low frequency filtering of image plane, the plane of delineation, horizontal direction High frequency filter in the plane of delineation.Often
There are different textural characteristics in the different filial generations of layer transformation.Temporal characteristics are often related with the characteristic of tumour, for example, for liver
Dirty primary tumor often embodies the feature of " F.F. goes out soon " in three phase images of CT scan, and this feature can use the time
Intensity curve is weighed.For certain specific structures, shape feature method tends to provide the position of structure.Such as with more rulers
The tensor computation of the Hessian matrix of degree can be used for measuring the specific structure in image.Hessian matrix is defined as follows：
Wherein x=(x, y, z) is the voxel location in three dimensions, and I is pending image, and σ is that Gaussian convolution core is big
Small i.e. scale.The characteristic value and feature vector of Hessian matrix are λ1(x),λ2(x),λ3(x) andEnable λ1
(x)>λ2(x)>λ3(x), then the sphericity under the scale is
In view of the sphericity feature of multiple scales, there is following formula
Wherein σminAnd σmaxFor the range of dimension calculation.Fig. 3 A to 3C are respectively multiple dimensioned sphericity feature unenhanced, dynamic
Result on arteries and veins, portal vein.The region of red color is the feature of high value, and the closed curve of yellow swells for what doctor sketched the contours manually
Tumor boundary.
In step S112, from feature pool selection for the optimal characteristics of the image and the tumour.Above-mentioned attribute vector tool
There is high-dimensional feature, it is redundancy there are some in these features.Before analysis and comment and information gain, we can not be one
It is useful to know which is characterized in secondary specific lesion segmentation.Some Feature Selections can be used and dimension-reduction algorithm solves this
Problem, such as principal component analysis (Principal component analysis, PCA).Attribute vector can be projected to one by PCA
A orthogonal plane, to which the attribute vector is changed into a new attribute vector, this new attribute vector will be used as classification
The input of device.After PCA is converted, the dimension of attribute vector reduces, and obtains orthogonal characteristic, effectively removes in feature
Redundancy.
In step S114, according to faultage image, the label of tumor boundaries, feature, one grader of training.In the present embodiment
In, the training dataset that support vector machines (SVM) is crossed from label can be used and find best hyperplane.This process is represented by
Solve the following optimization problem with constraints：
Wherein, αiIt is Lagrange multiplier
yiIt is the label of feature, usually -1 (normal structure), 1 (tumour)
xiIt is ith attribute vector
N is data set size
K(xi,xj) it is kernel function, nonlinear characteristic can be mapped to a high-dimensional space, and the feature after mapping by it
It is linear classification.Common Non-linear Kernel has polynomial kernel, Gaussian kernel, laplace kernel or Sigmoid cores.For example, high
This core is represented by：
σ is the window width of Gaussian kernel
In step S116, other faultage images are identified with grader, so that it is determined that corresponding tumor boundaries.It will adjust manually
Then whole regional expansion uses quick level set computational methods to other levels, it is only necessary to point minute in current curves
Then class expands boundary.
In step S118, the tumor boundaries of each faultage image are connected, so that it is determined that the three-dimensional boundaries of tumour.
In the present embodiment, method 100 may also include step S120.In the step s 120, it receives and three-dimensional boundaries is commented
Valence, if being evaluated as negatively, going to step S104.
Fig. 2 is according to the structure diagram of the device 200 on the determination tumour 3 D boundary of the second embodiment of the present invention, the dress
Setting can be as a part for a CT machines.As shown in Fig. 2, device 200 includes a data loading unit 202, faultage image choosing
Select unit 204, a boundary determination unit 206, a tag unit 208, a feature extraction unit 210, a Feature Selection unit
212, a classifier training unit 214, a boundary recognition unit 216 and a connection unit 218.
Data loading unit 202 loads the body layer data of a specific detected object.
Faultage image selecting unit 204 receives a plurality of faultage images selected from body layer data.
Boundary determination unit 206 determines tumor boundaries on each faultage image.
Tag unit 208 receives the label of tumor boundaries, which is the classification to tumour.
Feature extraction unit 210 extracts the feature of faultage image in voxel range, forms a feature pool.Using existing
The algorithm of extraction feature calculates the features such as density, texture, time and shape, and they are cascaded as an attribute vector.
For density feature, such as characteristics of mean, Variance feature.Wavelet character may be used in textural characteristics calculating, i.e., first does image
Multi-layer three-dimension wavelet transformation, then each transformation after filial generation image-region extract feature.Wavelet transformation be it is separable,
Therefore convolution successively can be done in three dimensions of image with low frequency and high frequency one-dimensional filtering device.Fig. 2A is a faultage image
Example.Fig. 2 B are result of the faultage image shown in Fig. 2A through two layers of three-dimensional Stationary Wavelet Transform, wherein the first one layer of behavior is flat
Steady wavelet transformation, the second two layers of behavior Stationary Wavelet Transform.For 3-D view, each layer of wavelet transformation can obtain 8 sons
Generation.Fig. 2 B list some filial generation images, wherein first is classified as the low frequency filtering in three directions, second be classified as it is flat perpendicular to image
Vertical direction low frequency filtering in the direction low frequency filtering in face, the plane of delineation, horizontal direction High frequency filter in the plane of delineation.Every layer of change
There are different textural characteristics in the different filial generations changed.Temporal characteristics are often related with the characteristic of tumour, for example, for liver
Primary tumor often embodies the feature of " F.F. goes out soon " in three phase images of CT scan, and this feature can use time intensity
Curve is weighed.For certain specific structures, shape feature method tends to provide the position of structure.Such as with multiple dimensioned
The tensor computation of Hessian matrix can be used for measuring the specific structure in image.Hessian matrix is defined as follows：
Wherein x=(x, y, z) is the voxel location in three dimensions, and I is pending image, and σ is that Gaussian convolution core is big
Small i.e. scale.The characteristic value and feature vector of Hessian matrix are λ1(x),λ2(x),λ3(x) andEnable λ1
(x)>λ2(x)>λ3(x), then the sphericity under the scale is
In view of the sphericity feature of multiple scales, there is following formula
Wherein σminAnd σmaxFor the range of dimension calculation.Fig. 3 A to 3C are respectively multiple dimensioned sphericity feature unenhanced, dynamic
Result on arteries and veins, portal vein.The region of red color is the feature of high value, and the closed curve of yellow swells for what doctor sketched the contours manually
Tumor boundary.
Feature Selection unit 212 is from feature pool selection for the optimal characteristics of the image and the tumour.Above-mentioned attribute vector
With high-dimensional feature, it is redundancy to have some in these features.Before analysis and comment and information gain, we can not be
It is useful to know which is characterized in primary specific lesion segmentation.Some Feature Selections can be used and dimension-reduction algorithm solves this
One problem, such as principal component analysis (Principal component analysis, PCA).Attribute vector can be projected to by PCA
One orthogonal plane, to which the attribute vector is changed into a new attribute vector, this new attribute vector, which will be used as, to be divided
The input of class device.After PCA is converted, the dimension of attribute vector reduces, and obtains orthogonal characteristic, effectively removes feature
In redundancy.
Classifier training unit 214 is according to faultage image, the label of tumor boundaries, feature, one grader of training.In this reality
It applies in example, the training dataset that support vector machines (SVM) is crossed from label can be used and find best hyperplane.This process can table
It is shown as solving the following optimization problem with constraints：
Wherein, αiIt is Lagrange multiplier
yiIt is the label of feature, usually -1 (normal structure), 1 (tumour)
xiIt is ith attribute vector
N is data set size
K(xi,xj) it is kernel function, nonlinear characteristic can be mapped to a high-dimensional space, and the feature after mapping by it
It is linear classification.Common Non-linear Kernel has polynomial kernel, Gaussian kernel, laplace kernel or Sigmoid cores.For example, high
This core is represented by：
σ is the window width of Gaussian kernel
Boundary Recognition unit 216 identifies other faultage images with grader, so that it is determined that corresponding tumor boundaries.
Connection unit 218 connects the tumor boundaries of each faultage image, so that it is determined that the three-dimensional boundaries of tumour.It will manually adjust
Regional expansion to other levels, then use quick level set computational methods, it is only necessary to in current curves point classify,
Then boundary is expanded.
In the present embodiment, device 200 may also include an evaluation unit 220, the evaluation to three-dimensional boundaries be received, if commenting
Valence is negative, and faultage image selecting unit 204 receives a plurality of faultage images selected from body layer data again, and after execution
The task of continuous unit.
The present invention, which extracts, is characterized in what tumour and normal structure characteristic for each patient were elected, traditional method
It is to be chosen from the sample set of multiple patients.Therefore compared to the feature selecting of more patient's samples, method of the invention is more
With specific aim.In addition, the case where never with the machine for being susceptible to different scanning parameter or even different manufacturers is selected in sample, need
Specific processing is done, and present method avoids this problems.In addition, doctor is every time to swollen on some faultage image
After tumor segmentation/modification, the feature of final partitioning algorithm is all different.These are characterized in the interaction by doctor and computerized algorithm
What common choice came out.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
With within principle, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention god.
Claims (5)
1. a kind of method on determining tumour 3 D boundary, including：
Step S102 loads the body layer data of a specific detected object；
Step S104 receives a plurality of faultage images selected from the body layer data；
Step S106 determines tumor boundaries on each faultage image；
Step S108, receives the label of the tumor boundaries, and the label is the classification to tumour；
Step S110 extracts the feature of the faultage image in voxel range, forms a feature pool；
Step S112 selects optimal characteristics from the feature pool；
Step S114, according to the faultage image, the label of the tumor boundaries, the optimal characteristics, one grader of training；
Step S116 identifies other faultage images with the grader, so that it is determined that corresponding tumor boundaries；
Step S118 connects the tumor boundaries of each faultage image, so that it is determined that the three-dimensional boundaries of tumour.
2. the method as described in claim 1, characterized in that further include：
Step S120 receives the evaluation to the three-dimensional boundaries, if being evaluated as negatively, going to step S104.
3. a kind of device on determining tumour 3 D boundary, including：
One data loading unit (202) loads the body layer data of a specific detected object；
One faultage image selecting unit (204) receives a plurality of faultage images selected from the body layer data；
One boundary determination unit (206) determines tumor boundaries on each faultage image；
One tag unit (208), receives the label of the tumor boundaries, and the label is the classification to tumour；
One feature extraction unit (210) extracts the feature of the faultage image in voxel range, forms a feature pool；
One Feature Selection unit (212) selects optimal characteristics from the feature pool；
One classifier training unit (214), according to the faultage image, the label of the tumor boundaries, the optimal characteristics,
One grader of training；
One boundary recognition unit (216), other faultage images are identified with the grader, so that it is determined that corresponding tumour side
Boundary；
One connection unit (218), connects the tumor boundaries of each faultage image, so that it is determined that the three-dimensional boundaries of tumour.
4. device as claimed in claim 3, characterized in that further include an evaluation unit (220), receive to the three-dimensional side
The evaluation on boundary, if being evaluated as negative, the faultage image selecting unit (204) receives again to be selected from the body layer data
A plurality of faultage images.
5. a kind of CT machines, including the device as described in claim 3 or 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710028776.8A CN108320277A (en) | 2017-01-16 | 2017-01-16 | Determine the method, apparatus and CT machines on tumour 3 D boundary |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710028776.8A CN108320277A (en) | 2017-01-16 | 2017-01-16 | Determine the method, apparatus and CT machines on tumour 3 D boundary |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108320277A true CN108320277A (en) | 2018-07-24 |
Family
ID=62891583
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710028776.8A Pending CN108320277A (en) | 2017-01-16 | 2017-01-16 | Determine the method, apparatus and CT machines on tumour 3 D boundary |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108320277A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109190690A (en) * | 2018-08-17 | 2019-01-11 | 东北大学 | The Cerebral microbleeds point detection recognition method of SWI image based on machine learning |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110317888A1 (en) * | 2010-06-25 | 2011-12-29 | Simon Richard A | Liver lesion segmentation |
| CN102473303A (en) * | 2009-08-12 | 2012-05-23 | 皇家飞利浦电子股份有限公司 | Generating object data |
| CN103854286A (en) * | 2014-03-05 | 2014-06-11 | 东南大学 | Renal artery blood-supply area segmenting method based on CT contrastographic picture |
| CN105139410A (en) * | 2015-09-11 | 2015-12-09 | 浙江工商大学 | Brain tumor MRI image segmentation algorithm utilizing space section projection |
| CN105261002A (en) * | 2014-07-10 | 2016-01-20 | 西门子公司 | Method and apparatus for displaying pathological changes in an examination object based on 3d data records |
| CN105957066A (en) * | 2016-04-22 | 2016-09-21 | 北京理工大学 | CT image liver segmentation method and system based on automatic context model |
-
2017
- 2017-01-16 CN CN201710028776.8A patent/CN108320277A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102473303A (en) * | 2009-08-12 | 2012-05-23 | 皇家飞利浦电子股份有限公司 | Generating object data |
| US20110317888A1 (en) * | 2010-06-25 | 2011-12-29 | Simon Richard A | Liver lesion segmentation |
| CN103854286A (en) * | 2014-03-05 | 2014-06-11 | 东南大学 | Renal artery blood-supply area segmenting method based on CT contrastographic picture |
| CN105261002A (en) * | 2014-07-10 | 2016-01-20 | 西门子公司 | Method and apparatus for displaying pathological changes in an examination object based on 3d data records |
| CN105139410A (en) * | 2015-09-11 | 2015-12-09 | 浙江工商大学 | Brain tumor MRI image segmentation algorithm utilizing space section projection |
| CN105957066A (en) * | 2016-04-22 | 2016-09-21 | 北京理工大学 | CT image liver segmentation method and system based on automatic context model |
Non-Patent Citations (2)
| Title |
|---|
| 刘洋: "基于改进支持向量机的脑部医学图像分割及肿瘤图像特征提取的研究", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》 * |
| 武倩: "磁共振脑图像中肿瘤化组织的特征提取和分割", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109190690A (en) * | 2018-08-17 | 2019-01-11 | 东北大学 | The Cerebral microbleeds point detection recognition method of SWI image based on machine learning |
| CN109190690B (en) * | 2018-08-17 | 2021-10-19 | 东北大学 | Detection and identification method of cerebral microbleeds in SWI images based on machine learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gonçalves et al. | Hessian based approaches for 3D lung nodule segmentation | |
| JP5864542B2 (en) | Image data processing method, system, and program for detecting image abnormality | |
| Cha et al. | Bladder cancer segmentation in CT for treatment response assessment: application of deep-learning convolution neural network—a pilot study | |
| CN112529834B (en) | Spatial distribution of pathological image patterns in 3D image data | |
| Perk et al. | Automated classification of benign and malignant lesions in 18F-NaF PET/CT images using machine learning | |
| Nie et al. | Quantitative analysis of lesion morphology and texture features for diagnostic prediction in breast MRI | |
| US10593035B2 (en) | Image-based automated measurement model to predict pelvic organ prolapse | |
| Taşcı et al. | Shape and texture based novel features for automated juxtapleural nodule detection in lung CTs | |
| Tan et al. | Optimization of breast mass classification using sequential forward floating selection (SFFS) and a support vector machine (SVM) model | |
| CN107767962B (en) | Determining result data based on medical measurement data from different measurements | |
| EP3444781B1 (en) | Image processing apparatus and image processing method | |
| US20160012604A1 (en) | Automatic background region selection for lesion delineation in medical images | |
| Shariaty et al. | Texture appearance model, a new model-based segmentation paradigm, application on the segmentation of lung nodule in the CT scan of the chest | |
| Sluimer et al. | Automated classification of hyperlucency, fibrosis, ground glass, solid, and focal lesions in high‐resolution CT of the lung | |
| Juang et al. | Psoriasis image identification using k-means clustering with morphological processing | |
| Riyahi et al. | Quantifying local tumor morphological changes with Jacobian map for prediction of pathologic tumor response to chemo-radiotherapy in locally advanced esophageal cancer | |
| US10949698B2 (en) | Image processing apparatus, image processing system, image processing method, and storage medium | |
| WO2007026598A1 (en) | Medical image processor and image processing method | |
| Zhang et al. | A supervised texton based approach for automatic segmentation and measurement of the fetal head and femur in 2D ultrasound images | |
| Gonçalves et al. | Learning lung nodule malignancy likelihood from radiologist annotations or diagnosis data | |
| Rey et al. | A hybrid CAD system for lung nodule detection using CT studies based in soft computing | |
| Egger et al. | A medical software system for volumetric analysis of cerebral pathologies in magnetic resonance imaging (MRI) data | |
| Jaffar et al. | Ensemble classification of pulmonary nodules using gradient intensity feature descriptor and differential evolution | |
| Hoogi et al. | A fully-automated pipeline for detection and segmentation of liver lesions and pathological lymph nodes | |
| CN105556567B (en) | Method and system for vertebral location detection |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180724 |