CN103914845A - Method for acquiring initial contour in ultrasonic image segmentation based on active contour model - Google Patents
Method for acquiring initial contour in ultrasonic image segmentation based on active contour model Download PDFInfo
- Publication number
- CN103914845A CN103914845A CN201410141568.5A CN201410141568A CN103914845A CN 103914845 A CN103914845 A CN 103914845A CN 201410141568 A CN201410141568 A CN 201410141568A CN 103914845 A CN103914845 A CN 103914845A
- Authority
- CN
- China
- Prior art keywords
- vector
- ultrasonoscopy
- contour
- ultrasonic image
- standard
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Ultra Sonic Daignosis Equipment (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method for acquiring an initial contour in ultrasonic image segmentation based on an active contour model. The method comprises the following steps that the textural features of an ultrasonic image tumor area are trained, and a normal vector is formed by the textural features, a standard elliptical experience value and the tumor prior size in an ultrasonic image to be detected; the ultrasonic image to be detected is preprocessed; dynamic threshold segmentation is carried out on the image; all closed contours generated in the dynamic threshold segmentation result are extracted to form corresponding sub-images; 24 textural features of the sub-images, an ellipse fitting result parameter and the number of pixels in the closed contours of the sub-images are calculated, and a vector is formed by 26 data; the distance between the obtained vector and the normal vector obtained from the tumor area through training is calculated, and the closed contour, corresponding to the vector with the smallest distance, in the sub-images is the contour of a tumor in the segmented ultrasonic image. The method solves the problem that the ultrasonic image has much noise and is fuzzy in boundary, and the initial contour accuracy is high.
Description
Technical field
A kind of method that the present invention relates to obtain in Ultrasound Image Segmentation based on active profile initial profile, belongs to ultrasonoscopy process field.
Background technology
Along with the appearance of high intensity focused ultrasound, non-invasively treating oncologic application more and more extensive, this therapy system has very large advantage if do not operated in clinical, preventing from scar, without wound or Wicresoft's wound, can detect treatment in real time, be not subject to that tumor size limits, total expenses is low etc.These advantages all determine that high-strength focus supersonic therapeutic system has very large development and application prospect.And in high-strength focus supersonic therapeutic system most critical exactly patient's tumour is carried out to real-time navigation.Navigation procedure need to position the tumour in real-time ultrasonic image.Traditional manual positioning mode is changed into automatic location, will the therapeutic efficiency of this therapy system and treatment accuracy be brought to very large help.
Ultrasound Image Segmentation is to be widely used in lesion detection, diagnosis and treatment, and its accuracy of cutting apart is directly connected to the location of tumour.Through long-term further investigation, the ultrasonic image division method having proposed have based on region, based on border and based on statistical information, dissimilar dividing method has the different features of cutting apart.Active contour model (Active Contour Model), the Snake that is otherwise known as, is a kind of objective contour describing method being proposed by Andrew professor Blake, is mainly used in the Target Segmentation based on shape.The superior part of this model is that it has provided unified solution for far-ranging a series of visual problems, in nearest more than ten years, it has successfully been applied to many fields of computer vision by increasing researcher, as edge extracting, image is cut apart and is classified, motion tracking, three-dimensional reconstruction, stereoscopic vision coupling etc.
But this model needs initial profile, restrain again if manually obtain initial profile, waste time and energy, accurate not again.Therefore the method that designs automatic acquisition initial profile is very important.
Summary of the invention
Technical matters to be solved by this invention is to provide a kind of obtain initial profile in Ultrasound Image Segmentation based on active contour model method.Operation is simple and reliable for the method, speed is fast, and the ultrasonoscopy tumour initial profile accuracy of being obtained by the method is also very high.
Automatically the method for extracting tumour initial profile in a kind of Ultrasound Image Segmentation of the present invention, comprises following steps:
(1) choose the above existing ultrasonoscopy composition training set of 60 width, described existing ultrasonoscopy contains the tumour identical with ultrasonoscopy to be split, from training set, in every width ultrasonoscopy, extract 24 textural characteristics of tumor region with gray level co-occurrence matrix, utilize in mathematics linear regression minimum distance method obtain the standard value of 24 textural characteristics, and form a standard vector A together with tumour priori size in standard ellipse empirical value, ultrasonoscopy to be measured
0;
(2) ultrasonoscopy to be split is carried out to the pre-service of anisotropy diffusion and histogram equalization;
(3) (2) gained image is carried out to dynamic threshold segmentation, and extract all closed contours that produce in dynamic threshold segmentation result, each closed contour forms a corresponding subimage;
(4) utilize gray level co-occurrence matrix to extract 24 textural characteristics of each number of sub images, calculate ellipse fitting result parameter and subimage closed contour interior pixels number, and with a vectorial A of this 26 data formations
1;
(5) calculate each subimage gained vector A
1standard vector A with tumor region training gained
0distance B, make the vectorial A of D minimum
1the closed contour in corresponding subimage be the initial profile of tumour in ultrasonoscopy to be split.
In step (1), standard ellipse empirical value is defined as 0.7.
Above-mentioned 24 textural characteristics comprise each six features of contrast, energy, correlativity, unfavourable balance square, entropy and non-similarity on 0 ° of image, 45 °, 90 °, 135 ° four directions.
Make obtaining of initial profile more accurate, at the vectorial A that settles the standard
0process in, need to use mathematical measure reliably.Such as supersonic tumor image in training process has 60 width, every piece image all will be tried to achieve a texture feature vector A
i, so how to process these 60 vectors to obtain a standard vector A
0need to think better of.In conjunction with utilizing A below
0method used when determining tumor region: calculate the proper vector A extracting from each number of sub images
1with A
0distance, so now also available range this with reference to obtaining standard vector A
0.Concrete manner of execution is the linear regression in mathematics, 60 eigenwerts that have by certain textural characteristics correspond in coordinate system, then obtain data in coordinate system, make these 60 values the shortest to the distance sum of this data place straight line, these data are exactly the standard value of this textural characteristics so, can obtain successively this 24 standard values corresponding to textural characteristics difference, these standard values form a standard vector A together with tumour priori size in standard ellipse data, ultrasonoscopy to be measured
0.
When wherein in ultrasonoscopy to be measured, tumour priori size is by Ultrasonic Detection, doctor or expert's Direct Mark go out tumor region major and minor axis, unit is generally millimeter, then according to physical length in ultrasonoscopy and pixel ratio, for example every millimeter of 4 pixels, can calculate the priori size of tumor region.
Above-mentioned every number of sub images can only comprise a closed contour.
Above-mentioned calculating distance B method used is χ
2range formula is tried to achieve:
A
0and A (k)
1(k) represent respectively vectorial A
0and A
1in k parameter, k gets the integer between 0～25.
Usefulness of the present invention is:
(1) changed conventional manual and obtained the clumsy part of initial profile.
(2) initial profile acquisition methods is simple, quick, is applicable to movable contour model
(3) initial profile of gained is clear accurately, has efficient accurate advantage as active contour afterwards for convergence, has greatly improved efficiency and the accuracy of Ultrasound Image Segmentation.
Brief description of the drawings
Fig. 1 is the process flow diagram of automatic profile acquisition methods of the present invention.
Fig. 2 is a width liver tumour image initial profile result figure of example 1 of the present invention.Fig. 2 (a) is liver neoplasm ultrasonoscopy original graph; Fig. 2 (b) is the pre-service result of original image after anisotropy diffusion and histogram equalization; Fig. 2 (c) is pretreatment image dynamic threshold segmentation result; The initial profile result of Fig. 2 (d) for obtaining.
Fig. 3 is two width fibroid ultrasonoscopy initial profile result figure of example 2 of the present invention, and result is as the initial profile of active contour model, obtains the result figure of final tumor boundaries through the GGVF field of force after the convergence under driving.
Embodiment
Carry out more detailed description the present invention by some embodiments below, but the present invention is not limited to these embodiment.
Embodiment 1:
Through the existing liver neoplasm ultrasonoscopy of 60 width being trained to the standard feature vector A that obtains liver tumour ultrasonoscopy
0.First each width imagery exploitation gray level co-occurrence matrixes is calculated its 24 textural characteristics values, can obtain so 60 groups of proper vectors that contain 24 elements, these all proper vectors are carried out to the processing of mathematics linear regression method, obtain a standard feature vector, by finding a shortest vectorial A of mean distance to this stack features vector, each textural characteristics value corresponding in this vector A is exactly the standard value that we need to obtain.Then two elements of liver tumour priori size that the tumor region major and minor axis being marked by doctor in Ultrasonic Detection process in this proper vector A interpolation standard ellipse fitting data 0.7 and ultrasonoscopy to be measured calculated, have so just obtained a standard feature vector A who contains 26 elements
0.
Fig. 2 (a) is liver tumour ultrasonoscopy I to be split, and Fig. 2 (b) processes the pre-service result of gained later for original image being carried out to anisotropy diffusion and histogram equalization; Fig. 2 (c) carries out the result of dynamic threshold segmentation to gained image in Fig. 2 (b), and extracts all closed contours that produce in dynamic threshold segmentation result, forms corresponding subimage; Utilize gray level co-occurrence matrix to calculate 24 of the textural characteristics of each number of sub images, calculate every number of sub images ellipse fitting result parameter and subimage closed contour interior pixels number, and form a proper vector A by these 26 data
1; Utilize χ
2range formula calculates each subimage gained vector A
1standard vector A with tumor region training gained
0distance B, determine and make the vectorial A of D minimum
1closed contour in corresponding subimage is the initial profile of tumour in ultrasonoscopy to be split, is the initial profile result of obtaining shown in Fig. 2 (d).
Embodiment 2:
Through the existing fibroid ultrasonoscopy of 60 width being trained to the standard feature vector A that obtains fibroid ultrasonoscopy
0.First each width imagery exploitation gray level co-occurrence matrixes is calculated its 24 textural characteristics values, can obtain so 60 groups of proper vectors that contain 24 elements, these all proper vectors are carried out to the processing of mathematics linear regression method, obtain a standard feature vector, by finding a shortest vectorial A of mean distance to this stack features vector, each textural characteristics value corresponding in this vector A is exactly the standard value that we need to obtain.Then this proper vector A is added to the tumor region major and minor axis that marked by doctor in Ultrasonic Detection process in standard ellipse fitting data 0.7 and ultrasonoscopy to be measured calculates two elements of fibroid priori size, so just obtained a standard feature vector A who contains 26 elements
0.
Fig. 3 (a1) is fibroid ultrasonoscopy I to be split
1, Fig. 3 (b1) processes the pre-service result of gained later for original image being carried out to anisotropy diffusion and histogram equalization; Fig. 3 (c1) carries out the result of dynamic threshold segmentation to gained image in Fig. 3 (b1), and extracts all closed contours that produce in dynamic threshold segmentation result, forms corresponding subimage; Utilize gray level co-occurrence matrix to calculate 24 of the textural characteristics of each number of sub images, calculate every number of sub images ellipse fitting result parameter and subimage closed contour interior pixels number, and form a proper vector A by these 26 data
1; Utilize χ
2range formula calculates each subimage gained vector A
1standard vector A with tumor region training gained
0distance B, determine and make the vectorial A of D minimum
1closed contour in corresponding subimage is the initial profile of tumour in ultrasonoscopy to be split, is the initial profile result of obtaining shown in Fig. 3 (d1).Fig. 3 (e1) is the initial profile using the initial profile of gained in Fig. 3 (d1) as active contour model, then restrains the final fibroid border result figure of gained through GGVF external force field.
Fig. 3 (a2) is fibroid ultrasonoscopy I to be split
2, Fig. 3 (b2) processes the pre-service result of gained later for original image being carried out to anisotropy diffusion and histogram equalization; Fig. 3 (c2) carries out the result of dynamic threshold segmentation to gained image in Fig. 3 (b2), and extracts all closed contours that produce in dynamic threshold segmentation result, forms corresponding subimage; Utilize gray level co-occurrence matrix to calculate 24 of the textural characteristics of each number of sub images, calculate every number of sub images ellipse fitting result parameter and subimage closed contour interior pixels number, and form a proper vector A by these 26 data
1; Utilize χ
2range formula calculates each subimage gained vector A
1standard vector A with tumor region training gained
0distance B, determine and make the vectorial A of D minimum
1closed contour in corresponding subimage is the initial profile of tumour in ultrasonoscopy to be split, is the initial profile result of obtaining shown in Fig. 3 (d2).Fig. 3 (e2) is the initial profile using the initial profile of gained in Fig. 3 (d2) as active contour model, then restrains the final fibroid border result figure of gained through GGVF external force field.
In embodiment, the extraction of initial profile does not need implementer that initial profile is manually provided, and the workload that this has clearly alleviated implementer has improved the efficiency of Ultrasound Image Segmentation.
Claims (3)
1. a method of obtaining initial profile in the Ultrasound Image Segmentation based on active contour model, is characterized in that, comprises following steps:
(1) choose the above existing ultrasonoscopy composition training set of 60 width, described existing ultrasonoscopy contains the tumour identical with ultrasonoscopy to be split, from training set, in every width ultrasonoscopy, extract 24 textural characteristics of tumor region with gray level co-occurrence matrix, utilize in mathematics linear regression minimum distance method obtain the standard value of 24 textural characteristics, and form a standard vector together with tumour priori size in standard ellipse empirical value, ultrasonoscopy to be measured
a 0;
(2) ultrasonoscopy to be split is carried out to the pre-service of anisotropy diffusion and histogram equalization;
(3) (2) gained image is carried out to dynamic threshold segmentation, and extract all closed contours that produce in dynamic threshold segmentation result, each closed contour forms a corresponding subimage;
(4) utilize gray level co-occurrence matrix to extract 24 textural characteristics of each number of sub images, calculate ellipse fitting result parameter and subimage closed contour interior pixels number, and with a vector of this 26 data formations
a 1;
(5) calculate each subimage gained vector
a 1standard vector with tumor region training gained
a 0distance B, make the vector of D minimum
a 1the closed contour in corresponding subimage be the initial profile of tumour in ultrasonoscopy to be split.
2. in method according to claim 1, it is characterized in that, in step (1), standard ellipse empirical value is defined as 0.7.
3. in method according to claim 1 and 2, it is characterized in that, in step (1), described 24 textural characteristics comprise contrast, energy, correlativity, unfavourable balance square, entropy and six features of non-similarity on 0 ° of image, 45 °, 90 °, 135 ° four directions.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410141568.5A CN103914845B (en) | 2014-04-09 | 2014-04-09 | The method obtaining initial profile in Ultrasound Image Segmentation based on active contour model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410141568.5A CN103914845B (en) | 2014-04-09 | 2014-04-09 | The method obtaining initial profile in Ultrasound Image Segmentation based on active contour model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103914845A true CN103914845A (en) | 2014-07-09 |
| CN103914845B CN103914845B (en) | 2016-08-17 |
Family
ID=51040501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410141568.5A Expired - Fee Related CN103914845B (en) | 2014-04-09 | 2014-04-09 | The method obtaining initial profile in Ultrasound Image Segmentation based on active contour model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103914845B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104504689A (en) * | 2014-12-11 | 2015-04-08 | 深圳先进技术研究院 | Method and device for segmenting inner and outer membranes in intravascular ultrasound image |
| CN105719278A (en) * | 2016-01-13 | 2016-06-29 | 西北大学 | Organ auxiliary positioning segmentation method based on statistical deformation model |
| CN105931226A (en) * | 2016-04-14 | 2016-09-07 | 南京信息工程大学 | Automatic cell detection and segmentation method based on deep learning and using adaptive ellipse fitting |
| CN111242965A (en) * | 2020-01-10 | 2020-06-05 | 西安电子科技大学 | Genetic algorithm-based breast tumor contour dynamic extraction method |
| CN112166474A (en) * | 2018-05-16 | 2021-01-01 | 皇家飞利浦有限公司 | Automated tumor identification during surgery using machine learning |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1793350A1 (en) * | 2005-12-01 | 2007-06-06 | Medison Co., Ltd. | Ultrasound imaging system and method for forming a 3D ultrasound image of a target object |
| CN101599174A (en) * | 2009-08-13 | 2009-12-09 | 哈尔滨工业大学 | Method for outline extraction of level set medical ultrasonic image area based on edge and statistical nature |
| US20110164831A1 (en) * | 2010-01-05 | 2011-07-07 | Stmicroelectronics (Grenoble 2) Sas | Method for detecting orientation of contours |
| CN102509286A (en) * | 2011-09-28 | 2012-06-20 | 清华大学深圳研究生院 | Target region sketching method for medical image |
| CN102855483A (en) * | 2011-06-30 | 2013-01-02 | 北京三星通信技术研究有限公司 | Method and device for processing ultrasonic images and breast cancer diagnosis equipment |
| CN103413307A (en) * | 2013-08-02 | 2013-11-27 | 北京理工大学 | Method for image co-segmentation based on hypergraph |
-
2014
- 2014-04-09 CN CN201410141568.5A patent/CN103914845B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1793350A1 (en) * | 2005-12-01 | 2007-06-06 | Medison Co., Ltd. | Ultrasound imaging system and method for forming a 3D ultrasound image of a target object |
| CN101599174A (en) * | 2009-08-13 | 2009-12-09 | 哈尔滨工业大学 | Method for outline extraction of level set medical ultrasonic image area based on edge and statistical nature |
| US20110164831A1 (en) * | 2010-01-05 | 2011-07-07 | Stmicroelectronics (Grenoble 2) Sas | Method for detecting orientation of contours |
| CN102855483A (en) * | 2011-06-30 | 2013-01-02 | 北京三星通信技术研究有限公司 | Method and device for processing ultrasonic images and breast cancer diagnosis equipment |
| CN102509286A (en) * | 2011-09-28 | 2012-06-20 | 清华大学深圳研究生院 | Target region sketching method for medical image |
| CN103413307A (en) * | 2013-08-02 | 2013-11-27 | 北京理工大学 | Method for image co-segmentation based on hypergraph |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104504689A (en) * | 2014-12-11 | 2015-04-08 | 深圳先进技术研究院 | Method and device for segmenting inner and outer membranes in intravascular ultrasound image |
| CN104504689B (en) * | 2014-12-11 | 2017-06-06 | 深圳先进技术研究院 | Adventitia dividing method and device in ivus image |
| CN105719278A (en) * | 2016-01-13 | 2016-06-29 | 西北大学 | Organ auxiliary positioning segmentation method based on statistical deformation model |
| CN105719278B (en) * | 2016-01-13 | 2018-11-16 | 西北大学 | A kind of medical image cutting method based on statistics deformation model |
| CN105931226A (en) * | 2016-04-14 | 2016-09-07 | 南京信息工程大学 | Automatic cell detection and segmentation method based on deep learning and using adaptive ellipse fitting |
| CN112166474A (en) * | 2018-05-16 | 2021-01-01 | 皇家飞利浦有限公司 | Automated tumor identification during surgery using machine learning |
| CN111242965A (en) * | 2020-01-10 | 2020-06-05 | 西安电子科技大学 | Genetic algorithm-based breast tumor contour dynamic extraction method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103914845B (en) | 2016-08-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Almajalid et al. | Development of a deep-learning-based method for breast ultrasound image segmentation | |
| US8098918B2 (en) | Method and system for measuring left ventricle volume | |
| Kim et al. | Machine-learning-based automatic identification of fetal abdominal circumference from ultrasound images | |
| CN106683104B (en) | Prostate Magnetic Resonance Image Segmentation method based on integrated depth convolutional neural networks | |
| CN103914823B (en) | The method of the quick exact non-linear registration solid medical image based on rarefaction representation | |
| CN104809723A (en) | Three-dimensional liver CT (computed tomography) image automatically segmenting method based on hyper voxels and graph cut algorithm | |
| CN103150723B (en) | The stomach CT image lymph node detection system of Shape-based interpolation and ellipse fitting and method | |
| CN102542556B (en) | Method for automatically extracting ultrasonic breast tumor image | |
| CN103914845A (en) | Method for acquiring initial contour in ultrasonic image segmentation based on active contour model | |
| CN103824295A (en) | Segmentation method of adhesion vascular pulmonary nodules in lung CT (computed tomography) image | |
| CN102831614B (en) | Sequential medical image quick segmentation method based on interactive dictionary migration | |
| CN104899851A (en) | Lung nodule image segmentation method | |
| CN102682449A (en) | Automatic segmentation method based on self-adaptive external force level set for magnetic resonance images (MRIs) of soft tissue and realization method thereof | |
| CN103871066A (en) | Method for constructing similarity matrix in ultrasound image Ncut segmentation process | |
| CN107507189A (en) | Mouse CT image kidney dividing methods based on random forest and statistical model | |
| CN105389821A (en) | Medical image segmentation method based on combination of cloud module and image segmentation | |
| CN104881858A (en) | Method and device for extracting background parenchymal enhancement in breast | |
| CN103544695A (en) | Efficient medical image segmentation method based on game framework | |
| Bi et al. | Hyper-fusion network for semi-automatic segmentation of skin lesions | |
| Ma et al. | Random walk based segmentation for the prostate on 3D transrectal ultrasound images | |
| CN102881021B (en) | Based on the aorta petal ultrasonic image division method of probability distribution and continuous max-flow | |
| Furuzuki et al. | A detection method for liver cancer region based on faster R-CNN | |
| CN102509296B (en) | Maximum-likelihood-region-merging-based interactive segmentation method for stomach computed tomography (CT) image | |
| CN103208114A (en) | Stomach adipose tissue extraction method based on interactive segmentation | |
| CN103793916B (en) | Fibroid ultrasonic image division method in a kind of HIFU Treatment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160817 Termination date: 20200409 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |