CN107204001A - Film automatic division method in a kind of carotid ultrasound image - Google Patents
Film automatic division method in a kind of carotid ultrasound image Download PDFInfo
- Publication number
- CN107204001A CN107204001A CN201610149681.7A CN201610149681A CN107204001A CN 107204001 A CN107204001 A CN 107204001A CN 201610149681 A CN201610149681 A CN 201610149681A CN 107204001 A CN107204001 A CN 107204001A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- msup
- profile
- edge
- 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
Classifications
-
- 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
- G06T2207/10132—Ultrasound image
-
- 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
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
Abstract
The present invention relates to image processing field, film automatic division method, comprises the following steps in disclosing in a kind of carotid ultrasound image：1) background information beyond interior middle film, the edge graph that membrane boundary is highlighted in obtaining in one are suppressed using multiple dimensioned Gaussian kernel multiplication method；2) the initial profile detection method based on Otsu threshold methods and Sobel operators is used, according to step 2) obtained edge graph, film epicardial border and the partial contour line segment on inner membrance tube chamber border in acquisition；3) in step 2) on the basis of the initial profile that obtains, a kind of both legs ant colony optimization algorithm is designed, initial profile line segment is connected, a complete profile is obtained；4) using the contour optimization algorithm based on Snake models come further Optimization Steps 3) obtained profile, make that final profile is more smooth, continuous, approaching to reality edge.
Description
Technical field
The present invention relates to middle film automatic division method in a kind of carotid ultrasound image, belong to Medical Image Processing neck
Domain.
Background technology
Angiocardiopathy is the primary disease for threatening human health.Atherosclerosis can cause arterial blood tube wall to increase
Thickness, causes luminal stenosis, thus is the main cause that angiocardiopathy occurs.Vascular wall is divided into outer from inside to outside
The film layer that film layer, middle film layer and theca interna three are mutually close to.Internal-media thickness (intima-media thickness,
IMT) refer to for from tube chamber-inner membrance (Lumen-Intima, LI) border to middle film-outer membrane (Media-Adventitia,
MA) distance on border, i.e. inner membrance are added with middle film two parts thickness.IMT increase is atherosclerosis
An early clinic symptom.Numerous studies prove that Carotid arterial IMT showed and cardiovascular and cerebrovascular disease have notable phase
Guan Xing, and can as Future Cardiovascular Events one strong prediction index.Nowadays, middle thickness in arteria carotis
Degree has been counted as an important finger of principium identification degree of carotid and cardiovascular pathological changes situation
Mark.
Medical ultrasound image technology can carry out more visible imaging to arteria carotis, be a kind of effective arteria carotis congee
Sample hardens generaI investigation means.By means of ultrasonic imaging technique, middle film in arteria carotis can be imaged, Jin Erke
To be partitioned into the border of interior middle film, IMT measurement is realized.However, it is clinical use by doctor's manual segmentation
It is time-consuming, uninteresting come the method that obtains interior middle membrane boundary, and can have the difference between larger observer, therefore
It can split the image partition method of interior middle membrane boundary automatically in the urgent need to a kind of.
Film partitioning algorithm in having at present in some arteria carotis, film segmentation in can realizing in accurately.So
And, ultrasonoscopy is influenceed by intrinsic speckle noise, and signal to noise ratio is relatively low, and anatomical details are not enough.Serious
Speckle noise even can cover interior middle film layer, have impact on interior middle film automatic division method success rate.The present invention is directed to
This problem, based on ant group optimization scheduling algorithm, it is proposed that a kind of interior middle film with more excellent robustness is split automatically
Method.
Ant colony optimization algorithm (ant colony optimization, ACO) be it is a kind of by ant colony creep feature inspire and
A kind of intelligent optimization algorithm of the biomimetic type grown up, is characterized in by bionical adaptive individual part most
Dominance determines the total optimization solution of problem jointly.The algorithm has the strong search capability that self-learning function is conciliate, tool
There are parallelization, strong robustness, positive feedback, be introduced among image segmentation field.For interior
Middle film segmentation, can use ant group algorithm, the contours extract problem on MA and LI borders is converted into optimization
Problem is solved.
The content of the invention
In order to substitute manual segmentation step, film automatic division method during the present invention is proposed in a kind of ultrasonoscopy.
The present invention is according to the unique dual-layer Parallel interfacial structure of interior middle membrane structure, it is proposed that a kind of both legs ant colony optimization algorithm,
And multiple dimensioned Gaussian kernel multiplication method, edge detection operator and Snake models are combined, realize interior middle film oneself
Dynamic segmentation.
Technical scheme is as follows：
Film automatic division method, comprises the following steps in a kind of ultrasonoscopy：
1) in being suppressed using multiple dimensioned Gaussian kernel multiplication method in the background information beyond interior middle film, enhancing
Membrane boundary, the edge graph that membrane boundary is highlighted in obtaining in one；
2) the initial profile detection method based on Otsu threshold methods and Sobel operators is used, according to step
2) edge graph obtained, obtains the partial contour line segment on LI and MA borders；
3) in step 2) on the basis of the initial profile that obtains, a kind of both legs ACO algorithms are designed, will
Initial profile line segment is connected, and obtains a complete profile；
4) using the contour optimization algorithm based on Snake models come further Optimization Steps 3) obtained wheel
Exterior feature, makes that final profile is more smooth, continuous, approaching to reality edge.
Above-mentioned steps 1) suppress the background information beyond interior middle film using multiple dimensioned Gaussian kernel multiplication method, highlight
Go out LI the and MA borders of interior middle film, obtain an edge graph.Edge graph, which is defined as two, has different scale
Gaussian density core wave filter and image convolution result product, and only retention is positive part：
Wherein,
It is the two-dimensional Gaussian function of a small yardstick, σ1Value 1~3.And Gσ2(y) it is one big chi
The one-dimensional Gaussian function of degree, σ2Value 10~20.
Step 2) the initial profile detection method based on Otsu threshold methods and Sobel operators is used, it is broadly divided into
Following steps：
(1) it is based on step 1) obtained edge graph, use Otsu threshold methods to obtain a binary image；
(2) top edge at MA interfaces is extracted from binary image and following using horizontal Sobel operators
Edge, and LI interfaces top edge and lower edge；
(3) edge line for having redundancy and the edge line for having defect are left out；
(4) for the part retained, the midpoint of top edge line and lower edge line is taken, each pair edge line is used as
Final single edge line after merging.
Step 3) employ a kind of both legs ACO algorithms, by initial profile line segment connected into one it is complete
Profile.It is characterized in that, middle film starting point placed two ants inside, make them be crawled toward the right side from the left end of blood vessel
End.Wherein one ant is by fixed placement in the top of another ant, and then the two creeps simultaneously, thus can
Originally to regard the two legs of an ant as.In addition, definition detects that the profile line segment obtained is by initial profile
The pressure path of ant, when ant creeps to this part, will be forced to creep along initial profile line segment, and line
Space between section will be connected by ACO algorithms.Its feature includes：
(1) transition probability equation
Ant is from the P chosen manuallys(xs,ys) point starts, creep from left to right, fixed every time to move right 1
Individual abscissa.In t, l (l=1,2) leg of kth ant is moved to adjacent pixel from pixel (x, i)
(x+1, process j) is according to lower transition probability equation：
Wherein τ refers to pheromones, and η refers to the density of pixel.α and β determine pheromones and the phase of heuristic information respectively
To influence power.Wherein, α values are that 1~3, β values are 3~5.
The l leg positions permitted set in next step is creeped of kth ant is represented, its correspondence
The right neighborhood of pixel (x, i).It must be noted that neighborhood territory pixel can not be shared by two legs simultaneously.Therefore,
If there is coincidence situation in the right neighborhood of two legs, the pixel of coincidence will by fromIt is middle to exclude.
(2) it is global to update rule
After all ants complete to creep, pheromones are substituted according to equation below：
τ(x,i)(t+1)=ρ τ(x,i)(t)+Δτ(x,i)
Wherein, τ(x,i)(t) when being the t times iteration pixel (x, i) place pheromones quantity, τ(x,i)(t+1) it is next time
The pheromones quantity at pixel (x, i) place during iteration.The initial value τ of pheromones0Value is 0.1~1.ρ is that decay is normal
Number, for the volatilization of artificial intelligence element, value is 0.5~1.Δτ(x,i)It is the information this time discharged in iteration
Prime number amount, its calculation formula is：
Wherein Q is a constant, and m is the quantity of ant.Q=1, m value are 10~50.C (k) is kth
Cost function of the ant during searching route, it is defined as follows：
Wherein IN(x,y)(0≤IN(xi,yi)≤1) is the gray value after pixel (x, i) normalization.D (k) is ant
antkThe terminal P that the terminal distance in path of creeping is manually selected beforee(xe,ye) distance.A is a punishment
Coefficient, sets weights of the error distance D (k) in cost function, and value is 1~2.
Above-mentioned steps 4) using the contour optimization algorithm based on Snake models come further profile, Snake moulds
Type is (boundary energy) and homogeneous comprising smoothed energy (smoothing energy), edge energy
Energy term (uniform energy), and realized by minimizing following energy functional：
Wherein y1And y (x)2(x) profile of LI and MA interfaces, parameter μ control smoothed energy are represented respectively
The weight of item, ν controls the weight of homogeneous energy term.Homogeneous energy term is connected to LI and MA two mutually
Independent profile, makes them keep homogeneous distance.μ values are that 0.1~0.3, ν values are 1~2.
The present invention has advantages below：
The present invention devises a kind of multiple dimensioned Gaussian kernel multiplication method first according to carotid ultrasound characteristics of image,
Image is pre-processed, membrane boundary in suppressing in the background information beyond interior middle film, enhancing.Then, in warp
In the image basis for crossing pretreatment, devise a kind of both legs ant colony optimization algorithm, and with Otsu threshold methods, Sobel
Edge detection operator, ant colony optimization algorithm, Snake models scheduling algorithm are combined, and realize carotid ultrasound image
The automatic segmentation of interior middle film.It is based on clinical image test result indicates that, this method achieve accurately segmentation knot
Really, its error is less than error between the observer of manual segmentation；At the same time, this method takes in clinical data test
Obtained 98.7% success rate, with good robustness, and cope with by the figure of speckle noise severe contamination
Picture.
Brief description of the drawings
Fig. 1 is the flow chart of middle film in present invention segmentation arteria carotis.
Fig. 2 is the result of each step in initial profile detection process.
During Fig. 3 is ant colony optimization algorithm, the allowance of leg (upper leg) and lower leg (lower leg) on ant
The typical case of location sets.
Fig. 4 is segmentation result exemplary plot of the present invention.
Embodiment
The present invention will be further described by the following examples, to more fully understand technical scheme.
Step is as follows：
1. the edge graph f (x, y) that membrane boundary is highlighted in being obtained using multiple dimensioned Gaussian kernel multiplication method in one.
F (x, y) computational methods are, by the wave filter and image convolution of two gaussian density core with different scale
As a result multiplication, and only retention is positive part：
Wherein,
It is σ in the two-dimensional Gaussian function of a small yardstick, the present embodiment1Value is 1.5.And
Gσ2(y) be a large scale one-dimensional Gaussian function, σ in the present embodiment2Value is 15.
2. using the initial profile detection method based on Otsu threshold methods and Sobel operators, LI and MA sides are obtained
The part initial profile line segment on boundary.As shown in Fig. 2 comprising the steps of：
1) it is based on by edge graph (a) obtained in the previous step, a binary image is obtained using Otsu threshold methods,
Such as figure (b)；
2) top edge at MA interfaces is extracted from binary image (b) using horizontal Sobel operators with
Edge, and LI interfaces top edge and lower edge；As shown in figure (c), white wire represents top edge
, grey lines represent lower edge.
3) after 2) step, desired result should correspond to LI respectively containing 2 to (4) contour line
With the lower edges at MA interfaces.However, the edge line obtained using Sobel operators generally all can
Contain unnecessary false edges line or breach.Therefore, seen along longitudinal direction, more or less than 2 pairs edges
The part of line will be left out, as a result as shown in (d).
4) finally, for the part retained, the midpoint of top edge line and lower edge line is taken, every opposite side is used as
Final single edge line after the merging of edge line, as shown in (e).
3. on the basis of the initial profile that step 2 is obtained, a kind of both legs ACO algorithms are designed, by initial profile line
Section is connected, and obtains a complete profile.Its step includes：
1) initialization information prime matrix.
2) ant always is placed on to the initial point P manually selecteds(xs,ys) nearby (upper leg is placed on a top 2
At individual pixel, lower leg is placed at the pixel of a lower section 2).
3) ant is from Ps(xs,ys) start to creep to the right, 1 abscissa that moves right is fixed every time, and that creeps is total
Step number is equal to xe-xs。
In t, l (l=1,2) leg of kth ant is moved to adjacent pixel from pixel (x, i)
(x+1, process j) is according to lower transition probability equation：
Wherein τ refers to pheromones, and η refers to the density of pixel.α and β determine pheromones and heuristic information respectively
Relative influence.Value is α=1, β=4 in the present embodiment.
The l leg positions permitted set in next step is creeped of kth ant is represented, it is right
The right neighborhood of pixel (x, i) is answered；But, if there is coincidence situation, the picture of coincidence in the right neighborhood of two legs
Element will by fromIt is middle to exclude.The upper leg of ant can so be ensured all the time in the top of lower leg, thus can
To ensure that final profile will not intersect or overlap.Fig. 3 illustrates an allusion quotation in both legs ant crawling process
Type neighborhoodAnt is this moment just from the position that abscissa is x to x+1 position creepings, light color side in figure
Lattice are the allowance location sets of the upper leg (upper leg) of ant and lower leg (lower leg), due to dark color side
Lattice are present in the allowance location sets of creeping of leg and lower leg simultaneously, thus are excluded from the set of the two
。
4) pheromones of pixel on ant path are calculated.
5) repeat step 2-4, until all ants complete their task of creeping.
6) fresh information prime matrix, and calculate the volatile quantity of pheromones.
Pheromones are substituted according to equation below：
τ(x,i)(t+1)=ρ τ(x,i)(t)+Δτ(x,i)
Wherein, τ(x,i)(t) when being the t times iteration pixel (x, i) place pheromones quantity, τ(x,i)(t+1) it is next time
The pheromones quantity at pixel (x, i) place during iteration.In the present embodiment, the initial value of pheromones is τ0=0.5.ρ is
Attenuation constant, for the meeting hair of artificial intelligence element, ρ=0.6 is chosen in the present embodiment.Δτ(x,i)It is this time to change
The pheromones quantity that discharges in generation, its calculation formula is：
Wherein Q is a constant, and m is the quantity of ant.Q=1, m=20 are chosen in the present embodiment.C(k)
It is cost function of the kth ant during searching route, it is defined as follows：
Wherein IN(x,y)(0≤IN(xi,yi)≤1) is the gray value after pixel (x, i) normalization.D (k) is ant
antkThe terminal P that the terminal distance in path of creeping is manually selected beforee(xe,ye) distance.A is a punishment
Coefficient, sets in weights of the error distance D (k) in cost function, the present embodiment and takes a=1.
7) 2-6 steps are repeated, a number of iteration is carried out.Iterations is set to 10 in the present embodiment.
8) the ant path with minimal consumption equation is chosen as optimal path, as LI and MA interfaces
Profile.
4. using profile is further optimized based on the contour optimization algorithm of Snake models, Snake models pass through most
Smallization following energy functional is realized：
Wherein y1And y (x)2(x) profile of LI and MA interfaces, parameter μ control smoothed energy are represented respectively
The weight of item, ν controls the weight of homogeneous energy term.The present embodiment chooses μ=0.1, ν=1.4.
Fig. 4 shows segmentation results of the CGACO in three different exemplary plots.Figure a is that a noise is serious
Image.Due to the pollution of noise, interior middle film is very unintelligible.As schemed shown in b, serious speckle noise is simultaneously
Without influence on the accurate segmentation of interior middle film.It is the carotid images that there is bending situation to scheme c, schemes e
It is that interior middle film the typical carotid images substantially thickened occurs.Figure d and figure f correspond to their segmentation knot respectively
Really.
Claims (7)
1. film automatic division method in a kind of ultrasonoscopy, it is characterised in that comprise the following steps：
1) film edge in being suppressed using multiple dimensioned Gaussian kernel multiplication method in the background information beyond interior middle film, enhancing, is obtained
The edge graph that membrane boundary is highlighted in one；
2) the initial profile detection method based on Otsu threshold methods and Sobel operators is used, according to step 2) obtained side
Edge figure, obtains the partial contour line segment of inner membrance-tube chamber border and inner membrance-middle membrane boundary；
3) in step 2) on the basis of the initial profile that obtains, a kind of both legs ant colony optimization algorithm is designed, by initial profile
Line segment is connected, and obtains a complete profile；
4) using the contour optimization algorithm based on Snake models come further Optimization Steps 3) obtained profile, make final
Profile is more smooth, continuous, approaching to reality edge.
2. the method as described in claim 1, it is characterised in that the step 1) in, edge graph computational methods are, by two tools
There are the wave filter of gaussian density core and being multiplied for image convolution result of different scale, and only retention is positive part.
3. method as claimed in claim 2, it is characterised in that in two gaussian density core wave filters of use, one is small yardstick
Two-dimensional Gaussian function, one be a large scale one-dimensional Gaussian function, they definition difference it is as follows：
<mrow>
<msub>
<mi>G</mi>
<msub>
<mi>&sigma;</mi>
<mn>1</mn>
</msub>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>2</mn>
<msup>
<msub>
<mi>&pi;&sigma;</mi>
<mn>1</mn>
</msub>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<msup>
<mi>e</mi>
<mfrac>
<mrow>
<mo>-</mo>
<mrow>
<mo>(</mo>
<msup>
<mi>x</mi>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mi>y</mi>
<mn>2</mn>
</msup>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<msub>
<mi>&sigma;</mi>
<mn>1</mn>
</msub>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
</msup>
<mo>,</mo>
<msub>
<mi>G</mi>
<msub>
<mi>&sigma;</mi>
<mn>2</mn>
</msub>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<msqrt>
<mrow>
<mn>2</mn>
<mi>&pi;</mi>
</mrow>
</msqrt>
<msub>
<mi>&sigma;</mi>
<mn>2</mn>
</msub>
</mrow>
</mfrac>
<msup>
<mi>e</mi>
<mfrac>
<mrow>
<mo>-</mo>
<msup>
<mi>y</mi>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<msub>
<mi>&sigma;</mi>
<mn>2</mn>
</msub>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
</msup>
</mrow>
σ1Value is 1~3, σ2Value is 10~20.
4. the method as described in claim 1, the step 2) acquisition of initial profile, it is characterised in that comprise the steps of：
1) a binary image is obtained using Otsu threshold methods；2) extracted using horizontal Sobel operators from binary image
Go out the top edge and lower edge of inner membrance-middle membrane interface, and inner membrance-tube chamber interface top edge and lower edge；3) leaving out has
The edge line of redundancy and the edge line for having defect；4) for the part retained, take in top edge line and lower edge line
Point, the final single edge line after merging as each pair edge line.
5. the method as described in claim 1, it is characterised in that the step 3) in,tMoment, the l (l=1,2) of kth ant
Leg is moved to adjacent pixel from pixel (x, i), and (x+1, the transition probability equation of process foundation j) is：
<mrow>
<msubsup>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
</mrow>
<mrow>
<mi>k</mi>
<mo>,</mo>
<mi>l</mi>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mfrac>
<mrow>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mrow>
<msub>
<mi>&tau;</mi>
<mrow>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>+</mo>
<mn>1</mn>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
</mrow>
<mo>&rsqb;</mo>
</mrow>
<mi>&alpha;</mi>
</msup>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mrow>
<msub>
<mi>&eta;</mi>
<mrow>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>+</mo>
<mn>1</mn>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
</mrow>
<mo>&rsqb;</mo>
</mrow>
<mi>&beta;</mi>
</msup>
</mrow>
<mrow>
<msub>
<mo>&Sigma;</mo>
<mi>s</mi>
</msub>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mrow>
<msub>
<mi>&tau;</mi>
<mrow>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>+</mo>
<mn>1</mn>
<mo>,</mo>
<mi>s</mi>
</mrow>
<mo>)</mo>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
</mrow>
<mo>&rsqb;</mo>
</mrow>
<mi>&alpha;</mi>
</msup>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mrow>
<msub>
<mi>&eta;</mi>
<mrow>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>+</mo>
<mn>1</mn>
<mo>,</mo>
<mi>s</mi>
</mrow>
<mo>)</mo>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
</mrow>
<mo>&rsqb;</mo>
</mrow>
<mi>&beta;</mi>
</msup>
</mrow>
</mfrac>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>j</mi>
<mo>&Element;</mo>
<msubsup>
<mi>N</mi>
<mrow>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>i</mi>
</mrow>
<mo>)</mo>
</mrow>
<mrow>
<mi>k</mi>
<mo>,</mo>
<mi>l</mi>
</mrow>
</msubsup>
<mo>,</mo>
<mi>s</mi>
<mo>&Element;</mo>
<msubsup>
<mi>N</mi>
<mrow>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>i</mi>
</mrow>
<mo>)</mo>
</mrow>
<mrow>
<mi>k</mi>
<mo>,</mo>
<mi>l</mi>
</mrow>
</msubsup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>0</mn>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>o</mi>
<mi>t</mi>
<mi>h</mi>
<mi>e</mi>
<mi>r</mi>
<mi>w</mi>
<mi>i</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein τ refers to pheromones, and η refers to the density of pixel.α and β determine the relative influence of pheromones and heuristic information respectively.Its
In, α values are that 1~3, β values are 3~5.Represent what the l leg of kth ant was allowed in next step is creeped
Location sets.
6. method as claimed in claim 5, it is characterised in that what the l leg of kth ant was allowed in next step is creeped
Location setsFor the right neighborhood of pixel (x, i)；But if there is coincidence situation, the picture of coincidence in the right neighborhood of two legs
Element will by fromIt is middle to exclude.
7. the method as described in claim 1, it is characterised in that the step 4) in, Snake models are by minimizing following energy
Functional is measured to realize：
Wherein y1And y (x)2(x) profile of LI and MA interfaces, parameter μ control smoothed energy (smoothing are represented respectively
Energy weight), ν controls the weight of homogeneous energy term (uniform energy).μ values are that 0.1~0.3, ν values are
1~2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610149681.7A CN107204001B (en) | 2016-03-16 | 2016-03-16 | Automatic midium segmentation method in carotid artery ultrasonic image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610149681.7A CN107204001B (en) | 2016-03-16 | 2016-03-16 | Automatic midium segmentation method in carotid artery ultrasonic image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107204001A true CN107204001A (en) | 2017-09-26 |
| CN107204001B CN107204001B (en) | 2020-03-31 |
Family
ID=59904195
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610149681.7A Active CN107204001B (en) | 2016-03-16 | 2016-03-16 | Automatic midium segmentation method in carotid artery ultrasonic image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107204001B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108765432A (en) * | 2018-05-07 | 2018-11-06 | 山东大学 | Middle membrane boundary automatic division method and system in a kind of arteria carotis |
| CN109394268A (en) * | 2018-12-07 | 2019-03-01 | 余姚市华耀工具科技有限公司 | Polyp extent of injury Mapping Platform |
| CN109961424A (en) * | 2019-02-27 | 2019-07-02 | 北京大学 | A kind of generation method of hand x-ray image data |
| CN110047086A (en) * | 2019-04-24 | 2019-07-23 | 飞依诺科技(苏州)有限公司 | Carotic Intima-media Thinkness method for automatic measurement and system |
| CN111986139A (en) * | 2019-05-23 | 2020-11-24 | 深圳市理邦精密仪器股份有限公司 | Method and device for measuring intima-media thickness in carotid artery and storage medium |
| CN117557460A (en) * | 2024-01-12 | 2024-02-13 | 济南科汛智能科技有限公司 | Angiography image enhancement method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102163326A (en) * | 2010-12-22 | 2011-08-24 | 武汉沃生科学技术研究中心有限公司 | Method for automatic computerized segmentation and analysis on thickness uniformity of intima media of carotid artery blood wall in sonographic image |
| CN102800089A (en) * | 2012-06-28 | 2012-11-28 | 华中科技大学 | Main carotid artery blood vessel extraction and thickness measuring method based on neck ultrasound images |
| CN104599271A (en) * | 2015-01-20 | 2015-05-06 | 中国科学院半导体研究所 | CIE Lab color space based gray threshold segmentation method |
| CN105105741A (en) * | 2015-07-15 | 2015-12-02 | 无锡海鹰电子医疗系统有限公司 | Envelope line extracting and feature point tracking method of pulse wave image |
-
2016
- 2016-03-16 CN CN201610149681.7A patent/CN107204001B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102163326A (en) * | 2010-12-22 | 2011-08-24 | 武汉沃生科学技术研究中心有限公司 | Method for automatic computerized segmentation and analysis on thickness uniformity of intima media of carotid artery blood wall in sonographic image |
| CN102800089A (en) * | 2012-06-28 | 2012-11-28 | 华中科技大学 | Main carotid artery blood vessel extraction and thickness measuring method based on neck ultrasound images |
| CN104599271A (en) * | 2015-01-20 | 2015-05-06 | 中国科学院半导体研究所 | CIE Lab color space based gray threshold segmentation method |
| CN105105741A (en) * | 2015-07-15 | 2015-12-02 | 无锡海鹰电子医疗系统有限公司 | Envelope line extracting and feature point tracking method of pulse wave image |
Non-Patent Citations (1)
| Title |
|---|
| 张丽娜等: "基于蚁群聚类算法的动脉硬化无创检测", 《计算机工程与应用》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108765432A (en) * | 2018-05-07 | 2018-11-06 | 山东大学 | Middle membrane boundary automatic division method and system in a kind of arteria carotis |
| CN108765432B (en) * | 2018-05-07 | 2020-08-07 | 山东大学 | Automatic carotid intima-media boundary segmentation method and system |
| CN109394268A (en) * | 2018-12-07 | 2019-03-01 | 余姚市华耀工具科技有限公司 | Polyp extent of injury Mapping Platform |
| CN109394268B (en) * | 2018-12-07 | 2021-05-11 | 刘志红 | Polyp harm degree mapping platform |
| CN109961424A (en) * | 2019-02-27 | 2019-07-02 | 北京大学 | A kind of generation method of hand x-ray image data |
| CN109961424B (en) * | 2019-02-27 | 2021-04-13 | 北京大学 | Hand X-ray image data generation method |
| CN110047086A (en) * | 2019-04-24 | 2019-07-23 | 飞依诺科技(苏州)有限公司 | Carotic Intima-media Thinkness method for automatic measurement and system |
| CN111986139A (en) * | 2019-05-23 | 2020-11-24 | 深圳市理邦精密仪器股份有限公司 | Method and device for measuring intima-media thickness in carotid artery and storage medium |
| CN117557460A (en) * | 2024-01-12 | 2024-02-13 | 济南科汛智能科技有限公司 | Angiography image enhancement method |
| CN117557460B (en) * | 2024-01-12 | 2024-03-29 | 济南科汛智能科技有限公司 | Angiography image enhancement method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107204001B (en) | 2020-03-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107204001A (en) | Film automatic division method in a kind of carotid ultrasound image | |
| CN110930397B (en) | Magnetic resonance image segmentation method and device, terminal equipment and storage medium | |
| CN107766894B (en) | Remote sensing image natural language generation method based on attention mechanism and deep learning | |
| CN109242844B (en) | Pancreatic cancer tumor automatic identification system based on deep learning, computer equipment and storage medium | |
| CN108764006B (en) | SAR image target detection method based on deep reinforcement learning | |
| CN107563983A (en) | Image processing method and medical imaging devices | |
| CN110443815B (en) | Semi-supervised retina OCT image layer segmentation method combined with generation of countermeasure network | |
| CN110675406A (en) | CT image kidney segmentation algorithm based on residual double-attention depth network | |
| CN105825509A (en) | Cerebral vessel segmentation method based on 3D convolutional neural network | |
| Tang et al. | Segmentation-by-detection: A cascade network for volumetric medical image segmentation | |
| CN104217422B (en) | A kind of sonar image detection method of adaptive narrow-band level set | |
| CN109376756B (en) | System, computer device and storage medium for automatically identifying lymph node transferred from upper abdomen based on deep learning | |
| CN102068281A (en) | Processing method for space-occupying lesion ultrasonic images | |
| US11302094B2 (en) | System and method for segmenting normal organ and/or tumor structure based on artificial intelligence for radiation treatment planning | |
| CN110390665A (en) | Knee joint illness ultrasound diagnostic method based on deep learning multichannel and the embedding method of figure | |
| CN106056596A (en) | Fully-automatic three-dimensional liver segmentation method based on local apriori information and convex optimization | |
| CN114820520A (en) | Prostate image segmentation method and intelligent prostate cancer auxiliary diagnosis system | |
| CN111784701A (en) | Ultrasonic image segmentation method and system combining boundary feature enhancement and multi-scale information | |
| CN112820399A (en) | Method and device for automatically diagnosing benign and malignant thyroid nodules | |
| CN109712161A (en) | A kind of image partition method, device, equipment and storage medium | |
| CN113920109A (en) | Medical image recognition model training method, recognition method, device and equipment | |
| CN112149691B (en) | Neural network searching method and device for binocular vision matching | |
| CN109034218A (en) | model training method, device, equipment and storage medium | |
| Zhou et al. | Subcellular structure segmentation from cryo-electron tomograms via machine learning | |
| US20220198670A1 (en) | Method and system for automated segmentation of biological object parts in mri |
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 | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 215123 5th floor, building a, 4th floor, building C, No. 27, Xinfa Road, Suzhou Industrial Park, Jiangsu Province Patentee after: Feiyinuo Technology Co.,Ltd. Patentee after: Peking University Address before: 215123 5th floor, building a, 4th floor, building C, No. 27, Xinfa Road, Suzhou Industrial Park, Jiangsu Province Patentee before: Feiyinuo Technology (Suzhou) Co.,Ltd. Patentee before: Peking University |
|
| CP01 | Change in the name or title of a patent holder | ||
| CP03 | Change of name, title or address |
Address after: 215123 5th floor, building a, 4th floor, building C, No. 27, Xinfa Road, Suzhou Industrial Park, Jiangsu Province Patentee after: Feiyinuo Technology (Suzhou) Co.,Ltd. Patentee after: Peking University Address before: 215123 5F, building a, No. 27, Xinfa Road, Suzhou Industrial Park, Jiangsu Province Patentee before: VINNO TECHNOLOGY (SUZHOU) Co.,Ltd. Patentee before: Peking University |
|
| CP03 | Change of name, title or address |