CN106570871A - Fuzzy C mean value carotid ultrasonic image intima-media thickness measuring method and system - Google Patents
Fuzzy C mean value carotid ultrasonic image intima-media thickness measuring method and system Download PDFInfo
- Publication number
- CN106570871A CN106570871A CN201610951289.4A CN201610951289A CN106570871A CN 106570871 A CN106570871 A CN 106570871A CN 201610951289 A CN201610951289 A CN 201610951289A CN 106570871 A CN106570871 A CN 106570871A
- Authority
- CN
- China
- Prior art keywords
- image
- roi
- algorithm
- media thickness
- fuzzy
- 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
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 70
- 230000011218 segmentation Effects 0.000 claims abstract description 33
- 238000001914 filtration Methods 0.000 claims abstract description 25
- 238000002604 ultrasonography Methods 0.000 claims description 23
- 238000005070 sampling Methods 0.000 claims description 19
- 230000002146 bilateral effect Effects 0.000 claims description 8
- 210000004204 blood vessel Anatomy 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 4
- 238000012805 post-processing Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 2
- 238000005259 measurement Methods 0.000 description 13
- 230000008569 process Effects 0.000 description 10
- 208000024172 Cardiovascular disease Diseases 0.000 description 5
- 238000003384 imaging method Methods 0.000 description 5
- 239000012528 membrane Substances 0.000 description 4
- 210000000709 aorta Anatomy 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 210000001715 carotid artery Anatomy 0.000 description 2
- 210000002808 connective tissue Anatomy 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005315 distribution function Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 230000002792 vascular Effects 0.000 description 2
- 206010003210 Arteriosclerosis Diseases 0.000 description 1
- 201000001320 Atherosclerosis Diseases 0.000 description 1
- 206010008190 Cerebrovascular accident Diseases 0.000 description 1
- 208000006011 Stroke Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 208000011775 arteriosclerosis disease Diseases 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 210000001168 carotid artery common Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 206010025482 malaise Diseases 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 210000004165 myocardium Anatomy 0.000 description 1
- 230000005311 nuclear magnetism Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 231100000915 pathological change Toxicity 0.000 description 1
- 230000036285 pathological change Effects 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000005309 stochastic process Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 231100000216 vascular lesion Toxicity 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20028—Bilateral filtering
-
- 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
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Geometry (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the ultrasonic image processing field, and provides a carotid ultrasonic image intima-media thickness measuring method and system, thus accurately measuring the intima-media thickness, and providing good theory values and usage values; the fuzzy C mean value carotid ultrasonic image intima-media thickness measuring method comprises the following steps: 1, extracting an interested area; 2, using an image filter selection double-side filtering algorithm to respectively filter two ROI images; 3, ROI segmentation, importing a HMRF model so as to build a space constraint field between pixels on the basis of a standard FCM algorithm. The fuzzy C mean value carotid ultrasonic image intima-media thickness measuring method and system are mainly applied to the ultrasonic image processing occasions.
Description
Technical field
The present invention relates to ultrasonoscopy process field, specifically, is related to carotid ultrasound image Internal-media thickness measurement side
Method and system.
Background technology
The statistical study of World Health Organization (WHO) shows：Cardiovascular disease is with sickness rate height, disguised strong, course of disease length, death
The high characteristic of rate becomes harm human life and health primary factor.The year two thousand thirty is expected, the whole world loses life because of cardiovascular disease
Number be up to 23,600,000 people.Only vascular lesion early detection, early discovery, early stage are rationally intervened, could be from basic
The upper disability rate and mortality rate for reducing cardiovascular disease.The pathologic basis (initial stage performance) of cardiovascular disease are Atherosclerosis
Change, in the presence of various paathogenic factors, cause Internal-media thickness (Intima Media Thickness, IMT) in arterial wall
Increase.Here Internal-media thickness refers to tube chamber-intima boundary (Lumen-Intima Interface, LII) and middle film-outer
The distance between membrane boundary (Media-Adventitia Interface, MAI).Due to cervical region aorta (Common
Carotid Artery, CCA) blood vessel be arteriosclerosis common site, its position is approximately parallel with skin, easily measures, because
The IMT of this cervical region aorta can be as the important indicator of the early lesion degree of assessment cardiovascular disease, to burst cardiac muscle stalk
Plug, the diagnosis of apoplexy and prevention have critically important value.
Relative to imaging modes such as nuclear-magnetism, CT, ultra sonic imaging has real-time, repeatability, Noninvasive and low cost
Advantage, therefore become carotid artery inspection first-selected imaging mode.Imitated based on picture the characteristics of ultra sonic imaging, is clinically often chosen to
Fruit preferably distal vessels wall measurement Internal-media thickness.Traditionally, the Internal-media thickness of carotid ultrasound image is generally by artificial
Manual measurement is obtained, and then the LII and MAI in gauger's hand drawing image is obtained by calculating the distance between two borders
Obtain IMT.But this method workload is big, training, individual Jing suffered by the result test subject for taking very much and finally obtaining
Go through and its subjective judgment impact.In order to solve to ask present in the manual measurement process of carotid ultrasound image Internal-media thickness
Topic, in past 20 years, experts and scholars both domestic and external propose various computer assisted IMT Measurement Algorithm.
Fuzzy C-mean algorithm (Fuzzy C Means, FCM) can effectively process medical science figure as a kind of fuzzy clustering algorithm
As present in partial volume effect phenomenon, but using standard FCM segmentation figures as when, it is possible that following shortcoming：
To the noise-sensitive in image；Cluster process does not make full use of image information, does not consider the space constraint of image；Algorithm iteration
Convergence time is longer, especially many situations of the quantity of pending pixel in image.Therefore, realized using FCM algorithms
During the computer aided measurement of the Internal-media thickness of carotid ultrasound image, need to be correspondingly improved which.In the present invention
Space in image between pixel is set up about using hidden Markov random field (Hidden Markov Random Field, HMRF)
Beam, and standard FCM algorithm is introduced, improve its segmentation performance.
The content of the invention
To overcome the deficiencies in the prior art, the present invention is intended to provide carotid ultrasound image Internal-media thickness measuring method and
System, to realize accurate measurement Internal-media thickness, and has preferable theory value and use value.What the present invention was adopted
Technical scheme is, fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measuring method, and step is,
First, extract area-of-interest
(1) image cropping：Cutting raw ultrasound is needed before extracting region of interest ROI (Region of Interesting)
Image, removes words identification part；
(2) image drop sampling, wherein decimation factor down-sampled twice to the image after cutting using bi-cubic interpolation algorithm
Respectively 2 and 4, that is, after sampling, the row, column number of image is respectively 1/2nd and a quarter of original image；
(3) the statistic sampling factor is the gray scale rule of 4 sampled images, automatic selected threshold to its binaryzation；
(4) apparent position of distal vessels wall is found, to bianry image closing operation of mathematical morphology, the hole filled up in image,
And the characteristics of be located at below ultrasonoscopy according to distal vessels wall, it is larger using connected domain area and barycenter vertical coordinate is larger
Principle, selects desired white portion, and its coboundary intersected with black region is regarded as the apparent position of LII；
(5) ROI is extracted, the apparent position with LII takes upwards, downwards certain pixel dimension as standard, respectively as
The upper and lower border of ROI, obtains two in this step and corresponds to original image, the ROI that decimation factor is 2 sampled images respectively；
2nd, image filtering
Bilateral filtering algorithm is chosen respectively to two ROI image filtering；
3rd, ROI segmentations, on the basis of standard FCM algorithm, introducing HMRF models is used to build the space constraint between pixel
.
To image f (x), it is output as based on filtering of the pel spacing from similarity and grey similarity
Wherein g (x) is filtered image,It is picture
The similarity function of plain ξ and x, wherein σ1It is domain of definition variance, σ2It is codomain variance,
For normalized function.
Introduce HMRF models to comprise the concrete steps that for the ROI that FCM algorithms split original image：
(1) obtain initial label field：Initial label field is obtained using the ROI image of 2 sampling of drop, concrete operations are as follows：It is logical
Cross peakvalue's checking thought initial cluster center is obtained from the rectangular histogram of testing image；Image is gathered for 3 with standard FCM algorithm
Class；There are many holes in the regional of the figure for now obtaining, fill up hole by closing operation of mathematical morphology；To on segmentation figure picture 2
Interpolation, and using image after interpolation as improved FCM algorithm initial label field；
(2) film segmentation in：Split blood vessel wall using FCM algorithms, wherein iteration is required for updating pixel and cluster every time
In in the heart apart from ds
ds=d2(-E1(X)-E2(X,Y))
E1(X), E2(X, Y) is the space constraint set up, and is the condition distribution with the distribution of label field X, gray scale field Y respectively
Related function.Can be calculated by condition iterative algorithm (Iterative Conditional Mode, ICM)；
(3) post processing:Due to the presence of speckle noise in image, after causing to split, on image, tube chamber location may
Some discrete zonules occur, is processed using largest connected domain criterion, removes discrete zonule, finally split
As a result；
(4) mark boundaries：According to final segmentation figure picture, using discrete first difference operator --- in Sobel operator extractions
Middle membrane boundary.
Fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measuring system, structure include：
First, area-of-interest module is extracted, is used for：
(1) image cropping, needs cutting raw ultrasound before extracting region of interest ROI (Region of Interesting)
Image, removes words identification part；
(2) image drop sampling, wherein decimation factor down-sampled twice to the image after cutting using bi-cubic interpolation algorithm
Respectively 2 and 4, that is, after sampling, the row, column number of image is respectively 1/2nd and a quarter of original image；
(3) the statistic sampling factor is the gray scale rule of 4 sampled images, automatic selected threshold to its binaryzation；
(4) apparent position of distal vessels wall is found, to bianry image closing operation of mathematical morphology, the hole filled up in image,
And the characteristics of be located at below ultrasonoscopy according to distal vessels wall, it is larger using connected domain area and barycenter vertical coordinate is larger
Principle, selects desired white portion, and its coboundary intersected with black region is regarded as the apparent position of LII；
(5) ROI is extracted, the apparent position with LII takes upwards, downwards certain pixel dimension as standard, respectively as
The upper and lower border of ROI, obtains two in this step and corresponds to original image, the ROI that decimation factor is 2 sampled images respectively；
2nd, image filtering module
Bilateral filtering algorithm is chosen respectively to two ROI image filtering；
3rd, interior middle film splits module, and introducing HMRF models is used to build the space constraint field between pixel, calculates in standard FCM
Method base
Interior middle film segmentation is carried out on plinth.
ROI splits module, is the ROI for splitting original image using improved FCM algorithms, specifically refers to, and calculates in standard FCM
On the basis of method, introducing HMRF models is used to build the space constraint field between pixel.
The characteristics of of the invention and beneficial effect are：
FCM clustering algorithms can effectively process partial volume effect phenomenon present in medical image, but FCM algorithms
Have the shortcomings that noise-sensitive, algorithmic statement is slow and loses relevant information between image pixel, therefore how to overcome standard FCM
These shortcomings of algorithm, effective segmentation figure picture, are the keys for being introduced into the measurement of carotid ultrasound image Internal-media thickness.This
Shortcoming of the design for FCM algorithms is improved using following methods：Before classification, ＆ apos to image filtering；There is provided for algorithm and align
True initial cluster center；HMRF space constraints are introduced on the basis of former algorithm, to obtain effective segmentation result.Experiment card
Bright, the algorithm proposed in the design with accurate segmentation figure picture, and can obtain accurate IMT results.With standard FCM algorithm
Compare, performance has larger lifting.In strong support clinic, the measurement of carotid ultrasound image Internal-media thickness, is IMT
The further optimized development of computer aided measurement technology provides reference, is to mend well to the mode of expert's manual measurement
Fill.
Description of the drawings：
Fig. 1 is that the flow process of the one embodiment for the carotid ultrasound image Internal-media thickness measuring method that the present invention is provided is shown
It is intended to；
Fig. 2 is the detailed process schematic diagram of the HMRF model refinement FCM algorithms that the present invention is provided；
Fig. 3 is the image after cutting；
Fig. 4 is 2 sampled images of drop；
Fig. 5 is 4 sampled images of drop；
Fig. 6 is bianry image；
Fig. 7 is original image ROI；
Fig. 8 is 2 image ROI of drop；
Fig. 9 is the filtering image of original ROI；
Figure 10 is the filtering image for dropping 2ROI；
Figure 11 is standard FCM dendrogram picture；
Figure 12 is image after Morphological scale-space；
Figure 13 is initial label field；
Figure 14 is to improve image after FCM segmentations；
Figure 15 is final segmentation figure picture；
Figure 16 is the mark boundaries of IMC.
Specific embodiment
The purpose of the present invention is, for defect present in prior art, to introduce fuzzy C-mean algorithm of the HMRF models to standard
Algorithm is improved, and with reference to the preprocessing process of image, pretreated image is split using the algorithm for improving, most
The Internal-media thickness measurement of carotid ultrasound image is realized eventually.The accurate measurement Internal-media thickness of program energy, with preferable
Theory value and use value.Implement example and description of the drawings with reference to specific, do further detailed to the present invention
Description.
First, extract area-of-interest
(1) image cropping.Cutting raw ultrasound is needed before extracting area-of-interest (Region of Interesting, ROI)
Image, removes words identification part：Under conditions of ensureing that distal vessels wall is present, the image of 320 × 400 pixel size of cutting
Region (referring to Fig. 3).
(2) image drop sampling.Down-sampled twice, decimation factor is carried out to the image after cutting using bi-cubic interpolation algorithm
Respectively 2 and 4 (the row, column number of image is respectively 1/2nd and a quarter of original image after sampling), the image after sampling
Referring to Fig. 4, Fig. 5.
(3) the gray scale rule of 4 sampled images of statistics drop, automatic selected threshold to its binaryzation.Ideally, ultrasound
The characteristics of gray scale of the tube chamber and blood vessel wall of image medium vessels presents dark, bright respectively, therefore advised according to the statistics of gradation of image
Rule, obtains gray threshold, bianry image, (referring to Fig. 6) using maximum kind differences method self adaptation.
(4) find the apparent position of distal vessels wall.The difference of ultrasonoscopy medium vessels position, the presence of speckle noise,
Bianry image is made to there may be multiple white portions.In order to find white portion corresponding with distal vessels wall, to bianry image
Closing operation of mathematical morphology, the hole filled up in image.The characteristics of according to distal vessels wall in ultrasonoscopy, lower section, using even
The logical principle that domain area is larger and barycenter vertical coordinate is larger, selects desired white portion, and it is upper which is intersected with black region
Border is regarded as the apparent position of LII.
(5) extract ROI.The arterial vascular IMT of Normal Cervical about in the range of 0.5mm to 1mm, correspondence ultrasonoscopy in about
8-16 pixel.In view of blood vessel there may be bending and pathological changes situation, we with the apparent position of LII as standard, to
It is upper, take downwards certain pixel dimension (be 40,20 respectively to original image, be 20 respectively to dropping 2 images, 10), respectively as
The upper and lower border of ROI.Can obtain in this step two respectively correspond to original image, drop 2 sampled images ROI Fig. 7 and
Fig. 8.
2nd, image filtering
Noise in ultrasonoscopy can bring appreciable impact to the extraction of marginal information, therefore the present invention chooses bilateral filtering
Algorithm reduces the speckle noise of image.Bilateral filtering algorithm is a kind of non-linear filtering method, and amount of calculation is little, and it is in medical image
Denoising in have preferably performance, and the marginal information of image can be preferably kept while noise is removed.Should
Algorithm considers the distance similarity and gray value similarity between image pixel simultaneously, and can be carved by Gauss distribution
Draw.To image f (x), it is output as based on filtering of the pel spacing from similarity and grey similarity
Wherein g (x) is filtered image,It is picture
The similarity function of plain ξ and x, wherein σ1It is domain of definition variance, σ2It is codomain variance,
For normalized function.Fig. 9, Figure 10 are the filter results of two kinds of ROI.
3rd, interior middle film segmentation
According to the imaging characteristicses of carotid artery vascular, the present invention choose FCM clustering algorithms by blood vessel wall be divided into tube chamber, it is interior in
3 class of film and adventitia, is numbered 1,2,3 respectively.Standard FCM algorithm is by minimizing pixel to cluster centre in essence
Realizing the purpose of image segmentation, the Weighted distance is otherwise known as object function Weighted distance, and its mathematic(al) representation is as follows：
In above formula, C is the clusters number for pre-setting；N is the number of pixels in image；μikIt is pixel xkPerson in servitude to the i-th class
Category degree, meets conditionM is default fuzzy factor；dik(xk,vi)=| | xk-vi| | it is pixel xk
To cluster centre viEuclidean distance.Formula (2) can be obtained by Lagrangian method minimum object function and obtain minimum
The condition of value：
Reusable (3) and formula (4), update the degree of membership of cluster centre and pixel, until meeting conditionStop calculating, algorithmic statement.That is when the difference of the degree of membership for obtaining adjacent to iterative calculation twice is less than
During default threshold value, iteration ends realize that the fuzzy clustering of image pixel is divided.
The shortcoming of image space information is lost to make up standard FCM algorithm, present invention introduces HMRF models are used to build
Space constraint field between pixel.In HMRF, commonly use two random fields to describe image to be split, one of random field is mark
Number field X, can describe its local correlations with prior distribution P (X).Label field X is Markov Random Fields, meets Markov characteristics, its
Prior distribution is represented by
Wherein S=(1 ... s } be pixel in image set, xi∈ (1 ..., q } be pixel i category label, NiIt is picture
The neighborhood of plain i.According to Hammersley-Clifford equivalence theorems, the joint probability distribution of all pixels is
Wherein Z is normaliztion constant, and U (x) is priori energy
VcX () is defined in the potential function on group c, it is only relevant with the neighborhood territory pixel of pixel.Consider calculating to disappear
Consumption and practical function, we select second order neighborhood, and estimate V with Potts modelsc(x).So with the elder generation of pseudo- likelihood close approximation
Test probability can be expressed as：
Wherein β be control neighborhood between action intensity function, ni(xi) it is pixel i neighborhood territory pixel.
Another random field Y is referred to as gray scale field or Characteristic Field, can describe its observation data with distribution function P (Y | X)
Or the distribution of characteristic vector.Characteristic Field Y refers to image to be split, and it is observable stochastic process.Image segmentation is based on pixel
Characteristic attribute and area attribute distribute the process of label to each pixel.According to Bayes theorem, its Formal Representation is such as
Under：
P(X|Y)∝P(X)P(Y|X) (9)
Here P (X | Y) is the posterior probability of image, and conditional distribution function P (Y | X), generally described with Gauss distribution：
WhereinFor the variance of same tag area pixel gray scale, μmFor the average of pixel grey scale.Chatzis points out,
The Posterior distrbutionp of HMRF models also meets the form of fuzzy division, based on this characteristic we using P (Y | X) as image to be split
Space constraint field, and introduce the distance function of FCM, in the present invention, the distance function form of improved FCM algorithm definition is：
ds=d2(-ln P(X|Y))∝d2(-ln P(X)-ln P(Y|X)) (11)
Wherein
OrderThen distance function
Finally it is defined as：
ds=d2(-E1(X)-E2(X,Y)) (14)
The FCM algorithms constrained based on HMRF realize that step is as follows：
(5) obtain initial label field.In order to realize the purpose of Fast Segmentation, obtain initial using the ROI image of 2 sampling of drop
Label field.Concrete operations are as follows：Initial cluster center is obtained by peakvalue's checking thought from the rectangular histogram of testing image；With mark
Quasi- FCM algorithms gather image for 3 classes；There are many holes in the regional of the Figure 11 for now obtaining, by closing operation of mathematical morphology
Fill up hole (referring to Figure 12)；To 2 interpolation on segmentation figure picture, and using image after interpolation as improved FCM algorithm initial label
Field (referring to Figure 13).
(6) film segmentation in.Image after down-sampled and original image have a similar intensity profile, therefore by primary segmentation
During the cluster centre that obtains as improved FCM algorithm initial cluster center.Split blood vessel using improved FCM algorithms
Wall.Wherein every time iteration be required for updating between pixel and cluster centre apart from ds, space constraint E1And E (X)2(X, Y) can be by bar
Part iterative algorithm (Iterative Conditional Mode, ICM) is calculated, and this is a kind of efficient algorithm, and to noise
There is very strong robustness.
(7) post processing.Due to the presence of speckle noise in image, after causing to split, on image, tube chamber location may
Occur some discrete zonules (referring to Figure 14), and these zonules are not it is desirable that.We are using largest connected
Domain criterion is processed, and removes discrete zonule, obtains final segmentation result Figure 15.
(8) mark boundaries.According to final segmentation figure picture, using discrete first difference operator --- in Sobel operator extractions
Middle membrane boundary.
The flow process of example of the present invention is as follows：
(1) image cropping.The words identification part around ultrasonoscopy is cropped first, is ensureing the presence of distal vessels wall
Under conditions of, intercept the image-region of 320 × 400 pixel sizes.
(2) ROI is extracted.Using the down-sampled image of bi-cubic interpolation algorithm.To dropping 4 image manipulations, distal vessels wall is found,
Obtain the ROI of the ROI and original image of 2 sampled images of drop.
(3) set up initial label field.To dropping 2ROI bilateral filterings, the impact that noise brings is reduced.By peakvalue's checking
Thought obtains initial cluster center from the rectangular histogram of testing image.Standard FCM image is adopted to gather filtered image for pipe
Chamber, 3 class of interior middle film and adventitia.In order that initial label field is more complete, morphology operations are used to sorted image, and to place
On image after reason 2 sampling, using result as improved FCM algorithm initial label field.
(4) film segmentation in.To original ROI bilateral filterings, the impact that noise brings is reduced.By what is obtained after initial segmentation
Initial cluster center of the cluster centre as subsequent cutting procedure.Using the FCM algorithm segmentation figure pictures of HMRF model refinements.Remove
Discrete zonule in classification results figure.Obtain final segmentation result.
(5) post processing.According to segmentation figure picture, interior middle membrane boundary is obtained using Sobel operators, IMT is subsequently calculated.
Claims (5)
1. a kind of fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measuring method, is characterized in that, step is as follows：
First, extract area-of-interest
(1) image cropping：Cutting raw ultrasound image is needed before extracting region of interest ROI (Region of Interesting),
Remove words identification part；
(2) image drop sampling, down-sampled twice to the image after cutting using bi-cubic interpolation algorithm, wherein decimation factor is distinguished
For 2 and 4, that is, after sampling, the row, column number of image is respectively 1/2nd and a quarter of original image；
(3) the statistic sampling factor is the gray scale rule of 4 sampled images, automatic selected threshold to its binaryzation；
(4) apparent position of distal vessels wall, to bianry image closing operation of mathematical morphology, the hole filled up in image, and root are found
The characteristics of being located at below ultrasonoscopy according to distal vessels wall, using the original that connected domain area is larger and barycenter vertical coordinate is larger
Then, desired white portion is selected, its coboundary intersected with black region is regarded as the apparent position of LII；
(5) ROI is extracted, the apparent position with LII takes upwards, downwards certain pixel dimension, respectively as ROI's as standard
Upper and lower border, obtains two in this step and corresponds to original image, the ROI that decimation factor is 2 sampled images respectively；
2nd, image filtering
Bilateral filtering algorithm is chosen respectively to two ROI image filtering；
3rd, ROI segmentations, on the basis of standard FCM algorithm, introducing HMRF models is used to build the space constraint field between pixel.
2. fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measuring method as claimed in claim 1, is characterized in that, right
Image f (x), is output as based on filtering of the pel spacing from similarity and grey similarity：
Wherein g (x) is filtered image,It is pixel
The similarity function of ξ and x, wherein σ1It is domain of definition variance, σ2It is codomain variance,To return
One changes function.
3. fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measuring method as claimed in claim 1, is characterized in that, draw
Enter HMRF models for FCM algorithms split original image ROI comprise the concrete steps that：
(1) obtain initial label field：Initial label field is obtained using the ROI image of 2 sampling of drop, concrete operations are as follows：By peak
Value detection thought obtains initial cluster center from the rectangular histogram of testing image；Image is gathered for 3 classes with standard FCM algorithm；This
When the regional of figure that obtains there are many holes, hole is filled up by closing operation of mathematical morphology；To 2 interpolation on segmentation figure picture,
And using image after interpolation as improved FCM algorithm initial label field；
(2) film segmentation in：Split blood vessel wall using FCM algorithms, wherein iteration is required for updating pixel and cluster centre every time
Between apart from ds
ds=d2(-E1(X)-E2(X,Y))
E1(X), E2(X, Y) is the space constraint set up, and is related to the condition distribution of the distribution of label field X, gray scale field Y respectively
Function.Can be calculated by condition iterative algorithm (Iterative Conditional Mode, ICM)；
(3) post processing:Due to the presence of speckle noise in image, after causing segmentation, on image, tube chamber location may go out
Existing some discrete zonules, are processed using largest connected domain criterion, are removed discrete zonule, obtain final segmentation knot
Really；
(4) mark boundaries：According to final segmentation figure picture, using discrete first difference operator --- film in Sobel operator extractions
Border.
4. a kind of fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measuring system, is characterized in that, by what is arranged on computer
As lower module is realized：
First, area-of-interest module is extracted, is used for：
(1) image cropping, needs cutting raw ultrasound image before extracting region of interest ROI (Region of Interesting),
Remove words identification part；
(2) image drop sampling, down-sampled twice to the image after cutting using bi-cubic interpolation algorithm, wherein decimation factor is distinguished
For 2 and 4, that is, after sampling, the row, column number of image is respectively 1/2nd and a quarter of original image；
(3) the statistic sampling factor is the gray scale rule of 4 sampled images, automatic selected threshold to its binaryzation；
(4) apparent position of distal vessels wall, to bianry image closing operation of mathematical morphology, the hole filled up in image, and root are found
The characteristics of being located at below ultrasonoscopy according to distal vessels wall, using the original that connected domain area is larger and barycenter vertical coordinate is larger
Then, desired white portion is selected, its coboundary intersected with black region is regarded as the apparent position of LII；
(5) ROI is extracted, the apparent position with LII takes upwards, downwards certain pixel dimension, respectively as ROI's as standard
Upper and lower border, obtains two in this step and corresponds to original image, the ROI that decimation factor is 2 sampled images respectively；
2nd, image filtering module
Bilateral filtering algorithm is chosen respectively to two ROI image filtering；
3rd, interior middle film splits module, and introducing HMRF models is used to build the space constraint field between pixel, in standard FCM algorithm base
Interior middle film segmentation is carried out on plinth.
5. fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measuring system as claimed in claim 4, is characterized in that, ROI
Segmentation module, is the ROI for being split original image using improved FCM algorithms, specifically refers to, on the basis of standard FCM algorithm, draw
Enter the space constraint field that HMRF models are used to build between pixel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610951289.4A CN106570871B (en) | 2016-11-02 | 2016-11-02 | Fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measurement method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610951289.4A CN106570871B (en) | 2016-11-02 | 2016-11-02 | Fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measurement method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106570871A true CN106570871A (en) | 2017-04-19 |
| CN106570871B CN106570871B (en) | 2019-10-18 |
Family
ID=58535029
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610951289.4A Expired - Fee Related CN106570871B (en) | 2016-11-02 | 2016-11-02 | Fuzzy C-mean algorithm carotid ultrasound image Internal-media thickness measurement method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106570871B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108182436A (en) * | 2017-12-29 | 2018-06-19 | 辽宁工程技术大学 | A kind of method for segmentation of high resolution remote sensing image |
| CN108230341A (en) * | 2018-03-07 | 2018-06-29 | 汕头大学 | A kind of eye fundus image blood vessel segmentation method that nomography is scratched based on layering |
| CN108492272A (en) * | 2018-03-26 | 2018-09-04 | 西安交通大学 | Cardiovascular vulnerable plaque recognition methods based on attention model and multitask neural network and system |
| CN110288604A (en) * | 2019-06-12 | 2019-09-27 | 三峡大学 | Image partition method and device based on K-means |
| CN110517263A (en) * | 2019-09-02 | 2019-11-29 | 青岛海信医疗设备股份有限公司 | Determine the method, apparatus and storage medium of Internal-media thickness |
| CN111192255A (en) * | 2019-12-30 | 2020-05-22 | 上海联影智能医疗科技有限公司 | Index detection method, computer device, and storage medium |
| CN111833348A (en) * | 2020-08-10 | 2020-10-27 | 上海工程技术大学 | Automatic detection method for vascular sediments based on image processing |
| CN113951930A (en) * | 2021-09-16 | 2022-01-21 | 李世岩 | Three-dimensional neck ultrasonic automatic scanning and evaluation system and method |
| CN117333529A (en) * | 2023-11-30 | 2024-01-02 | 之江实验室 | Template matching-based vascular ultrasonic intima automatic measurement method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102469987A (en) * | 2010-02-10 | 2012-05-23 | 松下电器产业株式会社 | Ultrasonic diagnostic device, and method for measuring intima-media complex thickness |
| 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 |
| CN103971363A (en) * | 2014-03-31 | 2014-08-06 | 南京信息工程大学 | Hierarchical fuzzy C-means based image segmenting method |
| CN104665872A (en) * | 2014-12-29 | 2015-06-03 | 深圳开立生物医疗科技股份有限公司 | Ultrasonic image-based carotid intima-media thickness measuring method and device |
| WO2015130231A1 (en) * | 2014-02-27 | 2015-09-03 | Agency For Science, Technology And Research | Segmentation of cardiac magnetic resonance (cmr) images using a memory persistence approach |
-
2016
- 2016-11-02 CN CN201610951289.4A patent/CN106570871B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102469987A (en) * | 2010-02-10 | 2012-05-23 | 松下电器产业株式会社 | Ultrasonic diagnostic device, and method for measuring intima-media complex thickness |
| 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 |
| WO2015130231A1 (en) * | 2014-02-27 | 2015-09-03 | Agency For Science, Technology And Research | Segmentation of cardiac magnetic resonance (cmr) images using a memory persistence approach |
| CN103971363A (en) * | 2014-03-31 | 2014-08-06 | 南京信息工程大学 | Hierarchical fuzzy C-means based image segmenting method |
| CN104665872A (en) * | 2014-12-29 | 2015-06-03 | 深圳开立生物医疗科技股份有限公司 | Ultrasonic image-based carotid intima-media thickness measuring method and device |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108182436A (en) * | 2017-12-29 | 2018-06-19 | 辽宁工程技术大学 | A kind of method for segmentation of high resolution remote sensing image |
| CN108230341A (en) * | 2018-03-07 | 2018-06-29 | 汕头大学 | A kind of eye fundus image blood vessel segmentation method that nomography is scratched based on layering |
| CN108230341B (en) * | 2018-03-07 | 2021-12-17 | 汕头大学 | Eyeground image blood vessel segmentation method based on layered matting algorithm |
| CN108492272A (en) * | 2018-03-26 | 2018-09-04 | 西安交通大学 | Cardiovascular vulnerable plaque recognition methods based on attention model and multitask neural network and system |
| CN108492272B (en) * | 2018-03-26 | 2021-01-19 | 西安交通大学 | Cardiovascular vulnerable plaque identification method and system based on attention model and multitask neural network |
| CN110288604B (en) * | 2019-06-12 | 2022-03-22 | 三峡大学 | Image segmentation method and device based on K-means |
| CN110288604A (en) * | 2019-06-12 | 2019-09-27 | 三峡大学 | Image partition method and device based on K-means |
| CN110517263A (en) * | 2019-09-02 | 2019-11-29 | 青岛海信医疗设备股份有限公司 | Determine the method, apparatus and storage medium of Internal-media thickness |
| CN111192255A (en) * | 2019-12-30 | 2020-05-22 | 上海联影智能医疗科技有限公司 | Index detection method, computer device, and storage medium |
| CN111192255B (en) * | 2019-12-30 | 2024-04-26 | 上海联影智能医疗科技有限公司 | Index detection method, computer device, and storage medium |
| CN111833348A (en) * | 2020-08-10 | 2020-10-27 | 上海工程技术大学 | Automatic detection method for vascular sediments based on image processing |
| CN113951930B (en) * | 2021-09-16 | 2023-11-28 | 杭州影想未来科技有限公司 | Three-dimensional neck ultrasonic automatic scanning and evaluating system and method |
| CN113951930A (en) * | 2021-09-16 | 2022-01-21 | 李世岩 | Three-dimensional neck ultrasonic automatic scanning and evaluation system and method |
| CN117333529A (en) * | 2023-11-30 | 2024-01-02 | 之江实验室 | Template matching-based vascular ultrasonic intima automatic measurement method and system |
| CN117333529B (en) * | 2023-11-30 | 2024-03-15 | 之江实验室 | Template matching-based vascular ultrasonic intima automatic measurement method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106570871B (en) | 2019-10-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106570871A (en) | Fuzzy C mean value carotid ultrasonic image intima-media thickness measuring method and system | |
| Chen et al. | A review of thyroid gland segmentation and thyroid nodule segmentation methods for medical ultrasound images | |
| EP2378978B1 (en) | Method and system for automated generation of surface models in medical images | |
| JP6570145B2 (en) | Method, program, and method and apparatus for constructing alternative projections for processing images | |
| CN104794708A (en) | Atherosclerosis plaque composition dividing method based on multi-feature learning | |
| WO2022105623A1 (en) | Intracranial vascular focus recognition method based on transfer learning | |
| CN113012173A (en) | Heart segmentation model and pathology classification model training, heart segmentation and pathology classification method and device based on cardiac MRI | |
| CN112562058B (en) | Method for quickly establishing intracranial vascular simulation three-dimensional model based on transfer learning | |
| CN106485203A (en) | Carotid ultrasound image Internal-media thickness measuring method and system | |
| Kumar et al. | Image processing in diabetic related causes | |
| Qiu et al. | Lung Nodules Detection in CT Images Using Gestalt‐Based Algorithm | |
| Guo et al. | A novel myocardium segmentation approach based on neutrosophic active contour model | |
| Rocha et al. | Automatic detection of the carotid lumen axis in B-mode ultrasound images | |
| CN113066054B (en) | Cervical OCT image feature visualization method for computer-aided diagnosis | |
| CN114170151A (en) | Intracranial vascular lesion identification method based on transfer learning | |
| Xia et al. | Automatic plane of minimal hiatal dimensions extraction from 3D female pelvic floor ultrasound | |
| CN111986216B (en) | RSG liver CT image interactive segmentation algorithm based on neural network improvement | |
| CN112509076A (en) | Intracranial vascular lesion marking and three-dimensional display system based on intelligent medical treatment | |
| CN111292285B (en) | Automatic screening method for diabetes mellitus based on naive Bayes and support vector machine | |
| Hemalatha et al. | Segmentation of 2D and 3D images of carotid artery on unified technology learning platform | |
| CN113409275B (en) | Method for determining thickness of transparent layer behind fetal neck based on ultrasonic image and related device | |
| Suganya et al. | Classification of liver diseases from ultrasound images using a hybrid kohonen SOM and LPND speckle reduction method | |
| CN112862787B (en) | CTA image data processing method, device and storage medium | |
| CN114170337A (en) | Method for establishing intracranial vascular enhancement three-dimensional model based on transfer learning | |
| Bharath et al. | Multi-modal framework for automatic detection of diagnostically important regions in nonalcoholic fatty liver ultrasonic images |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20191018 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |