CN110400313A - A kind of the soft tissue separation method and separation system of nuclear magnetic resonance image - Google Patents
A kind of the soft tissue separation method and separation system of nuclear magnetic resonance image Download PDFInfo
- Publication number
- CN110400313A CN110400313A CN201910708072.4A CN201910708072A CN110400313A CN 110400313 A CN110400313 A CN 110400313A CN 201910708072 A CN201910708072 A CN 201910708072A CN 110400313 A CN110400313 A CN 110400313A
- Authority
- CN
- China
- Prior art keywords
- pixel
- tissue
- gray level
- image
- level image
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides the soft tissue separation methods and separation system of a kind of nuclear magnetic resonance image, solve the technical problem that the lesser soft tissue preciseness of separation of posture is low in conventional images separation process.Method includes: that the class categories of pixel in the gray level image are determined by the random fern disaggregated model of gray level image；Correlation between the pixel is determined according to the class categories of the pixel, forms tissue contours；According to the stereo profile of the tissue contours position and tissue contours variation tendency formation tissue in the adjacent gray level image.The automatic processing and classification of associated pixel are formed using relevant image and information characteristics, accuracy is promoted.It verifies to form classification check and Vector contour description using contour generating method, so that image quantization is regular, pixel intension information and modeling framework effective integration, soft tissue realization body in gray level image is divided automatically, profile be accurately positioned and automatic three-dimensional modeling, effectively increase recognition efficiency.
Description
Technical field
The present invention relates to medical image recognition technical fields, and in particular to a kind of soft tissue separation side of nuclear magnetic resonance image
Method and separation system.
Background technique
For smooth muscle, striated muscle, blood vessel, ligament and lymphatic vessel etc. have the soft tissue of diversity but finite volume by
The imaging gray level resolution influence of MRI (Magnetic Resonance Imaging, Magnetic resonance imaging) is often unintelligible, respectively
Edge gray areas between tissue combines, and image segmentation between tissue is caused to have difficulties.Such as normal soft tissue with it is normal soft
Early stage benign tumour between tissue.
In the prior art, medical image it is usually necessary to use manual mode to various texture patterns carry out selective editing,
Defect makes up processing, artifact and tedious data separating, and it is complete to establish then to generate segmentation result with the method that region increases
Digital model.Operator's a large amount of time can be exhausted in this way, and when data to be treated are more, professional resources can not
Meet timeliness.In the prior art using the determination object bounds problem in random forests algorithm processing image, but due to random
The processing feature of forest is based on reflection local gray level or textural characteristics, can not effectively be directed in image pixel in larger range
Between Attribute Association information identification, be unfavorable for the determination of RANDOM BOUNDARY.
Summary of the invention
In view of the above problems, the embodiment of the present invention provides the soft tissue separation method and segregative line of a kind of nuclear magnetic resonance image
System solves the technical problem that the lesser soft tissue preciseness of separation of posture is low in conventional images separation process.
The soft tissue separation method of the nuclear magnetic resonance image of the embodiment of the present invention, comprising:
The class categories of pixel in the gray level image are determined by the random fern disaggregated model of gray level image；
Correlation between the pixel is determined according to the class categories of the pixel, forms tissue contours；
According in the adjacent gray level image the tissue contours position and the tissue contours variation tendency formation group
The stereo profile knitted.
It is described that picture in the gray level image is determined by the random fern disaggregated model of gray level image in one embodiment of the invention
The class categories of element include random fern disaggregated model forming process:
Determine that two-value test is to as training set in standard picture block region；
Form the random fern model of pixel classifications；
The class probability that classification obtains fern feature is carried out to the random fern model using the training set.
It is described that picture in the gray level image is determined by the random fern disaggregated model of gray level image in one embodiment of the invention
Element class categories include:
Pixel classifications process:
Sequence calculates the binary feature of each pixel in the gray level image；
The binary feature of single pixel is combined to form binary sequence；
The single pixel classifications classification is determined compared with the posterior probability of character pair in the random fern model.
In one embodiment of the invention, the class categories according to the pixel determine correlation between the pixel,
Forming tissue contours includes:
Other adjacent pixels of each pixel are determined in the gray level image, form the related pixel collection of each pixel
It closes；
An initial pixel is randomly selected, determines that classification is consistent similar in the related pixel set of initial pixel
Other pixels；
Other consistent similar pixel conducts of classification are determined in the related pixel set of other similar pixels
Other new similar pixels；
It repeats other described similar pixels to determine to the classification of other similar pixels, forms pixel recurrence and identified
Journey；
Compared in all generic pixels of acquisition by position and obtains preliminary tissue contours；
It excludes the pixel that the preliminary tissue contours include and returns identification process by forming profile, obtain all described
Preliminary tissue contours；
The exclusive PCR profile in all preliminary tissue contours of formation, determines effective institute in the gray level image
State tissue contours.
In one embodiment of the invention, the profile returns identification process and includes:
An initial pixel is randomly selected, determines that classification is consistent similar in the related pixel set of initial pixel
Other pixels；
Other consistent similar pixel conducts of classification are determined in the related pixel set of other similar pixels
Other new similar pixels；
It repeats other described similar pixels to determine to the classification of other similar pixels, forms pixel recurrence and identified
Journey；
Compared in all generic pixels of acquisition by position and obtains preliminary tissue contours；
Exclude the pixel that the preliminary tissue contours include.
In one embodiment of the invention, the interference profile includes the multi-thread profile of dot profile, single line profile, single intersection point.
In one embodiment of the invention, the tissue contours position according in the adjacent gray level image and described group
Driving wheel exterior feature variation tendency formed tissue stereo profile include:
Establish the relative seat feature between tissue contours described in every gray level image；
Institute is formed according to the variation tendency of the relative seat feature between each tissue contours of neighboring gradation image
State the fitting coefficient between neighboring gradation image；
In conjunction in gray level image the tissue contours and the fitting coefficient form the stereo profile of each tissue.
The soft tissue separation system of the nuclear magnetic resonance image of the embodiment of the present invention, comprising:
Memory, for storing the soft tissue separation method treatment process corresponding program generation of above-mentioned nuclear magnetic resonance image
Code；
Processor, for executing said program code.
The soft tissue separation system of the nuclear magnetic resonance image of the embodiment of the present invention, comprising:
Pixel classifications device determines pixel in the gray level image for the random fern disaggregated model by gray level image
Class categories；
Face profile determining device determines correlation between the pixel for the class categories according to the pixel,
Form tissue contours；
Stereo profile determining device, for according in the adjacent gray level image the tissue contours position and described group
The stereo profile of driving wheel exterior feature variation tendency formation tissue.
The soft tissue separation method and system of the nuclear magnetic resonance image of the embodiment of the present invention utilize relevant image and information
Feature forms the automatic processing and classification of associated pixel, and accuracy is promoted.It verifies to form classification school using contour generating method
Test with Vector contour describe so that image quantization is regular, pixel intension information and model framework effective integration so that soft tissue
Can in gray level image realization body divide automatically, profile be accurately positioned and automatic three-dimensional modeling, effectively increase professional people
The recognition efficiency of power identification resource.
Detailed description of the invention
Fig. 1 show the flow diagram of the soft tissue separation method of one embodiment of the invention nuclear magnetic resonance image.
Fig. 2 show random fern disaggregated model in the soft tissue separation method of one embodiment of the invention nuclear magnetic resonance image
Establishing process schematic diagram.
Fig. 3 show the stream that tissue contours are determined in the soft tissue separation method of one embodiment of the invention nuclear magnetic resonance image
Journey schematic diagram.
Fig. 4, which is shown in the soft tissue separation method of one embodiment of the invention nuclear magnetic resonance image, forms showing for stereo profile
It is intended to.
Fig. 5 show the configuration diagram of the soft tissue separation system of one embodiment of the invention nuclear magnetic resonance image.
Specific embodiment
To be clearer and more clear the objectives, technical solutions, and advantages of the present invention, below in conjunction with attached drawing and specific embodiment party
The invention will be further described for formula.Obviously, described embodiments are only a part of the embodiments of the present invention, rather than all
Embodiment.Based on the embodiments of the present invention, those of ordinary skill in the art institute without creative efforts
The every other embodiment obtained, shall fall within the protection scope of the present invention.
The soft tissue separation method of the nuclear magnetic resonance image of one embodiment of the invention is as shown in Figure 1.In Fig. 1, this implementation
Example include:
Step 100: the class categories of pixel in gray level image are determined by the random fern disaggregated model of gray level image.
It will be understood by those skilled in the art that utilizing semi-naive Bayes principle (semi- in random fern sorting algorithm
Naive Bayes), form the decision tree structure of more flattening.Therefore be conducive to reinforce pixel using random fern sorting algorithm
Between feature association, the variation of texture between prominent pixel.
Step 200: determining correlation between pixel according to the class categories of pixel, form tissue contours.
Exact outline is done to sorted pixel using cluster and Extension algorithm to be formed and verified, and excludes error in classification.
Step 300: according in neighboring gradation image tissue contours position and tissue contours variation tendencies form tissue
Stereo profile.
It will be understood by those skilled in the art that can be according to right in parallel cutting surfaces quantity and section during three-dimensional modeling
The profile of elephant forms the three-D profile of same target, can be improved three by the variation tendency of affiliated partner in quantization different section
Tie up the smoothness and precision of profile.
The soft tissue separation method of nuclear magnetic resonance image of the embodiment of the present invention is formed using relevant image and information characteristics
The automatic processing and classification of associated pixel, accuracy are promoted.It verifies to form classification check and vector using contour generating method
Change profile description so that image quantization is regular, pixel intension information and model framework effective integration, allow soft tissue in ash
Realization body is divided automatically in degree image, profile is accurately positioned and automatic three-dimensional modeling, effectively increases professional manpower identification money
The recognition efficiency in source.
Formation such as Fig. 2 institute of random fern model in the soft tissue separation method of the nuclear magnetic resonance image of one embodiment of the invention
Show.In Fig. 2, the present embodiment includes:
Random fern disaggregated model forming process:
Step 110: determining that two-value test is to as training set in standard picture block region.
Fern feature used in random fern is tested by a series of two-value in relatively large image block areas to (two-value
Feature) it forms, these simple binary features connect each other, can be very good that the texture of this image-region, structure is depicted
Information, particularly with the determination on the object boundary being difficult to differentiate between.In a piece of image-region, two defined in two-value test
Value tag is as follows:
Ii1And Ii2For any two pixel in current block (patch).Due to binary feature very simple, for essence
Really description image information needs use a large amount of such features.Correlation between these features is also required to be used.If one
A a total of N number of binary feature of patch, is equally divided into M group, every group of size is S=N/M.These feature groups are exactly fern spy
Sign.
Step 120: forming the random fern model of pixel classifications.
The expression formula that pixel v classifies to it using fern feature by random fern can be write as:
Wherein, Fm={ fσ(m,1),fσ(m,2),…,fσ(m,S),, m=1,2 ..., M are m-th of fern characteristic, and σ (m, j) indicates model
Enclose the random sequence from 1 to N.That is, FmIt is to extract S from N number of binary feature and constitute fern feature without putting back to.Fern
Binary feature sequence in feature has fixed sequence, once it is determined that no longer changing.Such characteristic Design is by half simple shellfish
This principle (semi-naive Bayesian) of leaf is determined that this is also meaned that, in these binary features between fern feature
It is no longer mutually indepedent, but still mutually indepedent in each fern feature.In order to improve storage efficiency and computational efficiency, in fern feature
Binary feature sequence can decimally express.Standard and original image are inputed to random fern together to be trained.It trained
Journey needs to estimate each fern feature FmIn every one kind CkUnder the conditions of conditional probability P (Fm| C=Ci).For each fern feature Fm, In
Here remember its conditional probability:
K in formula is determined that k is that (binary sequence is actually a string to binary sequence by the binary sequence in a fern feature
Binary digit) the decimal system expression, then a total of K=2 of ksKind possibility.For each possibility, it will be calculated
ProbabilityAnd there is following constraint:
Step 130: the class probability that classification obtains fern feature being carried out to random fern model using training set.
Class probabilityIt can be denoted as
Classification is c in training samplei, while fern characteristic value is the number of samples of k,Be then classification be ciSample
This total number.AllIt can Computing acquisition.
Random fern category of model such as Fig. 2 is utilized in the soft tissue separation method of the nuclear magnetic resonance image of one embodiment of the invention
It is shown.In Fig. 2, the present embodiment includes:
Pixel classifications process:
Step 140: sequence calculates the binary feature of each pixel in gray level image.
Step 150: the binary feature of pixel is combined to form binary sequence.
Binary feature sequence in fern feature can be expressed decimally.
Step 160: pixel classifications classification is determined compared with the posterior probability of character pair in random fern model.
Obtained decimal code can be obtained to point of pixel to be measured compared with the posterior probability that random fern model training obtains
Class classification.
The soft tissue separation method of the nuclear magnetic resonance image of the embodiment of the present invention is special using the posterior probability of random fern model
Property handles pixel classifications forms binary feature correlation in different fern features, improves the correlation between pixel classifications
Property.So that pixel classifications are more suitable for the description texture of image-region, structural information, for the object boundary that is difficult to differentiate between
Determination has very great help.
Determine that tissue contours are as shown in Figure 3 in the soft tissue separation method of the nuclear magnetic resonance image of one embodiment of the invention.
In Fig. 3, the present embodiment includes:
Step 210: determining other adjacent pixels of each pixel in gray level image, form the related pixel of each pixel
Set.
Each pixel includes at least 3 other adjacent pixels (side or angle), includes up to 8 other adjacent pixels
(middle part).
Step 220: randomly selecting an initial pixel, determine that classification is consistent in the related pixel set of initial pixel
Other similar pixels.
The selection of one initial pixel can guarantee that outline identification process has necessary confirmation filtering sequence, meet profile
Return the initial procedure of identification.Other similar pixels are played as the extension of initial pixel the same category to be expanded and quantifies just
The effect of beginning pixel profile.
Step 230: other consistent similar pixel conducts of classification are determined in the related pixel set of other similar pixels
Other new similar pixels.
Using other consistent similar pixels of classification in the related pixel set of other similar pixels be equivalent to it is similar other
Pixel, so that the pixel of constantly extension has classification continuity.
Step 240: repeating other similar pixels and determine that forming pixel returns identification process to the classification of other similar pixels.
The cluster that identification process forms generic pixel in a certain part of gray level image is returned by pixel.
Step 250: being compared in all generic pixels of acquisition by position and obtain preliminary tissue contours.
The position of generic pixel is compared the Vector Message formed in coordinate space with coordinate origin by pixel and is identified,
Preliminary tissue contours are obtained using the variation in length and direction.
Step 260: excluding the pixel that preliminary tissue contours include and form profile recurrence identification process, obtain all preliminary
Tissue contours.Profile returns identification process and repeats step 220 to step 260 formation by sequence.
In gray level image, the pixel that the preliminary tissue contours of identification include gradually is excluded, until all pixels belong to certain
A preliminary tissue contours.
Step 270: the exclusive PCR profile in all preliminary tissue contours of formation determines in gray level image effective group
Driving wheel is wide.
Interference profile includes the multi-thread profile etc. of dot profile, single line profile, single intersection point, according to random fern disaggregated model
It is verified using truthful data, these profiles will not be formed by live tissue.
The soft tissue separation method of the nuclear magnetic resonance image of the embodiment of the present invention can automate to form tissue contours, avoid
Tissue contours interference information.Meanwhile it can be in outline identification mistake using pixel recurrence identification process and profile recurrence identification process
The vector description information that profile is extracted in journey is conducive to the position and the form that determine corresponding contour.
It is as shown in Figure 4 that stereo profile is formed in the soft tissue separation method of the nuclear magnetic resonance image of one embodiment of the invention.
In Fig. 4, the present embodiment includes:
Step 310: establishing the relative seat feature in every gray level image between tissue contours.
There is determining relative positional relationship, relative positional relationship includes each tissue between each tissue contours in gray level image
It determines immediate position and the spacing between shape, each tissue contours, the relative seat feature to form quantization is described using vector.
Step 320: phase is formed according to the variation tendency of relative seat feature between each tissue contours of neighboring gradation image
Fitting coefficient between adjacent gray level image.
The face profile of identical tissue between neighboring gradation image has similitude, and same group is woven between neighboring gradation image
Variation it is limited, the detailed vector parameters that opposite variation occurs for relative seat feature between neighboring gradation image can be obtained, in turn
Fitting coefficient is respectively organized between formation neighboring gradation image.
Step 330: in conjunction in gray level image tissue contours and fitting coefficient form the stereo profile of each tissue.
The solid of soft tissue is formed using the fitting coefficient in each gray level image between relative seat feature and each gray level image
Profile completes the segmentation of soft tissue.
Soft tissue is separated into two-dimensional silhouette and three dimensional object by the soft tissue separation method of nuclear magnetic resonance image, is realized point
Automation and organization object are cut, the professional manpower level of resources utilization and observation dimension are substantially increased.
The soft tissue separation system of the nuclear magnetic resonance image of one embodiment of the invention, comprising:
Memory, for storing the soft tissue separation method treatment process corresponding program generation of above-mentioned nuclear magnetic resonance image
Code；
Processor, for executing the soft tissue separation method treatment process corresponding program generation of above-mentioned nuclear magnetic resonance image
Code.
Processor can use DSP (Digital Signal Processing) digital signal processor, FPGA
(Field-Programmable Gate Array) field programmable gate array, MCU (Microcontroller Unit) system
Plate, SoC (system on a chip) system board or PLC (Programmable Logic Controller) including I/O are most
Mini system.
The soft tissue separation system of the nuclear magnetic resonance image of one embodiment of the invention is as shown in Figure 5.In Fig. 5, this implementation
Example include:
Pixel classifications device 1100, for determining pixel in gray level image by the random fern disaggregated model of gray level image
Class categories；
Face profile determining device 1200 determines correlation between pixel for the class categories according to pixel, forms tissue
Profile；
Stereo profile determining device 1300, for according in neighboring gradation image tissue contours position and tissue contours become
The stereo profile of change trend formation tissue.
As shown in figure 5, in one embodiment of the invention, pixel classifications device 1100 includes:
Training set forms module 1110, for determining that two-value test is to as training set in standard picture block region；
Model forms module 1120, is used to form the random fern model of pixel classifications；
Probability forms module 1130, and the classification for carrying out classification acquisition fern feature to random fern model using training set is general
Rate.
As shown in figure 5, in one embodiment of the invention, pixel classifications device 1100 includes:
Two-value forms module 1140, for sequentially calculating the binary feature of each pixel in gray level image；
Sequence forms module 1150, is combined to form binary sequence for the binary feature to the pixel；
Match stop module 1160, for determining classification class compared with the posterior probability of character pair in random fern model
Not.
As shown in figure 5, in one embodiment of the invention, face profile determining device 1200 includes:
Related pixel forms module 1210, for determining other adjacent pixels of each pixel in gray level image, is formed
The related pixel set of each pixel；
Pixel comparison module 1220, for randomly selecting an initial pixel, in the related pixel set of initial pixel
Determine other consistent similar pixels of classification；
Pixel extension of module 1230, for determining that classification is consistent similar in the related pixel set of other similar pixels
Other pixels are as other new similar pixels；
Extend comparison module 1240, forms pixel to the determination of the classification of other similar pixels for repeating other similar pixels
Return identification process；
Preliminary profile forms module 1250, preliminary for comparing acquisition by position in all generic pixels of acquisition
Tissue contours；
Preliminary profile filtering module 1260, for excluding pixel that preliminary tissue contours include and repeating step by sequence
220, which form profile to step 260, returns identification process, obtains all preliminary tissue contours；
Filtering module 1270 is interfered to determine gray scale for the exclusive PCR profile in all preliminary tissue contours of formation
Effective tissue contours in image.
As shown in figure 5, in one embodiment of the invention, stereo profile determining device 1300 includes:
Position feature forms module 1310, for establishing the relative seat feature in every gray level image between tissue contours；
Fitting coefficient forms module 1320, for relative seat feature between each tissue contours according to neighboring gradation image
Variation tendency formed neighboring gradation image between fitting coefficient；
Constructing module 1330, for combining tissue contours in gray level image and fitting coefficient to form the stereoscopic wheel of each tissue
It is wide.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by anyone skilled in the art,
It should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with the protection model of claims
Subject to enclosing.
Claims (9)
1. a kind of soft tissue separation method of nuclear magnetic resonance image characterized by comprising
The class categories of pixel in the gray level image are determined by the random fern disaggregated model of gray level image；
Correlation between the pixel is determined according to the class categories of the pixel, forms tissue contours；
According in the adjacent gray level image the tissue contours position and the tissue contours variation tendency form tissue
Stereo profile.
2. the soft tissue separation method of nuclear magnetic resonance image as described in claim 1, which is characterized in that described to pass through grayscale image
The random fern disaggregated model of picture determines that the class categories of pixel in the gray level image include random fern disaggregated model forming process:
Determine that two-value test is to as training set in standard picture block region；
Form the random fern model of pixel classifications；
The class probability that classification obtains fern feature is carried out to the random fern model using the training set.
3. the soft tissue separation method of nuclear magnetic resonance image as described in claim 1, which is characterized in that described to pass through grayscale image
The random fern disaggregated model of picture determines that the class categories of pixel in the gray level image include:
Pixel classifications process:
Sequence calculates the binary feature of each pixel in the gray level image；
The binary feature of single pixel is combined to form binary sequence；
The single pixel classifications classification is determined compared with the posterior probability of character pair in the random fern model.
4. the soft tissue separation method of nuclear magnetic resonance image as described in claim 1, which is characterized in that described according to the picture
The class categories of element determine correlation between the pixel, form tissue contours and include:
Other adjacent pixels of each pixel are determined in the gray level image, form the related pixel set of each pixel；
Randomly select an initial pixel, in the related pixel set of initial pixel determine classification it is consistent it is similar other
Pixel；
In the related pixel set of other similar pixels determine other consistent similar pixels of classification as newly
Other similar pixels；
It repeats other described similar pixels to determine to the classification of other similar pixels, forms pixel and return identification process；
Compared in all generic pixels of acquisition by position and obtains preliminary tissue contours；
It excludes the pixel that the preliminary tissue contours include and returns identification process by forming profile, obtain all described preliminary
Tissue contours；
The exclusive PCR profile in all preliminary tissue contours of formation determines described group effective in the gray level image
Driving wheel is wide.
5. the soft tissue separation method of nuclear magnetic resonance image as claimed in claim 4, which is characterized in that the profile, which returns, to be known
Other process includes:
Randomly select an initial pixel, in the related pixel set of initial pixel determine classification it is consistent it is similar other
Pixel；
In the related pixel set of other similar pixels determine other consistent similar pixels of classification as newly
Other similar pixels；
It repeats other described similar pixels to determine to the classification of other similar pixels, forms pixel and return identification process；
Compared in all generic pixels of acquisition by position and obtains preliminary tissue contours；
Exclude the pixel that the preliminary tissue contours include.
6. the soft tissue separation method of nuclear magnetic resonance image as claimed in claim 4, which is characterized in that the interference profile packet
Include the multi-thread profile of dot profile, single line profile, single intersection point.
7. the soft tissue separation method of nuclear magnetic resonance image as described in claim 1, which is characterized in that described according to adjacent institute
It states the tissue contours position and the tissue contours variation tendency in gray level image and forms the stereo profile of tissue and include:
Establish the relative seat feature between tissue contours described in every gray level image；
The phase is formed according to the variation tendency of the relative seat feature between each tissue contours of neighboring gradation image
Fitting coefficient between adjacent gray level image；
In conjunction in gray level image the tissue contours and the fitting coefficient form the stereo profile of each tissue.
8. a kind of soft tissue separation system of nuclear magnetic resonance image, comprising:
Memory, the soft tissue separation method for storing the nuclear magnetic resonance image as described in claim 1 to 7 is any are processed
The corresponding program code of journey；
Processor, for executing said program code.
9. a kind of soft tissue separation system of nuclear magnetic resonance image characterized by comprising
Pixel classifications device determines the classification of pixel in the gray level image for the random fern disaggregated model by gray level image
Classification；
Face profile determining device determines correlation between the pixel for the class categories according to the pixel, is formed
Tissue contours；
Stereo profile determining device, for according in the adjacent gray level image the tissue contours position and described group of driving wheel
Wide variation tendency forms the stereo profile of tissue.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910708072.4A CN110400313B (en) | 2019-08-01 | 2019-08-01 | Soft tissue separation method and separation system for nuclear magnetic resonance image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910708072.4A CN110400313B (en) | 2019-08-01 | 2019-08-01 | Soft tissue separation method and separation system for nuclear magnetic resonance image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110400313A true CN110400313A (en) | 2019-11-01 |
| CN110400313B CN110400313B (en) | 2021-01-01 |
Family
ID=68327170
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910708072.4A Active CN110400313B (en) | 2019-08-01 | 2019-08-01 | Soft tissue separation method and separation system for nuclear magnetic resonance image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110400313B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100067761A1 (en) * | 2006-03-24 | 2010-03-18 | Exini Diagnostics Ab | Automatic interpretation of 3-d medicine images of the brain and methods for producing intermediate results |
| CN102436638A (en) * | 2010-08-24 | 2012-05-02 | 奥林巴斯株式会社 | Image processing apparatus, image processing method, and computer-readable recording medium |
| CN103729875A (en) * | 2013-12-09 | 2014-04-16 | 深圳先进技术研究院 | Method and system for reconstructing three-dimensional left ventricular profile of cardiac magnetic resonance image |
| CN105825233A (en) * | 2016-03-16 | 2016-08-03 | 中国地质大学（武汉） | Pedestrian detection method based on random fern classifier of online learning |
| CN107305634A (en) * | 2016-04-21 | 2017-10-31 | 杭州凌绝科技有限公司 | A kind of license plate locating method returned based on integrated random fern and shape |
| CN109801360A (en) * | 2018-12-24 | 2019-05-24 | 北京理工大学 | Stomach and intestine three-dimensionalreconstruction and method for visualizing based on image |
-
2019
- 2019-08-01 CN CN201910708072.4A patent/CN110400313B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100067761A1 (en) * | 2006-03-24 | 2010-03-18 | Exini Diagnostics Ab | Automatic interpretation of 3-d medicine images of the brain and methods for producing intermediate results |
| CN102436638A (en) * | 2010-08-24 | 2012-05-02 | 奥林巴斯株式会社 | Image processing apparatus, image processing method, and computer-readable recording medium |
| CN103729875A (en) * | 2013-12-09 | 2014-04-16 | 深圳先进技术研究院 | Method and system for reconstructing three-dimensional left ventricular profile of cardiac magnetic resonance image |
| CN105825233A (en) * | 2016-03-16 | 2016-08-03 | 中国地质大学（武汉） | Pedestrian detection method based on random fern classifier of online learning |
| CN107305634A (en) * | 2016-04-21 | 2017-10-31 | 杭州凌绝科技有限公司 | A kind of license plate locating method returned based on integrated random fern and shape |
| CN109801360A (en) * | 2018-12-24 | 2019-05-24 | 北京理工大学 | Stomach and intestine three-dimensionalreconstruction and method for visualizing based on image |
Non-Patent Citations (3)
| Title |
|---|
| YUDONG GUAN ETAL.: "Study of Surface Reconstruction Based on Contours of CT Image Slices", 《2009 INTERNATIONAL WORKSHOP ON INTELLIGENT SYSTEMS AND APPLICATIONS》 * |
| 姚民仓等: "由断层图像轮廓重构三维表面的新算法", 《微计算机信息》 * |
| 杨国庆: "基于空间邻域信息的高光谱遥感影像半监督分类", 《中国优秀硕士学位论文全文数据库 基础科学辑》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110400313B (en) | 2021-01-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Attribute-aware face aging with wavelet-based generative adversarial networks | |
| CN110766051A (en) | Lung nodule morphological classification method based on neural network | |
| CN106780751A (en) | Three-dimensional point cloud method for reconstructing based on improved shielding Poisson algorithm | |
| CN111738351B (en) | Model training method and device, storage medium and electronic equipment | |
| CN109886970B (en) | Detection segmentation method for target object in terahertz image and computer storage medium | |
| CN110648337A (en) | Hip joint segmentation method, hip joint segmentation device, electronic apparatus, and storage medium | |
| US9330336B2 (en) | Systems, methods, and media for on-line boosting of a classifier | |
| CN104636721B (en) | A kind of palm grain identification method based on profile Yu Edge texture Fusion Features | |
| CN106228528B (en) | A kind of multi-focus image fusing method based on decision diagram and rarefaction representation | |
| CN110135227B (en) | Laser point cloud outdoor scene automatic segmentation method based on machine learning | |
| CN104200240A (en) | Sketch retrieval method based on content adaptive Hash encoding | |
| Nie et al. | Two-dimensional extension of variance-based thresholding for image segmentation | |
| CN104766275B (en) | Sparse disparities figure denseization method and apparatus | |
| Demir et al. | Coupled segmentation and similarity detection for architectural models | |
| Morard et al. | Parsimonious path openings and closings | |
| CN112085675A (en) | Depth image denoising method, foreground segmentation method and human motion monitoring method | |
| CN117011274A (en) | Automatic glass bottle detection system and method thereof | |
| CN104112145A (en) | Facial expression recognition method based on PWLD and D-S evidence theory | |
| Omidalizarandi et al. | Segmentation and classification of point clouds from dense aerial image matching | |
| CN110443790A (en) | Cartilage recognition methods and identifying system in a kind of medical image | |
| CN116758212A (en) | 3D reconstruction method, device, equipment and medium based on self-adaptive denoising algorithm | |
| CN110400313A (en) | A kind of the soft tissue separation method and separation system of nuclear magnetic resonance image | |
| CN108256578B (en) | Gray level image identification method, device, equipment and readable storage medium | |
| CN115861717A (en) | Breast tumor benign and malignant classification model method, system, terminal and storage medium | |
| Cuadros-Vargas et al. | Generating segmented tetrahedral meshes from regular volume data for simulation and visualization applications |
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 |