CN107292346B - A kind of MR image hippocampus partitioning algorithm based on Local Subspace study - Google Patents
A kind of MR image hippocampus partitioning algorithm based on Local Subspace study Download PDFInfo
- Publication number
- CN107292346B CN107292346B CN201710541559.9A CN201710541559A CN107292346B CN 107292346 B CN107292346 B CN 107292346B CN 201710541559 A CN201710541559 A CN 201710541559A CN 107292346 B CN107292346 B CN 107292346B
- Authority
- CN
- China
- Prior art keywords
- training sample
- hippocampus
- image
- sample
- label
- 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.)
- Expired - Fee Related
Links
- 210000001320 hippocampus Anatomy 0.000 title claims abstract description 53
- 238000000638 solvent extraction Methods 0.000 title claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 82
- 238000000034 method Methods 0.000 claims abstract description 39
- 230000011218 segmentation Effects 0.000 claims abstract description 27
- 230000000644 propagated effect Effects 0.000 claims abstract description 9
- 239000000284 extract Substances 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 35
- 238000012360 testing method Methods 0.000 claims description 11
- 238000013507 mapping Methods 0.000 claims description 7
- 238000012512 characterization method Methods 0.000 claims description 4
- 238000003064 k means clustering Methods 0.000 claims description 3
- 238000001228 spectrum Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 2
- 210000005013 brain tissue Anatomy 0.000 abstract description 2
- 230000000694 effects Effects 0.000 abstract description 2
- 210000004556 brain Anatomy 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 2
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 2
- 208000024827 Alzheimer disease Diseases 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 239000012141 concentrate Substances 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 239000003814 drug Substances 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 238000003786 synthesis reaction Methods 0.000 description 2
- 101000831205 Danio rerio Dynein axonemal assembly factor 11 Proteins 0.000 description 1
- 102100024282 Dynein axonemal assembly factor 11 Human genes 0.000 description 1
- 241001559542 Hippocampus hippocampus Species 0.000 description 1
- 101000831210 Homo sapiens Dynein axonemal assembly factor 11 Proteins 0.000 description 1
- 208000028017 Psychotic disease Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000001419 dependent effect 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
- 238000005516 engineering process Methods 0.000 description 1
- 206010015037 epilepsy Diseases 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 208000024714 major depressive disease Diseases 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000001272 neurogenic effect Effects 0.000 description 1
- 210000001769 parahippocampal gyrus Anatomy 0.000 description 1
- 201000000980 schizophrenia Diseases 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 210000003478 temporal lobe Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- 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
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- 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
-
- 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
- G06T2207/30016—Brain
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of MR image hippocampus partitioning algorithms based on Local Subspace study, will be registrated map on marked atlas registration to target image first；Neighborhood is selected centered on target voxel in N number of be registrated in map, obtains training sample set；It extracts the characteristic informations such as texture and gray scale and constitutes original feature space, the sub-space learning of MFA is carried out to original feature space；Then it is propagated by label, obtains the segmentation result of target image.The present invention is based on the hippocampus dividing methods of Local Subspace study can preferably distinguish hippocampus structure and other brain tissue structures independent of the efficiency that registration accuracy and label are propagated, have more stable segmentation effect.
Description
Technical field
The present invention relates to medicine brain Image Segmentation field more particularly to a kind of MR image seas based on Local Subspace study
Horse body partitioning algorithm.
Background technique
Hippocampus also known as hippocampal gyrus are the important components of human brain nervous system, are mainly responsible for memory, study
And space orientation.Studies have shown that the variation of hippocampus volume and form and Alzheimer syndrome, schizophrenia, temporal lobe
A variety of neurogenic diseases such as epilepsy, major depressive disorder have close connection.In clinical application, by algorithm appropriate to sea
Horse body is accurately divided, to analyze the variation of its volume, form, can provide diagnosis basis for above-mentioned disease.However,
Since hippocampus has, in irregular shape, small in size, to be easy partial volume effect, edge big without obvious boundary, individual difference
The features such as, so that carrying out Accurate Segmentation to hippocampus becomes an extremely challenging task.Further, since between different images
The otherness that gray scale, texture, the diversity and complexity of marginal information and different images segmentation require, so far still without one kind
Unified method is all effective to all images to be split.Traditional hippocampus dividing method is many kinds of, relatively conventional
Have threshold method, region-growing method, clustering procedure, the dividing method based on edge, Markov random field model, based on small echo become
The dividing method etc. changed.
From the mechanism of hippocampus, the influence to human brain memory and spatialization function and its with Alzheimer syndrome etc.
Since being closely connected between psychotic disorder finds for researcher, hippocampus cutting techniques experienced manual segmentation, semi-automatic point
It cuts, divide three phases automatically.Due to the backwardness of technological means, the researcher of early stage generallys use threshold method, region-growing method
Hippocampus is split.And with the development of science and technology the hippocampus dividing method based on atlas registration also results in recent years
Researchers greatly pay close attention to.Dividing method based on atlas registration uses image registration as an important step, it is established that pre-
The corresponding relationship between atlas image and target image first marked.Then using obtained Deformation Field, by the label in map
Target image space is traveled to, the final segmentation result an of target image is obtained.Dividing method based on atlas registration can be with
It is divided into based on single map, based on average shape map and based on three kinds of multiple maps.
Dividing method based on single map is registrated with single map with target image, and therefore, it is difficult to adapt to difference
Otherness between individual, to be difficult to obtain ideal segmentation result.
Researchers propose the segmentation side based on average shape map on the basis of based on the dividing method of single map
Method, this method use multiple maps and certain means, obtain the information of all maps, and generating one has multiple frequency spectrums flat
The synthesis map of equal shape, then the synthesis map is registrated with target image to realize final segmentation.
Similar with based on average shape map dividing method, the dividing method based on multiple atlas registrations equally uses multiple
Map.Multiple maps are registrated with target image by this method respectively, obtain multiple groups deformation parameter, corresponding to each map respectively
Tag image carry out deformation, corresponding map label is traveled in target image space, then merge the map of all propagation
Label and the final segmentation result for generating a target image.
Compared with traditional hippocampus dividing method, the segmentation performance of the hippocampus dividing method based on atlas registration has
Biggish promotion.By taking the optimal dividing method based on multispectral figure registration of wherein segmentation performance as an example, by merging target image
The propagation label of multichannel chromatogram in space, this method can obtain robustness, the higher segmentation result of accuracy.However, being based on
The efficiency that the performance of the dividing method of multichannel chromatogram registration is propagated dependent on the precision and label of registration, therefore it is unable to get stabilization
Segmentation result.
Therefore, the efficiency that the segmentation performance for how solving existing hippocampus body method depends on registration accuracy and label to propagate
The problem of problem is current medicine brain Image Segmentation field urgent need to resolve.
Summary of the invention
For the deficiencies of the prior art, the present invention proposes a kind of MR image hippocampus segmentation based on Local Subspace study
Algorithm, comprising the following steps:
Step 1: will be obtained on marked atlas registration to target hippocampus MR image it is N number of be registrated map, it is described to have matched
Positive sample in quasi- map is the hippocampus labeled as+1, and the negative sample being registrated in map is the non-hippocampus labeled as -1
Body；
Step 2: selecting size for w × w × w neighborhood body centered on target voxel in N number of be registrated in map respectively
Element, and N × w × w × w candidate training sample is obtained, q is chosen respectively from the positive and negative samples of the candidate training sample1+q2
A sample constructs positive and negative training sample set, wherein Positive training sample collection and negative training sample collection composing training sample set, it is described
Training sample set includes 2 (q1+q2) a training sample；
Step 3: being concentrated from training sample and extract the original spy of characteristic information composition for including at least texture and gray value of image
Levy space；
Step 4: the sub-space learning analyzed based on marginal Fisher is carried out to the original feature space；
Step 4.1: by the type label of training sample expand for by center t × t of the training sample × neighborhood of t size
The structure type tag block of interior all training samples, the corresponding structure type mark of all training samples that the training sample is concentrated
It signs block and constitutes structure type tally set；
Step 4.2: k means clustering method being applied on the structure type tally set, is obtained and the structure type
The corresponding subtype tally set of tally set replaces the structure type tally set with the subtype tally set and by the subtype
Tally set obtains the training sample set of new tape label in conjunction with corresponding training sample set；
Step 4.3: constructing intrinsic figure GIG is schemed with punishmentP；
Step 4.4: defining the intrinsic figure GISimilar matrix SIAnd G is schemed in the punishmentPSimilar matrix SP；
Step 4.5: defining the intrinsic figure GIDiagonal matrix DIWith Laplacian Matrix LIAnd G is schemed in the punishmentP's
Diagonal matrix DPWith Laplacian Matrix LP；
Step 4.6: calculating subspace mapping matrix, the voxel maps that training sample concentration characteristic information is characterized are to
In Learning Subspaces；
Step 5: label is propagated；
Step 6: input target hippocampus MR image；
Step 7: extracting the characteristic information of each test sample voxel in the target hippocampus MR image；
Step 8: the test sample voxel with characteristic information characterization in the target hippocampus MR image is projected into institute
It states in Learning Subspaces；
Step 9: using based on k nearest neighbor classification device in the Learning Subspaces to the target hippocampus MR image
The classification of each test sample voxel is differentiated and obtains final segmentation result.
The beneficial effects of the present invention are:
The segmentation performance of existing dividing method excessively relies on the precision of registration and rich, the present invention of extracted feature
The dividing method of proposition can effectively improve voxel feature by carrying out the sub-space learning based on MFA to feature space
Resolution makes segmentation performance obtain biggish promotion, and segmentation performance is more stable, and the scope of application is wider.
Detailed description of the invention
Fig. 1 is the method flow diagram of partitioning algorithm of the present invention；
Fig. 2 is the schematic diagram that a multiscale image block is constructed by Gaussian filter；With
Fig. 3 is the syntople schematic diagram of intrinsic figure and punishment figure of the invention.
Specific embodiment
It is described in detail with reference to the accompanying drawing.
Voxel in the present invention refers to: voxel i.e. volume element, is the minimum unit in data and three-dimensional space segmentation.
Target voxel in the present invention refers to: the voxel of classification to be determined in feeling the pulse with the finger-tip mark hippocampus MR image.
Label in the present invention refers to: label is the corresponding label of sample voxel in the present invention, and wherein positive sample is mark
It is denoted as+1 hippocampus, negative sample is the non-hippocampus labeled as -1.
For the prior art hippocampus is divided existing insufficient, the present invention proposes a kind of based on Local Subspace study
MR image hippocampus partitioning algorithm.Fig. 1 is the method flow diagram of invention partitioning algorithm.Now in conjunction with attached drawing, the present invention will be described in detail
A kind of MR image hippocampus partitioning algorithm based on Local Subspace study, hippocampus partitioning algorithm proposed by the present invention includes
Following steps:
Step 1: inputting marked map, spatial alternation is carried out to these marked maps, such as affine transformation, piecewise affine
Transformation or nonlinear transformation, map that on target hippocampus MR image, obtain N number of Registration of MR map.
Step 2: select size for w × w × w neighboring voxels centered on target voxel N number of atlas image that has been registrated,
Obtain N × w × w × w candidate training sample.Q is selected respectively from the positive and negative samples of candidate training sample1+q2A sample structure
Training sample set is built, training sample set includes Positive training sample collection and negative training sample collection.
A kind of specific embodiment, it is all to be registrated figure for the voxel x to set the goal on hippocampus MR image
All voxels in spectrum in neighboring voxels V (x) corresponding with voxel x all be used to generate training sample set, neighboring voxels V (x)
Size is w × w × w.Therefore, N × w × w × w training sample can be obtained from N number of be registrated in mapCandidate training sample set is constructed, whereinTo be derived from i-th of map
J-th of sample voxel, lI, j∈ {+1, -1 } is the type label of each candidate training sample.
Candidate training sample concentration includes the voxel sample x difference similarity degree with type to be discriminated in target image
Sample.Training sample concentrates positive and negative samples number to reach balance when in order to make sub-space learning, from the positive and negative of candidate training sample
Q is chosen in sample respectively1+q2A sample constructs positive and negative training sample set.Specifically, selecting q for Positive training sample1A and mesh
The highest sample of standard type element sample x similarity, and q is randomly selected in remaining Positive training sample2A positive sample constitutes positive instruction
Practice sample set.Similarly, for negative training sample, also selection selects q1A and highest sample of target voxel sample x similarity,
And q is randomly selected in remaining negative training sample2A negative sample constitutes negative training sample collection.Positive training sample collection and negative training
Sample set, which combines, obtains training sample set.
Similitude is calculated by following structural similarity calculation formula (1):
Wherein the size centered on μ and σ difference in voxel x and j is the mean value and standard deviation of w × w × w neighborhood.
Step 3: extracting the characteristic information for including at least texture and gray value of image for constructing original feature space.The original
Beginning feature space it is subsequent by the training for being used for subspace to improve the voxel area in the target image characterized with characteristic information
Indexing.
Step 31: the texture feature information that this algorithm extracts includes:
First-order difference filter (FODs) output, such as formula (2)
{ H (z+u)-H (z-u), u=(r cos θ sin φ, r sin θ sin φ, r cos φ) } (2)
Second differnce filter (SODs) output, such as formula (3)
{ H (z+u)+H (z-u) -2H (z), u=(r cos θ sin φ, r sin θ sin φ, r cos φ) } (3)
Three-dimensional hyperplane filter output, such as formula (4)
3D-Sobel filter output such as formula (5)
Laplace filter output, such as formula (6)
Tonal range filter output, such as formula (7)
Wherein, CA, b, c(z) one central point of expression is the rectangular area a × b × c in z volume, and the u amount of being biased towards, r is u
Length, θ and φ are two rotation angles of u, and Ψ is filter operator, and H is corresponding filter function, Op(z) the p neighborhood of z is indicated
In voxel, * indicate convolution algorithm.In texture information, FODs and SODs are for detecting the variation of intensity on a cut-off rule
Situation.
Here, it enables r ∈ { 1,2,3 }, θ ∈ { 0, π/4, pi/2,3 π/4 }, φ ∈ { 0, π/4, pi/2 }.Three-dimensional hyperplane filtering
Device and 3D-Sobel filter are the extensions in the plane to FODs and SODs.In addition, also using other two kinds of filters, i.e.,
Laplace filter and tonal range filter.Laplace filter be it is isotropic, for detect second-order intensity become
Change.Tonal range filter is then used to calculate the minimum gray scale difference between maximum value of each voxel in given neighborhood.Of the invention
In technical solution, p neighborhood is dimensioned to p ∈ { 7,19,27 }.
Step 32: the step of extracting gray value of image characteristic information is as follows:
The image block not overlapped each other is divided an image into first.In the present invention, using different Gaussian filters come by
Green strength information replaces with the strength information after convolution.
Fig. 2 is the schematic diagram that a multiscale image block is constructed by Gaussian filter.As shown in Fig. 2, due to segmentation
Final purpose is the type label of determining target image voxel, is caught in a range (subgraph on the right in such as Fig. 2) appropriate
Obtain the central information (internal layer area in such as Fig. 2) of voxel block, and this region be gradually expanded, with image block centre distance
Increase obtain rough multiple dimensioned strength information.
Step 4: for the anatomical information for saving MR image in segmentation, the type label of training sample being expanded as knot
Structure type label block, then based on k mean cluster obtain subtype tally set corresponding with the structure type tally set and with
This constructs new training sample set, then carries out limit Fisher to original feature space and analyze (Margin Fisher
Analysis, MFA) sub-space learning:
Step 4.1: the type label of training sample is expanded as the big small neighbourhood of t × t × t centered on the training sample
The structure type tag block of interior all training samples, the corresponding structure type tag block of all training samples that training sample is concentrated
Constitute structure type tally set.
Specifically, with U=[u1, u2..., uM] indicate training sample set,Indicate training sample, Y=[y1,
y2..., yM] indicate voxel sample type tally set, yi∈ {+1, -1 } is indicated and uiThe type label of corresponding voxel sample,
Positive sample namely the sample type label for belonging to hippocampus are+1, and the type label of negative sample namely non-hippocampus sample is -1.M
=2 (q1+q2) it is the quantity that training sample concentrates training sample, d is sample characteristics dimension.By the type mark of each training sample
Label It expands as using training sample as the knot of all training samples in the big small neighbourhood of center t × t × t
Structure type label blockWherein t × t × t indicates the volume of tag block.
Step 4.2: structure type tally set corresponding with its for training sample set UEach trained sample
This uiA structure type tag block is assignedIt is right since MFA cannot be directly used to train subspaceUsing k means clustering method, subtype corresponding with structure type tally set is obtained under the conditions of unsupervised
Tally set { y '1..., y 'M, y 'i∈ {+1, -1 } indicates subtype label, by subtype tally set replacing structure type label
Collection, and new training sample set is generated in conjunction with training sample set, it is used for subsequent MFA sub-space learning.Specifically, described in
Subtype tally set replaces the structure type tally set and by the subtype tally set in conjunction with corresponding training sample set
To the training sample set of new tape label.
Step 4.3: constructing intrinsic figure GIG is schemed with punishmentP；
Two non-directed graphs, i.e., intrinsic figure G are constructed according to figure embedding theoryI={ U, SIAnd punishment figure GP={ U, SP, it uses
Manifold structure and manifold boundary in the different types of class of description, wherein U indicates candidate training set,For
GIAnd GPCorresponding similarity matrix；
Step 4.4: defining intrinsic figure similarity matrix SIAnd punishment figure similarity matrix SP；
In order to describe the compactness of sample in class, intrinsic figure GICorresponding similar matrix SIIt can be defined by following formula (8):
WhereinIndicate u in same classjK1The indexed set of arest neighbors.
In addition, MFA defines punishment figure GPSimilarity matrix SPFor portraying the discrete case of sample between class, definition
As shown in formula (9):
WhereinIt is set { (i, j), yi=y, yj≠ y } in k2A nearest data set.
Fig. 3 is the syntople figure of intrinsic figure and punishment.For MFA, the syntople of intrinsic figure and punishment figure can be by Fig. 3
It obtains.
Step 4.5: passing through intrinsic figure similarity matrix SIDefine the diagonal matrix D of intrinsic figureIWith Laplacian Matrix LI,
And pass through punishment figure similarity matrix SPThe diagonal matrix D of definition punishment figurePWith Laplacian Matrix LP。
Specifically, the corresponding diagonal matrix D of intrinsic figureIWith Laplacian Matrix LIIt can be defined and be obtained by following formula (10):
Punishment figure GPCorresponding Laplacian Matrix LPIt can be defined with the method similar with formula (10).
Step 4.6: the L obtained by step 4.5IAnd LPSubspace mapping matrix is calculated, it will by subspace mapping matrix
The voxel maps that training sample concentration characteristic information characterizes are into Learning Subspaces.
Specifically, being embedded in frame, subspace mapping matrix according to figure(wherein p is subspace dimension
Degree) it can be calculated by solving objective function (11):
Wherein Tr () is the mark of matrix, mapping matrix ΦMFAIncluding matrix The corresponding feature vector of preceding several larger characteristic values.
After obtaining subspace mapping matrix, training sample u that training sample concentration characteristic information characterizesi∈ U, i=1,
2 ..., M is mapped in Learning Subspaces in the form of formula (12), and formula (12) is as follows:
In Learning Subspaces, distance is compressed between the sample with similar tags, and the sample with different labels
Between distance be extended, be effectively promoted with characteristic information characterize image voxel resolution.
Step 5: label is propagated.The basic ideas that label is propagated are: going prediction not mark with the label information of marked sample
Remember that the label information of sample, the label of each sample are broadcast to adjacent node by similarity.Label propagate have it is simple, efficiently,
Quick advantage.
Step 6: input target hippocampus MR image.
Step 7: the characteristic information of each test sample voxel in target hippocampus MR image is extracted, to target hippocampus
Each test sample voxel in MR image extracts characteristic information according to the consistent method of training sample, i.e., in primitive character sky
Between in characterized.
Step 8: the test sample voxel with characteristic information characterization in target hippocampus MR image is projected into step 4.6
In in the subspace that has learnt；
Step 9: using based on k nearest neighbor classification device in the Learning Subspaces that step 4.6 obtains to target hippocampus MR
The classification of the test sample voxel of image is differentiated, final segmentation result is obtained.The basic ideas of k nearest neighbor classification method are:
If most of in K of the sample in feature space most like samples belong to some classification, the sample
Belong to this classification.Judge that test sample voxel be hippocampus is also non-hippocampus.
By the present invention with two kinds of traditional partitioning algorithms based on multichannel chromatogram, i.e., non local piecemeal label propagation algorithm with it is dilute
Dredge piecemeal label propagation algorithm compare it is found that algorithm proposed by the present invention by using MFA sub-space learning to marked map
Label propagated, and in the subspace of acquistion carry out k nearest neighbor classification method, can be by the hippocampus in MR brain image
Body structure is preferably distinguished with other brain tissue structures.It is a large amount of the experimental results showed that, the final segmentation arrived of the present invention
As a result above two point is superior on Dice coefficient, Jaccard coefficient, the evaluation indexes such as PI, RI and Hausdroff distance
Cut algorithm.The efficiency that technical solution of the present invention is propagated independent of the precision and label of registration, so that segmentation performance is more stable,
The algorithm scope of application is wider.
It should be noted that above-mentioned specific embodiment is exemplary, those skilled in the art can disclose in the present invention
Various solutions are found out under the inspiration of content, and these solutions also belong to disclosure of the invention range and fall into this hair
Within bright protection scope.It will be understood by those skilled in the art that description of the invention and its attached drawing are illustrative and are not
Constitute limitations on claims.Protection scope of the present invention is defined by the claims and their equivalents.
Claims (1)
1. a kind of MR image hippocampus partitioning algorithm based on Local Subspace study, which is characterized in that
Step 1: will be obtained on marked atlas registration to target hippocampus MR image it is N number of be registrated map, it is described to be registrated figure
Positive sample in spectrum is the hippocampus labeled as+1, and the negative sample being registrated in map is the non-hippocampus labeled as -1；
Step 2: select size for w × w × w neighboring voxels in N number of be registrated in map centered on target voxel respectively, and
N × w × w × w candidate training sample is obtained, chooses q respectively from the positive and negative samples of the candidate training sample1+q2A sample
The positive and negative training sample set of this building, wherein Positive training sample collection and negative training sample collection composing training sample set, the training
Sample set includes 2 (q1+q2) a training sample, target voxel refers to: the body of classification to be determined in feeling the pulse with the finger-tip mark hippocampus MR image
Element；
Step 3: being concentrated from training sample and extract the characteristic information composition primitive character sky for including at least texture and gray value of image
Between；
Step 4: the sub-space learning analyzed based on marginal Fisher is carried out to the original feature space；
Step 4.1: by the type label of training sample expand for by center t × t of the training sample × neighborhood of t size in institute
There are the structure type tag block of training sample, the corresponding structure type tag block of all training samples that the training sample is concentrated
Constitute structure type tally set；
Step 4.2: k means clustering method being applied on the structure type tally set, is obtained and the structure type label
Collect corresponding subtype tally set, replaces the structure type tally set with the subtype tally set and by the subtype label
Collection obtains the training sample set of new tape label in conjunction with corresponding training sample set；
Step 4.3: constructing intrinsic figure GIG is schemed with punishmentP；
Step 4.4: defining the intrinsic figure GISimilar matrix SIAnd G is schemed in the punishmentPSimilar matrix SP；
Step 4.5: defining the intrinsic figure GIDiagonal matrix DIWith Laplacian Matrix LIAnd G is schemed in the punishmentPIt is diagonal
Matrix DPWith Laplacian Matrix LP；
Step 4.6: subspace mapping matrix is calculated, by the voxel maps of training sample concentration characteristic information characterization to having learnt
In subspace；
Step 5: label is propagated；
Step 6: input target hippocampus MR image；
Step 7: extracting the characteristic information of each test sample voxel in the target hippocampus MR image；
Step 8: by the target hippocampus MR image with the test sample voxel of characteristic information characterization project to it is described
In Learning Subspaces；
Step 9: using based on k nearest neighbor classification device in the Learning Subspaces to each of the target hippocampus MR image
The classification of test sample voxel is differentiated and obtains final segmentation result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710541559.9A CN107292346B (en) | 2017-07-05 | 2017-07-05 | A kind of MR image hippocampus partitioning algorithm based on Local Subspace study |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710541559.9A CN107292346B (en) | 2017-07-05 | 2017-07-05 | A kind of MR image hippocampus partitioning algorithm based on Local Subspace study |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107292346A CN107292346A (en) | 2017-10-24 |
| CN107292346B true CN107292346B (en) | 2019-11-15 |
Family
ID=60100778
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710541559.9A Expired - Fee Related CN107292346B (en) | 2017-07-05 | 2017-07-05 | A kind of MR image hippocampus partitioning algorithm based on Local Subspace study |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107292346B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107895373B (en) * | 2017-11-14 | 2021-04-27 | 西安建筑科技大学 | Image segmentation method based on local region consistency manifold constraint MRF model |
| CN109584244B (en) * | 2018-11-30 | 2023-05-23 | 无锡本希奥智能技术有限公司 | Hippocampus segmentation method based on sequence learning |
| CN112183639B (en) * | 2020-09-30 | 2022-04-19 | 四川大学 | Mineral image identification and classification method |
| CN112674720B (en) * | 2020-12-24 | 2022-03-22 | 四川大学 | Alzheimer disease pre-judgment method based on 3D convolutional neural network |
| CN117557821A (en) * | 2024-01-11 | 2024-02-13 | 兰州大学 | Semi-supervised subspace clustering method and device based on soft MFA |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103714536A (en) * | 2013-12-17 | 2014-04-09 | 深圳先进技术研究院 | Sparse-representation-based multi-mode magnetic resonance image segmentation method and device |
| CN105205810A (en) * | 2015-08-30 | 2015-12-30 | 南方医科大学 | Distance-field-fusion-based hippocampus segmentation method of MR image |
| WO2016020008A1 (en) * | 2014-08-07 | 2016-02-11 | Mcmri Ag | System and method for image processing and analysis |
| CN106204562A (en) * | 2016-07-04 | 2016-12-07 | 西安交通大学 | A kind of method of the arched roof white matter segmentation merged based on fMRI Yu DTI |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9714288B2 (en) * | 2014-09-30 | 2017-07-25 | The Regents Of The University Of California | Antisense compounds and uses thereof |
-
2017
- 2017-07-05 CN CN201710541559.9A patent/CN107292346B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103714536A (en) * | 2013-12-17 | 2014-04-09 | 深圳先进技术研究院 | Sparse-representation-based multi-mode magnetic resonance image segmentation method and device |
| CN103714536B (en) * | 2013-12-17 | 2017-06-16 | 深圳先进技术研究院 | The dividing method and device of the multi-modal MRI based on rarefaction representation |
| WO2016020008A1 (en) * | 2014-08-07 | 2016-02-11 | Mcmri Ag | System and method for image processing and analysis |
| CN105205810A (en) * | 2015-08-30 | 2015-12-30 | 南方医科大学 | Distance-field-fusion-based hippocampus segmentation method of MR image |
| CN106204562A (en) * | 2016-07-04 | 2016-12-07 | 西安交通大学 | A kind of method of the arched roof white matter segmentation merged based on fMRI Yu DTI |
Non-Patent Citations (9)
| Title |
|---|
| Jianzhou Feng et al.."Learning dictionary via subspace segmentation for sparse representation".《 2011 18th IEEE International Conference on Image Processing》.2011, * |
| Xiaoxiao Liu et al.."A large-to-fine-scale shape prior for probabilistic segmentations using a deformable m-rep".《 2008 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops》.2008, * |
| 刘颖 等."基于稀疏编码与方向-尺度描述子的海马体自动分割".《计算机科学》.2017,第44卷(第1期), * |
| 孙俊 等."基于MFICSC算法的生菜图像目标聚类分割".《农业工程学报》.2012,第28卷(第13期), * |
| 张明慧 等."基于多图谱活动轮廓模型的脑部图像分割".《计算机学报》.2016,第39卷(第7期), * |
| 李小平 等."图像分割的改进稀疏子空间聚类方法".《***工程与电子技术》.2015,第37卷(第10期), * |
| 沈维政 等."基于遗传神经网络的大豆叶片病斑图像分割技术研究".《控制理论与引用》.2013, * |
| 温锐 等."基于引导滤波的多图谱医学图像分割".《南方医科大学学报》.2015, * |
| 程有娥."基于多图谱最优标签融合的脑MR图像分割".《浙江工贸职业技术学院学报》.2017,第17卷(第2期), * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107292346A (en) | 2017-10-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107292346B (en) | A kind of MR image hippocampus partitioning algorithm based on Local Subspace study | |
| Xie et al. | Beyond classification: structured regression for robust cell detection using convolutional neural network | |
| CN105009171B (en) | For learning the method and apparatus split automatically based on collection of illustrative plates of enhancing | |
| CN106682633B (en) | The classifying identification method of stool examination image visible component based on machine vision | |
| CN105335716B (en) | A kind of pedestrian detection method extracting union feature based on improvement UDN | |
| EP3101594A1 (en) | Saliency information acquisition device and saliency information acquisition method | |
| CN107506761A (en) | Brain image dividing method and system based on notable inquiry learning convolutional neural networks | |
| Pan et al. | Cell detection in pathology and microscopy images with multi-scale fully convolutional neural networks | |
| CN110889852A (en) | Liver segmentation method based on residual error-attention deep neural network | |
| CN104933420B (en) | A kind of scene image recognition methods and scene image identify equipment | |
| CN107133651A (en) | The functional magnetic resonance imaging data classification method of subgraph is differentiated based on super-network | |
| CN107066972B (en) | Natural scene Method for text detection based on multichannel extremal region | |
| CN109492668B (en) | MRI (magnetic resonance imaging) different-phase multimode image characterization method based on multi-channel convolutional neural network | |
| CN109409384A (en) | Image-recognizing method, device, medium and equipment based on fine granularity image | |
| CN104123554B (en) | SIFT image characteristic extracting methods based on MMTD | |
| He et al. | Automatic magnetic resonance image prostate segmentation based on adaptive feature learning probability boosting tree initialization and CNN-ASM refinement | |
| CN110503613A (en) | Based on the empty convolutional neural networks of cascade towards removing rain based on single image method | |
| CN107944403A (en) | Pedestrian's attribute detection method and device in a kind of image | |
| CN110705565A (en) | Lymph node tumor region identification method and device | |
| CN109447963A (en) | A kind of method and device of brain phantom identification | |
| CN115546605A (en) | Training method and device based on image labeling and segmentation model | |
| CN115496720A (en) | Gastrointestinal cancer pathological image segmentation method based on ViT mechanism model and related equipment | |
| CN113096184A (en) | Diatom positioning and identifying method under complex background | |
| CN110956157A (en) | Deep learning remote sensing image target detection method and device based on candidate frame selection | |
| KR101821770B1 (en) | Techniques for feature extraction |
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: 20191115 |