CN109409403A - Brain network clustering method based on local attribute and topological structure - Google Patents
Brain network clustering method based on local attribute and topological structure Download PDFInfo
- Publication number
- CN109409403A CN109409403A CN201811061148.0A CN201811061148A CN109409403A CN 109409403 A CN109409403 A CN 109409403A CN 201811061148 A CN201811061148 A CN 201811061148A CN 109409403 A CN109409403 A CN 109409403A
- Authority
- CN
- China
- Prior art keywords
- brain
- network
- similarity
- indicate
- brain network
- 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
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of brain network clustering method based on local attribute and topological structure, this method specifically follow the steps below: pre-processing first to brain function network, and extract each brain area average time sequence；Then Pearson correlation coefficient is calculated, unbiased brain function network is constructed；Calculate brain function network similarity；Finally brain function network similarity is clustered, and is tested to cluster result；The present invention is by obtaining the higher similarity of rate of precision for local attribute's similarity of brain function network and the fusion of topological structure Similarity-Weighted, and after the similarity cluster based on local attribute and topological structure, obtained cluster result is accurate, zero deflection.
Description
Technical field
The invention belongs to machine learning techniques fields, more particularly to a kind of brain net based on local attribute and topological structure
Network clustering method.
Background technique
Currently, important tool of the machine learning as brain network analysis, due to can from data learning law and to not
Primary data is predicted, it has also become the new research hotspot in one, brain network analysis field in recent years；Machine learning is according to study shape
The difference of formula is divided into supervised learning (classification) and unsupervised learning (cluster).
Current research, it is most of all to use supervised learning, i.e., using the training data training classification mould with label
Then type classifies to test data with disaggregated model；But since the label of data is by professional according to some elder generations
It tests what knowledge was labeled, has subjectivity, and it is possible that mistake, final to influence classification knot during mark
Fruit reduces the accuracy of brain network class.
Based on this, it is necessary to a kind of brain network clustering method is invented, to solve classification present in existing brain network class
There is deviation, the problem of classification results inaccuracy.
Summary of the invention
The purpose of the present invention is to provide a kind of brain network clustering method based on local attribute and topological structure, to overcome
The problem of the inaccuracy of label for labelling present in supervised learning, so that brain network similarity cluster result is more accurate.
The technical scheme adopted by the invention is that the brain network clustering method based on local attribute and topological structure, specifically
The following steps are included:
Step S1: pre-processing brain function MRI, then carries out brain area segmentation, and extract each brain area
Average time sequence；
Step S2: the Pearson correlation coefficient between each brain area average time sequence is calculated, Pearson came Correlation Moment is obtained
Battle array；Unbiased brain function network is obtained using kruskal algorithm；
Step S3: the local attribute's similarity and topological structure similarity of unbiased brain function network, and the subordinate that plays a game are calculated
Property similarity and topological structure similarity are weighted fusion, obtain the similarity of brain function network；
Step S4: similar matrix is constructed using the similarity of brain function network, similar matrix is calculated using multi-path spectral clustering
Method realizes the cluster of brain network.
Further, in step S1, brain function MRI is pre-processed using Dparsf software, pretreatment tool
Body include: remove preceding 10 time points, time point correction, head move correction, Spatial normalization, smoothly, remove linear drift and filtering；
Then according to selected standardization brain map-AAL template, pretreated brain function MRI is split, it will be big
Brain is divided into 90 brain areas；Finally, the data according to pretreated brain function MRI, extract respectively and calculate each brain
Inside area each voxel in different time points on activation value and its average value, obtain the average time sequence of each brain area；Institute
State activation value refer to each voxel in different time points on Blood oxygen level dependence intensity.
Further, in step S2, construct unbiased brain function network the following steps are included:
Step S21: the Pearson correlation coefficient r between two brain areas is calculated using formula (1)xy,
In formula (1), 1≤n≤N, N indicate time point number, xnIndicate activation value of the brain area x n-th of time point,
Indicate brain area x in the average value of all sweep time point activation values；ynIndicate activation value of the brain area y n-th of time point,Table
Show brain area y in the average value of all sweep time point activation values；rxyIt indicates the Pearson correlation coefficient between brain area x and y, obtains
The Pearson relevance matrix R of 90*90；
Step S22: unbiased brain function network is constructed as described below using kruskal algorithm: (1) by Pierre
Related coefficient in inferior correlation matrix R carries out descending sort；(2) the maximum node of related coefficient is connected, until all
Until node is connected in the form of acyclic subgraph；(3) if occurring loop after adding the connection in step (2), then abandoning
The connection.
Further, in step S3, steps are as follows for the calculating of unbiased brain function network similarity:
Step S31: calculating the betweenness value b of each node in network using formula (2), calculates brain network using formula (3)
The similarity S of local attributeAtr(G,H)；
In formula (2), v indicates the number of brain nodes, and V indicates the set of all node compositions, ρadIndicate node a
Shortest path length between node d,Indicate the shortest path length for passing through node e between node a and node d；
In formula (3), 1≤m≤v, bm(G) betweenness of m-th of node in brain network G, b are indicatedm(H) it indicates in brain network H
The betweenness of m-th of node, SAtr(G, H) indicates brain network G and brain network H local attribute similarity；
Step S32: the similarity of brain network on the topology is calculated using the Weisfeiler-Lehman kernel of graph；
Weisfeiler-Lehman kernel of graph calculation method is as follows: the initial labels for 1. defining each node in brain network are
The angle value of node；2. being ranked up to the label of the adjacent node of each node, the node is then extended to, and this length
Tag update is one and new does not occur label；3. step is repeated 2., until the number of iterations reaches predetermined value h；4. using formula
(4) the Weisfeiler-Lehman subtree core of brain network G and brain network H, the i.e. topological structure of brain network G and brain network H are calculated
Similarity SStr(G, H):
In formula (4), k (G, H) indicates the kernel of graph value of brain network G and brain network H,Indicate that brain network G is mapped to height
The mapping function of dimensional feature space,Indicate that brain network H is mapped to the mapping function of high-dimensional feature space, SStr(G, H) table
Show the topological structure similarity of brain network G and brain network H；
Wherein,
In formula (5), h indicates the number of iteration, σ0(G,s01) indicate the label s in the 0th iteration01Occur in figure G
Number,Indicate the label in the 0th iterationThe number occurred in figure G, σh(G,sh1) indicate at the h times
Label s when iterationh1The number occurred in figure G,Indicate the label in the h times iterationOccur in figure G
Number,
In formula (6), σ0(H,s01) indicate the label s in the 0th iteration01The number occurred in brain network H,Indicate the label in the 0th iterationThe number occurred in brain network H, σh(H,sh1) indicate at the h times repeatedly
For when label sh1The number occurred in brain network H,Indicate the label in the h times iterationIn brain network H
The number of appearance；
Step S33: setting weight δ, δ ∈ (0,1) calculate local attribute's similarity in conjunction with topological structure similarity
The similarity S (G, H) of brain network G and brain network H, formula (7) are as follows:
S (G, H)=δ SAtr(G,H)+(1-δ)SStr(G,H) (7)。
Further, in step S4, realize brain network clustering the step of it is as follows:
Step S41: using the similar matrix S of the similarity building brain network of brain network, as shown in formula (8):
In formula (8), sGHIndicate the similarity between brain network G and brain network H, U indicates the quantity of brain network；
Step S42: utilizing multi-path spectral clustering algorithm, that is, NJW algorithm, realizes the cluster of brain network, and specific algorithm is as follows: first
Adjacency matrix W and degree matrix D are first constructed according to similar matrix S, Laplacian Matrix L=D-W is calculated with this；Then to La Pula
This matrix standardizes to obtainCalculate LsymPreceding k eigen vector, construction feature vector is empty
Between；Finally clustered using feature vector of the K-means algorithm to characteristic vector space；
Step S43: the rate of precision P, recall rate R of brain network clustering result are calculated separately using formula (9), (10) and (11)
With F1 value；
In formula (9), (10) and (11), TP indicates true positives number, and TN indicates true negative number, and FP indicates false positive
Number, FN indicate false negative number.
The beneficial effects of the present invention are: (1) calculating the similarity between brain network by building brain network model and making
The automatic cluster of brain network is realized with spectral clustering；(2) fully consider brain network in local attribute and global Topological Structure two
The similitude of aspect constructs a kind of method for calculating brain network similarity, accurate can calculate similar between brain network
Degree improves the accuracy of brain network clustering.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the brain network clustering model flow figure based on local attribute and topological structure.
Fig. 2 is the cluster result comparison diagram of the present invention with only consideration local attribute or topological similarity.
Fig. 3 is influence of the different values of weight δ to cluster result rate of precision.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Embodiment 1
(1) experimental data
By taking Alzheimer's disease and normal aging people as an example, clustering is carried out to the brain network of two classes subject；Study number
(ADNI) database, including 28 Alzheimer Disease patient (ages: 72.2 are proposed according to from Alzheimer disease neuroimaging
± 7.5, women: 17, mini-mental situation inspection method (MMSE): 22.8 ± 2.5, clinical dementia evaluation scale (CRD): 0.84
± 0.23), 28 normal aging peoples (age: 74.3 ± 6.3, women: 17, MMSE:28.9 ± 1.3, CRD:0 ± 0)；
(2) experimentation
According to process shown in FIG. 1, clustered using the brain network that following steps are tested 56:
Step S1: pre-processing brain function MRI, then carries out brain area segmentation, and extract each brain area
Average time sequence；
Brain function MRI is pre-processed using Dparsf software, pretreated step specifically includes: being removed
Preceding 10 time points, time point correction, head move correction, Spatial normalization, smoothly, remove linear drift and filtering；Then according to choosing
Fixed standardization brain map, is split pretreated brain function MRI, and brain is divided into 90 brain areas；Research
Show that the impaired brain area of Alzheimer's disease is concentrated mainly on default mode network (DMN), so extracting quilt in the present embodiment
The default mode network of examination is studied, and 32 brain areas that extraction default mode network includes from 90 brain areas are (on pars orbitalis volume
Return, middle frontal gyrus, pars orbitalis middle frontal gyrus, return rectus, preceding cingulum and other cingulum gyrus, posterior cingutate, quader, hippocampus, parahippocampal gyrus,
Top lower edge angular convolution, angular convolution, superior temporal gyrus, temporo pole: superior temporal gyrus, gyrus temporalis meduus, temporo pole: gyrus temporalis meduus, inferior temporal gyrus) it is studied；
Finally, the data according to pretreated brain function MRI, extract respectively and calculate inside each brain area
Each voxel goes up the average value of activation value in different time points, obtains the average time sequence of each brain area；Activation value refers to respectively
A voxel in different time points on Blood oxygen level dependence intensity；
Step S2: the Pearson correlation coefficient between each brain area average time sequence is calculated, is calculated using Kruskal
Method obtains unbiased brain function network；
The Pearson correlation coefficient r between brain area two-by-two is calculated using formula (1)xy:
In formula (1), 1≤n≤N, N indicate time point number, xnIndicate activation value of the brain area x n-th of time point,
Indicate brain area x in the average value of all sweep time point activation values；ynIndicate activation value of the brain area y n-th of time point,Table
Show brain area y in the average value of all sweep time point activation values；rxyIt indicates the Pearson correlation coefficient between brain area x and y, obtains
The Pearson relevance matrix R of 32*32；
Unbiased brain network is constructed according to following description using kruskal algorithm: (1) by Pearson relevance matrix
Related coefficient carries out descending sort；(2) the maximum node of related coefficient is connected, until all nodes are with acyclic subgraph
Until form connects；(3) if occurring loop after adding the connection in step (2), then abandoning the connection；
Step S3: the similarity between unbiased brain function network is calculated；
The betweenness value b that each node in network is calculated using formula (2) calculates brain network local attribute using formula (3)
Similarity SAtr(G,H)；
In formula (2), v indicates the number of brain nodes, and V indicates the set of all node compositions, ρadIndicate node a
Shortest path length between node d,Indicate the shortest path length for passing through node e between node a and node d；
In formula (3), 1≤m≤v, bm(G) betweenness of m-th of node in figure G, b are indicatedm(H) m-th of section in figure H is indicated
The betweenness of point, SAtr(G, H) indicates figure G and figure H local attribute similarity；
It is assessed using Weisfeiler-Lehman subtree and calculates the similarity of brain network on the topology, Weisfeiler-
Lehman kernel of graph calculation method is as follows:
1. the initial labels for defining each node in brain network are the angle value of node；2. to the adjacent node of each node
Label is ranked up, and then extends to the node, and the tag update of this length is one and new does not occur label；3. repeating
Step 2., until the number of iterations reaches predetermined value h；4. calculating the Weisfeiler- of brain network G and brain network H using formula (4)
The topological structure similarity S of Lehman subtree core, i.e. brain network G and brain network HStr(G, H):
In formula (4), k (G, H) indicates the kernel of graph value of brain network G and brain network H,Indicate that brain network G is mapped to height
The mapping function of dimensional feature space,Indicate that brain network H is mapped to the mapping function of high-dimensional feature space, SStr(G, H) table
Show the topological structure similarity of brain network G and brain network H；
Wherein,
In formula (5), h indicates the number of iteration, σ0(G,s01) indicate the label s in the 0th iteration01Occur in figure G
Number,Indicate the label in the 0th iterationThe number occurred in figure G, σh(G,sh1) indicate at the h times
Label s when iterationh1The number occurred in figure G,Indicate the label in the h times iterationOccur in figure G
Number,
In formula (6)；σ0(H,s01) indicate the label s in the 0th iteration01The number occurred in figure H,
Indicate the label in the 0th iterationThe number occurred in figure H, σh(H,sh1) indicate the label s in the h times iterationh1?
The number occurred in figure H,Indicate the label in the h times iterationThe number occurred in figure H；
Weight δ, δ ∈ (0,1) are set, local attribute's similarity is calculated to the phase of brain network in conjunction with topological similarity
Like degree, formula (7) is as follows:
S (G, H)=δ SAtr(G,H)+(1-δ)SStr(G,H) (7)；
Step S4: using the similar matrix S of the similarity building brain network of brain network, as shown in formula (8):
In formula (8), sGHIndicate the similarity between brain network G and brain network H, U indicates the quantity of brain network；
Using multi-path spectral clustering algorithm, that is, NJW algorithm, the cluster of brain network is realized, and use rate of precision, recall rate and F1
Evaluate clustering performance:
Similar matrix S is calculated first, adjacency matrix W and degree matrix D are constructed according to similar matrix S, drawing is calculated with this
This matrix L=D-W of pula；Then Laplacian Matrix is standardized to obtainCalculate LsymFirst k it is special
Value indicative and feature vector, construction feature vector space；Finally using K-means algorithm to the feature vector of characteristic vector space into
Row cluster；
The rate of precision P of brain network clustering result, recall rate R and F1 value are calculated separately using formula (9), (10) and (11)；
In formula (9), (10) and (11), TP indicates true positives number, and TN indicates true negative number, and FP indicates false positive
Number, FN indicate false negative number；
(3) experimental result
In order to examine the value of weight δ to the influence of cluster result, set the weight δ in step 3 to from 0.1 to 0.9,
Step-length is 0.1, obtains the different brain network similarities based on local attribute and topological structure, carries out cluster point to each similarity
Analysis, the rate of precision of cluster result is as shown in figure 3, from the figure 3, it may be seen that according to δ=0.7 by local attribute's similarity and topological structure phase
The similarity that can more accurately describe between brain network is combined like degree.
Embodiment 2
The weight of step 3 takes 0.7 in embodiment 1, each similarity obtained to step 3: SAtr(G,H)、SStr(G, H) and S
(G, H) is clustered using clustering algorithm, and calculates the rate of precision, recall rate and F1 value of cluster result, is tested to it；
Rate of precision, recall rate and the F1 value of each cluster result as shown in Fig. 2, as shown in Figure 2, by local attribute's similarity and
Brain network similarity Clustering Effect after topological structure Similarity-Weighted combines is best, rate of precision 0.64, recall rate 0.62,
F1 is 0.63.
Each embodiment in this specification is all made of relevant mode and describes, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for system reality
For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method
Part explanation.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the scope of the present invention.It is all
Any modification, equivalent replacement, improvement and so within the spirit and principles in the present invention, are all contained in protection scope of the present invention
It is interior.
Claims (5)
1. the brain network clustering method based on local attribute and topological structure, which is characterized in that specifically includes the following steps:
Step S1: pre-processing brain function MRI, then carries out brain area segmentation, and extract being averaged for each brain area
Time series；
Step S2: the Pearson correlation coefficient between each brain area average time sequence is calculated, Pearson relevance matrix is obtained；Benefit
Unbiased brain function network is obtained with kruskal algorithm；
Step S3: the local attribute's similarity and topological structure similarity of unbiased brain function network are calculated, and to local attribute's phase
It is weighted fusion like degree and topological structure similarity, obtains the similarity of brain function network；
Step S4: constructing similar matrix using the similarity of brain function network, uses multi-path spectral clustering algorithm to similar matrix, real
The cluster of existing brain network.
2. the brain network clustering method according to claim 1 based on local attribute and topological structure, which is characterized in that institute
It states in step S1, brain function MRI is pre-processed using Dparsf software, pretreatment specifically includes: removing preceding 10
A time point, time point correction, head move correction, Spatial normalization, smoothly, remove linear drift and filtering；Then basis is selected
Brain map-AAL template is standardized, pretreated brain function MRI is split, brain is divided into 90 brain areas；
Finally, the data according to pretreated brain function MRI, extract respectively and calculate each voxel inside each brain area
Activation value and its average value in different time points obtain the average time sequence of each brain area；The activation value refers to respectively
A voxel in different time points on Blood oxygen level dependence intensity.
3. the brain network clustering method according to claim 1 based on local attribute and topological structure, which is characterized in that institute
State in step S2, construct unbiased brain function network the following steps are included:
Step S21: the Pearson correlation coefficient r between two brain areas is calculated using formula (1)xy,
In formula (1), 1≤n≤N, N indicate time point number, xnIndicate activation value of the brain area x n-th of time point,It indicates
Average value of the brain area x in all sweep time point activation values；ynIndicate activation value of the brain area y n-th of time point,Indicate brain
Average value of the area y in all sweep time point activation values；rxyIt indicates the Pearson correlation coefficient between brain area x and brain area y, obtains
The Pearson relevance matrix R of 90*90；
Step S22: unbiased brain function network is constructed as described below using kruskal algorithm: (1) by Pearson came phase
The related coefficient closed in matrix R carries out descending sort；(2) the maximum node of related coefficient is connected, until all nodes
Until being connected in the form of acyclic subgraph；(3) if occurring loop after adding the connection in step (2), then abandoning the company
It connects.
4. the brain network clustering method according to claim 1 based on local attribute and topological structure, which is characterized in that institute
It states in step S3, steps are as follows for the calculating of unbiased brain function network similarity:
Step S31: calculating the betweenness value b of each node in network using formula (2), calculates brain network part using formula (3)
The similarity S of attributeAtr(G,H)；
In formula (2), v indicates the number of brain nodes, and V indicates the set of all node compositions, ρadIndicate node a and section
Shortest path length between point d,Indicate the shortest path length for passing through node e between node a and node d；
In formula (3), 1≤m≤v, bm(G) betweenness of m-th of node in brain network G, b are indicatedm(H) m in brain network H is indicated
The betweenness of a node, SAtr(G, H) indicates brain network G and brain network H local attribute similarity；
Step S32: the similarity of brain network on the topology is calculated using the Weisfeiler-Lehman kernel of graph；
Weisfeiler-Lehman kernel of graph calculation method is as follows: the initial labels for 1. defining each node in brain network are node
Angle value；2. being ranked up to the label of the adjacent node of each node, the node, and the label that this is grown then are extended to
It is updated to one and new does not occur label；3. step is repeated 2., until the number of iterations reaches predetermined value h；4. being counted using formula (4)
The Weisfeiler-Lehman subtree core of brain network G and brain network H is calculated, i.e. brain network G is similar with the topological structure of brain network H
Spend SStr(G, H):
In formula (4), k (G, H) indicates the kernel of graph value of brain network G and brain network H,Indicate that brain network G is mapped to higher-dimension spy
The mapping function in space is levied,Indicate that brain network H is mapped to the mapping function of high-dimensional feature space, SStr(G, H) indicates brain net
The topological structure similarity of network G and brain network H；
Wherein,
In formula (5), h indicates the number of iteration, σ0(G,s01) indicate the label s in the 0th iteration01Time occurred in figure G
Number,Indicate the label in the 0th iterationThe number occurred in figure G, σh(G,sh1) indicate in the h times iteration
When label sh1The number occurred in figure G,Indicate the label in the h times iterationTime occurred in figure G
Number,
In formula (6), σ0(H,s01) indicate the label s in the 0th iteration01The number occurred in brain network H,
Indicate the label in the 0th iterationThe number occurred in brain network H, σh(H,sh1) indicate the label in the h times iteration
sh1The number occurred in brain network H,Indicate the label in the h times iterationTime occurred in brain network H
Number；
Step S33: local attribute's similarity is calculated brain net by setting weight δ, δ ∈ (0,1) in conjunction with topological structure similarity
The similarity S (G, H) of network G and brain network H, formula (7) are as follows:
S (G, H)=δ SAtr(G,H)+(1-δ)SStr(G,H) (7)。
5. the brain network clustering method according to claim 1 based on local attribute and topological structure, which is characterized in that institute
The step of stating in step S4, realizing brain network clustering is as follows:
Step S41: using the similar matrix S of the similarity building brain network of brain network, as shown in formula (8):
In formula (8), sGHIndicate the similarity between brain network G and brain network H, U indicates the quantity of brain network；
Step S42: utilizing multi-path spectral clustering algorithm, that is, NJW algorithm, realizes the cluster of brain network, and specific algorithm is as follows: root first
According to similar matrix S building adjacency matrix W and degree matrix D, Laplacian Matrix L=D-W is calculated with this；Then to Laplce's square
Battle array standardization obtainsCalculate LsymPreceding k eigen vector, construction feature vector space；Most
It is clustered afterwards using feature vector of the K-means algorithm to characteristic vector space；
Step S43: the rate of precision P, recall rate R and F1 of brain network clustering result are calculated separately using formula (9), (10) and (11)
Value；
In formula (9), (10) and (11), TP indicates true positives number, and TN indicates true negative number, and FP indicates false positive number,
FN indicates false negative number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811061148.0A CN109409403B (en) | 2018-09-12 | 2018-09-12 | Brain network clustering method based on local attribute and topological structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811061148.0A CN109409403B (en) | 2018-09-12 | 2018-09-12 | Brain network clustering method based on local attribute and topological structure |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109409403A true CN109409403A (en) | 2019-03-01 |
| CN109409403B CN109409403B (en) | 2019-07-26 |
Family
ID=65464724
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811061148.0A Active CN109409403B (en) | 2018-09-12 | 2018-09-12 | Brain network clustering method based on local attribute and topological structure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109409403B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110084381A (en) * | 2019-04-22 | 2019-08-02 | 天津工业大学 | A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph |
| CN111161226A (en) * | 2019-12-20 | 2020-05-15 | 西北工业大学 | Method for uniformly segmenting cerebral cortex surface based on spectral clustering algorithm |
| CN111259849A (en) * | 2020-01-22 | 2020-06-09 | 深圳大学 | Method and device for detecting resting brain network by functional near infrared spectrum imaging |
| CN111949812A (en) * | 2020-07-10 | 2020-11-17 | 上海联影智能医疗科技有限公司 | Brain image classification method and storage medium |
| CN113398422A (en) * | 2021-07-19 | 2021-09-17 | 燕山大学 | Rehabilitation training system and method based on motor imagery-brain-computer interface and virtual reality |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106548206A (en) * | 2016-10-27 | 2017-03-29 | 太原理工大学 | Multi-modal nuclear magnetic resonance image data classification method based on minimum spanning tree |
| CN107316308A (en) * | 2017-06-27 | 2017-11-03 | 苏州大学 | A kind of clean robot map dividing method based on improved multi-path spectral clustering algorithm |
-
2018
- 2018-09-12 CN CN201811061148.0A patent/CN109409403B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106548206A (en) * | 2016-10-27 | 2017-03-29 | 太原理工大学 | Multi-modal nuclear magnetic resonance image data classification method based on minimum spanning tree |
| CN107316308A (en) * | 2017-06-27 | 2017-11-03 | 苏州大学 | A kind of clean robot map dividing method based on improved multi-path spectral clustering algorithm |
Non-Patent Citations (3)
| Title |
|---|
| ANVAR KURMUKOV等: "Classification of normal and pathological brain networks based on similarity in graph partitions", 《IEEE》 * |
| 崔晓红等: "静息态功能脑网络核心节点评价方法及其在抑郁症分类上的应用", 《太原理工大学学报》 * |
| 牛会兰等: "基于节点相似度的功能脑网络建模方法", 《计算机工程与设计》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110084381A (en) * | 2019-04-22 | 2019-08-02 | 天津工业大学 | A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph |
| CN111161226A (en) * | 2019-12-20 | 2020-05-15 | 西北工业大学 | Method for uniformly segmenting cerebral cortex surface based on spectral clustering algorithm |
| CN111259849A (en) * | 2020-01-22 | 2020-06-09 | 深圳大学 | Method and device for detecting resting brain network by functional near infrared spectrum imaging |
| CN111259849B (en) * | 2020-01-22 | 2023-05-12 | 深圳大学 | Functional near infrared spectrum imaging resting state brain network detection method and device |
| CN111949812A (en) * | 2020-07-10 | 2020-11-17 | 上海联影智能医疗科技有限公司 | Brain image classification method and storage medium |
| CN113398422A (en) * | 2021-07-19 | 2021-09-17 | 燕山大学 | Rehabilitation training system and method based on motor imagery-brain-computer interface and virtual reality |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109409403B (en) | 2019-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109409403B (en) | Brain network clustering method based on local attribute and topological structure | |
| CN105320944B (en) | A kind of human body behavior prediction method based on human skeleton motion information | |
| CN110163258A (en) | A kind of zero sample learning method and system reassigning mechanism based on semantic attribute attention | |
| CN107133651B (en) | The functional magnetic resonance imaging data classification method of subgraph is differentiated based on super-network | |
| CN105809672B (en) | A kind of image multiple target collaboration dividing method constrained based on super-pixel and structuring | |
| CN104881689B (en) | A kind of multi-tag Active Learning sorting technique and system | |
| CN101178703B (en) | Failure diagnosis chart clustering method based on network dividing | |
| CN104966105A (en) | Robust machine error retrieving method and system | |
| CN104298974B (en) | A kind of Human bodys' response method based on deep video sequence | |
| CN110236543A (en) | The more classification diagnosis systems of Alzheimer disease based on deep learning | |
| CN106991132A (en) | A kind of figure sorting technique reconstructed based on atlas with kernel of graph dimensionality reduction | |
| Manivannan et al. | Dengue fever prediction using K-means clustering algorithm | |
| CN105117731A (en) | Community partition method of brain functional network | |
| Hatami et al. | A machine learning approach to brain tumors segmentation using adaptive random forest algorithm | |
| CN110084381A (en) | A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph | |
| CN109344889A (en) | A kind of cerebral disease classification method, device and user terminal | |
| CN105354593B (en) | A kind of threedimensional model sorting technique based on NMF | |
| CN112418337B (en) | Multi-feature fusion data classification method based on brain function hyper-network model | |
| CN106127263A (en) | The human brain magnetic resonance image (MRI) classifying identification method extracted based on three-dimensional feature and system | |
| Wang et al. | Bag-of-features based classification of breast parenchymal tissue in the mammogram via jointly selecting and weighting visual words | |
| CN109902614A (en) | A kind of Human bodys' response method based on local space time's feature | |
| Yang et al. | Deep learning for cerebellar ataxia classification and functional score regression | |
| CN104182621A (en) | DBN based ADHD discriminatory analysis method | |
| CN111090764A (en) | Image classification method and device based on multitask learning and graph convolution neural network | |
| CN109033944A (en) | A kind of all-sky aurora image classification and crucial partial structurtes localization method and system |
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 |