CN110084381A - A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph - Google Patents
A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph Download PDFInfo
- Publication number
- CN110084381A CN110084381A CN201910332184.4A CN201910332184A CN110084381A CN 110084381 A CN110084381 A CN 110084381A CN 201910332184 A CN201910332184 A CN 201910332184A CN 110084381 A CN110084381 A CN 110084381A
- Authority
- CN
- China
- Prior art keywords
- matrix
- brain network
- network
- brain
- attribute
- 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.)
- Pending
Links
- 210000004556 brain Anatomy 0.000 title claims abstract description 84
- 238000000034 method Methods 0.000 title claims abstract description 58
- 230000004927 fusion Effects 0.000 title claims abstract description 19
- 239000013598 vector Substances 0.000 claims abstract description 18
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000012706 support-vector machine Methods 0.000 claims abstract description 13
- 230000000694 effects Effects 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 51
- 238000002474 experimental method Methods 0.000 claims description 14
- 238000000354 decomposition reaction Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 239000000284 extract Substances 0.000 claims description 2
- 238000010835 comparative analysis Methods 0.000 claims 1
- 208000010877 cognitive disease Diseases 0.000 abstract description 7
- 238000004458 analytical method Methods 0.000 abstract description 6
- 208000024827 Alzheimer disease Diseases 0.000 abstract description 5
- 230000003902 lesion Effects 0.000 abstract description 2
- 238000010801 machine learning Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 6
- 208000027061 mild cognitive impairment Diseases 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 238000002599 functional magnetic resonance imaging Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 208000018152 Cerebral disease Diseases 0.000 description 1
- 239000000090 biomarker Substances 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000004886 head movement Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000002610 neuroimaging Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000006213 oxygenation reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention belongs to machine learning and brain network field, specially a kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph.The various features attribute of brain network is merged, firstly, raw experimental data is completed brain network struction after pretreatment；Secondly, extracting a variety of brain network attribute values in brain network according to different threshold values；They the optimal threshold parameter of sort effect and are carried out the characteristic attribute based on weight and merged by third using all data of support vector machines training, according to the superiority and inferiority of training result, in every kind of network attribute value；4th, use the fused feature vector of support vector machines training based on the novel kernel of graph；Classified using the method for the present invention to brain network, improves the classification accuracy of brain network.The present invention is suitable for the fields such as the analysis of brain area lesion, Alzheimer's disease and cognitive disorder Analysis on Mechanism.
Description
Technical field
The invention belongs to machine learning and brain network field, be related to a kind of method of characteristic attribute fusion, in particular to one
The various features attribute of brain network is carried out the fusion based on weight while being divided using novel kernel of graph technology brain network by kind
The method of class.
Background technique
It has a wide range of applications in the diagnosis for being sorted in cerebral disease and brain science research of brain network.Brain structure and function
It is not high always that the complexity of energy results in the accuracy rate classified to brain network.Current existing brain network class method is most
It is the signal by processing collection to construct brain network, and is classified according to the brain network characterization attribute between brain area, this
A little classification methods only consider a characteristic attribute and have ignored other characteristic attributes of brain network, and ignored characteristic attribute is very
Large effect may be generated to experimental result.
Therefore, the various features attribute of brain network is carried out the fusion based on weight while using the novel kernel of graph by the present invention
Technology classifies to brain network, improves the accuracy rate of brain network class.
Summary of the invention
The technical problem to be solved by the present invention is to a kind of brains based on weight characteristic attribute fusion and novel kernel of graph technology
Network class method improves the accuracy rate of brain network class.
The technical scheme adopted by the invention is that the multi-form characteristic attribute based on weight merges and has used the novel kernel of graph
Brain network class method, which is made of following 4 steps: firstly, by raw experimental data after pretreatment
Complete the building of brain network；Secondly, extracting a variety of brain network attribute values in brain network according to different threshold values；Third utilizes
All data of support vector machines training, according to the superiority and inferiority of training result, sort effect is optimal in every kind of network attribute value
Threshold parameter, and by they carry out the characteristic attribute based on weight fusion；Finally, using the supporting vector based on the novel kernel of graph
The fused feature vector of machine training；
Original Pearson correlation coefficients matrix and different characteristic carry out linear fusion by the present invention, is mentioned with this
The accuracy rate of high-class；By improving figure decomposition method, that is, the subgraph structure of a small size is constructed, the several and node is included
The maximum point of contiguity (that is, the maximum point of Pearson correlation coefficients), is compared by identification function between subgraph structure
Similitude, to construct a kind of novel kernel of graph；
It is an object of the invention to construct a kind of brain network class side based on weight characteristic attribute fusion and the novel kernel of graph
Method can be improved the accuracy rate of brain network class, have good practicability.
Detailed description of the invention
Fig. 1 is the method for the present invention experiment flow figure；
Fig. 2 is to carry out brain network class acquired results figure using the method for the present invention；
Fig. 3 is data prediction flow chart；
Fig. 4 is the brain network class accuracy rate comparison diagram using the method for the present invention and other 8 kinds of methods.Related experimental methods
Classification accuracy comparison.
Specific embodiment
Brain network class method implementation process based on the fusion of weight characteristic attribute and the novel kernel of graph is as shown in Figure 1, this point
Class method is made of 4 steps: firstly, raw experimental data to be completed to the building of brain network after pretreatment；Secondly, according to
Different threshold values extracts a variety of brain network attribute values in brain network；Third trains all data using support vector machines, according to
The superiority and inferiority of training result, the optimal threshold parameter of sort effect in various network attribute values, and they are subjected to feature
Attribute fusion；4th, using the fused feature vector of support vector machines training based on the novel kernel of graph, realize brain network point
Class.
Wherein, characteristic attribute fusion method is, by original Pearson's (Pearson) correlation matrix and different spies
It levies data and carries out linear fusion, the accuracy rate of classification, concrete implementation step are improved with this are as follows: 1. according to original brain network
Matrix calculates different topological attributes.Such as: Pierre's Si correlation matrix P, cluster coefficients Matrix C C, characteristic path length square
Battle array CL and node degree matrix D；2. the matrix 1. obtained (P, CC, CL, D) is carried out thresholding processing, obtain under different threshold value t
Topological attribute matrix (P (t), CCb(t)、CLb(t)、Db(t))；3. training the square by 2. obtaining using multi-kernel support vector machine
Battle array records accurate number, calculates accuracy rate M, wherein the specific calculation of M are as follows: predict to different topology attribute
When, it is that every kind of topological attribute (code name L) distributes a counter CL(positive integer), it is accurate for recording this method prediction
Number.Then MLFor CLWith the ratio of all experiment number N (positive integer), it may be assumed that ML=Ck/N；Divide 4. choosing in every kind of topological attribute
The best thresholding matrix of class effect；5. various topological attributes to be fused into new matrix A.
If Pearson correlation coefficients matrix is P:
The cluster coefficients matrix of optimal threshold is CCb:
Optimal threshold characteristic path length matrix is CLb:
The node degree matrix of optimal threshold is Db:
Then, fused characteristic attribute matrix A are as follows:
So far, the various characteristic attributes of brain network are melted and is incorporated as new experiment sample.
Wherein, the building method of the new kernel of graph is by improving figure decomposition method, that is, to construct the subgraph knot of a small size
Structure passes through identification function comprising several and the maximum point of the node contiguity (that is, the maximum point of Pearson correlation coefficients)
Compare the similitude between subgraph structure, to construct a kind of novel kernel of graph.Specific configuration method is as follows:
Define a symmetric positive semidefinite matrix CG(i, j), the matrix cover the structured message of figure.In this way, brain network
It can be defined in the space of a regularization, convenient for carrying out across comparison analysis to each brain image.
If the adjacency matrix A ∈ R for the figure G that a number of nodes is mm×mWith a positive integer k, then a positive semidefinite square is defined
Battle array CG∈Rk×k:
Wherein, e indicates that unit vector, cov indicate the covariance between two vectors, AiE indicates the matrix A on vector e
I-th power iteration, k is the number of iterations.Algorithm is as follows:
1. input: adjacency matrix A ∈ Rm×m, k, initialization x0=e ∈ Rm×1；
2. for q=1 to k (circulation)
xq=M(:), (q)；
end；
3. μ=e ∈ Rk×1；
④
5. returning to CA∈Rk×k。
Wherein, ∑ is summation symbol, and m is positive integer.
Two figure A ∈ R are calculated with the identification function of the kernel of graphn1×n1With B ∈ Rn2×n2Between Gaussian Profile
Bhattacharya similitude, is defined as follows:
Wherein, det () indicates determinant,CA, CBIt is to be calculated according to formula (6).
In terms of scheming decomposition, the present invention constructs the subgraph structure of a small size, generally comprises and several connect with the node
The maximum point of degree (that is, the maximum point of Pearson correlation coefficient).Network G ∈ R is connected by a pairN×NWith H ∈ RN×NWith parameter b
(b indicates to choose and the maximum b node of the node Pearson correlation coefficient when building sub-network), can construct one group of subnet
Network:
Wherein,For with node ViCentered on, selection and ViThe subnet that constitutes of the highest b point of Pearson correlation coefficient
Network.Side collectionIt is derived from former connection network.It can correspond, and be remained to the greatest extent between each node in figure each in this way
The structural information of figure more than possible.
Using above-mentioned figure decomposition method and the novel kernel of graph of identification construction of function, is defined as:
Wherein, s function are as follows:
Wherein, WithIt is figure respectivelyAnd figureIt is calculated by formula (6).
Wherein, data set and experiment flow are that experimental data is from Alzheimer neuroimaging plan
(Alzheimer Disease neural imagin) ADNI.ADNI was in the neck in Michael doctor W.Weiner in 2004
Lead lower creation, it is intended to exploitation is clinical, iconography, science of heredity and biochemistry biomarker, for early detection and tracking Ah
Alzheimer's disease (Alzheimer Disease, AD) and mild cognitive impairment (Mild Cognitive Impairment,
MCI research).Experiment obtains fMRI (functional magnetic resonance imaging, functional magnetic resonance
Imaging) totally 123, sample, including 72 MCI (mild cognitive impairment) patients, 51 normal person (Normal
Control, NC).
The brain that experimental data image is pre-processed to obtain 116 × 116 as shown in figure 3, firstly, be first associated with by experiment flow
Matrix.Then, different threshold values is used using the matrix thresholding and calculates the attribute values of some networks as the feature of classification
The feature vector of different threshold values is classified using support vector machines, selects the best threshold value of classifying quality by vector.Most
Afterwards, then optimal solution obtained in the previous step is fused to new feature vector and carries out classification based training.
Wherein, the building of brain network is that collected raw experimental data is carried out data prediction, process master first
SPM8 (Statistical Parametric Mapping, statistics parameter figure) is used to carry out time complexity curve to initial data
Space error caused by (Slice Timing), head movement corrects (Head Motion), visual fusion (Co-
Registration), (Normalization) and smoothing (Smoothing) etc. are standardized, detailed process is as shown in Figure 3.Its
In, in space error amendment, if it find that the space of testee is mobile more than 3mm, then then thinking that experimental data loses
Validity and reliability and given up；Then IBASPM (Individual brain atlases using is used
Statistical Parametric Mapping, individual brain map morphological analysis) according to AAL (Automated
Anatomical Labeling) brain of each experiment sample is divided into 116 nodes of 116 brain areas composition brain networks；Again
According to the Pierre of BOLD (Blood Oxygenation Level Dependent Contrast) signal between each brain area
Gloomy correlation power defines the connection between each brain area, i.e. side between each node of brain network.After treatment, each
Sample is abstracted as the weighting fully-connected network of a 116*116.Gained matrix is the symmetrical matrix that diagonal entry is all 1,
And since brain network is to be positively correlated network, therefore all elements in matrix are positive.So far, brain network basic building is completed.It is real
Test operation design parameter are as follows: the scan period is 2 seconds, number of slices 33, totally 225 time points, bandpass filtering range=0.01Hz
~0.08Hz；
Wherein, characteristic attribute extraction is, according to the brain network being previously obtained, the parameter of brain network to be calculated and is made
For the feature vector of support vector machines sample.The network parameter studied has global clustering coefficient (Clustering
Coefficient Global, CCG), Local Clustering coefficient (Clustering Coefficient Local, CCL), node
Spend (Degree, D), average characteristics path length (Characteristic Path Length Global, CPG), local feature
Path length (Charactering Path Local, CPL) and the original Pearson correlation square based on BOLD signal
Battle array.Since CCG and CPG are one-dimensional matrix, so by CCG and CCL, CPG merged with CPL to be formed new matrix obtain CC and
CL；
Brain network has preferable classification performance when connection ratio between brain network node is between 25% to 75%, will use
Different numerical value carry out thresholding brain connection network, specially [0.20,0.25,0.30,0.35,0.38,0.40,0.45,0.50].Then
It can arrive 3 kinds of data of D, CC and CL, each 8 groups of every kind of data；
Wherein, model training is that 24 groups of experimental datas of acquired ` are used characteristic attribute fusion method set forth above
It is merged, is then classified using the support vector machines with the novel kernel of graph.At random by 123 experimental datas point when experiment
For two parts, wherein training set accounts for 2/3, test set and accounts for 1/3.Model parameter is adjusted by the way of cross validation to divide to improve
Class accuracy rate；
In experiment, use classification accuracy as evaluation criterion, since the limitation of experimental data number causes to test every time
Result difference it is larger, so this experiment measure 400 times experiment average classification accuracy variation, as shown in Figure 2.The present invention
Classification accuracy rate is about 84.21%, it is compared with the accuracy rate of following 8 kinds of existing classification methods, compares knot
Fruit is as shown in Figure 4.Wherein, method 1 is to use initial data (Perason relative coefficient) as the SVM brain network of classifier
Classification method, method 2 are the SVM brain network class method based on single network characteristic attribute (characteristic path length), method 3
It is the SVM brain network class method based on shortest path core, method 4 is based on WL (Weisfeiler-Lehman)-edge core
SVM brain network class method, method 5 are the SVM brain network class methods based on WL-substree core, and method 6 is based on WL-
The SVM brain network class method of shortest path core, method 7 are the SVM brain network class method of the kernel of graph dimensionality reduction of Wang Lipeng, method
8 be to connect target multicore SVM brain network class method.The classification accuracy of the method for the present invention is above other 8 kinds of methods.
It is merged the present invention is based on weight characteristic attribute and the novel kernel of graph is used to classify brain network, the method for the present invention
Classification accuracy is improved compared with other classification methods, demonstrates the superiority of the method for the present invention.The present invention is suitable for brain
The fields such as the analysis of area's lesion, Alzheimer's disease and cognitive disorder Analysis on Mechanism.
Claims (4)
1. a kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph, which is characterized in that the classification side
Method is made of following 4 steps: firstly, raw experimental data to be completed to the building of brain network after pretreatment；Secondly, according to
Different threshold values extracts a variety of brain network attribute values in brain network；Third trains all data using support vector machines, according to
The superiority and inferiority of training result, the optimal threshold parameter of sort effect in every kind of network attribute value, and they are based on
The characteristic attribute of weight merges；Finally, using the fused feature vector of support vector machines training based on the novel kernel of graph.
2. a kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph according to claim 1,
It is characterized in that, original Pearson's (Pearson) correlation matrix is subjected to linear fusion from different characteristics, with
This accuracy rate to improve classification, concrete implementation step are 1. to calculate different topologys according to original brain network matrix and belong to
Property, such as: the degree matrix D of Pierre's Si correlation matrix P, cluster coefficients Matrix C C, characteristic path length Matrix C L and node；
2. by the matrix 1. obtained (P, CC, CL, D) carry out thresholding processing, obtain under different threshold value t topological attribute matrix (P (t),
CCb(t)、CLb(t)、Db(t))；3. 2. using multicore SVM (Support Vector Machine, support vector machines) training
The matrix arrived, records accurate number, calculates accuracy rate M, and wherein the specific calculation of M is, to different topology attribute into
It is that every kind of topological attribute (code name L) distributes a counter C when row predictionL(positive integer), for recording this method prediction
Accurate number；Then, MLFor CLWith the ratio of all experiment number N (positive integer), it may be assumed that ML=Ck/N；4. choosing every kind of topology to belong to
The best thresholding matrix of classifying quality in property；5. every various topological attributes are fused into new matrix A；
If Pearson correlation coefficients matrix is P:
The cluster coefficients matrix of optimal threshold is CCb:
Optimal threshold characteristic path length matrix is CLb:
The node degree matrix of optimal threshold is Db:
Then, fused characteristic attribute matrix A are as follows:
So far, the various characteristic attributes of brain network are melted and is incorporated as new experiment sample.
3. a kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph according to claim 1,
It is characterized in that, constructing the subgraph structure of a small size by improving figure decomposition method, it is connect comprising several with the node
The maximum point of degree (that is, the maximum point of Pearson correlation coefficients), the phase between subgraph structure is compared by identification function
Like property, to construct a kind of novel kernel of graph；Specific configuration method is as follows:
Define a symmetric positive semidefinite matrix CG(i, j), the matrix cover some structured messages of figure.In this way, brain network can
To be defined in the space of a regularization, convenient for carrying out lateral comparative analysis to each brain image；
If the adjacency matrix A ∈ R for the figure G that a number of nodes is mm×mWith a positive integer k, then a positive semidefinite matrix C is definedG
∈Rk×k:
Wherein, e indicates that unit vector, cov indicate the covariance between two vectors, AiE indicates i-th of the matrix A on vector e
Secondary power iteration, k are the number of iterations.Algorithm is as follows:
1. input: adjacency matrix A ∈ Rm×m, k, initialization x0=e ∈ Rm×1；
2. for q=1 to k is recycled
xq=M(:), (q)
End:
3. μ=e ∈ Rk×1；
④
5. returning to CA∈Rk×k
Wherein, ∑ is summation symbol, and m is positive integer；
Two figure A ∈ R are calculated with the identification function of the kernel of graphn1×n1With B ∈ Rn2×n2Bhattacharya between Gaussian Profile
Similitude is defined as follows:
Wherein, det () indicates determinant,CA, CBIt is calculated according to formula (6)；
In terms of scheming decomposition, the present invention constructs the subgraph structure of a small size, generally comprises the several and node contiguity
Maximum point (that is, the maximum point of Pearson correlation coefficient)；Network G ∈ R is connected by a pairN×NWith H ∈ RN×NWith parameter b (b table
Chosen and the maximum b node of the node Pearson correlation coefficient when showing building sub-network), one group of sub-network can be constructed:
Wherein,For with node ViCentered on, selection and ViThe sub-network that constitutes of the highest b point of Pearson correlation coefficient.
Side collectionIt is derived from former connection network；It can be corresponded between each node in figure each in this way, and remain and to the greatest extent may be used
The structural information of figure more than energy；
Using above-mentioned figure decomposition method and the novel kernel of graph of identification construction of function, is defined as:
Wherein, s function are as follows:
Wherein, WithIt is figure respectivelyAnd figureIt is calculated by formula (6).
4. a kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph according to claim 1,
It is characterized in that, firstly, first experimental data image is pre-processed to obtain 116 × 116 brain incidence matrix；Then, it uses
Different threshold values is using the matrix thresholding and calculates the attribute values of some networks as the feature vector of classification, uses SVM will
The feature vector of different threshold values is classified, and the best threshold value of classifying quality is selected；Finally, again will be obtained in the previous step optimal
Solution be fused to new feature vector and carry out classification based training.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910332184.4A CN110084381A (en) | 2019-04-22 | 2019-04-22 | A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910332184.4A CN110084381A (en) | 2019-04-22 | 2019-04-22 | A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110084381A true CN110084381A (en) | 2019-08-02 |
Family
ID=67416392
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910332184.4A Pending CN110084381A (en) | 2019-04-22 | 2019-04-22 | A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110084381A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110459317A (en) * | 2019-08-13 | 2019-11-15 | 东北大学 | Alzheimer disease assistant diagnosis system and method based on the dynamic brain network kernel of graph |
| CN111009324A (en) * | 2019-11-25 | 2020-04-14 | 东北大学 | Mild cognitive impairment auxiliary diagnosis system and method based on brain network multi-feature analysis |
| CN111063423A (en) * | 2019-12-16 | 2020-04-24 | 哈尔滨工程大学 | Method for extracting specific structure of brain network of Alzheimer disease and mild cognitive impairment |
| CN111627553A (en) * | 2020-05-26 | 2020-09-04 | 四川大学华西医院 | Method for constructing individualized prediction model of first-onset schizophrenia |
| CN112784886A (en) * | 2021-01-11 | 2021-05-11 | 南京航空航天大学 | Brain image classification method based on multilayer maximum spanning tree image kernel |
| CN113313232A (en) * | 2021-05-19 | 2021-08-27 | 华南理工大学 | Functional brain network classification method based on pre-training and graph neural network |
| CN113902024A (en) * | 2021-10-20 | 2022-01-07 | 浙江大立科技股份有限公司 | Small-volume target detection and identification method based on deep learning and dual-band fusion |
| CN113902024B (en) * | 2021-10-20 | 2024-06-04 | 浙江大立科技股份有限公司 | Small-volume target detection and identification method based on deep learning and dual-band fusion |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103886328A (en) * | 2014-03-19 | 2014-06-25 | 太原理工大学 | Functional magnetic resonance image data classification method based on brain network modular structure characteristics |
| CN105726026A (en) * | 2016-01-28 | 2016-07-06 | 电子科技大学 | Mild cognitive impairment disease classifying method based on brain network and brain structure information |
| CN106650768A (en) * | 2016-09-27 | 2017-05-10 | 北京航空航天大学 | Gaussian image model-based brain network modeling and mode classification method |
| CN109409403A (en) * | 2018-09-12 | 2019-03-01 | 太原理工大学 | Brain network clustering method based on local attribute and topological structure |
-
2019
- 2019-04-22 CN CN201910332184.4A patent/CN110084381A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103886328A (en) * | 2014-03-19 | 2014-06-25 | 太原理工大学 | Functional magnetic resonance image data classification method based on brain network modular structure characteristics |
| CN105726026A (en) * | 2016-01-28 | 2016-07-06 | 电子科技大学 | Mild cognitive impairment disease classifying method based on brain network and brain structure information |
| CN106650768A (en) * | 2016-09-27 | 2017-05-10 | 北京航空航天大学 | Gaussian image model-based brain network modeling and mode classification method |
| CN109409403A (en) * | 2018-09-12 | 2019-03-01 | 太原理工大学 | Brain network clustering method based on local attribute and topological structure |
Non-Patent Citations (2)
| Title |
|---|
| NAN YANG: "Brain Network Classification Based on Multiple Brain Networks Attributes", 《INTERNATIONAL JOURNAL OF ENGINEERING RESEARCH AND MANAGEMENT》 * |
| 王立鹏 等: "基于子图选择和图核降维的脑网络分类方法", 《计算机科学与探索》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110459317A (en) * | 2019-08-13 | 2019-11-15 | 东北大学 | Alzheimer disease assistant diagnosis system and method based on the dynamic brain network kernel of graph |
| CN110459317B (en) * | 2019-08-13 | 2023-03-24 | 东北大学 | Alzheimer disease auxiliary diagnosis system and method based on dynamic brain network image core |
| CN111009324A (en) * | 2019-11-25 | 2020-04-14 | 东北大学 | Mild cognitive impairment auxiliary diagnosis system and method based on brain network multi-feature analysis |
| CN111009324B (en) * | 2019-11-25 | 2023-11-28 | 东北大学 | Auxiliary diagnosis system and method for mild cognitive impairment through multi-feature analysis of brain network |
| CN111063423A (en) * | 2019-12-16 | 2020-04-24 | 哈尔滨工程大学 | Method for extracting specific structure of brain network of Alzheimer disease and mild cognitive impairment |
| CN111063423B (en) * | 2019-12-16 | 2022-05-20 | 哈尔滨工程大学 | Method for extracting specific structure of brain network of Alzheimer disease and mild cognitive impairment |
| CN111627553A (en) * | 2020-05-26 | 2020-09-04 | 四川大学华西医院 | Method for constructing individualized prediction model of first-onset schizophrenia |
| CN112784886A (en) * | 2021-01-11 | 2021-05-11 | 南京航空航天大学 | Brain image classification method based on multilayer maximum spanning tree image kernel |
| CN112784886B (en) * | 2021-01-11 | 2024-04-02 | 南京航空航天大学 | Brain image classification method based on multi-layer maximum spanning tree graph core |
| CN113313232A (en) * | 2021-05-19 | 2021-08-27 | 华南理工大学 | Functional brain network classification method based on pre-training and graph neural network |
| CN113902024A (en) * | 2021-10-20 | 2022-01-07 | 浙江大立科技股份有限公司 | Small-volume target detection and identification method based on deep learning and dual-band fusion |
| CN113902024B (en) * | 2021-10-20 | 2024-06-04 | 浙江大立科技股份有限公司 | Small-volume target detection and identification method based on deep learning and dual-band fusion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110084381A (en) | A kind of brain network class method based on weight characteristic attribute fusion and the novel kernel of graph | |
| CN107909117B (en) | Classification device for early and late mild cognitive impairment based on brain function network characteristics | |
| CN107133651B (en) | The functional magnetic resonance imaging data classification method of subgraph is differentiated based on super-network | |
| Li et al. | Pooling regularized graph neural network for fmri biomarker analysis | |
| Poggi et al. | Supervised segmentation of remote sensing images based on a tree-structured MRF model | |
| CN110443798A (en) | A kind of self-closing disease detection method based on magnetic resonance image, apparatus and system | |
| CN113040715A (en) | Human brain function network classification method based on convolutional neural network | |
| CN104715261A (en) | FMRI dynamic brain function sub-network construction and parallel connection SVM weighted recognition method | |
| CN112418337B (en) | Multi-feature fusion data classification method based on brain function hyper-network model | |
| CN109409403B (en) | Brain network clustering method based on local attribute and topological structure | |
| CN113947157B (en) | Dynamic brain effect connection network generation method based on hierarchical clustering and structural equation model | |
| CN113558603B (en) | Multi-modal cognitive disorder recognition method based on deep learning | |
| CN102509123A (en) | Brain functional magnetic resonance image classification method based on complex network | |
| CN111090764A (en) | Image classification method and device based on multitask learning and graph convolution neural network | |
| Özbılge et al. | Tomato disease recognition using a compact convolutional neural network | |
| CN112634214A (en) | Brain network classification method combining node attributes and multilevel topology | |
| Vega-Pons et al. | Classification of inter-subject fMRI data based on graph kernels | |
| Sathish et al. | Exponential cuckoo search algorithm to radial basis neural network for automatic classification in MRI images | |
| CN108596236A (en) | It is a kind of to roll into a ball partition method based on the thalamic nuclei of global connection features and geodesic distance | |
| Jiao et al. | Module partitioning for multilayer brain functional network using weighted clustering ensemble | |
| Slavakis et al. | Clustering brain-network time series by Riemannian geometry | |
| Moreno et al. | Regional multiscale motion representation for cardiac disease prediction | |
| Costa et al. | Cluster analysis using self-organizing maps and image processing techniques | |
| Yang et al. | Automatic segmentation of left ventricular myocardium by deep convolutional and de-convolutional neural networks | |
| CN110604572A (en) | Brain activity state identification method based on human brain characteristic map |
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 | ||
| DD01 | Delivery of document by public notice | ||
| DD01 | Delivery of document by public notice |
Addressee: Zhang Dakun Document name: Notice of First Examination Opinion |
|
| DD01 | Delivery of document by public notice | ||
| DD01 | Delivery of document by public notice |
Addressee: Zhang Dakun Document name: Deemed withdrawal notice |
|
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190802 |