CN109730678A - A kind of multilayer cerebral function network module division methods - Google Patents
A kind of multilayer cerebral function network module division methods Download PDFInfo
- Publication number
- CN109730678A CN109730678A CN201910083223.1A CN201910083223A CN109730678A CN 109730678 A CN109730678 A CN 109730678A CN 201910083223 A CN201910083223 A CN 201910083223A CN 109730678 A CN109730678 A CN 109730678A
- Authority
- CN
- China
- Prior art keywords
- brain
- module
- cluster
- cerebral function
- function 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.)
- Pending
Links
Landscapes
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The present invention relates to a kind of module partition method of multilayer cerebral function network, key step is as follows: pre-processing to cerebral function magnetic resonance image, by treated, image carries out brain area division and time series extraction；Time series is divided into the identical subsegment of length using sliding window method, each brain area is calculated and corresponds to Pearson correlation between time subsegment, constructs the multilayer cerebral function network changed over time；It selects Different brain region as initialization cluster centre using k-means, carries out multiple base clustering, divide a corresponding Subject Matrix and similarity matrix every time；Each division is assessed using Cluster Validity Index, using assessed value as the weight of similarity matrix, construction weighting similarity matrix；Weighting similarity matrix is divided using Fuzzy C-means cluster, division result is evaluated using Q function, obtains the module division result of each layer cerebral function network.
Description
Technical field
The invention belongs to biomedical information processing technology fields, specifically, being a kind of multilayer cerebral function network mould
Block division methods.
Background technique
Brain is one of system most complicated in nature, about 10 in the brain of an adult11A neuron is thin
The neuronal cell of born of the same parents, these enormous amounts pass through about 1015A cynapse is connected with each other, and forms a highly complex brain net
Network.Brain is carrying out needing to interact between multiple brain areas or neuron when certain activities perhaps execute a certain task, phase
Mutual cooperation.Studies have shown that human cortical brain's thickness network has the group corresponding to cerebral function module (such as vision, language)
Mode is knitted, brain is completely embedded between the brain area of mutually coordinated work when executing certain tasks, constitutes the function mould of brain
Block.In complex network, module is made of the node being completely embedded, and is completely embedded between the node of inside modules, and module with
Connection between module is relatively sparse.
The research of functional network is carried out to brain, it is also necessary to be acquired to data dependent on brain image technology.Function magnetic
Resonance image-forming is a kind of emerging neuroimaging mode, compared with electroencephalogram, possesses better spatial resolution, space point
Resolution can achieve millimeter level.As a kind of non-damage brain imaging technique, magnetic resonance imaging plays in brain function research
Irreplaceable role.The Blood oxygen level dependence signal of functional mri measurement is by deoxyhemoglobin and oxygenated blood
Caused by the magnetic susceptibility variation of Lactoferrin, when the neuronal populations activity in some region, metabolic demand will be improved, and can be caused
The change of local cerebral blood flow and the oxygen content of blood.The change of functional mri energy real-time tracking signal, such as thinking activities
Or recognize the variation of signal in experiment.Large quantities of brain science researchers, which has begun, is engaged in grinding for magnetic resonance functional neurosurgery imaging
Study carefully, and it is applied to Cognitive Neuroscience.Tranquillization state functional MRI method is with its simple experimental design, higher noise
It is widely adopted than, simple flow chart of data processing.
In recent years, researcher recognizes the connection in cerebral function network between each brain area and brain area and brain area,
In different times with changes will occur under frequency.Compared with traditional cerebral function network, multilayer cerebral function network is examined
Cerebral function network is considered at any time or the variation of frequency.Cerebral function network modularization is studied on the basis of multitiered network
Feature helps to study in cerebral function network module at any time or the Variation Features of frequency, and to multilayer cerebral function net
The analysis of network progress module, it is necessary first to the division of module is carried out to multilayer cerebral function network.
Summary of the invention
In view of the shortcomings of the prior art with the needs of practical application, the problem to be solved in the present invention is:
A kind of module partition method of multilayer cerebral function network is provided, is realized at any time or the multilayer of frequency variation
The module of cerebral function network divides, and provides better analysis foundation for multilayer cerebral function network.
In order to achieve the above object, the present invention takes following technical scheme:
A kind of multilayer cerebral function network module division methods, comprising the following steps:
(1) magnetic resonance imaging data is pre-processed, finally carries out low frequency filtering, reduce the life of low frequency wonder and high frequency
Object noise；
(2) it selectes a kind of standardization brain region template and brain is divided into several brain areas, each brain area respectively corresponds
A node in cerebral function network；
(3) the time series average value for calculating each all voxels of brain area in magnetic resonance image, extracts various criterion subregion
Corresponding time series；
(4) time series of each brain area is divided into using sliding window method by several overlapped, the identical subsegment of window, it is right
Each time subsegment calculates Pearson correlation, constructs the multilayer cerebral function network changed over time；
(5) module division is carried out to the first layer network in multilayer cerebral function network, is selected using k-means clustering method
Different brain areas carries out M base cluster as initialization cluster centre, the corresponding Subject Matrix of each base cluster Pm:
Wherein, in (1) formula, Km indicates the division classification number of base cluster Pm, and N indicates that all data object numbers are i.e. all
Brain area number；
(6) similarity matrix between corresponding Subject Matrix Um building brain area is clustered according to base:
Sm={ 0,1 }N×N
Wherein, in (2) formula, element (Sm) ij in similarity matrix Sm indicates the Subject Matrix Um Midbrain Area i of base cluster
Whether belong to the same cluster with brain area j；If brain area i is the same cluster with brain area j, the value of element (Sm) ij is 1, is otherwise taken
Value is 0；
(7) Cluster Validity assessment is carried out to M base cluster, it is assumed that the validity evaluation index selected is Π, then corresponds to
The secondary base cluster Pm efficiency assessment of m (1≤m≤M) is Π (Pm), weight is carried out according to the assessed value at evaluation index Π
It calculates:
Wherein, in (3) formula,When expression is evaluated using Validity Index Π, the shared weight of the m times cluster；
(8) the cluster weight under efficiency assessment index Π is calculatedWeighting phase is constituted in conjunction with base cluster similarity matrix
Like property matrix:
(9) using Fuzzy C-means clustering method to weighting similarity matrix SΠLast clustering is carried out, by the layer
All brain areas of network are divided into c module；
(10) the modularity function Q value in the case of different c values, the expression formula of modularity Q function are calculated are as follows:
Wherein, in (5) formula, m indicates the number on the side in the layer network, and Aij indicates the element in network connection matrix, if
Node i and node j two o'clock have Bian Xianglian, then Aij=1, are otherwise equal to 0；Di is the number of module belonging to node i, δ (di,dj) be
δ function, for judging whether two nodes belong to the same module, if node i and node j belong to a module, i.e. di=
When dj, δ function value is 1, and otherwise value is 0；The degree of wi expression node i；
(11) this layer of final module division result is evaluated using modularity function Q；If the Q value under difference c value
The requirement of module division is not satisfied, return step (5) changes the value of Km, and the module for re-starting this layer of cerebral function network is drawn
Point；Otherwise the module chosen under optimal Q value is divided as the layer network in module division result, and to next layer of cerebral function net
Network carries out module division.
Further improvement of the present invention, the time series of all brain areas is divided into using sliding window method it is several overlapped,
The identical subsegment of window；If time window length is L, the interval steps between window are S, each brain area corresponding total time
Sequence length is P, and entire time series is divided into l sections, and expression formula is as follows:
The correlation for calculating each time subsegment constructs the l layer cerebral function network changed over time.
Further improvement of the present invention, for each layer of cerebral function network, each base cluster is selected using k-means
Different brain areas is as initialization cluster centre, and the similarity matrix that different cluster centres divide corresponds to different weights, finally
Weighting similarity matrix in merge multiple base cluster and cluster to obtain as a result, carrying out Fuzzy C-means again on this basis
Last module divides.
Beneficial effects of the present invention: (1) compared with traditional cerebral function network, the present invention, which considers, to be changed over time
Base is done in the division of module in multilayer cerebral function network, the research changed over time for module in subsequent multi-layer cerebral function network
Plinth.
(2) through the invention in weighted cluster integrate to multilayer cerebral function network carry out module division, help
Effect is divided in the module for improving cerebral function network.
Detailed description of the invention
Fig. 1 is implementation flow chart of the invention.
Specific embodiment
In order to deepen the understanding of the present invention, the present invention is done below in conjunction with drawings and examples and is further retouched in detail
It states, the embodiment is only for explaining the present invention, does not constitute and limits to protection scope of the present invention.
A kind of embodiment: specific embodiment of multilayer cerebral function network module division methods, comprising the following steps:
(1) the collected brain image of magnetic resonance equipment is read out and format is converted: using normal group subject 30
The tranquillization state magnetic resonance scanning data of (each 15 people of men and women).The magnetic resonance image data read is converted to by DICOM format
NIFTI format.The pretreatment such as time correction, the dynamic correction of head, registration, Spatial normalization, smooth is carried out again, finally carries out low frequency filter
Wave reduces the biological noise of low frequency wonder and high frequency.In the present embodiment, low frequency filtering range takes 0.01Hz~0.08Hz.
(2) a kind of standardization brain region template (such as AAL Partition Mask, Brodmann Partition Mask, CH are selected2Subregion
Template etc.) brain is divided into several brain areas, each brain area respectively corresponds a node in cerebral function network.This implementation
In example, the division that AAL brain Partition Mask carries out brain area is chosen, human brain is divided into 90 (left and right each 45 of half brain) brain areas, 90
A brain area corresponds to 90 nodes in cerebral function network.The time series for calculating each all voxels of brain area in magnetic resonance image is flat
Mean value extracts the corresponding time series of 90 brain areas.
(3) 30 normal subjects, the corresponding time sequence of 90 brain areas being normally tested to 30 are had chosen in the present embodiment
Column do double sample T inspection two-by-two, examine in this group of data of selection that whether there is or not abnormal datas between subject.
(4) to carrying out subject data without exception after double sample T inspection, using sliding window method by the time sequence of each brain area
Column are divided into several overlapped, identical subsegments of window.If time window length is L, the interval steps between window are
S, the corresponding total length of time series of each brain area is P, and entire time series is divided into l sections, then l are as follows:
In the present embodiment, each brain area time series overall length P is 137, and it is 5 that sliding window L, which takes 67, window interval step-length S,.
(5) Pearson correlation is calculated to each time subsegment, constructs the l layers of multilayer cerebral function net changed over time
Network.
(6) module division is carried out to the first layer network in multilayer cerebral function network, selects Different brain region using k-means
As initialization cluster centre, it is assumed that carry out M base cluster, the corresponding Subject Matrix of each base cluster Pm:
Wherein Km indicates that the division classification number of base cluster Pm, N indicate all data object number, that is, brain area numbers.This implementation
N is 90 in example.
(7) similarity matrix between corresponding Subject Matrix Um building brain area is clustered according to base:
Sm={ 0,1 }N×N (3)
Wherein, element (Sm) ij in the similarity matrix Sm between brain area indicates the Subject Matrix Um Midbrain Area i of base cluster
Whether belong to the same cluster with brain area j.If brain area i belongs to the same cluster with brain area j, the value of element (Sm) ij is 1, otherwise
Value is 0.
(8) Cluster Validity assessment is carried out to M base cluster, it is assumed that the validity evaluation index selected is Π, then corresponds to
The secondary base cluster Pm efficiency assessment of m (1≤m≤M) is Π (Pm), weight is carried out according to the assessed value at evaluation index Π
It calculates:Wherein,Indicating to utilize has
When effect property index Π is evaluated, the shared weight of the m times cluster.
(9) the cluster weight under efficiency assessment index Π is calculatedWeighting phase is constituted in conjunction with base cluster similarity matrix
Like property matrix:
(10) using Fuzzy C-means clustering method to weighting similarity matrix SπLast clustering is carried out, is obtained
The module division result of first layer network, i.e., be divided into c module for brain area.
(11) last division result is evaluated using modularity function Q, calculates the modularity in the case of different c values
Function Q value selects the module under optimal Q value to divide the module division result final as the layer network, Q function expression:
In formula, in formula, m indicates the number on the side in the layer network, and Aij indicates the element in network connection matrix, if section
Point i and node j two o'clock have Bian Xianglian, then Aij=1, are otherwise equal to 0；Di is the number of module belonging to node i, δ (di,dj) it is δ
Function, for judging whether two nodes belong to the same module, if node i and node j belong to a module, i.e. di=dj
When, δ function value is 1, and otherwise value is 0；The degree of wi expression node i；
(12) if module is not satisfied in the Q value of the module division result under difference c value, division is required, return step (6),
Change in base cluster and cluster classification number, re-starts the division of module；If meeting module divides condition, will be under optimal Q value
Module divides the module division result as the layer network, and the module for carrying out next layer of cerebral function network divides；
(13) drawing for module is carried out to each layer network in multilayer cerebral function network using above-mentioned module partition method
Point, obtain the module division result of entire multilayer cerebral function network.
Basic principles and main features and advantage of the invention have been shown and described above.The technical staff of the industry should
Understand, the present invention is not limited to the above embodiments, and the above embodiments and description only describe originals of the invention
Reason, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes and improvements
It all fall within the protetion scope of the claimed invention.The claimed scope of the invention is by appended claims and its equivalent circle
It is fixed.
Claims (3)
1. a kind of multilayer cerebral function network module division methods, which comprises the following steps:
(1) magnetic resonance imaging data is pre-processed, finally carries out low frequency filtering, the biology for reducing low frequency wonder and high frequency is made an uproar
Sound；
(2) it selectes a kind of standardization brain region template and brain is divided into several brain areas, each brain area respectively corresponds brain
A node in functional network；
(3) the time series average value for calculating each all voxels of brain area in magnetic resonance image, it is corresponding to extract various criterion subregion
Time series；
(4) time series of each brain area is divided into using sliding window method by several overlapped, the identical subsegment of window, to each
Time subsegment calculates Pearson correlation, constructs the multilayer cerebral function network changed over time；
(5) module division is carried out to the first layer network in multilayer cerebral function network, is selected using k-means clustering method different
Brain area carry out M base cluster as initialization cluster centre, each base cluster Pm corresponds to a Subject Matrix:
Wherein, in (1) formula, Km indicates the division classification number of base cluster Pm, and N indicates all data object number, that is, all brain areas
Number；
(6) similarity matrix between corresponding Subject Matrix Um building brain area is clustered according to base:
(2)
Sm={ 0,1 }N×N
Wherein, in (2) formula, element (Sm) ij in similarity matrix Sm indicates the Subject Matrix Um Midbrain Area i of base cluster with brain
Whether area j belongs to the same cluster；If brain area i is the same cluster with brain area j, the value of element (Sm) ij is 1, and otherwise value is
0；
(7) Cluster Validity assessment is carried out to M base cluster, it is assumed that the validity evaluation index selected is Π, then corresponds to m (1
≤ m≤M) secondary base cluster Pm efficiency assessment is Π (Pm), the calculating of weight is carried out according to the assessed value at evaluation index Π:
(3)
Wherein, in (3) formula,When expression is evaluated using Validity Index Π, the shared weight of the m times cluster；
(8) the cluster weight under efficiency assessment index Π is calculatedWeighting similitude is constituted in conjunction with base cluster similarity matrix
Matrix:(
(9) using Fuzzy C-means clustering method to weighting similarity matrix SΠLast clustering is carried out, by the layer network
All brain areas be divided into c module；
(10) the modularity function Q value in the case of different c values, the expression formula of modularity Q function are calculated are as follows:(
Wherein, in (5) formula, m indicates the number on the side in the layer network, and Aij indicates the element in network connection matrix, if node
I and node j two o'clock have Bian Xianglian, then Aij=1, are otherwise equal to 0；Di is the number of module belonging to node i, δ (di,dj) it is δ letter
Number, for judging whether two nodes belong to the same module, if node i and node j belong to a module, i.e. di=dj
When, δ function value is 1, and otherwise value is 0；The degree of wi expression node i；
(11) this layer of final module division result is evaluated using modularity function Q；If the Q value under difference c value is not
Meet the requirement of module division, return step (5) changes the value of Km, and the module for re-starting this layer of cerebral function network divides；
Otherwise the module chosen under optimal Q value is divided as the layer network in module division result, and to next layer of cerebral function network
Carry out module division.
2. a kind of multilayer cerebral function network module division methods according to claim 1, which is characterized in that utilize sliding window
The time series of all brain areas is divided into several overlapped, the identical subsegment of window by method；If time window length is L, window
Interval steps between mouthful are S, and the corresponding total length of time series of each brain area is P, and entire time series is divided into l sections,
Expression formula is as follows:
The correlation for calculating each time subsegment constructs the l layer cerebral function network changed over time.
3. a kind of multilayer cerebral function network module division methods according to claim 1, which is characterized in that for each
Layer cerebral function network, each base cluster select different brain areas as initialization cluster centre using k-means, and difference is poly-
The similarity matrix that class center divides corresponds to different weights, and multiple base cluster is merged in last weighting similarity matrix
It is divided as a result, carrying out the module that Fuzzy C-means clusters to the end again on this basis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910083223.1A CN109730678A (en) | 2019-01-28 | 2019-01-28 | A kind of multilayer cerebral function network module division methods |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910083223.1A CN109730678A (en) | 2019-01-28 | 2019-01-28 | A kind of multilayer cerebral function network module division methods |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109730678A true CN109730678A (en) | 2019-05-10 |
Family
ID=66366516
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910083223.1A Pending CN109730678A (en) | 2019-01-28 | 2019-01-28 | A kind of multilayer cerebral function network module division methods |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109730678A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111227827A (en) * | 2020-02-14 | 2020-06-05 | 广东司法警官职业学院 | Electroencephalogram signal analysis method based on community division algorithm |
| CN111241106A (en) * | 2020-01-15 | 2020-06-05 | 平安科技（深圳）有限公司 | Approximate data processing method, device, medium and electronic equipment |
| 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 |
| CN112580468A (en) * | 2020-12-11 | 2021-03-30 | 西安交通大学 | Method for detecting time-stable community in brain function network |
| CN112716477A (en) * | 2020-12-28 | 2021-04-30 | 聊城大学 | Estimation method of modularized brain function connection network |
| CN113657527A (en) * | 2021-08-23 | 2021-11-16 | 广东工业大学 | Brain functional region division method based on Bayes non-parametric model |
| CN114159043A (en) * | 2021-12-17 | 2022-03-11 | 天津大学 | Brain function network abnormal brain node data detection method based on Qcut algorithm |
| CN115219067A (en) * | 2022-09-20 | 2022-10-21 | 金乡县成启仓储服务有限公司 | Real-time state monitoring method for garlic storage |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104700120A (en) * | 2015-03-23 | 2015-06-10 | 南京工业大学 | Method for extracting and classifying fMRI features based on adaptive entropy algorithm for projection clustering (APEC) |
| CN105117731A (en) * | 2015-07-17 | 2015-12-02 | 常州大学 | Community partition method of brain functional network |
| CN105125213A (en) * | 2015-07-24 | 2015-12-09 | 常州大学 | Brain functional network activity level measurement method |
| CN107358022A (en) * | 2017-06-02 | 2017-11-17 | 常州大学 | A kind of Modularity analysis method of cerebral function network |
| CN108447546A (en) * | 2018-02-02 | 2018-08-24 | 常州大学 | Cerebral function link block division methods based on weighted network |
-
2019
- 2019-01-28 CN CN201910083223.1A patent/CN109730678A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104700120A (en) * | 2015-03-23 | 2015-06-10 | 南京工业大学 | Method for extracting and classifying fMRI features based on adaptive entropy algorithm for projection clustering (APEC) |
| CN105117731A (en) * | 2015-07-17 | 2015-12-02 | 常州大学 | Community partition method of brain functional network |
| CN105125213A (en) * | 2015-07-24 | 2015-12-09 | 常州大学 | Brain functional network activity level measurement method |
| CN107358022A (en) * | 2017-06-02 | 2017-11-17 | 常州大学 | A kind of Modularity analysis method of cerebral function network |
| CN108447546A (en) * | 2018-02-02 | 2018-08-24 | 常州大学 | Cerebral function link block division methods based on weighted network |
Non-Patent Citations (2)
| Title |
|---|
| 白亮 等: "基于聚类准则融合的加权聚类集成算法", 《山西大学学报（自然科学版）》 * |
| 赵学武 等: "面向fMRI数据的人脑功能划分", 《科学通报》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111241106A (en) * | 2020-01-15 | 2020-06-05 | 平安科技（深圳）有限公司 | Approximate data processing method, device, medium and electronic equipment |
| CN111241106B (en) * | 2020-01-15 | 2023-08-29 | 平安科技（深圳）有限公司 | Approximation data processing method, device, medium and electronic equipment |
| CN111227827A (en) * | 2020-02-14 | 2020-06-05 | 广东司法警官职业学院 | Electroencephalogram signal analysis method based on community division algorithm |
| 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 |
| CN112580468A (en) * | 2020-12-11 | 2021-03-30 | 西安交通大学 | Method for detecting time-stable community in brain function network |
| CN112716477A (en) * | 2020-12-28 | 2021-04-30 | 聊城大学 | Estimation method of modularized brain function connection network |
| CN112716477B (en) * | 2020-12-28 | 2024-03-08 | 聊城大学 | Method for estimating modularized brain function connection network |
| CN113657527A (en) * | 2021-08-23 | 2021-11-16 | 广东工业大学 | Brain functional region division method based on Bayes non-parametric model |
| CN113657527B (en) * | 2021-08-23 | 2023-07-07 | 广东工业大学 | Brain function partitioning method based on Bayesian non-parametric model |
| CN114159043A (en) * | 2021-12-17 | 2022-03-11 | 天津大学 | Brain function network abnormal brain node data detection method based on Qcut algorithm |
| CN115219067A (en) * | 2022-09-20 | 2022-10-21 | 金乡县成启仓储服务有限公司 | Real-time state monitoring method for garlic storage |
| CN115219067B (en) * | 2022-09-20 | 2023-01-03 | 金乡县成启仓储服务有限公司 | Real-time state monitoring method for garlic storage |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109730678A (en) | A kind of multilayer cerebral function network module division methods | |
| CN113616184B (en) | Brain network modeling and individual prediction method based on multi-mode magnetic resonance image | |
| CN109522894B (en) | Method for detecting dynamic covariation of fMRI brain network | |
| CN110598793B (en) | Brain function network feature classification method | |
| CN115359045B (en) | Image volume neural network disease prediction system based on multi-mode magnetic resonance imaging | |
| CN113558603B (en) | Multi-modal cognitive disorder recognition method based on deep learning | |
| CN112418337B (en) | Multi-feature fusion data classification method based on brain function hyper-network model | |
| Zhu et al. | Discovering dense and consistent landmarks in the brain | |
| Huang et al. | Coherent pattern in multi-layer brain networks: Application to epilepsy identification | |
| CN116503680A (en) | Brain image structured analysis and brain disease classification system based on brain atlas | |
| CN111513717A (en) | Method for extracting brain functional state | |
| CN111488934A (en) | Brain image data processing method, storage medium, computer device and apparatus | |
| CN111523617A (en) | Epilepsy detection system based on white matter fusion characteristic diagram and residual error attention network | |
| CN110689029A (en) | Method for determining sparsity of fMRI brain function connection network | |
| CN107256408B (en) | Method for searching key path of brain function network | |
| CN107832656B (en) | Brain function state information processing method based on dynamic function brain network | |
| Davuluri et al. | A survey of different machine learning models for alzheimer disease prediction | |
| CN114246588A (en) | Depression research method | |
| Jafadideh et al. | Topological analysis of brain dynamics in autism based on graph and persistent homology | |
| CN113197578B (en) | Schizophrenia classification method and system based on multi-center model | |
| CN114723937A (en) | Method and system for classifying blood vessel surrounding gaps based on nuclear magnetic resonance image | |
| Chen et al. | Attention-based node-edge graph convolutional networks for identification of autism spectrum disorder using multi-modal mri data | |
| Abenna et al. | Alcohol use disorders automatic detection based BCI systems: a novel EEG classification based on machine learning and optimization algorithms | |
| Wang | Indentification of autism spectrum disorder based on an improved convolutional neural networks | |
| CN111325268A (en) | Image classification method and device based on multi-level feature representation and integrated learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190510 |
|
| RJ01 | Rejection of invention patent application after publication |