CN106650768A - Gaussian image model-based brain network modeling and mode classification method - Google Patents
Gaussian image model-based brain network modeling and mode classification method Download PDFInfo
- Publication number
- CN106650768A CN106650768A CN201610857284.5A CN201610857284A CN106650768A CN 106650768 A CN106650768 A CN 106650768A CN 201610857284 A CN201610857284 A CN 201610857284A CN 106650768 A CN106650768 A CN 106650768A
- Authority
- CN
- China
- Prior art keywords
- brain network
- log
- model
- gauss
- feature
- 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
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
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention puts forward a brain network modeling and automatic identifying method for mild cognitive impairment function magnetic resonance data. The method comprises the following steps: a brain network is connected via a function magnetic resonance data construction function of a Gaussian image model, a lower triangular matrix of a sparse inverse covariance matrix representing the brain network is straightened as a characteristic vector, a t test and support vector machine recursion characteristic elimination method is used for screening above characteristics, a characteristic set representing maximum differences between people with mild cognitive impairment and normal people is obtained, the characteristic set is used for training a pattern classifier, a training set classification correct rate is obtained, and a brain network model and a classifier SVM used for classification can be finally determined. The brain network model built based on the brain network modeling and mode classification method disclosed in the invention is better than a conventional brain network modeling method in terms of mild cognitive impairment classification and identification performance and model generalization performance; the brain network model in the invention is of great significance in assistance in clinic diagnosis and automatic identification of mild cognitive impairment.
Description
Technical field
The present invention relates to cerebral function connection network struction, pattern classification and machine learning, belong to signal transacting and pattern
Technology of identification field.
Background technology
Alzheimer's disease is a kind of common nervous system degenerative disease, clinically with memory disorders, perform function
Obstacle, the infringement of visual space technical ability, aphasia etc. are characterized, and so far without effective remedy measures, serious harm people especially
The physical and mental health of the elderly.Mild cognitive impairment (Mild Cognitive Impairment, MCI) is used as between usual aging
A kind of intermediateness and dementia between, is the people at highest risk for developing into dementia, and only a minority finally can be converted into
Normally；Therefore, if early intervention can be carried out in the mild cognitive impairment stage and treatment is given, the morbidity of dementia will be greatly reduced
Rate, and slow down going down for cognition of mild cognitive impairment.However, the Precise Diagnosis of mild cognitive impairment are always one
Challenging world-famous puzzle, main reason is that the symptom of its Cognitive function damage is not obvious, it is difficult to qualitative analysis is done,
And brain network function link model is built, and searching mild cognitive impairment is connected with the difference of normal person from function connects model
Become the effective measures of auxiliary diagnosis mild cognitive impairment.In the past few years, Functional magnetic resonance imaging is because of its Noninvasive, height
The advantages of spatial resolution, it is widely used in structure and its function connects problem probed between cerebral neuron, and such as
What builds effective brain network function link model from fMRI data has become popular research field, and gradually receives research
The attention of persons.
At present, most brain network struction researchs are concentrated mainly on based on the method for correlation, i.e., related by estimating
Property (such as Pearson correlations, partial correlation) is defining the association between node.Pearson methods are calculated between each two brain area
Coefficient correlation, with this come constructing function connection brain network model, partial Correlation Analysis calculate two brain areas between correlation, and
And the impact of other brain areas is rejected building brain network.There are some researches show, the method compared with additive method, based on correlation
Two interactions between brain area or neuron can be preferably portrayed, but true brain network is often multiple brain areas or neuron colony
Between be connected with each other, therefore, the method based on correlation can not truly describe the connection mode between brain area；It is additionally based on
Brain network constructed by correlation is usually fully-connected network, and true brain network has sparse characteristic, therefore based on correlation
Constructed brain network wherein must be comprising some redundancies and unessential connection, and this brings certain being stranded to brain network analysis
It is difficult.
For the deficiency of above-mentioned structure brain network method, the present invention builds the brain net of rarefaction using Gauss graph model
Network, and determine the sparse degree of network by training accuracy, so that the network for building more truly goes back protocerebrum archicerebrum network
Internal connection.
Function connects network based on Gauss map model construction is characterizing being connected with each other between Different brain region inside brain
Aspect performance is excellent, but often there is substantial amounts of identical connection between patient and normal person, puies forward feature and is characterizing mild cognitive
There is redundancy or uncorrelated to normal person's difference aspect in obstacle.Therefore, the difference connection with maximum separating capacity how is found
Become crucial with feature, a variety of feature selection approach are proposed for this, wherein typical method has t inspections, variance point
Analysis method, characteristic weighing algorithm (ReliefF), SVMs recursive feature null method (Support Vector Machine-
Recursive Feature Elimination, SVM-RFE) etc..T is checked and thought of the method for analysis of variance based on statistics is checked
Difference of the characteristic statistic on two class samples.ReliefF can verify the feature larger with Category Relevance, and this phase
Closing property is not limited to linear correlation.Above-mentioned t methods of inspection, method of analysis of variance, ReliefF methods are all a kind of features of filtering type
Selection algorithm, feature selecting is unrelated with specific sorting algorithm, with calculating the features such as simple, speed is fast, but lacks and considers special
Correlation and redundancy between levying.SVM-RFE is used as typical packaged type feature selection approach, it is contemplated that to table between multiple features
Levy the impact of differences between samples and pattern classification, index such as the accuracy rate of reference pattern classification is reflecting the importance of feature, selection
Feature can obtain preferable pattern classification effect.Based on filtering type and packaged type feature selection approach, two classes can be effectively combined
The advantage of feature selection approach, can be based on first filtering type feature selection approach and choose the spy for characterizing differences between samples to a certain extent
Levy, then the character subset for causing pattern classification effect optimum is chosen based on packaged type feature selection approach.
The present invention is applied to Gauss graph model in mild cognitive impairment functional MRI data, and based on a kind of combined type
Feature selection approach, realizes accurate structure, the automatic identification of mild cognitive impairment of brain network, to aiding in mild cognitive impairment to face
Bed diagnosis, mitigates doctor's burden raising diagnosis efficiency and has important practical usage.
The content of the invention
According to an aspect of the present invention, Gauss Graphical modeling is applied to mild cognitive impairment (MCI) work(by the present invention
In energy magnetic resonance image data (fMRI), there is provided a kind of side for building mild cognitive impairment function connects network and feature extraction
Method, carries in pattern classification of the feature between Patients with Mild Cognitive Impairment and normal control and achieves excellent classifying quality,
And disaggregated model has good Generalization Capability.
For achieving the above object, the invention provides the brain network modelling and method for classifying modes based on Gauss graph model,
Comprise the steps：
1. brain network struction：Using the sparse inverse covariance matrix of Gauss graph model assessment function MR data, build
Brain network function link model；
2. feature construction：Take the weight on side in brain network, i.e., the lower triangular matrix of sparse inverse covariance matrix, alternately
Feature set S0；
3. feature selecting：From alternative features collection S0It is middle to choose the optimal feature subset S with good discrimination ability1；
4. disaggregated model：Based on optimal feature subset S1Training grader Nonlinear Support Vector Machines；
5. pattern classification：Functional MRI data is classified using grader, verifies character subset S1Validity.
Wherein, in affiliated step 1, test sample is finally determined with the parameter value on training set with maximum training accuracy
With the function connects brain network of training sample.
In the step 2, the lower triangular matrix for taking the inverse covariance matrix of rarefaction stretches composition row vector, used as spy
Levy.
In the step 3, using the combined feature selection approach of a kind of filtering type and packaged type from feature set S0Middle choosing
Select optimal feature subset S1。
In the step 4, from non-linear of RBF (Radial Basis Function, RBF) kernel function
Vector machine (Support Vector Machine, SVM) is held as grader, and using a kind of grid for taking into account global and local
Searching method determines the optimized parameter of grader.
Brain network modelling based on Gauss graph model provided by the present invention includes with the advantage of method for classifying modes：
1. network constructed by is sparse network, can more exact representation brain connection；
2. combined type feature selection approach, convenient and swift, can quickly select optimal feature subset；
3. the difference that the difference connection found based on this method can be accurately reflected between mild cognitive impairment and normal person
Different, the auxiliary diagnosis for mild cognitive impairment provide quantitative technological guidance.
Description of the drawings
Fig. 1 be according to the brain network struction of the embodiment of the present invention go forward side by side row mode classification schematic flow sheet.
Fig. 2 is based on the two-layer leave one cross validation exhaustive division flow chart of the embodiment of the present invention.
Fig. 3 is mild cognitive impairment function connects brain network in the case of different sparse degree.
Fig. 4 is combined type secondary filter feature selecting schematic flow sheet in the present invention.
Fig. 5 is to put forward experimenter work of the feature in mild cognitive impairment pattern classification according to different brain network establishing methods
Make characteristic curve.
Fig. 6 to stay every time in a cross validation, size of the institute eventually for the parameter lambda of structure brain network.
Fig. 7 (a) and Fig. 7 (b) are the distribution map of two optimal characteristics carried based on the present invention.
Specific embodiment
With reference to the accompanying drawings and detailed description the present invention is described in further detail.
According to one embodiment of present invention, Gauss graph model is applied to build brain network in functional MRI data,
There are some researches show, human brain has sparse characteristic, and Gauss graph model is by way of sparse inverse covariance matrix so that inverse
Some elements in covariance matrix are contracted to zero to simulate the sparse characteristic of brain, and carry from sparse inverse covariance matrix
The connection for characterizing mild cognitive impairment is taken and recognized, pattern classification is carried out.Fig. 1 illustrates according to an embodiment of the invention
Method flow diagram, including：
First, rarefaction inverse covariance matrix is obtained to functional MRI data application Gauss graph model, constructing function connects
Connect network (step 1)；Then, the lower triangular matrix of sparse inverse covariance matrix is taken as feature (step 2)；Then, examined using t
Test the secondary filter feature selection approach combined with SVM-RFE to screen feature set, obtain optimal feature subset (step
3)；Then, using optimal feature subset trainable pattern classifier SVMs (step 4)；Finally, trained non-thread is adopted
Property support vector machine classifier carry out pattern classification, and verify the validity and the Generalization Capability (step 5) of model of selected feature.
Fig. 2 further illustrates in detail the classification process figure based on the present invention.As shown in Fig. 2 the present invention is handed over using two-layer
Fork checking carries out model training and pattern classification.Internal layer leave one cross validation is used to find optimal brain network model, i.e. parameter lambda
Value, and assign it to outer loop for building brain network model；Outer layer leave one cross validation is based on internal layer optimal parameter
The brain network model of structure finds optimal feature subset S1, and trainable pattern classifier SVM, carry out pattern classification.
The introduction of lower mask body is according to the brain network modelling based on Gauss graph model provided by the present invention and pattern classification side
The concrete steps of method：
1. brain network struction：Brain network function link model is built to functional MRI data using Gauss graph model；
Gauss graph model to maximal possibility estimation by adding L1Norm constraint item so that some units in inverse covariance matrix
Element is contracted to zero, and its expression formula is as follows：
In formula：θ is inverse covariance matrix,For the estimation of inverse covariance matrix, S is sample covariance matrix, and λ is canonical
Change parameter, for adjusting sparse degree.λ takes different values, you can obtain the brain network of different sparse degree, and λ is bigger constructed
Network is more sparse.When Fig. 3 shows different sparse degree, the function connects brain network of Patients with Mild Cognitive Impairment.For final
For the brain network of classification, i.e. parameter λ value will be determined by step 4.
2. feature construction：Take side in brain network alternately feature set S0；
Rarefaction inverse covariance matrixFor symmetrical matrix, each of which nonzero term can regard a line as, used as two
A connection between brain area, and the size of each nonzero element, can regard as " weight " of this edge.In the present invention using side as
Feature, takesLower triangular matrix constitutive characteristic vector, alternately characteristic set S0。
3. feature selecting：Selection can at utmost characterize the character subset S of mild cognitive impairment and normal person's difference1。
The present invention selects optimal feature subset S using the mode of secondary filter1, filter checked to spy using t for the first time
Collection conjunction is screened, and selects the maximum character subset of average difference；Then, the character subset of selection is filtered based on first time,
Reapplying SVM-RFE feature selecting algorithms carries out second filtration, selects and most can characterize mild cognitive impairment and normal person's difference
Feature set S1.Fig. 4 shows the flow process of t inspections and SVM-RFE combined type feature selection approach.
Wherein, SVM-RFE feature selecting algorithms, specifically comprise the steps of：
A) initialization feature ordering vector U is sky；
If b) character subset S is sky, step f) is gone to；
C) otherwise, svm classifier model is trained with character subset S；
D) score of each feature is calculated；
E) the minimum feature of score is moved into the stem of ordering vector U from S, step b) is back to
F) in output vector U it is front k vector as optimal feature subset S1, terminate.
4. disaggregated model：Based on optimal feature subset and training data, trainable pattern classifier SVM
The present invention uses the SVMs for being based on RBF kernel functions as final disaggregated model, and uses grid search
Method is determining the best of breed of penalty factor c and kernel function width g.I.e. first with larger step size log (step)=0.5 at one
Larger scope (log (c):- 8～0, log (g):- 8～1) in search one rough optimum combination (c0,g0), then with less
Step-length log (step)=0.1 in (c0,g0) territory in (log (c0) -2～log (c0)+2, log (g0) -2～log
(g0)+2) the fine optimum combination (c of search1,g1)。
For the brain network constructed by each parameter λ value in step 1, a SVM classifier can be trained, the present invention
Used in possess the SVM classifier of maximum training accuracy on training set as final disaggregated model, and with ginseng now
Number λ value is that test sample constructing function connects brain network.
5. pattern classification：The grader trained based on optimal feature subset is classified, and the Generalization Capability of analysis model
Carried feature can indirectly be embodied in terms of mild cognitive impairment and normal person's difference degree is characterized by classifying quality,
And classifying quality can be by accuracy (Accuracy), susceptibility (Sensitivity), specificity (Specificity), balance
Area under accuracy (Balanced Accuracy) and the ROC curve of characterization model Generalization Capability (Area Under Curve,
) etc. AUC index embodies, and each index of the above is bigger, shows that classifying quality is better, carries feature difference and becomes apparent from.The present invention is used
Leave one cross validation obtains above-mentioned each index, and original fMRI data are big from U.S. north Carolina Luo Na states Durham city Du Ke
Learn Brian Imaging and analysis center, including 12 MCI patients and 25 normal persons.
1). accuracy
, used as the most basic amount for reflecting a grader classifying quality, grader classifying quality is better, necessarily has for accuracy
Higher classification accuracy rate.In classification problem, accuracy computing formula is as follows：
Wherein, ACC represents accuracy, and TP represents the positive sample number correctly classified, and TN represents the negative sample correctly classified
This number, FP represents that by the negative sample number of mistake classification FN is represented by the positive sample number of mistake classification.
2). susceptibility
Susceptibility medically represents the percentage that actual diseased (positive) is diagnosed also known as True Positive Rate.Its meter
Calculate formula as follows：
Wherein, SEN represents susceptibility.
3). specificity
Specificity, also known as true negative rate, i.e., reality is disease-free is correctly judged to disease-free percentage by diagnostic criteria, is reflected
Screening experiment determines the ability of non-patient.Its computing formula is as follows：
Wherein, SPE represents specificity.
4). balance accuracy
Balance accuracy, it is contemplated that signal concentrates positive and negative sample number to there is impact of the skewness situation to accuracy, its calculating
Formula is as follows：
Wherein, BAC represents balance accuracy.
5) area under .ROC lines
ROC curve be according to a series of different two mode classifications (cut off value determines threshold), with kidney-Yang rate as ordinate,
False sun rate is the curve that abscissa is drawn.Can be by ROC curve, automatic identification energy of the different diagnostic test of comparison to disease
Power.The closer to the upper left corner ROC curve representated by experimenter work it is more accurate, it is also possible to by calculate AUC value compared
Compared with AUC is bigger, and the diagnostic value of experiment is higher.Fig. 5 is illustrated based on Gauss graph model brain network and partial correlation and Pearson came
Brain network carries out the ROC curve of classification diagnosis, the svm classifier as can be seen from the figure trained based on Gauss graph model brain network
Device has good diagnosis capability to mild cognitive impairment.
For the functional MRI data from U.S.'s north Carolina Luo Na states Duke University's Brian Imaging and analysis center, table
1 lists the classification results contrast based on the brain network constructed by the present invention with Pearson came brain network and partial correlation brain network.
As can be seen from the table based on the present invention constructed by brain network, rate of accuracy reached to 91.89%, while also have good SEN,
The indexs such as SPE, BAC, AUC, show good Generalization Capability, and its classifying quality is substantially better than Pearson came and partial correlation brain net
Network.
The pattern classification Comparative result of the different brain network modeling methods of table 1
For the functional MRI data from U.S.'s north Carolina Luo Na states Duke University's Brian Imaging and analysis center, figure
The 6 exhibition formulas present invention is stayed for 37 times in a classification eventually for the parameter λ value size for building brain network model, it can be seen that and λ=
The 0.03 and 0.04 frequency highest for occurring, illustrates that brain network now based on Gauss map model construction is characterizing mild cognitive impairment
Difference aspect performance between normal person is more excellent.Two optimums when Fig. 7 (a) and Fig. 7 (b) respectively show λ=0.03
Feature, as can be seen from the figure there is obvious difference between mild cognitive impairment and normal person.
Brain network modelling and method for classifying modes based on Gauss graph model provided by the present invention, is mainly accurate structure
Brain network is built, the diagnosis proposition of mild cognitive impairment cerebral disease is aided in.It will be clear that the brain network described in this specification is built
Mould classifying and analyzing method is also applied for the diagnostic analysis of other cerebral diseases.
The brain network modelling based on Gauss graph model provided by the present invention and method for classifying modes have been carried out in detail above
Thin explanation, it is apparent that the scope of the present invention is not limited thereto.In the protection model limited without departing from appended claims
In the case of enclosing, the various changes of above-described embodiment are within the scope of the present invention.
Claims (5)
1. the brain network modelling and method for classifying modes of Gauss graph model are based on, it is characterised in that included：
Step 1. brain network struction：Using the sparse inverse covariance matrix of Gauss graph model assessment function MR data, build
Brain network function link model；
Step 2. feature construction：Take the weight on side in brain network, i.e., the lower triangular matrix of sparse inverse covariance matrix, alternately
Feature set S0；
Step 3. feature selecting：From alternative features collection S0It is middle to choose the optimal feature subset S with good discrimination ability1；
Step 4. disaggregated model：Based on optimal feature subset S1Training grader Nonlinear Support Vector Machines；
Step 5. pattern classification：Functional MRI data is classified using grader, verifies character subset S1Validity.
2. the brain network modelling and method for classifying modes based on Gauss graph model as claimed in claim 1, is characterised by：
The use of the parameter λ value for possessing maximum training accuracy in step 4 is that test sample builds brain network mould in the step 1
Type.
3. the brain network modelling and method for classifying modes based on Gauss graph model as claimed in claim 1, is characterised by：
In the step 2, the lower triangular matrix of sparse inverse covariance matrix is stretched into composition row vector alternately characteristic set.
4. the brain network modelling and method for classifying modes based on Gauss graph model as claimed in claim 1, is characterised by：
In the step 3, concentrated from alternative features with SVM-RFE combined secondary filter mode using t inspections and select optimum
Character subset.
5. the brain network modelling and method for classifying modes based on Gauss graph model as claimed in claim 1, is characterised by：
In the step 4, grader SVMs selects RBF RBF as kernel function, and using grid search side
Method determines parameter c and g.Initial search frequency range is (log (c):- 8～0, log (g):- 8～1), step-length be log (step)=
0.5, obtain optimum combination (c0,g0) after, Local Search scope is (log (c0) -2～log (c0)+2, log (g0) -2～log
(g0)+2), step-length is log (step)=0.1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610857284.5A CN106650768A (en) | 2016-09-27 | 2016-09-27 | Gaussian image model-based brain network modeling and mode classification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610857284.5A CN106650768A (en) | 2016-09-27 | 2016-09-27 | Gaussian image model-based brain network modeling and mode classification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106650768A true CN106650768A (en) | 2017-05-10 |
Family
ID=58854556
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610857284.5A Pending CN106650768A (en) | 2016-09-27 | 2016-09-27 | Gaussian image model-based brain network modeling and mode classification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106650768A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108898096A (en) * | 2018-06-27 | 2018-11-27 | 重庆交通大学 | A kind of quick accurate extracting method of the information towards high score image |
| CN109036556A (en) * | 2018-08-29 | 2018-12-18 | 王雁 | A method of keratoconus case is diagnosed based on machine learning |
| CN109065128A (en) * | 2018-09-28 | 2018-12-21 | 郑州大学 | A kind of sparse brain network establishing method of weighted graph regularization |
| CN109344889A (en) * | 2018-09-19 | 2019-02-15 | 深圳大学 | A kind of cerebral disease classification method, device and user terminal |
| 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 |
| CN110298364A (en) * | 2019-06-27 | 2019-10-01 | 安徽师范大学 | Based on the feature selection approach of multitask under multi-threshold towards functional brain network |
| CN110522463A (en) * | 2019-08-28 | 2019-12-03 | 常州大学 | A kind of depression assistant diagnosis system based on brain function linking parsing |
| 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 |
| CN111343032A (en) * | 2020-05-18 | 2020-06-26 | 中国航空油料集团有限公司 | Industrial control network abnormal session detection method, device, electronic equipment and storage medium |
| CN111652193A (en) * | 2020-07-08 | 2020-09-11 | 中南林业科技大学 | Wetland classification method based on multi-source images |
| CN113052800A (en) * | 2021-03-09 | 2021-06-29 | 山东大学 | Alzheimer disease image analysis method and device |
| CN113243891A (en) * | 2021-05-31 | 2021-08-13 | 平安科技（深圳）有限公司 | Mild cognitive impairment recognition method and device, computer equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104680176A (en) * | 2015-02-09 | 2015-06-03 | 北京邮电大学 | Electroencephalography (EEG) signal classification method based on non-Gaussian neutral vector feature selection |
| CN104887224A (en) * | 2015-05-29 | 2015-09-09 | 北京航空航天大学 | Epileptic feature extraction and automatic identification method based on electroencephalogram signal |
| US20150290460A1 (en) * | 2014-03-04 | 2015-10-15 | Advanced Neuromodulation Systems, Inc. dba St. Jude Neuromodulation Division | Methods for Treating Mild Cognitive Impairment and Alzheimer's Disease |
| 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 |
-
2016
- 2016-09-27 CN CN201610857284.5A patent/CN106650768A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150290460A1 (en) * | 2014-03-04 | 2015-10-15 | Advanced Neuromodulation Systems, Inc. dba St. Jude Neuromodulation Division | Methods for Treating Mild Cognitive Impairment and Alzheimer's Disease |
| CN104680176A (en) * | 2015-02-09 | 2015-06-03 | 北京邮电大学 | Electroencephalography (EEG) signal classification method based on non-Gaussian neutral vector feature selection |
| CN104887224A (en) * | 2015-05-29 | 2015-09-09 | 北京航空航天大学 | Epileptic feature extraction and automatic identification method based on electroencephalogram signal |
| 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 |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108898096A (en) * | 2018-06-27 | 2018-11-27 | 重庆交通大学 | A kind of quick accurate extracting method of the information towards high score image |
| CN108898096B (en) * | 2018-06-27 | 2022-04-08 | 重庆交通大学 | High-resolution image-oriented information rapid and accurate extraction method |
| CN109036556B (en) * | 2018-08-29 | 2021-06-04 | 王雁 | Method for diagnosing keratoconus case based on machine learning |
| CN109036556A (en) * | 2018-08-29 | 2018-12-18 | 王雁 | A method of keratoconus case is diagnosed based on machine learning |
| CN109344889A (en) * | 2018-09-19 | 2019-02-15 | 深圳大学 | A kind of cerebral disease classification method, device and user terminal |
| CN109065128A (en) * | 2018-09-28 | 2018-12-21 | 郑州大学 | A kind of sparse brain network establishing method of weighted graph regularization |
| 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 |
| CN110298364A (en) * | 2019-06-27 | 2019-10-01 | 安徽师范大学 | Based on the feature selection approach of multitask under multi-threshold towards functional brain network |
| CN110522463A (en) * | 2019-08-28 | 2019-12-03 | 常州大学 | A kind of depression assistant diagnosis system based on brain function linking parsing |
| 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 |
| CN111343032A (en) * | 2020-05-18 | 2020-06-26 | 中国航空油料集团有限公司 | Industrial control network abnormal session detection method, device, electronic equipment and storage medium |
| CN111343032B (en) * | 2020-05-18 | 2020-09-01 | 中国航空油料集团有限公司 | Industrial control network abnormal session detection method, device, electronic equipment and storage medium |
| CN111652193A (en) * | 2020-07-08 | 2020-09-11 | 中南林业科技大学 | Wetland classification method based on multi-source images |
| CN111652193B (en) * | 2020-07-08 | 2024-03-19 | 中南林业科技大学 | Wetland classification method based on multi-source images |
| CN113052800A (en) * | 2021-03-09 | 2021-06-29 | 山东大学 | Alzheimer disease image analysis method and device |
| CN113052800B (en) * | 2021-03-09 | 2022-02-22 | 山东大学 | Alzheimer disease image analysis method and device |
| CN113243891A (en) * | 2021-05-31 | 2021-08-13 | 平安科技（深圳）有限公司 | Mild cognitive impairment recognition method and device, computer equipment and storage medium |
| CN113243891B (en) * | 2021-05-31 | 2024-10-18 | 平安科技（深圳）有限公司 | Mild cognitive impairment recognition method, device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106650768A (en) | Gaussian image model-based brain network modeling and mode classification method | |
| CN113616184B (en) | Brain network modeling and individual prediction method based on multi-mode magnetic resonance image | |
| CN112418337B (en) | Multi-feature fusion data classification method based on brain function hyper-network model | |
| CN110236543B (en) | Alzheimer disease multi-classification diagnosis system based on deep learning | |
| CN107909117A (en) | A kind of sorting technique and device based on brain function network characterization to early late period mild cognitive impairment | |
| CN111009324B (en) | Auxiliary diagnosis system and method for mild cognitive impairment through multi-feature analysis of brain network | |
| CN109840554B (en) | Alzheimer's disease MRI image classification method based on SVM-RFE-MRMR algorithm | |
| WO2020103683A1 (en) | Method and system for individualized prediction of mental illness on basis of brain function map monkey-human cross-species migration | |
| Wei et al. | The segmentation of retinal layer and fluid in SD-OCT images using mutex dice loss based fully convolutional networks | |
| CN109325942A (en) | Eye fundus image Structural Techniques based on full convolutional neural networks | |
| CN101667289B (en) | Retinal image segmentation method based on NSCT feature extraction and supervised classification | |
| CN109509552A (en) | A kind of mental disease automatic distinguishing method of the multi-level features fusion based on function connects network | |
| CN108629369A (en) | A kind of Visible Urine Sediment Components automatic identifying method based on Trimmed SSD | |
| CN108511055A (en) | Ventricular premature beat identifying system and method based on Multiple Classifier Fusion and diagnostic rule | |
| Sharma et al. | Machine learning approach for detection of diabetic retinopathy with improved pre-processing | |
| CN106980815A (en) | Facial paralysis objective evaluation method under being supervised based on H B rank scores | |
| CN114998651A (en) | Skin lesion image classification and identification method, system and medium based on transfer learning | |
| Firke et al. | Convolutional neural network for diabetic retinopathy detection | |
| Sanjana et al. | Automated diabetic retinopathy detection using transfer learning models | |
| CN110444294A (en) | A kind of prostate cancer aided analysis method and equipment based on layered perception neural networks | |
| CN106778830A (en) | Staging method based on double cluster results and AdaBoost | |
| CN114305387A (en) | Magnetic resonance imaging-based method, equipment and medium for classifying small cerebral vascular lesion images | |
| Kanse et al. | HG-SVNN: harmonic genetic-based support vector neural network classifier for the glaucoma detection | |
| CN118098567A (en) | Multi-view sparse typical correlation analysis method based on brain network structure constraint | |
| CN114821176B (en) | Viral encephalitis classification system for MR (magnetic resonance) images of children brain |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170510 |