CN110459317A - Alzheimer disease assistant diagnosis system and method based on the dynamic brain network kernel of graph - Google Patents
Alzheimer disease assistant diagnosis system and method based on the dynamic brain network kernel of graph Download PDFInfo
- Publication number
- CN110459317A CN110459317A CN201910742381.3A CN201910742381A CN110459317A CN 110459317 A CN110459317 A CN 110459317A CN 201910742381 A CN201910742381 A CN 201910742381A CN 110459317 A CN110459317 A CN 110459317A
- Authority
- CN
- China
- Prior art keywords
- graph
- kernel
- network
- dynamic
- dynamic brain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0033—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room
- A61B5/004—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room adapted for image acquisition of a particular organ or body part
- A61B5/0042—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room adapted for image acquisition of a particular organ or body part for the brain
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
- A61B5/055—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/40—Detecting, measuring or recording for evaluating the nervous system
- A61B5/4076—Diagnosing or monitoring particular conditions of the nervous system
- A61B5/4088—Diagnosing of monitoring cognitive diseases, e.g. Alzheimer, prion diseases or dementia
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B2576/00—Medical imaging apparatus involving image processing or analysis
- A61B2576/02—Medical imaging apparatus involving image processing or analysis specially adapted for a particular organ or body part
- A61B2576/026—Medical imaging apparatus involving image processing or analysis specially adapted for a particular organ or body part for the brain
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Pathology (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Veterinary Medicine (AREA)
- Animal Behavior & Ethology (AREA)
- Heart & Thoracic Surgery (AREA)
- Neurology (AREA)
- Radiology & Medical Imaging (AREA)
- Artificial Intelligence (AREA)
- Psychiatry (AREA)
- Physiology (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Psychology (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Child & Adolescent Psychology (AREA)
- Developmental Disabilities (AREA)
- High Energy & Nuclear Physics (AREA)
- Neurosurgery (AREA)
- Hospice & Palliative Care (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The present invention provides a kind of Alzheimer disease assistant diagnosis system and method based on the dynamic brain network kernel of graph.The diagnostic system includes pretreatment unit, dynamic brain network struction unit, dynamic brain network core computing unit and classification diagnosis unit, carry out the pretreatment of image to Functional imagnetic resonance imaging by pretreatment unit first, then the matching of brain area is successively carried out to pretreated Functional imagnetic resonance imaging by dynamic brain network struction unit, the segmentation of period, the calculating of association relationship and the excavation of Frequent tree mining, then the Optimum Matching of bipartite graph is passed sequentially through to the Frequent tree mining dynamic brain function network of reconstruction by dynamic brain network core computing unit, the calculating of the kernel of graph, the combination of kernel of graph matrix and the distribution of weight, obtain a fused dynamic brain function network nuclear matrix, by the training for carrying out data by data training aids in conjunction with core SVM, it realizes eventually by auxiliary diagnosis device to Alzheimer disease Diagnosis.
Description
Technical field
The present invention relates to computer-aided diagnosis technical field more particularly to a kind of A Er based on the dynamic brain network kernel of graph
Ci Haimo disease assistant diagnosis system and method.
Background technique
Alzheimer disease (Alzheimer Disease, AD) is that a kind of nerve of influence people cognition even behavior moves back
Row disease, the reason of causing this disease are that the brain areas of some control cognitions are impaired, cause the connection between brain area weaken or
It disappears, therefore, brain connection mode plays key effect during the diagnosis of AD.Tranquillization state functional mri (resting
State functional Magnetic Resonance Imaging, rs-fMRI) provide a kind of method of Noninvasive
To measure functional activity and the variation of brain.Using brain area as node, using the connection between brain area as side, building is based on rs-fMRI
Brain network, classify to brain network, realize the diagnosis of AD.
Traditional static brain network is calculated based on the function connects of entire time-series image.In recent years, have perhaps
More classification method brain networks, the topology metric (such as cluster coefficients) for such as extracting brain network classified, deep learning and kernel of graph side
Method.In these methods, kernel of graph method is proved to have good adaptive classification effect.The kernel of graph is able to solve in assorting process
Higher-dimension challenge, have good generalization ability, be the important tool classified in machine learning to structural data.
In addition, establishing being associated between kernel function and core machine by the kernel function for calculating figure, may be directly applied to simple base
In the machine learning method of kernel function, the classification of AD is realized, such as support vector machines (Support Vector Machine, SVM)
Deng.
Traditional static brain network is based on by calculating the global correlation coefficient of signal between any two brain area
The function connects of whole image time series construct single brain network.But the studies have found that function connects of brain signal
Dynamic change is shown in each period, in the case where Dynamic link library, more passes can be shown within per a period of time
In the local message of brain function activity.However, the existing kernel of graph is all based on static brain network, only consider between brain area
GF global functions connection.Therefore it is considered as the auxiliary diagnosis that the dynamic brain network kernel of graph carries out AD.
Summary of the invention
It is a kind of based on dynamic brain network the technical problem to be solved by the present invention is in view of the above shortcomings of the prior art, propose
The Alzheimer disease assistant diagnosis system and method for the kernel of graph, the kernel of graph of dynamic brain network can capture function connects each
The dynamic change of period.Therefore, the calculating of the dynamic brain network kernel of graph can sufficiently reflect the similitude of dynamic change, to obtain
Obtain higher nicety of grading.
In order to solve the above technical problems, the invention proposes a kind of Alzheimer disease based on the dynamic brain network kernel of graph is auxiliary
Help diagnostic system and method, a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph includes: that processing is single
Member, dynamic brain network struction unit, dynamic brain network core computing unit and classification diagnosis unit；The pretreatment unit is used for
By the Functional imagnetic resonance imaging of input, passes sequentially through time adjustment, the dynamic correction of head and smoothing and noise-reducing process and obtain one later
Functional imagnetic resonance imaging after serial noise reduction；What the dynamic brain network struction unit was used to export the pretreatment unit
Functional imagnetic resonance imaging after noise reduction successively carries out the matching of brain area, the segmentation of period, the calculating of association relationship and frequency
The excavation of numerous subgraph；The dynamic brain network core computing unit is used for the reconstruction for exporting the dynamic brain network struction unit
Frequent tree mining dynamic brain function network, pass sequentially through the Optimum Matching of bipartite graph, the calculating of the kernel of graph, the combination of kernel of graph matrix with
And the distribution of weight, finally obtain a fused dynamic brain function network nuclear matrix；The classification diagnosis unit is used for
Whether obtained dynamic brain function network nuclear matrix and core SVM knot merga pass data training aids are subjected to Alzheimer disease
Then the auxiliary diagnosis of Alzheimer disease is realized in the training of illness using obtained training data and by auxiliary diagnosis device.
The dynamic brain network struction unit includes brain area adaptation, time series segmentation device, mutual information calculator, brain
The brain area adaptation of network struction device and Frequent tree mining delver, dynamic brain network struction unit receives from pretreatment unit
Functional imagnetic resonance imaging after noise reduction, and by the AAL template of Functional imagnetic resonance imaging and 90 brain areas after the noise reduction
It is matched, the Functional imagnetic resonance imaging after being matched, is then used the Functional imagnetic resonance imaging after the matching
The mode of overlapping carries out the segmentation of period, obtains the time-series image of each period, secondly will be by time series point
All time-series images after section device segmentation, which export, gives mutual information calculator, calculates any two by the mutual information calculator
Association relationship between a brain area, the association relationship be used to indicate relevance between brain area, then obtained association relationship is defeated
Enter into brain network struction device, construct the dynamic brain function network of the time-series image of each period, finally by sometimes
Between section time-series image dynamic brain function network inputs into Frequent tree mining delver, the Frequent tree mining rebuild is dynamic
State brain function network image；The brain area adaptation is used for the standard AAL template of pretreated image and 90 brain areas
Functional imagnetic resonance imaging after being matched；
The time series that the time series segmentation device is used to obtain Functional imagnetic resonance imaging is segmented, and is obtained
The time series of s period；
The mutual information calculator be used under any one period time series calculate any two brain area it
Between association relationship, to indicate the relevance between brain area；
The brain network struction device is used for the association relationship by calculating for the s width Functional imagnetic resonance imaging after segmentation
Building brain network obtains the brain network of s period；
The Frequent tree mining delver is used to carry out obtained all dynamic brain function networks the excavation of Frequent tree mining,
The Frequent tree mining dynamic brain network rebuild.
The dynamic brain network core computing unit includes Optimum Matching device, kernel of graph matrix generator, weight generator, moves
State brain network kernel of graph device, the frequent son for being rebuild any pair of any two Functional imagnetic resonance imaging by Optimum Matching device
The number of isomorphism Frequent tree mining in figure dynamic brain function network image is as connecting this to dynamic brain function network in bipartite graph
Side weight, carry out optimal of bipartite graph for the Frequent tree mining dynamic brain function network image of each reconstruction as node
Match, obtain matched dynamic brain function network pair, the dynamic brain function network of each period after matching is then passed through into figure
Core device obtains the kernel of graph under each period between any two dynamic brain function network, each period that secondly will be obtained
The kernel of graph between lower any two dynamic brain function network obtains the kernel of graph square under each period by kernel of graph matrix generator
Battle array, then be that the kernel of graph under each period generates one by the kernel of graph matrix under obtained all periods by weight generator
A weight coefficient finally obtains the kernel of graph matrix with weight coefficient of all periods, is melted by dynamic brain network kernel of graph device
Synthesize a dynamic brain function network nuclear matrix；
The Optimum Matching device is used for the frequent son of any pair of reconstruction of any two Functional imagnetic resonance imaging
The number of isomorphism Frequent tree mining in figure dynamic brain function network image is as connecting this to dynamic brain function network in bipartite graph
Side weight, optimal of bipartite graph is carried out using the Frequent tree mining dynamic brain function network image of each reconstruction as node
Match；
The kernel of graph device is used for any two dynamic brain function network query function figure under the period after each matching
Core；
The kernel of graph matrix generator is used to all dynamic brain network kernel of graphs of each period being combined into a square
Battle array, results in the kernel of graph matrix under s period；
The weight generator is used to utilize the method for Multiple Kernel Learning to the kernel of graph matrix allocation weight of each period；
The side that the dynamic brain network kernel of graph device is used to pass through linear combination according to the weight distributed under each period
Formula calculates the kernel of graph, obtains the dynamic brain function network kernel of graph.
A kind of application method of the Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph, including following step
It is rapid:
Step 1: the pretreatment of Functional imagnetic resonance imaging；
Step 2: utilizing pretreated picture construction dynamic brain function network；
Step 3: to the Frequent tree mining dynamic brain network query function kernel of graph of reconstruction, and finally obtaining the dynamic brain network kernel of graph；
Step 4: being realized with the dynamic brain network nuclear matrix X generated and auxiliary diagnosis is carried out to Alzheimer disease.
The pretreatment of step 1 Functional imagnetic resonance imaging, comprising the following steps:
1) i Functional imagnetic resonance imaging is obtained into the image (I after time adjustment by time adjustment device1,I2,...,
Ii), wherein i belongs to natural number；
2) by the image (I after time adjustment1,I2,...,Ii) the dynamic correction of head is carried out, removal head moves bring error, passes through
Dynamic corrector obtains the image (H moved after correction to the end1,H2,...,Hi)；
3) by the image (H after the dynamic correction of head1,H2,...,Hi) by image noise reduction device progress the disposal of gentle filter, it obtains
Image (S after the disposal of gentle filter1,S2,...,Si)。
The step 2 utilizes pretreated picture construction dynamic brain function network, comprising the following steps:
1) by the image (S after the disposal of gentle filter obtained after the step 1 pretreatment1,S2,...,Si) and 90
The dissection of a brain area marks (Anatomical Automatic Labeling, AAL) template to be matched automatically, is included
The image of 90 brain areas；
2) by time series segmentation device, by it is described include 90 brain areas image (A1,A2,...,Ai) it is divided into s weight
The folded period obtains the time-series image (T comprising s period1,T2,...,Ts), wherein s belongs to natural number；
3) by mutual information calculator to the time series T of any one periodjIt calculates mutual between any two brain area
The value of information, wherein Tj∈(T1,T2,...,Ts)；
4) association relationship is passed sequentially through into brain network struction device, constructs dynamic brain function network, obtained i*s and move
State brain function network imageWherein i indicates of Functional imagnetic resonance imaging
Number, s indicate the period number that time series divider is divided；
5) by Frequent tree mining device to i*s obtained dynamic brain function network image Carry out Frequent tree mining excavation, the Frequent tree mining dynamic brain function network image rebuild
The Frequent tree mining dynamic brain function network query function kernel of graph that the step 3 pair is rebuild, and finally obtain dynamic brain function
The network kernel of graph, comprising the following steps:
1) with the Frequent tree mining dynamic brain function network of any pair of reconstruction of any two Functional imagnetic resonance imaging
The number of isomorphism Frequent tree mining as in is as this weight to the side of dynamic brain function network is connected in bipartite graph, with each heavy
The Frequent tree mining dynamic brain function network image built carries out the Optimum Matching of bipartite graph as node, obtains matched dynamic brain function
It can network pairWherein m, n indicate any pair of matched dynamic brain function network pair；
2) by the dynamic brain function network of matched each period againBy kernel of graph device,
Obtain the kernel of graph under each period between any two dynamic brain function network
3) by the kernel of graph under each period between any two dynamic brain function networkIt is raw by kernel of graph matrix
It grows up to be a useful person, obtains the kernel of graph matrix (X under s period1,X2,...,Xs)；
4) by the kernel of graph matrix (X under the s period1,X2,...,Xs) by weight generator, it is each period
Under the kernel of graph generate a weight coefficient；
5) by dynamic brain network kernel of graph device, by the kernel of graph matrix (X with weight coefficient of all periods1,X2,...,
Xs) merge as a dynamic brain network nuclear matrix X.
The step 4 is realized with the dynamic brain network nuclear matrix X generated carries out auxiliary diagnosis to Alzheimer disease,
The following steps are included:
1) A Erci is carried out using obtained dynamic brain network nuclear matrix X and core SVM knot merga pass data training aids
The silent disease in sea whether the training of illness；
2) auxiliary diagnosis of Alzheimer disease is realized using the training data and by auxiliary diagnosis device.
The beneficial effects of the present invention are:
The present invention is a kind of Alzheimer disease aided diagnosis method of dynamic brain network kernel of graph, solves traditional brain function
The problem of energy network and the kernel of graph can not indicate the brain information of dynamic change, so that the dynamic change of Functional imagnetic resonance imaging is believed
Breath is fully used, and can preferably realize the auxiliary diagnosis of disease.
Detailed description of the invention
Fig. 1 is the Alzheimer disease assistant diagnosis system structure of one of the embodiment of the present invention dynamic brain network kernel of graph
Block diagram.
Fig. 2 is the preprocess method flow chart of the Functional imagnetic resonance imaging in the embodiment of the present invention.
Fig. 3 is the dynamic brain function network establishing method flow chart in the embodiment of the present invention.
Fig. 4 is the aided diagnosis method flow chart of the dynamic brain network kernel of graph in the embodiment of the present invention.
Specific embodiment
With reference to the accompanying drawings and examples, specific embodiments of the present invention will be described in further detail.
Traditional static brain network kernel of graph is that single brain network is constructed based on the function connects of whole image time series,
And according to this brain network query function kernel of graph.But the function connects of brain signal show dynamic change in each period, often
The section time suffers from the local message of different brain function activities, therefore is considered as the dynamic brain network kernel of graph and carries out the auxiliary of AD
Help diagnosis.
Such as the Alzheimer disease assistant diagnosis system structure of one of Fig. 1 embodiment of the present invention dynamic brain network kernel of graph
Shown in block diagram, a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph, including pretreatment unit, dynamic
Brain network struction unit, dynamic brain network core computing unit and classification diagnosis unit, the pretreatment unit will be for that will input
Functional imagnetic resonance imaging, pass sequentially through and obtain a series of drops after time adjustment, the dynamic correction of head and smoothing and noise-reducing process
Functional imagnetic resonance imaging after making an uproar；After the noise reduction that the dynamic brain network struction unit is used to export the pretreatment unit
Functional imagnetic resonance imaging, successively carry out the matching of brain area, the segmentation of period, the calculating of association relationship and Frequent tree mining
Excavation；The dynamic brain network core computing unit be used for the dynamic brain network struction unit is exported reconstruction it is frequent
Subgraph dynamic brain function network, passes sequentially through the Optimum Matching of bipartite graph, the calculating of the kernel of graph, the combination of kernel of graph matrix and weight
Distribution, finally obtain a fused dynamic brain function network nuclear matrix；The classification diagnosis unit will be for that will obtain
Dynamic brain function network nuclear matrix and core SVM knot merga pass data training aids carry out Alzheimer disease whether illness
Then the auxiliary diagnosis of Alzheimer disease is realized in training using obtained training data and by auxiliary diagnosis device.
The pretreatment unit includes time adjustment device, the dynamic corrector of head, image noise reduction device and brain area adaptation, is passed through
The input port input function nuclear magnetic resonance image of pretreatment unit, first by time adjustment device to Functional imagnetic resonance imaging
Carry out time adjustment, a series of Functional imagnetic resonance imaging after obtaining time adjustments, then by the function core after time adjustment
Magnetic resonance image exports and moves the dynamic correction of corrector progress head to head, the function nuclear magnetic resonance figures after obtaining a series of dynamic correction of heads
Picture, then the Functional imagnetic resonance imaging after the dynamic correction of head is subjected to smooth noise reduction by image noise reduction device, obtain a series of noise reductions
Functional imagnetic resonance imaging after noise reduction is finally exported and gives brain area adaptation by Functional imagnetic resonance imaging afterwards, and with 90
The AAL template of brain area is matched, the Functional imagnetic resonance imaging after being matched.
The Functional imagnetic resonance imaging that the time adjustment device is used to input carries out time adjustment, to obtain a system
Functional imagnetic resonance imaging after column time adjustment；
The head moves corrector and is used to carrying out the Functional imagnetic resonance imaging after time adjustment into the dynamic correction of head, obtains one
Functional imagnetic resonance imaging after the dynamic correction of serial head；
The image noise reduction device is used to the Functional imagnetic resonance imaging after the dynamic correction of head carrying out smooth noise reduction, obtains one
Functional imagnetic resonance imaging after serial noise reduction.
The dynamic brain network struction unit includes brain area adaptation, time series segmentation device, mutual information calculator, brain
The brain area adaptation of network struction device and Frequent tree mining delver, dynamic brain network struction unit receives from pretreatment unit
Functional imagnetic resonance imaging after noise reduction, and by the AAL template of Functional imagnetic resonance imaging and 90 brain areas after the noise reduction
It is matched, the Functional imagnetic resonance imaging after being matched, is then used the Functional imagnetic resonance imaging after the matching
The mode of overlapping carries out the segmentation of period, obtains the time-series image of each period, secondly will be by time series point
All time-series images after section device segmentation, which export, gives mutual information calculator, calculates any two by the mutual information calculator
Association relationship between a brain area, the association relationship be used to indicate relevance between brain area, then obtained association relationship is defeated
Enter into brain network struction device, construct the dynamic brain function network of the time-series image of each period, finally by sometimes
Between section time-series image dynamic brain function network inputs into Frequent tree mining delver, the Frequent tree mining rebuild is dynamic
State brain network image；
The brain area adaptation is for match with the standard AAL template of 90 brain areas pretreated image
Functional imagnetic resonance imaging after to matching；
The time series that the time series segmentation device is used to obtain Functional imagnetic resonance imaging is segmented, and is obtained
The time series of s period；
The mutual information calculator be used under any one period time series calculate any two brain area it
Between association relationship, to indicate the relevance between brain area；
The brain network struction device is used for the association relationship by calculating for the s width Functional imagnetic resonance imaging after segmentation
Building brain network obtains the brain network of s period；
The Frequent tree mining delver is used to carry out obtained all dynamic brain function networks the excavation of Frequent tree mining,
The Frequent tree mining dynamic brain network rebuild.
The dynamic brain network core computing unit includes Optimum Matching device, kernel of graph matrix generator, weight generator, moves
State brain network kernel of graph device, the frequent son for being rebuild any pair of any two Functional imagnetic resonance imaging by Optimum Matching device
The number of isomorphism Frequent tree mining in figure dynamic brain function network image is as connecting this to dynamic brain function network in bipartite graph
Side weight, carry out optimal of bipartite graph for the Frequent tree mining dynamic brain function network image of each reconstruction as node
Match, obtain matched dynamic brain function network pair, the dynamic brain function network of each period after matching is then passed through into figure
Core device obtains the kernel of graph under each period between any two dynamic brain function network, each period that secondly will be obtained
The kernel of graph between lower any two dynamic brain function network obtains the kernel of graph square under each period by kernel of graph matrix generator
Battle array, then be that the kernel of graph under each period generates one by the kernel of graph matrix under obtained all periods by weight generator
A weight coefficient finally obtains the kernel of graph matrix with weight coefficient of all periods, is melted by dynamic brain network kernel of graph device
Synthesize a dynamic brain network nuclear matrix；
The Optimum Matching device is used for the frequent son of any pair of reconstruction of any two Functional imagnetic resonance imaging
The number of isomorphism Frequent tree mining in figure dynamic brain network image is as connecting this power to the side of dynamic brain network in bipartite graph
Weight carries out the Optimum Matching of bipartite graph using the Frequent tree mining dynamic brain network image of each reconstruction as node；
The kernel of graph device is used for any two dynamic brain network query function kernel of graph under the period after each matching；
The kernel of graph matrix generator is used to all dynamic brain network kernel of graphs of each period being combined into a square
Battle array, results in the kernel of graph matrix under s period；
The weight generator is used to utilize the method for Multiple Kernel Learning to the kernel of graph matrix allocation weight of each period；
The side that the dynamic brain network kernel of graph device is used to pass through linear combination according to the weight distributed under each period
Formula calculates the kernel of graph, obtains the dynamic brain network kernel of graph.
The classification diagnosis unit includes the dynamic brain function that will be obtained first including data training aids and auxiliary diagnosis device
Network nuclear matrix and core SVM knot merga pass data training aids carry out Alzheimer disease whether the training of illness, it is then sharp
The auxiliary diagnosis of Alzheimer disease is realized with obtained training data and by auxiliary diagnosis device.
The data training aids is used in core svm classifier, replaces the kernel function of core SVM simultaneously with the matrix that the kernel of graph is constituted
Realize the training to Functional imagnetic resonance imaging data；
The auxiliary diagnosis device is used for the data after principle and training using core SVM, realizes on core SVM to A Er
The auxiliary diagnosis of Ci Haimo disease.
A kind of application method of the Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph, including following step
It is rapid:
Step 1: the pretreatment of Functional imagnetic resonance imaging, such as the Functional imagnetic resonance imaging in Fig. 2 embodiment of the present invention
Preprocess method flow chart shown in, specifically includes the following steps:
1) i Functional imagnetic resonance imaging is obtained into the image (I after time adjustment by time adjustment device1,I2,...,
Ii), wherein i belongs to natural number；
2) by the image (I after time adjustment1,I2,...,Ii) the dynamic correction of head is carried out, removal head moves bring error, passes through
Dynamic corrector obtains the image (H moved after correction to the end1,H2,...,Hi)；
3) by the image (H after the dynamic correction of head1,H2,...,Hi) by image noise reduction device progress the disposal of gentle filter, it obtains
Image (S after the disposal of gentle filter1,S2,...,Si)。
Step 2: pretreated picture construction dynamic brain function network is utilized, such as the dynamic in Fig. 3 embodiment of the present invention
Shown in brain function network establishing method flow chart, specifically includes the following steps:
1) by the image (S after the disposal of gentle filter obtained after the step 1 pretreatment1,S2,…,Si) and 90
The dissection of brain area marks (Anatomical Automatic Labeling, AAL) template to be matched automatically, obtains comprising 90
The image of a brain area；
2) by time series segmentation device, by it is described include 90 brain areas image (A1,A2,…,Ai) it is divided into s weight
The folded period obtains the time-series image (T comprising s period1,T2,…,Ts), wherein s belongs to natural number；
3) by mutual information calculator to the time series T of any one periodjIt calculates mutual between any two brain area
The value of information, wherein Tj∈(T1,T2,…,Ts), it is specific to state are as follows: according to each Functional imagnetic resonance imaging on each period
Matching relationship between each node of obtained image is calculated, and is determined in the dynamic brain function network under the period
The expression on side is calculated the correlativity between each node using the method for mutual information, and is constituted incidence matrix.
4) association relationship is passed sequentially through into brain network struction device, constructs dynamic brain function network, obtained i*s and move
State brain function network imageWherein i indicates of Functional imagnetic resonance imaging
Number, s indicate the period number that time series divider is divided, specific building process are as follows: calculate function nuclear magnetic resonance figures
As any two node mutual information and after obtaining incidence matrix, the threshold value T of mutual information is set, association relationship and setting are taken
Threshold value T be compared, when association relationship is bigger than the threshold value T of setting, its side is set as 1, i.e. it is relevant between two nodes,
Otherwise its side is set as 0, i.e. onrelevant, thus converts adjacency matrix for incidence matrix, construct dynamic brain function network.
5) by Frequent tree mining device to i*s obtained dynamic brain function network image Carry out Frequent tree mining excavation, the Frequent tree mining dynamic brain function network image rebuildSpecific statement are as follows: Frequent tree mining digging is carried out to brain network using the method for gspan
Pick is arranged minimum support threshold minSup, frequency of occurrence of the subgraph in all figures is compared with minSup, if more than
MinSup, then it is assumed that it is Frequent tree mining, is not otherwise Frequent tree mining, then rebuilds each period using Frequent tree mining collection
Dynamic brain function network.
Present embodiment is to carry out computer-aided diagnosis to Functional imagnetic resonance imaging, and doctor is helped to diagnose.
As shown in the aided diagnosis method flow chart of the dynamic brain function network kernel of graph in Fig. 4 embodiment of the present invention, dynamic
The aided diagnosis method of the brain function network kernel of graph includes the following steps 3 and step 4 two parts.
Step 3: to the Frequent tree mining dynamic brain function network query function kernel of graph of reconstruction, and finally obtaining dynamic brain function network
The kernel of graph, the specific steps are as follows:
1) with the Frequent tree mining dynamic brain function network of any pair of reconstruction of any two Functional imagnetic resonance imaging
The number of isomorphism Frequent tree mining as in is as this weight to the side of dynamic brain function network is connected in bipartite graph, with each heavy
The Frequent tree mining dynamic brain function network image built carries out the Optimum Matching of bipartite graph as node, obtains matched dynamic brain function
It can network pairWherein m, n indicate any pair of matched dynamic brain function network pair；
2) by the dynamic brain function network of matched each period againBy kernel of graph device,
Obtain the kernel of graph under each period between any two dynamic brain function network
Centered on each node p in brain network χ, q is shortest path, wherein q=1, and 2 ..., h construct subnet respectivelyIt can be calculated on m-th of period in each sub-network group between each corresponding subnet χ and ζ according to the property of brain network
Similitude:
Wherein, | | expression does determinant computation to matrix,
Cov indicates covariance, and N is the number of brain nodes, and e indicates that all elements are all 1 vector, AyE is indicated to orientation
The y times power iteration on e to matrix A is measured, similarlyAlso it is defined in subnetOn symmetric positive semidefinite matrix.
Thus the kernel of graph of first of period is obtained are as follows:
Wherein, N indicates the number of brain nodes,It indicates in first of period upper each sub-network group
Similitude between each corresponding subnet χ and ζ.
3) by the kernel of graph under each period between any two dynamic brain function networkIt is raw by kernel of graph matrix
It grows up to be a useful person, obtains the kernel of graph matrix (X under s period1,X2,...,Xs)；
The g row h column element of nuclear matrix X can be expressed as g-th of brain networkWith h-th of brain networkIt calculates
The kernel of graph arrived, i.e.,
4) by the kernel of graph matrix (X under the s period1,X2,...,Xs) by weight generator, it is each period
Under the kernel of graph generate a weight coefficient；
The kernel of graph is expressed asWherein, xl(χ, ζ) indicates building in l period hypencephalon network
The kernel of graph, χ and ζ correspond to the brain network under the l period, and s is the number of period, μlIt is a nonnegative curvature vector, it is full
Foot constraintThe μ of each kernel of graph matrix is determined using the method for grid searchl。
5) by dynamic brain network kernel of graph device, by the kernel of graph matrix (X with weight coefficient of all periods1,X2,...,
Xs) merge as a dynamic brain network nuclear matrix X；
After generating weight to the kernel of graph of each period, the kernel of graph linear combination under multiple periods is become into a dynamic
The brain network kernel of graph, the kernel of graph can be expressed as
Step 4: being realized with the dynamic brain network nuclear matrix X generated and auxiliary diagnosis, tool are carried out to Alzheimer disease
Steps are as follows for body:
1) A Erci is carried out using obtained dynamic brain network nuclear matrix X and core SVM knot merga pass data training aids
The silent disease in sea whether the training of illness；
2) auxiliary diagnosis of Alzheimer disease is realized using the training data and by auxiliary diagnosis device.
Claims (8)
1. a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph, it is characterised in that: including pretreatment
Unit, dynamic brain network struction unit, dynamic brain network core computing unit and classification diagnosis unit, the pretreatment unit are used
In the Functional imagnetic resonance imaging that will be inputted, passes sequentially through time adjustment, the dynamic correction of head and smoothing and noise-reducing process and obtain later
A series of Functional imagnetic resonance imaging after noise reductions；The dynamic brain network struction unit is for exporting the pretreatment unit
Noise reduction after Functional imagnetic resonance imaging, successively carry out the matching of brain area, the segmentation of period, the calculating of association relationship and
The excavation of Frequent tree mining；The dynamic brain network core computing unit is used for the weight for exporting the dynamic brain network struction unit
The Frequent tree mining dynamic brain function network built, passes sequentially through the Optimum Matching of bipartite graph, the calculating of the kernel of graph, the combination of kernel of graph matrix
And the distribution of weight, finally obtain a fused dynamic brain function network nuclear matrix；The classification diagnosis unit is used
Carrying out Alzheimer disease with core SVM knot merga pass data training aids in the dynamic brain function network nuclear matrix that will be obtained is
Then the training of no illness realizes that the auxiliary of Alzheimer disease is examined using obtained training data and by auxiliary diagnosis device
It is disconnected.
2. a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph according to claim 1,
Be characterized in that: the dynamic brain network struction unit include brain area adaptation, time series segmentation device, mutual information calculator,
Brain network struction device and Frequent tree mining delver, the brain area adaptation of dynamic brain network struction unit receive to come from pretreatment unit
Noise reduction after Functional imagnetic resonance imaging, and by the AAL mould of Functional imagnetic resonance imaging and 90 brain areas after the noise reduction
Plate is matched, then the Functional imagnetic resonance imaging after being matched is adopted the Functional imagnetic resonance imaging after the matching
The segmentation that the period is carried out with the mode of overlapping obtains the time-series image of each period, secondly will pass through time series
All time-series images after sectionaliser segmentation, which export, gives mutual information calculator, is calculated by the mutual information calculator any
Association relationship between two brain areas, the association relationship are used to indicate the relevance between brain area, then the association relationship that will be obtained
It is input in brain network struction device, constructs the dynamic brain function network of the time-series image of each period, will finally own
The dynamic brain function network inputs of the time-series image of period are into Frequent tree mining delver, the Frequent tree mining rebuild
Dynamic brain function network image；The brain area adaptation is used for the standard AAL mould of pretreated image and 90 brain areas
Plate matched after Functional imagnetic resonance imaging；
The time series that the time series segmentation device is used to obtain Functional imagnetic resonance imaging is segmented, and obtains s
The time series of period；
The mutual information calculator is used to calculate between any two brain area the time series under any one period
Association relationship, to indicate the relevance between brain area；
The brain network struction device is used for the association relationship by calculating and constructs the s width Functional imagnetic resonance imaging after segmentation
Brain network obtains the brain network of s period；
The Frequent tree mining delver is used to carry out obtained all dynamic brain function networks the excavation of Frequent tree mining, obtains
The Frequent tree mining dynamic brain network of reconstruction.
3. a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph according to claim 1,
Be characterized in that: the dynamic brain network core computing unit include Optimum Matching device, kernel of graph matrix generator, weight generator,
Dynamic brain network kernel of graph device, by Optimum Matching device by the frequent of any pair of any two Functional imagnetic resonance imaging reconstruction
The number of isomorphism Frequent tree mining in subgraph dynamic brain function network image is as connecting this to dynamic brain function net in bipartite graph
The weight on the side of network, using the Frequent tree mining dynamic brain function network image of each reconstruction as optimal of node progress bipartite graph
Match, obtain matched dynamic brain function network pair, the dynamic brain function network of each period after matching is then passed through into figure
Core device obtains the kernel of graph under each period between any two dynamic brain function network, each period that secondly will be obtained
The kernel of graph between lower any two dynamic brain function network obtains the kernel of graph square under each period by kernel of graph matrix generator
Battle array, then be that the kernel of graph under each period generates one by the kernel of graph matrix under obtained all periods by weight generator
A weight coefficient finally obtains the kernel of graph matrix with weight coefficient of all periods, is melted by dynamic brain network kernel of graph device
Synthesize a dynamic brain function network nuclear matrix；
The Optimum Matching device is used for dynamic with the Frequent tree mining of any pair of reconstruction of any two Functional imagnetic resonance imaging
The number of isomorphism Frequent tree mining in state brain function network image is as connecting this side to dynamic brain function network in bipartite graph
Weight, using the Frequent tree mining dynamic brain function network image of each reconstruction as node carry out bipartite graph Optimum Matching；
The kernel of graph device is used for any two dynamic brain function network query function kernel of graph under the period after each matching；
The kernel of graph matrix generator is used to all dynamic brain network kernel of graphs of each period being combined into a matrix, by
This has obtained the kernel of graph matrix under s period；
The weight generator is used to utilize the method for Multiple Kernel Learning to the kernel of graph matrix allocation weight of each period；
The dynamic brain network kernel of graph device by according to the weight distributed under each period by way of linear combination based on
Nomogram core obtains the dynamic brain function network kernel of graph.
4. a kind of Alzheimer disease auxiliary diagnosis system based on the dynamic brain network kernel of graph described in claim 1-3 any one
The application method of system, which comprises the following steps:
Step 1: the pretreatment of Functional imagnetic resonance imaging；
Step 2: utilizing pretreated picture construction dynamic brain function network；
Step 3: to the Frequent tree mining dynamic brain network query function kernel of graph of reconstruction, and finally obtaining the dynamic brain network kernel of graph；
Step 4: being realized with the dynamic brain network nuclear matrix X generated and auxiliary diagnosis is carried out to Alzheimer disease.
5. a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph according to claim 4 makes
With method, which is characterized in that the pretreatment of step 1 Functional imagnetic resonance imaging the following steps are included:
1) i Functional imagnetic resonance imaging is obtained into the image (I after time adjustment by time adjustment device1,I2,...,Ii)；
2) by the image (I after time adjustment1,I2,...,Ii) the dynamic correction of head is carried out, removal head moves bring error, dynamic by head
Corrector obtains the image (H moved after correction to the end1,H2,...,Hi)；
3) by the image (H after the dynamic correction of head1,H2,...,Hi) by image noise reduction device progress the disposal of gentle filter, it obtains smooth
Image (S after filtering processing1,S2,...,Si)。
6. a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph according to claim 4 makes
With method, which is characterized in that the step 2 utilizes pretreated picture construction dynamic brain function network, including following step
It is rapid:
1) by the image (S after the disposal of gentle filter obtained after the step 1 pretreatment1,S2,...,Si) and 90 brains
The AAL template in area is matched, and the image comprising 90 brain areas is obtained；
2) by time series segmentation device, by it is described include 90 brain areas image (A1,A2,...,Ai) it is divided into s overlapping
Period obtains the time-series image (T comprising s period1,T2,...,Ts), wherein s belongs to natural number；
3) by mutual information calculator to the time series T of any one periodjCalculate the mutual information between any two brain area
It is worth, wherein Tj∈(T1,T2,...,Ts)；
4) association relationship is passed sequentially through into brain network struction device, constructs dynamic brain function network, obtains i*s dynamic brain
Functional network imageWherein i indicates the number of Functional imagnetic resonance imaging, s
Indicate the period number that time series divider is divided；
5) by Frequent tree mining device to i*s obtained dynamic brain function network image Carry out Frequent tree mining excavation, the Frequent tree mining dynamic brain function network image rebuild
7. a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph according to claim 4 makes
With method, which is characterized in that the Frequent tree mining dynamic brain function network query function kernel of graph that the step 3 pair is rebuild, and final
To the dynamic brain function network kernel of graph, comprising the following steps:
1) in the Frequent tree mining dynamic brain function network image of any pair of reconstruction of any two Functional imagnetic resonance imaging
Isomorphism Frequent tree mining number as this weight to the side of dynamic brain function network is connected in bipartite graph, with each reconstruction
Frequent tree mining dynamic brain function network image carries out the Optimum Matching of bipartite graph as node, obtains matched dynamic brain function net
Network pairWherein m, n indicate any pair of matched dynamic brain function network pair；
2) by the dynamic brain function network of matched each period againBy kernel of graph device, obtain
The kernel of graph under each period between any two dynamic brain function network
3) by the kernel of graph under each period between any two dynamic brain function networkIt is raw by kernel of graph matrix
It grows up to be a useful person, obtains the kernel of graph matrix (X under s period1,X2,...,Xs)；
4) by the kernel of graph matrix (X under the s period1,X2,...,Xs) by weight generator, it is under each period
The kernel of graph generates a weight coefficient；
5) by dynamic brain network kernel of graph device, by the kernel of graph matrix (X with weight coefficient of all periods1,X2,...,Xs)
Fusion becomes a dynamic brain network nuclear matrix X.
8. a kind of Alzheimer disease assistant diagnosis system based on the dynamic brain network kernel of graph according to claim 4 makes
With method, which is characterized in that the step 4 is realized with the dynamic brain network nuclear matrix X generated to Alzheimer disease
Carry out auxiliary diagnosis, comprising the following steps:
1) Alzheimer is carried out using obtained dynamic brain network nuclear matrix X and core SVM knot merga pass data training aids
Disease whether the training of illness；
2) auxiliary diagnosis of Alzheimer disease is realized using the training data and by auxiliary diagnosis device.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910742381.3A CN110459317B (en) | 2019-08-13 | 2019-08-13 | Alzheimer disease auxiliary diagnosis system and method based on dynamic brain network image core |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910742381.3A CN110459317B (en) | 2019-08-13 | 2019-08-13 | Alzheimer disease auxiliary diagnosis system and method based on dynamic brain network image core |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110459317A true CN110459317A (en) | 2019-11-15 |
| CN110459317B CN110459317B (en) | 2023-03-24 |
Family
ID=68486131
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910742381.3A Active CN110459317B (en) | 2019-08-13 | 2019-08-13 | Alzheimer disease auxiliary diagnosis system and method based on dynamic brain network image core |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110459317B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110916615A (en) * | 2019-12-05 | 2020-03-27 | 四川大学 | Alzheimer's disease detection device and computer equipment |
| 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 |
| CN112233805A (en) * | 2020-10-14 | 2021-01-15 | 河北工业大学 | Mining method for biomarkers based on multi-map neuroimaging data |
| CN112515653A (en) * | 2020-10-09 | 2021-03-19 | 天津大学 | Brain network construction method based on nuclear magnetic resonance image |
| CN112971778A (en) * | 2021-02-09 | 2021-06-18 | 北京师范大学 | Brain function imaging signal obtaining method and device and electronic equipment |
| CN113509148A (en) * | 2021-04-28 | 2021-10-19 | 东北大学 | Schizophrenia detection system based on mixed high-order brain network |
| EP4156200A1 (en) * | 2021-09-27 | 2023-03-29 | Universitätsmedizin der Johannes Gutenberg-Universität Mainz | Method for reconstructing a brain network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106991132A (en) * | 2017-03-08 | 2017-07-28 | 南京信息工程大学 | A kind of figure sorting technique reconstructed based on atlas with kernel of graph dimensionality reduction |
| WO2018048287A1 (en) * | 2016-09-12 | 2018-03-15 | 삼성전자 주식회사 | Image processing method and device for projecting image of virtual reality content |
| CN108921286A (en) * | 2018-06-29 | 2018-11-30 | 杭州电子科技大学 | A kind of tranquillization state function brain network establishing method for exempting from threshold value setting |
| CN109034263A (en) * | 2018-08-15 | 2018-12-18 | 东北大学 | The Alzheimer disease auxiliary diagnostic equipment and method of the brain network multi-frequency fusion kernel of graph |
| 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 |
-
2019
- 2019-08-13 CN CN201910742381.3A patent/CN110459317B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018048287A1 (en) * | 2016-09-12 | 2018-03-15 | 삼성전자 주식회사 | Image processing method and device for projecting image of virtual reality content |
| CN106991132A (en) * | 2017-03-08 | 2017-07-28 | 南京信息工程大学 | A kind of figure sorting technique reconstructed based on atlas with kernel of graph dimensionality reduction |
| CN108921286A (en) * | 2018-06-29 | 2018-11-30 | 杭州电子科技大学 | A kind of tranquillization state function brain network establishing method for exempting from threshold value setting |
| CN109034263A (en) * | 2018-08-15 | 2018-12-18 | 东北大学 | The Alzheimer disease auxiliary diagnostic equipment and method of the brain network multi-frequency fusion kernel of graph |
| 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 |
Non-Patent Citations (1)
| Title |
|---|
| 郭浩等: "基于超网络判别子图的阿尔兹海默症分类", 《计算机工程与设计》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN110916615A (en) * | 2019-12-05 | 2020-03-27 | 四川大学 | Alzheimer's disease detection device and computer equipment |
| CN112515653A (en) * | 2020-10-09 | 2021-03-19 | 天津大学 | Brain network construction method based on nuclear magnetic resonance image |
| CN112515653B (en) * | 2020-10-09 | 2024-03-26 | 天津大学 | Brain network construction method based on nuclear magnetic resonance image |
| CN112233805A (en) * | 2020-10-14 | 2021-01-15 | 河北工业大学 | Mining method for biomarkers based on multi-map neuroimaging data |
| CN112233805B (en) * | 2020-10-14 | 2022-03-01 | 河北工业大学 | Mining method for biomarkers based on multi-map neuroimaging data |
| CN112971778A (en) * | 2021-02-09 | 2021-06-18 | 北京师范大学 | Brain function imaging signal obtaining method and device and electronic equipment |
| CN113509148A (en) * | 2021-04-28 | 2021-10-19 | 东北大学 | Schizophrenia detection system based on mixed high-order brain network |
| CN113509148B (en) * | 2021-04-28 | 2022-04-22 | 东北大学 | Schizophrenia detection system based on mixed high-order brain network |
| EP4156200A1 (en) * | 2021-09-27 | 2023-03-29 | Universitätsmedizin der Johannes Gutenberg-Universität Mainz | Method for reconstructing a brain network |
| WO2023046960A1 (en) * | 2021-09-27 | 2023-03-30 | Universitätsmedizin Der Johannes Gutenberg-Universität Mainz | Method for reconstructing a brain network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110459317B (en) | 2023-03-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110459317A (en) | Alzheimer disease assistant diagnosis system and method based on the dynamic brain network kernel of graph | |
| CN111539467A (en) | GAN network architecture and method for data augmentation of medical image data set based on generation of countermeasure network | |
| CN110288609B (en) | Multi-modal whole-heart image segmentation method guided by attention mechanism | |
| CN111631688B (en) | Algorithm for automatic sleep staging | |
| CN109034263B (en) | Alzheimer disease auxiliary diagnosis device and method of brain network multi-frequency fusion image core | |
| CN110544239A (en) | Multi-modal MRI conversion method, system and medium for generating countermeasure network based on conditions | |
| CN105022934B (en) | A kind of artificial immunity method for being used to build brain effective connectivity network from fMRI data | |
| CN103985105A (en) | Contourlet domain multi-modal medical image fusion method based on statistical modeling | |
| CN110111325A (en) | Neuroimaging classification method, terminal and computer readable storage medium | |
| CN110934587A (en) | Alzheimer disease auxiliary diagnosis method based on atlas neural network | |
| CN109192298A (en) | Deep brain medical diagnosis on disease algorithm based on brain network | |
| CN113990494B (en) | Tic disorder auxiliary screening system based on video data | |
| Zhang et al. | Diffusion kernel attention network for brain disorder classification | |
| CN112883756A (en) | Generation method of age-transformed face image and generation countermeasure network model | |
| CN116503680B (en) | Brain image structured analysis and brain disease classification system based on brain atlas | |
| CN113947157B (en) | Dynamic brain effect connection network generation method based on hierarchical clustering and structural equation model | |
| CN112634214A (en) | Brain network classification method combining node attributes and multilevel topology | |
| CN106599903A (en) | Correlation weighed least squares-dictionary learning signal reconstruction method | |
| CN110192860A (en) | A kind of the Brian Imaging intelligent test analyzing method and system of network-oriented information cognition | |
| Kashtanova et al. | Deep learning for model correction in cardiac electrophysiological imaging | |
| CN114155952A (en) | Senile dementia illness auxiliary analysis system for elderly people | |
| CN113705680A (en) | Cancer associated depression identification method based on multi-modal magnetic resonance data | |
| CN107256408B (en) | Method for searching key path of brain function network | |
| CN109767429A (en) | A kind of image screening method and device | |
| Sellami et al. | Mapping individual differences in cortical architecture using multi-view representation 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |