CN111627553A - Method for constructing individualized prediction model of first-onset schizophrenia - Google Patents
Method for constructing individualized prediction model of first-onset schizophrenia Download PDFInfo
- Publication number
- CN111627553A CN111627553A CN202010454990.1A CN202010454990A CN111627553A CN 111627553 A CN111627553 A CN 111627553A CN 202010454990 A CN202010454990 A CN 202010454990A CN 111627553 A CN111627553 A CN 111627553A
- Authority
- CN
- China
- Prior art keywords
- network
- schizophrenia
- constructing
- node
- 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.)
- Pending
Links
- 201000000980 schizophrenia Diseases 0.000 title claims abstract description 68
- 238000000034 method Methods 0.000 title claims abstract description 34
- 210000004556 brain Anatomy 0.000 claims abstract description 69
- 238000009792 diffusion process Methods 0.000 claims abstract description 28
- 230000004927 fusion Effects 0.000 claims abstract description 23
- 238000012549 training Methods 0.000 claims abstract description 20
- 238000012216 screening Methods 0.000 claims abstract description 12
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 238000007500 overflow downdraw method Methods 0.000 claims abstract description 7
- 238000011156 evaluation Methods 0.000 claims abstract description 5
- 238000012795 verification Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 56
- 239000000835 fiber Substances 0.000 claims description 21
- 238000012360 testing method Methods 0.000 claims description 21
- 230000002776 aggregation Effects 0.000 claims description 19
- 238000004220 aggregation Methods 0.000 claims description 16
- 238000009826 distribution Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000000694 effects Effects 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 239000006185 dispersion Substances 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 6
- 238000003384 imaging method Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 5
- 238000007477 logistic regression Methods 0.000 claims description 5
- 238000002595 magnetic resonance imaging Methods 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000005054 agglomeration Methods 0.000 claims description 3
- 210000005013 brain tissue Anatomy 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 238000007635 classification algorithm Methods 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 230000004886 head movement Effects 0.000 claims description 3
- 210000001259 mesencephalon Anatomy 0.000 claims description 3
- 238000013421 nuclear magnetic resonance imaging Methods 0.000 claims description 3
- 238000003909 pattern recognition Methods 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 210000003625 skull Anatomy 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 238000012614 Monte-Carlo sampling Methods 0.000 claims description 2
- 238000001558 permutation test Methods 0.000 claims 1
- 238000003745 diagnosis Methods 0.000 abstract description 7
- 238000013473 artificial intelligence Methods 0.000 abstract description 4
- 238000002610 neuroimaging Methods 0.000 abstract description 2
- 238000005457 optimization Methods 0.000 description 7
- 239000000090 biomarker Substances 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 108010015046 cell aggregation factors Proteins 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 238000013399 early diagnosis Methods 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008309 brain mechanism Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 231100000518 lethal Toxicity 0.000 description 1
- 230000001665 lethal effect Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 208000020016 psychiatric disease Diseases 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- 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
-
- 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
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/70—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to mental therapies, e.g. psychological therapy or autogenous training
-
- 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
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Child & Adolescent Psychology (AREA)
- Developmental Disabilities (AREA)
- Hospice & Palliative Care (AREA)
- Psychiatry (AREA)
- Psychology (AREA)
- Social Psychology (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention belongs to the field of psychology, neuroimaging and artificial intelligence, and discloses a method for constructing an individualized prediction model of first-onset schizophrenia, which solves the problem of low accuracy of auxiliary diagnosis of the conventional SCH brain structure network model. The method comprises the following steps: A. acquiring a diffusion tensor image of a patient with first schizophrenia; B. preprocessing the acquired diffusion tensor image; C. constructing a sparse brain structure network based on the preprocessed image; D. constructing a multi-threshold fusion brain structure network after each tested sparse by adopting a similar network fusion method; E. extracting network topology attribute features of the multi-threshold fusion brain structure, and then performing feature screening; F. based on the screened features, performing classification training by adopting a classifier to obtain an individualized prediction model of the first schizophrenia; G. and (4) performing performance verification evaluation on the trained individualized prediction model of the first schizophrenia.
Description
Technical Field
The invention belongs to the fields of psychology, neuroimaging and artificial intelligence, and particularly relates to a method for constructing an individualized prediction model of first-onset schizophrenia.
Background
Schizophrenia (SCH) is a high disabling and lethal mental disorder, which is listed as one of ten diseases with the highest ranking list of global disease burden by the world health organization, however, the brain mechanism of the Schizophrenia is still not completely clear, the diagnosis lacks objective standards, and the cure rate is low. The clinical problem to be solved urgently is to seek an objective, effective, convenient and feasible biological marker for carrying out early individual classified diagnosis and treatment on SCH. Brain structure network changes are an important biological basis of SCH neuro-anatomical abnormalities, and a machine learning method serving as a prediction and analysis tool based on data driving can fully utilize structural information in biomarker data to construct an SCH individualized brain structure network model.
The current research situation of the SCH brain structure network model is as follows: 1) directly extracting a structural connection value in an original network as a feature, and incorporating pseudo-connection with lower weight; 2) the method for thinning the network based on the single fixed threshold has noise influence of different levels, and the selection of the single threshold has subjectivity; 3) the original structure connection value is directly applied as the information characteristic of the low-level structure, and the important network attribute with complex brain topology is ignored.
Therefore, the conventional SCH brain structure network model is very difficult to find schizophrenia sensitive biomarkers based on brain structure network characteristics, and the accuracy of the model for auxiliary diagnosis is low.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: a construction method of an individual prediction model of first-onset schizophrenia is provided, and the problem that the accuracy of auxiliary diagnosis of an existing SCH brain structure network model is low is solved.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the method for constructing the individualized prediction model of the first-onset schizophrenia comprises the following steps:
A. acquiring a diffusion tensor image of a patient with first schizophrenia;
B. preprocessing the acquired diffusion tensor image;
C. constructing a sparse brain structure network based on the preprocessed image;
D. constructing a multi-threshold fusion brain structure network after each tested sparse by adopting a similar network fusion method;
E. extracting network topology attribute features of the multi-threshold fusion brain structure, and then performing feature screening;
F. based on the screened features, performing classification training by adopting a classifier to obtain an individualized prediction model of the first schizophrenia;
G. and (4) performing performance verification evaluation on the trained individualized prediction model of the first schizophrenia.
As a further optimization, in step a, a nuclear magnetic resonance imaging scanner is used to scan and acquire a diffusion tensor image of the first schizophrenia patient by using a single-shot planar echo imaging (EPI) technology.
As a further optimization, in step B, the preprocessing of the obtained diffusion tensor image specifically includes:
b1, converting the dispersion tensor image of the DICOM data format into an image of NIFTI format by using MRI convert (magnetic resonance imaging conversion);
b2, performing eddy current correction and head movement correction on the diffusion tensor image after data format conversion;
b3 skull was removed by Brain Extraction Tool (FSL) and non-Brain tissue images were removed.
As a further optimization, in step C, the constructing a sparse brain structure network based on the preprocessed image specifically includes:
c1, registering the brain images after the pre-processing of each test to b0 images in a diffusion tensor space by a rotation and translation linear registration method; the registered b0 images are then registered to T1 images in standard MNI space; inverting the conversion matrix, and transforming the AAL template from the MNI space to the diffusion tensor space by using the obtained inverse matrix to obtain 90 brain area network nodes divided based on the AAL template;
c2, using a probabilistic fiber bundle imaging method, carrying out Bayesian estimation of diffusion parameters based on a BEDPOSTX tool, establishing distribution of the diffusion parameters of each voxel by using a Markov Chain Monte Carlo (MCMC) sampling method, presetting each voxel of a brain as a fiber cross model, and automatically judging how many kinds of crossed fiber bundles pass;
c3, carrying out probability tracking fiber bundle reconstruction based on a PROBTRACKX tool, repeatedly sampling the distribution of each voxel in the main diffusion direction, generating a streamline from the extracted local sample each time, and establishing a statistical graph of streamline position posterior distribution through multiple sampling to obtain the distribution situation of the structural connection probability between every two brain regions; defining the weight of each edge as the fiber bundle connection probability between every two node areas, and obtaining a symmetrical 90 x 90 fiber bundle connection probability weighting network matrix for each tested object;
and C4, setting a fiber bundle connection probability threshold, wherein structural connection exists between the two brain areas exceeding the threshold, testing the influence of different sparsity thresholds on the fusion effect, and constructing a sparse structure network by adopting a relatively narrower threshold range (5-40% and the step length of 1%).
As a further optimization, in step D, the constructing each tested sparse multi-threshold fusion brain structure network by using a similar network fusion method specifically includes:
d1, connecting the matrix W with the sparse structure to be defined as a full kernel (full kernel) matrixi jFor the ith tested jth threshold full-kernel matrix Wi jFurther constructing a corresponding sparse kernel (sparse kernel) matrix:
order touIs a full-core matrix Wi jK neighbors of the intermediate node u (including the node u itself), then the sparse kernel matrix Si jIs defined as:
d2, carrying out iterative updating of the full-kernel matrix based on the sparse kernel matrix corresponding to the full-kernel matrix:
wherein (W)i c)(m)Represents the full-kernel matrix under the ith tested c threshold value in the mth iteration, (W)i j)(m+1)Represents the m +1 thA full-kernel matrix in the secondary iteration, wherein N is the total sparse threshold number;
d3, judging whether the iteration convergence condition is met, if so, executing a step D4, otherwise, continuing the iteration;
wherein the convergence condition is: i (W)i j)(m+1)-(Wi j)(m)||≤0.01;
D4, and corresponding the updated N thresholds to the full-core matrix Wi jAveraging was performed to construct an averaged full kernel matrix for each test:
d5, mixing WiNormalized to the interval [0, 1 ]]Thus generating a final fused network for each test.
As a further optimization, in step E, the extracting the network topology attribute feature of the multi-threshold fusion brain structure specifically includes:
calculating AUC values of the fused network under all thresholds of 8 global topological attributes and 3 node topological attributes based on a graph theory analysis method as initial features of subsequent classification;
the 8 global topology attributes include network strength SpGlobal efficiency EglobLocal efficiency ElocLength of shortest path LpCoefficient of aggregation CpNormalized shortest path length λ, normalized clustering coefficient γ, small world attribute σ:
the network strength SpThe calculation formula of (2) is as follows:
wherein, S (i) is the weighted sum of edges connected with the ith node, and N is the number of brain areas in the whole brain network;
the shortest path length LpThe calculation formula of (2) is as follows:
wherein L isijRepresenting the shortest path between node i and node j, LpThe shortest path length for the entire network G;
the global efficiency EglobThe calculation formula (2) includes:
wherein E isglob_i(G) The global efficiency of the node i is the average value of the global efficiency of all nodes in the network G;
the local efficiency ElocThe calculation formula (2) includes:
wherein L isjkIs the shortest path length between region j and region k, GiIs a sub-network of nodes connected to area i, NGiIs a sub-network GiThe number of midbrain regions; eloc_i(G) Is the local efficiency of node i, the local efficiency E of network Gloc(G) The average value of the local efficiency of all nodes in the network is obtained;
the agglomeration coefficient CpThe calculation formula (2) includes:
wherein, C (i) is the aggregation coefficient of the node i, and the aggregation coefficient of the network G is the average value of the aggregation coefficients of all the nodes;
the normalized aggregation coefficient γ and the normalized shortest path length λ are calculated as follows:
wherein, Cp randAnd Lp randThe average values of the aggregation coefficients and the shortest path lengths of 100 random networks are respectively;
the formula for calculating the small world attribute sigma is as follows:
σ=γ/λ;
the 3 kinds of node topology attributes comprise a node degree Dnodal(i) Node efficiency Enodal(i) Center degree of node betweenness Bnodal(i) Respectively defined as follows:
wherein e isstRepresenting the number of all shortest paths from node s to node t in network G, esitIs the number of passing nodes i in these shortest paths.
As a further optimization, in step E, the performing feature screening specifically includes:
a Recursive Feature Elimination (RFE) algorithm based on a support vector machine performs Feature selection by continuously training a classifier and removing Feature dimensions with smaller Feature weights, and specifically includes:
secondly, taking the feature set as input, training a classifier, and obtaining a classification effect and the weight of each feature;
removing the features with the minimum weight to form a new feature set;
As a further optimization, in step F, the Classifier adopts an SVM Classifier, a Logistic Regression Classifier (Logistic Regression Classifier) or a plurality of classifiers ensemble learning based on a Radial Basis Function (RBF) kernel; during classification training, a plurality of pattern recognition classification methods are applied to find an optimal classifier model and screen key multi-threshold fusion brain structure network characteristics.
And G, evaluating the performances of various prediction models of the schizophrenia training set and the test set by adopting cross validation, wherein the performances comprise accuracy, sensitivity, specificity, ROC curve, AUC value and the like, and the system optimizes brain image feature selection and individual subtype prediction models by using a replacement test through a test feature screening and classification algorithm and identifies the core brain image group features related to the subtypes so as to improve the individual prediction accuracy.
The invention has the beneficial effects that:
(1) constructing a network after multi-threshold fusion, generating a topological attribute characteristic of the fusion network independent of a single threshold as an initial characteristic of classification by integrating complementary information provided by an original network under different topological views, and performing classification training after characteristic screening, wherein the method can improve the classification accuracy, considers the component of a structural index and has interpretability;
(2) the artificial intelligence model established based on the biological data fused with the multi-threshold brain structure network can automatically acquire the data updating model and establish an objective biological marker for early diagnosis of the schizophrenia patient, and the model accurately and qualitatively displays high reliability and high stability for classification of the first schizophrenia patient.
Drawings
Fig. 1 is a flow chart of a method for constructing an individualized prediction model of first-onset schizophrenia in the invention.
Detailed Description
The invention aims to provide a construction method of an individual prediction model of first-onset schizophrenia, and solves the problem of low accuracy of auxiliary diagnosis of the conventional SCH brain structure network model. The core idea is as follows: acquiring single-shot plane echo imaging of the diffusion tensor of the first schizophrenia patient; preprocessing a diffusion tensor image; constructing a sparse brain structure network based on the preprocessed image; constructing each tested sparse multi-threshold structure network by adopting a similar network fusion method; extracting the topological attribute characteristics of the network based on the processed fusion multi-threshold brain structure, performing classification training after characteristic screening to obtain an individualized prediction model of the first schizophrenia, and finally performing performance verification evaluation on the individualized prediction model of the first schizophrenia obtained by training.
For networks under different sparsity thresholds, the networks can be regarded as different types of feature expressions of the same tested brain network, and the invention fuses the networks with multiple sparsity thresholds, so that richer topological information can be obtained, the subsequent classification work is facilitated, and the problems that the method for thinning the networks by adopting a single fixed threshold has noise influence of different levels and the selection of the single threshold has subjectivity are avoided; the network topology attributes obtained based on the graph theory analysis can reflect the high-level attributes of the network, and the attributes are used as classification features, so that better classification results can be obtained compared with the original network connection values which can only reflect low-level information as classification features.
In particular implementation, the flow of the method for constructing the individualized prediction model for first-onset schizophrenia in the invention is shown in fig. 1, and the method comprises the following implementation steps:
1. acquiring a diffusion tensor image of a patient with first schizophrenia;
in the step, Philip 3.0T and GE 3.0T nuclear magnetic resonance imaging scanners are adopted to collect data as a training module test set and a test module data set as toolsIn the volume implementation, the scanning parameters are that in 32 axial plane directions, TR is 10295ms, TE is 91ms, and FOV is 128mm × 128mm2The flip angle is 90 °, the layer thickness is 4mm, the matrix is 256 × 256, and the individual voxel size is 2 × 2 × 2mm3Acquiring 3DT1 structural image data to optimize DTI data registration, wherein TR is 8.4ms, TE is 3.8ms, and FOV is 256 × 256mm2The flip angle is 90 °, the layer thickness is lmm, the scanning is continuous without intervals, the matrix is 256 × 256, and the size of a single voxel is 1 × 1 × 1mm3The whole brain acquires 188 layers of images.
2. Preprocessing the acquired diffusion tensor image;
in this step, as a specific implementation means, the pretreatment process is as follows:
firstly, a diffusion tensor image in a DICOM data format is converted into an NIFTI format image by using MRI convert;
performing eddy current correction and head movement correction on the diffusion tensor image after data format conversion;
③ removing the skull by using Brain Extraction Tool of FSL, and removing the non-Brain tissue image.
3. Constructing a sparse brain structure network based on the preprocessed image;
in this step, as a specific implementation means, the process of constructing the sparse brain structure network is as follows:
firstly, registering the brain images preprocessed by the tests to b0 images in a diffusion tensor space by a rotation and translation linear registration method; the registered b0 images are then registered to T1 images in standard MNI space; inverting the conversion matrix, and transforming the AAL template from the MNI space to the diffusion tensor space by using the obtained inverse matrix to obtain 90 brain area network nodes divided based on the AAL template;
secondly, a probability fiber bundle imaging method is used, Bayesian estimation of dispersion parameters is performed based on a BEDPOSTX tool, distribution of the dispersion parameters of each voxel is established by a Markov chain Monte Carlo sampling method, each voxel of a brain is preset as a fiber cross model, and how many kinds of cross-passing fiber bundles are automatically judged;
thirdly, probability tracking fiber bundle reconstruction is carried out based on a PROBTRACKX tool, a streamline is generated from an extracted local sample each time by repeatedly sampling the distribution of the main dispersion direction of each voxel, and a statistical graph of the posterior distribution of the streamline position is established by sampling for many times to obtain the distribution situation of the structural connection probability between every two brain regions; defining the weight of each edge as the fiber bundle connection probability between every two node areas, and obtaining a symmetrical 90 x 90 fiber bundle connection probability weighting network matrix for each tested object;
setting a fiber bundle connection probability threshold, and testing the influence of different sparsity thresholds on the fusion effect by the structural connection of two brain areas exceeding the threshold, wherein a relatively narrower threshold range (5-40% and 1% step length) is adopted to construct a sparse structure network.
4. Constructing a multi-threshold fusion brain structure network after each tested sparse by adopting a similar network fusion method;
in this step, as a specific implementation means, the process of constructing the multi-threshold fusion brain structure network is as follows:
① the sparse post-structure connection matrix is defined as the full-core matrix Wi jFor the ith tested jth threshold full-kernel matrix Wi jAnd further constructing a corresponding sparse kernel matrix, wherein the sparse kernel matrix is used for strong connection still reserved after the coding network is sparse:
order touIs a full-core matrix Wi jK neighbors of the intermediate node u (including the node u itself), then the sparse kernel matrix Si jIs defined as:
secondly, performing iterative updating of the full-kernel matrix based on the sparse kernel matrix corresponding to the full-kernel matrix:
wherein (W)i c)(m)Represents the full-kernel matrix under the ith tested c threshold value in the mth iteration, (W)i j)(m+1)Representing a full-kernel matrix in the (m + 1) th iteration, wherein N is the total sparse threshold number;
the full-core matrix W interacts with all but one of the other threshold networksi jComplementary information provided by the original network under other topological views can be integrated. At the same time, the sparse kernel matrix Si jBy corresponding to the full-core matrix Wi jThe strongest connection in (b) leads to an iterative process and thus noise can be effectively suppressed. From the perspective of matrix multiplication, the whole iterative process is represented by the above formula, which means the full-core matrix Wi jThe magnitude of the connection value of any two nodes depends on the k neighbors of the corresponding nodes in other threshold networks at the same time. Particularly if the respective k neighbors of two nodes are the strongest connections in other thresholded networks, the connections between them will strengthen after iterative updating (although they may themselves be weak connections) and vice versa.
Judging whether the iteration convergence condition is met, if so, executing the step IV, and otherwise, continuing the iteration;
wherein the convergence condition is: i (W)i j)(m+1)-(Wi j)(m)||≤0.01;
④, and corresponding the updated N thresholds to the full-core matrix Wi jAveraging was performed to construct an averaged full kernel matrix for each test:
⑤, mixing WiNormalized to the interval [0, 1 ]]Thus generating a final fused network for each test.
5. Extracting network topology attribute features of the multi-threshold fusion brain structure, and then performing feature screening;
in this step, as a concrete implementation means, the fusion is calculated based on a graph theory analysis methodAUC values of 8 global topology attributes and 3 node topology attributes of the post-network under all thresholds serve as initial features of subsequent classification. Wherein the 8 global topology attributes include network strength (node degree) SpGlobal efficiency EglobLocal efficiency ElocLength of shortest path LpCoefficient of aggregation CpNormalized shortest path length λ, normalized clustering coefficient γ, small world attribute σ. The specific definition is as follows:
network strength (node degree) SpReflecting important network evolution characteristics. The node degree is defined as the weighted sum of edges directly connected with the node, the node is more connected when the node degree is larger, and the position of the node in the network is more important. The definition formula is:
where S (i) is the sum of the weights of the edges connected to the ith node, and N is the number of brain regions in the whole brain network. The average of the degrees of all nodes in the network is the strength of the network.
Shortest path length Lp: the average value of the shortest paths of all nodes in the network is the shortest path of the network, and the operation efficiency of the whole network is reflected. Information can be transmitted faster through the shortest path, and system resources are saved. The definition formula is:
wherein L isi,jRepresenting the shortest path between node i and node j, LpIs the shortest path length of the entire network G. It can be seen that the calculation of the shortest path length must be based on the case where the network is fully connected: l if node i cannot reach node j by any way, Li,jAbsent or infinite, LpWill also not be present.
Global efficiency Eglob: describing the information transmission efficiency in the network, the global efficiency of the node i is based on the shortest path length according to the publicThe following definitions
It can be seen from the above formula that the smaller the shortest path of a node is, the faster the information transfer between the node and other nodes is, i.e. the higher the global efficiency of the node is.
And global efficiency E of network GglobDefined as the average of the global efficiency of all nodes in the network
Local efficiency ElocThe method is an important index for measuring the compactness of a cluster (clique) formed by adjacent nodes in the network and describing the redundancy (redundancy) and the tolerance (tolerance) of the network to external attacks. The local efficiency of the node i and the local efficiency of the network G are defined according to the formula:
wherein L isjkIs the shortest path length between region j and region k, GiIs a sub-network of nodes connected to area i, NGiIs a sub-network GiThe number of midbrain regions, N is the number of nodes in G in the whole brain network.
Agglomeration coefficient CpIs an important index for measuring the degree of small group (cliquescence) and local interconnection (interconnectivity) of the network, and the class coefficient c (i) of the node i is defined as the ratio of the number of edges between "other nodes" directly connected to the node i in the network G to the maximum possible number of edges between the "other nodes" and is defined according to the following formula. Aggregation factor C of network GpIs defined as the average of all node clustering coefficients.
A network is considered to have a small-world property (small-world) if it has both a high aggregation factor and a short shortest path length. To quantitatively determine whether a network has small-world attributes, the aggregation coefficient and shortest path length of the network are typically compared to the corresponding attributes of a random network.
The normalized aggregation coefficient γ and the normalized shortest path length λ are calculated respectively according to the following equations:
wherein C isp randAnd Lp randThe aggregation coefficients and the average of the shortest path lengths are for 100 random networks, respectively.
The small world attribute σ is defined as σ ═ γ/λ. If γ >1 and λ ≈ 1, i.e., σ >1, then this network is judged to have small-world properties.
The 3 node topology attributes include a node degree Dnodal(i) Node efficiency Enodal(i) Center degree of node betweenness Bnodal(i) Respectively defined as follows:
the median center degree (median center) of the intermediate number is the center degree of the node defined from the information flow perspective, and is shown in the formula estRepresenting nodes s to nodes in the network GNumber of all shortest paths of point t, esitIs the number of passing nodes i in these shortest paths.
In this step, the performing feature screening specifically includes:
a Recursive Feature Elimination (RFE) algorithm based on a support vector machine performs Feature selection by continuously training a classifier and removing Feature dimensions with smaller Feature weights, and specifically includes:
secondly, taking the feature set as input, training a classifier, and obtaining a classification effect and the weight of each feature;
removing the features with the minimum weight to form a new feature set;
6. Based on the screened features, performing classification training by adopting a classifier to obtain an individualized prediction model of the first schizophrenia;
in this step, as a specific implementation means, the Classifier adopts an SVM Classifier, a Logistic Regression Classifier (Logistic Regression Classifier) or a plurality of classifiers integrated learning based on a Radial Basis Function (RBF) kernel; during classification training, a plurality of pattern recognition classification methods are applied to find an optimal classifier model and screen key multi-threshold fusion brain structure network characteristics.
7. And (4) performing performance verification evaluation on the trained individualized prediction model of the first schizophrenia.
In the step, as a specific implementation means, cross validation is adopted to evaluate the performances of various prediction models of a schizophrenia training set and a testing set, including accuracy, sensitivity, specificity, ROC curve, AUC value and the like, and the system optimizes brain image feature selection and individual subtype prediction models and identifies subtype-related core brain image group features through testing feature screening and classification algorithms by applying displacement inspection so as to improve the individual prediction accuracy.
In conclusion, the invention adopts the technologies of artificial intelligence and machine learning, and constructs an individualized prediction model of the first-onset schizophrenia with good robustness by analyzing and mining the magnetic resonance imaging data of the brain structure of the first-onset schizophrenia, so as to realize accurate and objective auxiliary diagnosis for the early diagnosis and identification of the schizophrenia and improve the curative effect.
Claims (9)
1. The method for constructing the individualized prediction model of the first-onset schizophrenia is characterized by comprising the following steps of:
A. acquiring a diffusion tensor image of a patient with first schizophrenia;
B. preprocessing the acquired diffusion tensor image;
C. constructing a sparse brain structure network based on the preprocessed image;
D. constructing a multi-threshold fusion brain structure network after each tested sparse by adopting a similar network fusion method;
E. extracting network topology attribute features of the multi-threshold fusion brain structure, and then performing feature screening;
F. based on the screened features, performing classification training by adopting a classifier to obtain an individualized prediction model of the first schizophrenia;
G. and (4) performing performance verification evaluation on the trained individualized prediction model of the first schizophrenia.
2. The method for constructing an individualized model for predicting first-onset schizophrenia according to claim 1, wherein the first-onset schizophrenia is a single-dose schizophrenia,
in the step A, a nuclear magnetic resonance imaging scanner is used for scanning and obtaining a diffusion tensor image of the first schizophrenia patient by adopting a single-shot plane echo imaging technology.
3. The method for constructing an individualized model for predicting first-onset schizophrenia according to claim 1, wherein the first-onset schizophrenia is a single-dose schizophrenia,
in step B, the preprocessing of the acquired diffusion tensor image specifically includes:
b1, converting the dispersion tensor image in the DICOM data format into an image in an NIFTI format by using MRI convert;
b2, performing eddy current correction and head movement correction on the diffusion tensor image after data format conversion;
b3, removing skull by Brain Extraction Tool of FSL, and removing non-Brain tissue image.
4. The method for constructing an individualized model for predicting first-onset schizophrenia according to claim 1, wherein the first-onset schizophrenia is a single-dose schizophrenia,
in step C, the constructing a sparse brain structure network based on the preprocessed image specifically includes:
c1, registering the brain images after the pre-processing of each test to b0 images in a diffusion tensor space by a rotation and translation linear registration method; the registered b0 images are then registered to T1 images in standard MNI space; inverting the conversion matrix, and transforming the AAL template from the MNI space to the diffusion tensor space by using the obtained inverse matrix to obtain 90 brain area network nodes divided based on the AAL template;
c2, using a probabilistic fiber bundle imaging method, carrying out Bayesian estimation of dispersion parameters based on a BEDPOSTX tool, establishing distribution of the dispersion parameters of each voxel by using a Markov chain Monte Carlo sampling method, presetting each voxel of a brain as a fiber cross model, and automatically judging how many kinds of cross-passing fiber bundles pass;
c3, carrying out probability tracking fiber bundle reconstruction based on a PROBTRACKX tool, repeatedly sampling the distribution of each voxel in the main diffusion direction, generating a streamline from the extracted local sample each time, and establishing a statistical graph of streamline position posterior distribution through multiple sampling to obtain the distribution situation of the structural connection probability between every two brain regions; defining the weight of each edge as the fiber bundle connection probability between every two node areas, and obtaining a symmetrical 90 x 90 fiber bundle connection probability weighting network matrix for each tested object;
and C4, setting a fiber bundle connection probability threshold, wherein structural connection exists between the two brain areas exceeding the threshold, testing the influence of different sparsity thresholds on the fusion effect, and constructing a sparse structure network by adopting a relatively narrower threshold range.
5. The method for constructing an individualized model for predicting first-onset schizophrenia according to claim 1, wherein the first-onset schizophrenia is a single-dose schizophrenia,
in step D, the constructing of the multi-threshold fusion brain structure network after each tested sparse by using the similar network fusion method specifically includes:
d1, defining the sparse structure connection matrix as a full-core matrixFor the ith tested full-core matrix under the jth thresholdFurther constructing a corresponding sparse kernel matrix:
order touIs a full-core matrixK adjacent to the middle node u, then sparse kernel matrixIs defined as:
d2, carrying out iterative updating of the full-kernel matrix based on the sparse kernel matrix corresponding to the full-kernel matrix:
wherein (W)i c)(m)Represents the full-kernel matrix under the ith tested c threshold value in the mth iteration, (W)i j)(m+1)Representing a full-kernel matrix in the (m + 1) th iteration, wherein N is the total sparse threshold number;
d3, judging whether the iteration convergence condition is met, if so, executing a step D4, otherwise, continuing the iteration;
wherein the convergence condition is: i (W)i j)(m+1)-(Wi j)(m)||≤0.01;
D4, and corresponding the updated N thresholds to the full-core matrixAveraging was performed to construct an averaged full kernel matrix for each test:
d5, mixing WiNormalized to the interval [0, 1 ]]Thus generating a final fused network for each test.
6. The method for constructing an individualized model for predicting first-onset schizophrenia according to claim 1, wherein the first-onset schizophrenia is a single-dose schizophrenia,
in step E, the extracting of the multi-threshold fusion brain structure network topology attribute feature specifically includes:
calculating AUC values of the fused network under all thresholds of 8 global topological attributes and 3 node topological attributes based on a graph theory analysis method as initial features of subsequent classification;
the 8 global topology attributes include network strength SpGlobal efficiency EglobLocal efficiency ElocLength of shortest path LpCoefficient of aggregation CpNormalized shortest path length λ, normalized clustering coefficient γ, small world attribute σ:
the network strength SpThe calculation formula of (2) is as follows:
wherein, S (i) is the weighted sum of edges connected with the ith node, and N is the number of brain areas in the whole brain network;
the shortest path length LpThe calculation formula of (2) is as follows:
wherein L isijRepresenting the shortest path between node i and node j, LpThe shortest path length for the entire network G;
the global efficiency EglobThe calculation formula (2) includes:
wherein E isglob_i(G) The global efficiency of the node i is the average value of the global efficiency of all nodes in the network G;
the local efficiency ElocThe calculation formula (2) includes:
wherein L isjkIs the shortest path length between region j and region k, GiIs a sub-network of nodes connected to area i, NGiIs a sub-network GiThe number of midbrain regions; eloc_i(G) Is the local efficiency of node i, the local efficiency E of network Gloc(G) The average value of the local efficiency of all nodes in the network is obtained;
the agglomeration coefficient CpThe calculation formula (2) includes:
wherein, C (i) is the aggregation coefficient of the node i, and the aggregation coefficient of the network G is the average value of the aggregation coefficients of all the nodes;
the normalized aggregation coefficient γ and the normalized shortest path length λ are calculated as follows:
wherein, Cp randAnd Lp randThe average values of the aggregation coefficients and the shortest path lengths of 100 random networks are respectively;
the formula for calculating the small world attribute sigma is as follows:
σ=γ/λ;
the 3 kinds of node topology attributes comprise a node degree Dnodal(i) Node efficiency Enodal(i) Center degree of node betweenness Bnodal(i) Respectively defined as follows:
wherein e isstRepresenting the number of all shortest paths from node s to node t in network G, esitIs the number of passing nodes i in these shortest paths.
7. The method for constructing an individualized model for predicting first-onset schizophrenia according to claim 1, wherein the first-onset schizophrenia is a single-dose schizophrenia,
in step E, the performing feature screening specifically includes:
the recursive feature elimination algorithm based on the support vector machine performs feature selection by continuously training a classifier and removing feature dimensions with smaller feature weights, and specifically comprises the following steps:
secondly, taking the feature set as input, training a classifier, and obtaining a classification effect and the weight of each feature;
removing the features with the minimum weight to form a new feature set;
8. The method for constructing an individualized model for predicting first-onset schizophrenia according to claim 1, wherein the first-onset schizophrenia is a single-dose schizophrenia,
in the step F, the classifier adopts an SVM classifier, a logistic regression classifier or a plurality of classifiers integrated learning based on a radial basis function kernel; during classification training, a plurality of pattern recognition classification methods are applied to find an optimal classifier model and screen key multi-threshold fusion brain structure network characteristics.
9. The method for constructing the individualized prediction model for the first-onset schizophrenia according to any one of claims 1 to 8, wherein in the step G, cross validation is adopted to evaluate the performances of various prediction models of the training set and the test set of schizophrenia, including accuracy, sensitivity, specificity, ROC curve, AUC value and the like, and the system optimizes the brain image feature selection and the individual subtype prediction models by using the test feature screening and classification algorithm and applies the permutation test to identify the core brain image group features related to the subtypes so as to improve the individualized prediction accuracy.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010454990.1A CN111627553A (en) | 2020-05-26 | 2020-05-26 | Method for constructing individualized prediction model of first-onset schizophrenia |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010454990.1A CN111627553A (en) | 2020-05-26 | 2020-05-26 | Method for constructing individualized prediction model of first-onset schizophrenia |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111627553A true CN111627553A (en) | 2020-09-04 |
Family
ID=72260015
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010454990.1A Pending CN111627553A (en) | 2020-05-26 | 2020-05-26 | Method for constructing individualized prediction model of first-onset schizophrenia |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111627553A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112190269A (en) * | 2020-12-04 | 2021-01-08 | 兰州大学 | Construction method of depression auxiliary identification model based on multi-source electroencephalogram data fusion |
CN113197578A (en) * | 2021-05-07 | 2021-08-03 | 天津医科大学 | Schizophrenia classification method and system based on multi-center model |
CN113261973A (en) * | 2021-04-13 | 2021-08-17 | 山东师范大学 | Method and system for identifying atrial fibrillation electrocardiosignals through weighted multi-scale finite-crossing visual image |
CN113283465A (en) * | 2021-04-02 | 2021-08-20 | 电子科技大学 | Diffusion tensor imaging data analysis method and device |
CN114334130A (en) * | 2021-12-25 | 2022-04-12 | 浙江大学 | PET molecular image computer-aided diagnosis system based on brain symmetry |
CN118335319A (en) * | 2024-06-12 | 2024-07-12 | 四川大学 | Early prediction method for common major diseases based on virtual person simulation |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH01238696A (en) * | 1988-03-19 | 1989-09-22 | Nozomi Hoshimiya | Collation and recognition system of time series pattern |
US20130231552A1 (en) * | 2012-03-05 | 2013-09-05 | Siemens Corporation | Method and System for Diagnosis of Attention Deficit Hyperactivity Disorder from Magnetic Resonance Images |
CN103425983A (en) * | 2013-07-25 | 2013-12-04 | 电子科技大学 | Brain network topology difference fast extracting method based on network synchronicity |
US20150339570A1 (en) * | 2014-05-22 | 2015-11-26 | Lee J. Scheffler | Methods and systems for neural and cognitive processing |
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 |
WO2016168724A1 (en) * | 2015-04-15 | 2016-10-20 | Sync-Think, Inc. | System and method for concussion detection and quantification |
CN106096616A (en) * | 2016-06-08 | 2016-11-09 | 四川大学华西医院 | Magnetic resonance image feature extraction and classification method based on deep learning |
CN106845137A (en) * | 2017-02-27 | 2017-06-13 | 浙江工业大学 | A kind of sacred disease analysis method based on brain network structure |
US20170319123A1 (en) * | 2016-05-06 | 2017-11-09 | The Board Of Trustees Of The Leland Stanford Junior University | Systems and Methods for Using Mobile and Wearable Video Capture and Feedback Plat-Forms for Therapy of Mental Disorders |
CN109376751A (en) * | 2018-08-21 | 2019-02-22 | 北京工业大学 | A kind of human functional network's classification method based on convolutional neural networks |
US20190090749A1 (en) * | 2017-09-26 | 2019-03-28 | Washington University | Supervised classifier for optimizing target for neuromodulation, implant localization, and ablation |
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 |
-
2020
- 2020-05-26 CN CN202010454990.1A patent/CN111627553A/en active Pending
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH01238696A (en) * | 1988-03-19 | 1989-09-22 | Nozomi Hoshimiya | Collation and recognition system of time series pattern |
US20130231552A1 (en) * | 2012-03-05 | 2013-09-05 | Siemens Corporation | Method and System for Diagnosis of Attention Deficit Hyperactivity Disorder from Magnetic Resonance Images |
CN103425983A (en) * | 2013-07-25 | 2013-12-04 | 电子科技大学 | Brain network topology difference fast extracting method based on network synchronicity |
US20150339570A1 (en) * | 2014-05-22 | 2015-11-26 | Lee J. Scheffler | Methods and systems for neural and cognitive processing |
WO2016168724A1 (en) * | 2015-04-15 | 2016-10-20 | Sync-Think, Inc. | System and method for concussion detection and quantification |
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 |
US20170319123A1 (en) * | 2016-05-06 | 2017-11-09 | The Board Of Trustees Of The Leland Stanford Junior University | Systems and Methods for Using Mobile and Wearable Video Capture and Feedback Plat-Forms for Therapy of Mental Disorders |
CN106096616A (en) * | 2016-06-08 | 2016-11-09 | 四川大学华西医院 | Magnetic resonance image feature extraction and classification method based on deep learning |
CN106845137A (en) * | 2017-02-27 | 2017-06-13 | 浙江工业大学 | A kind of sacred disease analysis method based on brain network structure |
US20190090749A1 (en) * | 2017-09-26 | 2019-03-28 | Washington University | Supervised classifier for optimizing target for neuromodulation, implant localization, and ablation |
CN109376751A (en) * | 2018-08-21 | 2019-02-22 | 北京工业大学 | A kind of human functional network's classification method based on convolutional neural networks |
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 |
Non-Patent Citations (7)
Title |
---|
HONGWEI WEN ET AL.: "Multi-threshold white matter structural networks fusion for accurate diagnosis of early tourette syndrome children" * |
余敏;陶玲;钱志余;朱威州;: "基于弥散张量成像的脑结构网络参数研究", no. 05 * |
吴斌: "《复杂网络和科技文献知识发现》", 科学技术文献出版社, pages: 28 - 29 * |
杜欣;李嘉慧;熊冬生;潘智林;吴逢春;宁玉萍;陈军;吴凯;: "认知负载状态下的精神分裂症患者脑电图特异性研究", no. 01 * |
温洪;郭浩;李越;陈俊杰;: "重度抑郁症患者脑功能网络的分类研究", no. 08 * |
胡茂林;郑俊杰;唐劲松;廖艳辉;李宗昌;贺莹;陈华富;陈晓岗;宗小芬;: "首发未服药精神分裂症全脑及局部白质脑网络拓扑改变", no. 02 * |
薛绍伟;唐一源;李健;张兰华;曹宸;: "一种基于fMRI数据的脑功能网络构建方法", no. 11 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112190269A (en) * | 2020-12-04 | 2021-01-08 | 兰州大学 | Construction method of depression auxiliary identification model based on multi-source electroencephalogram data fusion |
CN112190269B (en) * | 2020-12-04 | 2024-03-12 | 兰州大学 | Depression auxiliary identification model construction method based on multisource brain electric data fusion |
CN113283465A (en) * | 2021-04-02 | 2021-08-20 | 电子科技大学 | Diffusion tensor imaging data analysis method and device |
CN113283465B (en) * | 2021-04-02 | 2022-04-29 | 电子科技大学 | Diffusion tensor imaging data analysis method and device |
CN113261973A (en) * | 2021-04-13 | 2021-08-17 | 山东师范大学 | Method and system for identifying atrial fibrillation electrocardiosignals through weighted multi-scale finite-crossing visual image |
CN113261973B (en) * | 2021-04-13 | 2022-09-02 | 山东师范大学 | Method and system for identifying atrial fibrillation electrocardiosignals through weighted multi-scale finite-crossing visual image |
CN113197578A (en) * | 2021-05-07 | 2021-08-03 | 天津医科大学 | Schizophrenia classification method and system based on multi-center model |
CN114334130A (en) * | 2021-12-25 | 2022-04-12 | 浙江大学 | PET molecular image computer-aided diagnosis system based on brain symmetry |
CN114334130B (en) * | 2021-12-25 | 2023-08-22 | 浙江大学 | Brain symmetry-based PET molecular image computer-aided diagnosis system |
CN118335319A (en) * | 2024-06-12 | 2024-07-12 | 四川大学 | Early prediction method for common major diseases based on virtual person simulation |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111627553A (en) | Method for constructing individualized prediction model of first-onset schizophrenia | |
CN113616184B (en) | Brain network modeling and individual prediction method based on multi-mode magnetic resonance image | |
CN109584254B (en) | Heart left ventricle segmentation method based on deep full convolution neural network | |
CN113040715B (en) | Human brain function network classification method based on convolutional neural network | |
He et al. | Automated model design and benchmarking of deep learning models for covid-19 detection with chest ct scans | |
CN111009324B (en) | Auxiliary diagnosis system and method for mild cognitive impairment through multi-feature analysis of brain network | |
CN111090764B (en) | Image classification method and device based on multitask learning and graph convolution neural network | |
CN101517602A (en) | Methods for feature selection using classifier ensemble based genetic algorithms | |
CN106650818B (en) | Resting state function magnetic resonance image data classification method based on high-order hyper-network | |
CN109935321B (en) | Risk prediction system for converting depression patient into bipolar affective disorder based on functional nuclear magnetic resonance image data | |
CN112418337A (en) | Multi-feature fusion data classification method based on brain function hyper-network model | |
CN112489029B (en) | Medical image segmentation method and device based on convolutional neural network | |
CN109255029A (en) | A method of automatic Bug report distribution is enhanced using weighted optimization training set | |
Han et al. | Multi-scale 3D convolution feature-based broad learning system for Alzheimer’s disease diagnosis via MRI images | |
CN113781385A (en) | Joint attention-seeking convolution method for brain medical image automatic classification | |
Misra et al. | PETLFC: Parallel ensemble transfer learning based framework for COVID-19 differentiation and prediction using deep convolutional neural network models | |
Tummala et al. | Few-shot learning using explainable Siamese twin network for the automated classification of blood cells | |
Luo et al. | Multi-view brain network analysis with cross-view missing network generation | |
Liu et al. | An Interpretable CNN for the Segmentation of the Left Ventricle in Cardiac MRI by Real-Time Visualization. | |
CN117765530A (en) | Multi-mode brain network classification method, system, electronic equipment and medium | |
Song et al. | Simultaneous cell detection and classification with an asymmetric deep autoencoder in bone marrow histology images | |
CN116523839A (en) | Parkinson's disease auxiliary analysis system | |
CN115760785A (en) | Brain somatotropin morphology high-order feature extraction method | |
Murugesh et al. | Provisioning a risk predictor model for Alzheimers disease using an improved deep network model | |
US11580390B2 (en) | Data processing apparatus and method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200904 |