CN104408713B - A kind of method and system of dispersion tensor image characteristics extraction - Google Patents

A kind of method and system of dispersion tensor image characteristics extraction Download PDF

Info

Publication number
CN104408713B
CN104408713B CN201410627576.0A CN201410627576A CN104408713B CN 104408713 B CN104408713 B CN 104408713B CN 201410627576 A CN201410627576 A CN 201410627576A CN 104408713 B CN104408713 B CN 104408713B
Authority
CN
China
Prior art keywords
tensor
feature
characteristic value
dti
subspace
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.)
Active
Application number
CN201410627576.0A
Other languages
Chinese (zh)
Other versions
CN104408713A (en
Inventor
王书强
谈维棋
胡金星
申妍燕
尹凌
曾春霞
朱英涛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to CN201410627576.0A priority Critical patent/CN104408713B/en
Publication of CN104408713A publication Critical patent/CN104408713A/en
Application granted granted Critical
Publication of CN104408713B publication Critical patent/CN104408713B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G06T2207/10092—Diffusion tensor magnetic resonance imaging [DTI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30016—Brain

Abstract

The invention belongs to technical field of medical image processing, more particularly to a kind of method and system of dispersion tensor image characteristics extraction.The method of the dispersion tensor image characteristics extraction includes：Step a：The tensor characteristic value in DTI image principal directions is calculated, and analyzes the distribution characteristics of the tensor characteristic value of all images；Step b：Weight matrix is constructed according to the distribution characteristics of tensor characteristic value, establishes and optimizes based on the object function of weight matrix to construct new feature space；Step c：Solve object function and obtain polyteny projection matrix, the space of matrices based on DTI tensor characteristic values is mapped to multilinear subspace, and feature extraction and dimensionality reduction are carried out to the DTI images of each subspace.The present invention can effectively reflect Different categories of samples feature, and can, which reduces, calculates cost, on the basis of original image structure information is kept, fully excavates dispersive test of the dispersion tensor image in three principal directions, the purpose of image dimensionality reduction and feature extraction is realized, effectively avoids dimension disaster.

Description

A kind of method and system of dispersion tensor image characteristics extraction
Technical field
The invention belongs to technical field of medical image processing, more particularly to a kind of method of dispersion tensor image characteristics extraction And system.
Background technology
Diffusion tensor (Diffusion Tensor Imaging, DTI), it is a kind of new side for describing brain structure Method, it is Magnetic resonance imaging (MRI) special shape.It is to follow the trail of the hydrogen atom in hydrone different from Magnetic resonance imaging, more Scattered tensor imaging is according to the drawing of hydrone moving direction.Diffusion tensor figure (presentation mode is different from former image) can To disclose how brain tumor influences nerve cell connection, guiding healthcare givers carries out operation on brain, can also disclose same apoplexy, multiple Property the trickle unusual change about brain and spinal cord such as sclerosis, schizophrenia, Dyslexia.
Diffusion tensor data are substantially second-order tensor structures, and its each voxel contains hydrone in white matter The three-dimensional spatial information of disperse in nerve fibre bundle.Traditional learning algorithm Treatment Analysis dispersion tensor figure by vector pattern Some row problems such as initial data structure are destroyed as that can trigger, and it is computationally intensive, it is high to calculate cost.
The content of the invention
The invention provides a kind of method and system of dispersion tensor image characteristics extraction, it is intended to solves existing vectorial mould The learning algorithm Treatment Analysis dispersion tensor image of formula, which exists, destroys initial data structure, and it is computationally intensive, to calculate cost high Technical problem.
The present invention is achieved in that a kind of method of dispersion tensor image characteristics extraction, including：
Step a：The tensor characteristic value in DTI image principal directions is calculated, and analyzes point of the tensor characteristic value of all images Cloth feature；
Step b：Weight matrix is constructed according to the distribution characteristics of tensor characteristic value, establishes and optimizes the mesh based on weight matrix Scalar functions are to construct new feature space；
Step c：Solve object function and obtain polyteny projection matrix, the space of matrices based on DTI tensor characteristic values is reflected Multilinear subspace is mapped to, and feature extraction and dimensionality reduction are carried out to the DTI images of each subspace.
The technical scheme that the embodiment of the present invention is taken also includes：It is described to establish and optimize based on weight in the step b The object function of matrix is specially：The object function based on weight matrix is established using the multiple power series method of development.
The technical scheme that the embodiment of the present invention is taken also includes：In the step c, the solution object function obtains more Linear projection matrix is specially：Alternating least-squares Optimization Solution object function, obtains polyteny projection matrix.
The technical scheme that the embodiment of the present invention is taken also includes：In the step c, it is described will be based on DTI tensors it is intrinsic The space of matrices of value is mapped to multilinear subspace：According to polyteny projection matrix, by tensor product computing, by base Multilinear subspace is mapped in the space of matrices of DTI tensor characteristic values so that each subspace can capture most of orthogonal The amount of variability of multidimensional is in new feature space；The DTI images to each subspace carry out feature extraction and dimensionality reduction is specially： Feature extraction and dimensionality reduction are carried out to the DTI images of each subspace with polyteny core principle component analysis method.
The technical scheme that the embodiment of the present invention is taken also includes：Also include after the step c：Calculate the DTI after projection The tensor characteristic value of image, and the tensor property after dimensionality reduction is determined according to the signal to noise ratio of tensor characteristic value.
Another technical scheme provided in an embodiment of the present invention is：A kind of system of dispersion tensor image characteristics extraction, including Characteristic value processing module, weight matrix constructing module, feature space constructing module, Feature Mapping module and characteristic extracting module, The characteristic value processing module is used to calculate the tensor characteristic value in DTI image principal directions, and analyzes the tensor sheet of all images The distribution characteristics of value indicative；The weight matrix constructing module is used to construct weight matrix according to the distribution characteristics of tensor characteristic value； The feature space constructing module is used to establish and optimize based on the object function of weight matrix to construct new feature space；Institute State Feature Mapping module and obtain polyteny projection matrix for solving object function, the matrix based on DTI tensor characteristic values is empty Between be mapped to multilinear subspace；The characteristic extracting module be used to carrying out the DTI images of each subspace feature extraction and Dimensionality reduction.
The technical scheme that the embodiment of the present invention is taken also includes：The characteristic value processing module includes characteristic value computing unit And characteristic analysis unit；
The characteristic value computing unit is used to calculate tensor characteristic value of the DTI images in three principal directions；
The characteristic analysis unit is used for the tensor design feature according to dispersion tensor view data, analyzes all images The distribution characteristics of tensor characteristic value.
The technical scheme that the embodiment of the present invention is taken also includes：The feature space constructing module is established including object function Unit and feature space structural unit；
The object function establishes unit and is used to establish the target letter based on weight matrix using the multiple power series method of development Number；
The feature space structural unit is used to construct new feature space by objective function optimization.
The technical scheme that the embodiment of the present invention is taken also includes：The Feature Mapping module includes projection matrix computing unit With Feature Mapping unit；
The projection matrix computing unit is used for alternating least-squares Optimization Solution object function, obtains polyteny projection Matrix；
The Feature Mapping unit is used for according to polyteny projection matrix, by tensor product computing, will be based on DTI tensor sheets The space of matrices of value indicative is mapped to multilinear subspace so that each subspace can capture the amount of variability of most of orthogonal multidimensional In new feature space.
The technical scheme that the embodiment of the present invention is taken also includes：The characteristic extracting module includes feature extraction unit and spy Levy computing unit；
The feature extraction unit is used to carry out spy to the DTI images of each subspace with polyteny core principle component analysis method Sign extraction and dimensionality reduction；
The feature calculation unit is used for the tensor characteristic value for calculating the DTI images after projecting, and according to tensor characteristic value Signal to noise ratio determine the tensor property after dimensionality reduction.
The method and system of the dispersion tensor image characteristics extraction of the embodiment of the present invention are existed by calculating and analyzing DTI images Tensor characteristic value in three principal directions is based on weight matrix feature space to construct, and is constructed using the multiple power series method of development The object function of weight matrix, optimal feature space is constructed by optimizing the object function, and will be all by tensor product Feature Mapping to respective multilinear subspace, it is effectively reflected Different categories of samples feature, and can is reduced and is calculated as This；And characteristic vector and depth can be ensured to subspace representation and dimensionality reduction with polyteny core principle component analysis method Correlation, mutual independence, can fully excavate dispersion tensor figure on the basis of original image structure information is kept As the dispersive test in three principal directions, the purpose of image dimensionality reduction and feature extraction is realized, effectively prevent dimension disaster.
Brief description of the drawings
Fig. 1 is the flow chart of the method for the dispersion tensor image characteristics extraction of the embodiment of the present invention；
Fig. 2 is the structural representation of the system of the dispersion tensor image characteristics extraction of the embodiment of the present invention.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and It is not used in the restriction present invention.
Referring to Fig. 1, it is the flow chart of the method for the dispersion tensor image characteristics extraction of the embodiment of the present invention.It is of the invention real The method for applying the dispersion tensor image characteristics extraction of example comprises the following steps：
Step 100：Calculate tensor characteristic value of the DTI images in three principal directions；
Step 200：According to the tensor design feature of dispersion tensor view data, the tensor characteristic value of all images is analyzed Distribution characteristics；
Step 300：Weight matrix is constructed according to the distribution characteristics of tensor characteristic value；
Step 400：The object function based on weight matrix is established using the multiple power series method of development；
Step 500：New feature space is constructed by objective function optimization；
Step 600：Alternating least-squares Optimization Solution object function, obtains polyteny projection matrix；
In step 600, least square method (also known as least squares method) is a kind of mathematical optimization techniques, and it passes through minimum The quadratic sum of error finds the optimal function matching of data；Unknown data can be easily tried to achieve using least square method, and So that the quadratic sum of error is minimum between these data and real data for trying to achieve.Least square method can be additionally used in curve plan Close, some other optimization problem can also be expressed by minimizing energy or maximizing entropy with least square method；The present invention is according to friendship For least square principle, space to a series of space-filling curve subspace of script is decomposed, mining analysis DTI images are in three masters The influence of the disperse level of interaction in direction.
Step 700：It is by tensor product computing, the matrix based on DTI tensor characteristic values is empty according to polyteny projection matrix Between be mapped to multilinear subspace so that the amount of variability that each subspace can capture most of orthogonal multidimensional is empty in new feature Between in；
In step 700, tensor product：In mathematics, every objects multiple in category obtain an object, and meet one The operation for determining binding rule and exchange regulation can be seen as " tensor product ", such as the cartesian product gathered, and simultaneously, topology is empty for no friendship Between product；The present invention by tensor product by all Feature Mappings to respective multilinear subspace, make it effective Reflect Different categories of samples feature, and can, which reduces, calculates cost.
Step 800：Feature is carried out to the DTI images of each subspace with polyteny core principle component analysis method (MKPCA) Extraction and dimensionality reduction；
In step 800, core of the polyteny core principle component analysis method (MKPCA) proposed by the present invention for feature extraction The heart is：A kind of non-linear popularization carried out using target kernel function to the principal component analysis method of classics；With it is traditional it is main into Componential analysis is compared, and MKPCA has and can effectively catch the nonlinear characteristic of data, not have to the distribution situations of luv space data Require, there is more preferable popularity, can also ensure that the correlation of characteristic vector and depth, mutual independence in addition, It is special can fully to excavate disperse of the dispersion tensor image in three principal directions on the basis of original image structure information is kept Sign, realizes the purpose of image dimensionality reduction and feature extraction, and can reduce amount of calculation, avoids dimension disaster.
Step 900：The tensor characteristic value of the DTI images after projection is calculated, and is determined according to the signal to noise ratio of tensor characteristic value Tensor property after dimensionality reduction.
Referring to Fig. 2, it is the structural representation of the system of the dispersion tensor image characteristics extraction of the embodiment of the present invention.This hair The system of the dispersion tensor image characteristics extraction of bright embodiment includes characteristic value processing module, weight matrix constructing module, feature Spatial configuration module, Feature Mapping module and characteristic extracting module；Specifically：
Characteristic value processing module includes characteristic value computing unit and characteristic analysis unit；
Characteristic value computing unit is used to calculate tensor characteristic value of the DTI images in three principal directions；
Characteristic analysis unit is used for the tensor design feature according to dispersion tensor view data, analyzes the tensor of all images The distribution characteristics of characteristic value；
Weight matrix constructing module is used to construct weight matrix according to the distribution characteristics of tensor characteristic value；
Feature space constructing module establishes unit and feature space structural unit including object function；
Object function establishes unit and is used to establish the object function based on weight matrix using the multiple power series method of development；
Feature space structural unit is used to construct new feature space by objective function optimization；
Feature Mapping module includes projection matrix computing unit and Feature Mapping unit；
Projection matrix computing unit is used for alternating least-squares Optimization Solution object function, obtains polyteny projection square Battle array；Wherein, least square method is a kind of mathematical optimization techniques, and it finds the optimal letter of data by minimizing the quadratic sum of error Number matching；Unknown data can be easily tried to achieve using least square method, and cause these data and real data for trying to achieve Between error quadratic sum for minimum；Least square method can be additionally used in curve matching, and some other optimization problem also can be by most Smallization energy or maximization entropy are expressed with least square method；The present invention decomposes the sky of script according to least square principle is replaced Between to a series of space-filling curve subspace, influence of the mining analysis DTI images in the disperse level of interaction of three principal direction.
Feature Mapping unit is used for according to polyteny projection matrix, by tensor product computing, will be based on DTI tensor characteristic values Space of matrices be mapped to multilinear subspace so that each subspace can capture the amount of variability of most of orthogonal multidimensional new Feature space in；Wherein, tensor product：In mathematics, every objects multiple in category obtain an object, and meet certain The operation of binding rule and exchange regulation can be seen as " tensor product ", such as the cartesian product gathered, no to hand over simultaneously, manifold Product；The present invention by tensor product by all Feature Mappings to respective multilinear subspace, make it effectively anti- Different categories of samples feature is reflected, and can, which reduces, calculates cost.
Characteristic extracting module includes feature extraction unit and feature calculation unit；
Feature extraction unit is used to carry out feature to the DTI images of each subspace with polyteny core principle component analysis method Extraction and dimensionality reduction；Wherein, polyteny core principle component analysis method proposed by the present invention is for the core of feature extraction：Utilize A kind of non-linear popularization that target kernel function is carried out to the principal component analysis method of classics；With traditional principal component analysis method phase The nonlinear characteristic of data can effectively be caught, not require have to the distribution situations of luv space data by having than, MKPCA More preferable popularity, the correlation of characteristic vector and depth, mutual independence are can also ensure that in addition, can keep former On the basis of having image structure information, dispersive test of the dispersion tensor image in three principal directions is fully excavated, realizes image Dimensionality reduction and the purpose of feature extraction, and amount of calculation can be reduced, avoid dimension disaster.
Feature calculation unit is used for the tensor characteristic value for calculating the DTI images after projecting, and according to the letter of tensor characteristic value Make an uproar than the tensor property after determination dimensionality reduction.
The method and system of the dispersion tensor image characteristics extraction of the embodiment of the present invention are existed by calculating and analyzing DTI images Tensor characteristic value in three principal directions is based on weight matrix feature space to construct, and is constructed using the multiple power series method of development The object function of weight matrix, optimal feature space is constructed by optimizing the object function, and will be all by tensor product Feature Mapping to respective multilinear subspace, it is effectively reflected Different categories of samples feature, and can is reduced and is calculated as This；And characteristic vector and depth can be ensured to subspace representation and dimensionality reduction with polyteny core principle component analysis method Correlation, mutual independence, can fully excavate dispersion tensor figure on the basis of original image structure information is kept As the dispersive test in three principal directions, the purpose of image dimensionality reduction and feature extraction is realized, effectively prevent dimension disaster.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention All any modification, equivalent and improvement made within refreshing and principle etc., should be included in the scope of the protection.

Claims (6)

1. a kind of method of dispersion tensor image characteristics extraction, including：
Step a：The tensor characteristic value in DTI image principal directions is calculated, and the distribution for analyzing the tensor characteristic value of all images is special Sign；
Step b：Weight matrix is constructed according to the distribution characteristics of tensor characteristic value, establishes and optimizes the target letter based on weight matrix Count to construct new feature space；
Step c：Solve object function and obtain polyteny projection matrix, the space of matrices based on DTI tensor characteristic values is mapped to Multilinear subspace, and feature extraction and dimensionality reduction are carried out to the DTI images of each subspace；
Wherein：It is described to establish and optimize the object function based on weight matrix and be specially in the step b：Using Multiple Power Series Expansion Method establishes the object function based on weight matrix；
In the step c, the solution object function obtains polyteny projection matrix and is specially：Alternating least-squares optimize Object function is solved, obtains polyteny projection matrix；It is described that the space of matrices based on DTI tensor characteristic values is mapped to multiplets Subspace is specially：It is by tensor product computing, the matrix based on DTI tensor characteristic values is empty according to polyteny projection matrix Between be mapped to multilinear subspace so that each subspace can capture the amount of variability of most of orthogonal multidimensional；It is described to each son The DTI images in space carry out feature extraction and dimensionality reduction is specially：DTI with polyteny core principle component analysis method to each subspace Image carries out feature extraction and dimensionality reduction.
2. the method for dispersion tensor image characteristics extraction according to claim 1, it is characterised in that after the step c Also include：The tensor characteristic value of the DTI images after projection is calculated, and after dimensionality reduction is determined according to the signal to noise ratio of tensor characteristic value Measure feature.
A kind of 3. system of dispersion tensor image characteristics extraction, it is characterised in that：Including characteristic value processing module, weight matrix structure Modeling block, feature space constructing module, Feature Mapping module and characteristic extracting module, the characteristic value processing module are used to calculate Tensor characteristic value in DTI image principal directions, and analyze the distribution characteristics of the tensor characteristic value of all images；The weight matrix Constructing module is used to construct weight matrix according to the distribution characteristics of tensor characteristic value；The feature space constructing module is used to establish And optimize based on the object function of weight matrix to construct new feature space；The Feature Mapping module is used to solve target letter Number obtains polyteny projection matrix, and the space of matrices based on DTI tensor characteristic values is mapped into multilinear subspace；The spy Extraction module is levied to be used to carry out feature extraction and dimensionality reduction to the DTI images of each subspace；
Wherein：The feature space constructing module establishes unit including object function, the object function establish unit be used for should The object function based on weight matrix is established with the multiple power series method of development；
The Feature Mapping module includes projection matrix computing unit and Feature Mapping unit；The projection matrix computing unit is used In alternating least-squares Optimization Solution object function, polyteny projection matrix is obtained；The Feature Mapping unit is used for basis Polyteny projection matrix, by tensor product computing, the space of matrices based on DTI tensor characteristic values is mapped to multiplets temper sky Between so that each subspace can capture the amount of variability of most of orthogonal multidimensional in new feature space；
The characteristic extracting module includes feature extraction unit, and the feature extraction unit is used for polyteny core principle component point Analysis method carries out feature extraction and dimensionality reduction to the DTI images of each subspace.
4. the system of dispersion tensor image characteristics extraction according to claim 3, it is characterised in that the characteristic value processing Module includes characteristic value computing unit and characteristic analysis unit；
The characteristic value computing unit is used to calculate tensor characteristic value of the DTI images in three principal directions；
The characteristic analysis unit is used for the tensor design feature according to dispersion tensor view data, analyzes the tensor of all images The distribution characteristics of characteristic value.
5. the system of dispersion tensor image characteristics extraction according to claim 3, it is characterised in that the feature space structure Modeling block also includes feature space structural unit；
The feature space structural unit is used to construct new feature space by objective function optimization.
6. the system of dispersion tensor image characteristics extraction according to claim 3, it is characterised in that the feature extraction mould Block also includes feature calculation unit；
The feature calculation unit is used for the tensor characteristic value for calculating the DTI images after projecting, and according to the letter of tensor characteristic value Make an uproar than the tensor property after determination dimensionality reduction.
CN201410627576.0A 2014-11-10 2014-11-10 A kind of method and system of dispersion tensor image characteristics extraction Active CN104408713B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410627576.0A CN104408713B (en) 2014-11-10 2014-11-10 A kind of method and system of dispersion tensor image characteristics extraction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410627576.0A CN104408713B (en) 2014-11-10 2014-11-10 A kind of method and system of dispersion tensor image characteristics extraction

Publications (2)

Publication Number Publication Date
CN104408713A CN104408713A (en) 2015-03-11
CN104408713B true CN104408713B (en) 2018-01-05

Family

ID=52646342

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410627576.0A Active CN104408713B (en) 2014-11-10 2014-11-10 A kind of method and system of dispersion tensor image characteristics extraction

Country Status (1)

Country Link
CN (1) CN104408713B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105184794B (en) * 2015-09-07 2018-04-17 中国科学院深圳先进技术研究院 A kind of CSM Computer Aided Analysis Systems and method based on tensor image
CN107292329B (en) * 2017-04-19 2020-10-16 西北工业大学 Event idea classification method based on CI-CSP algorithm
CN112183599B (en) * 2020-09-22 2024-02-27 东南大学 UMPCA-based multi-station forging data feature extraction method

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101763638A (en) * 2009-12-14 2010-06-30 西北工业大学 Method for classifying cerebral white matter fiber tracts in diffusion tensor nuclear magnetic resonance image
CN103049901A (en) * 2012-08-03 2013-04-17 上海理工大学 Magnetic resonance diffusion tensor imaging fiber bundle tracking device
CN103440512A (en) * 2013-09-17 2013-12-11 西安电子科技大学 Identifying method of brain cognitive states based on tensor locality preserving projection

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8098068B2 (en) * 2007-07-20 2012-01-17 Medical College Of Georgia Research Institute, Inc. Systems and methods for parametric mapping for correction of noise-based systematic bias of DTI metrics, using a DTI mapping phantom

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101763638A (en) * 2009-12-14 2010-06-30 西北工业大学 Method for classifying cerebral white matter fiber tracts in diffusion tensor nuclear magnetic resonance image
CN103049901A (en) * 2012-08-03 2013-04-17 上海理工大学 Magnetic resonance diffusion tensor imaging fiber bundle tracking device
CN103440512A (en) * 2013-09-17 2013-12-11 西安电子科技大学 Identifying method of brain cognitive states based on tensor locality preserving projection

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
SAR图像的快速核主成分分析识别方法;于红芸 等;《中国图象图形学报》;20120115;第17卷(第1期);摘要、第138段第3段至第140页第1段 *
基于多线性核主成分分析的掌纹识别;郭金玉 等;《光电子 激光》;20110215;第22卷(第2期);全文 *

Also Published As

Publication number Publication date
CN104408713A (en) 2015-03-11

Similar Documents

Publication Publication Date Title
CN104361318B (en) A kind of medical diagnosis on disease accessory system based on diffusion tensor technology
Duong et al. Feature significance for multivariate kernel density estimation
Brechmann et al. Flexible dependence modeling of operational risk losses and its impact on total capital requirements
Duong ks: Kernel density estimation and kernel discriminant analysis for multivariate data in R
CN106295694B (en) A kind of face identification method of iteration weight set of constraints rarefaction representation classification
CN102737250B (en) To the method and system of spinal injury automatic detection in 3D medical images
CN109949296A (en) A kind of transmission line of electricity defect identification method, device and storage medium
CN104657776B (en) Nerve network system, method for analyzing image and device based on nerve network system
CN110033440A (en) Biological cell method of counting based on convolutional neural networks and Fusion Features
Zipunnikov et al. Longitudinal high-dimensional principal components analysis with application to diffusion tensor imaging of multiple sclerosis
EP2434434A2 (en) Method and system for training a landmark detector using multiple instance learning
CN104573742A (en) Medical image classification method and system
CN104408713B (en) A kind of method and system of dispersion tensor image characteristics extraction
CN109657567B (en) Weak supervision characteristic analysis method and system based on 3D fingerprint image
CN108182412A (en) For the method and device of detection image type
CN104281856B (en) For the image pre-processing method and system of brain Medical Images Classification
CN103034778A (en) Method for extracting brain function network of individual based on analysis of multiple tested brain function data
CN109948439A (en) A kind of biopsy method, system and terminal device
CN105184794B (en) A kind of CSM Computer Aided Analysis Systems and method based on tensor image
CN106510736A (en) Psychological state judging method and system based on multidimensional psychological state indexes
CN108701360A (en) Image processing system and method
CN103345741A (en) Non-rigid multimode medical image precise registering method
CN107004259A (en) Statistical weight regularization in many contrast imagings
CN102201038B (en) Method for detecting P53 protein expression in brain tumor
WO2013040076A1 (en) Dynamics and control of state-dependent networks for probing genomic organization

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant