CN104361318B - A kind of medical diagnosis on disease accessory system based on diffusion tensor technology - Google Patents

A kind of medical diagnosis on disease accessory system based on diffusion tensor technology Download PDF

Info

Publication number
CN104361318B
CN104361318B CN201410627555.9A CN201410627555A CN104361318B CN 104361318 B CN104361318 B CN 104361318B CN 201410627555 A CN201410627555 A CN 201410627555A CN 104361318 B CN104361318 B CN 104361318B
Authority
CN
China
Prior art keywords
tensor
image
dispersion
projection
disaggregated model
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
CN201410627555.9A
Other languages
Chinese (zh)
Other versions
CN104361318A (en
Inventor
王书强
谈维棋
申妍燕
胡金星
尹凌
曾春霞
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 CN201410627555.9A priority Critical patent/CN104361318B/en
Publication of CN104361318A publication Critical patent/CN104361318A/en
Application granted granted Critical
Publication of CN104361318B publication Critical patent/CN104361318B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/757—Matching configurations of points or features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G06F18/24133—Distances to prototypes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/24765—Rule-based classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/03—Recognition of patterns in medical or anatomical images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/03—Recognition of patterns in medical or anatomical images
    • G06V2201/032—Recognition of patterns in medical or anatomical images of protuberances, polyps nodules, etc.

Abstract

The invention belongs to medical image aided diagnosis technique field, more particularly to a kind of medical diagnosis on disease accessory system and method based on diffusion tensor technology.The medical diagnosis on disease accessory system based on diffusion tensor technology includes image pre-processing module, expertise library module, tensor study module and diagnostic result output module；Described image pretreatment module is used for the collection of dispersion tensor image, dispersion tensor image weight registration and dispersion tensor image characteristics extraction；The expertise library module is used for the expert knowledge library for establishing brain and spinal cord relevant disease；The tensor study module is used to carry out dispersion tensor image the training of tensor disaggregated model using expert knowledge library and tensor learning algorithm and tensor disaggregated model optimizes and test；The diagnostic result output module is used to export diagnostic result.The of the invention original information for fully excavating image, improves pattern classification precision, greatly reduces amount of calculation, and the real-time for effectively having protected medical diagnosis on disease whole.

Description

A kind of medical diagnosis on disease accessory system based on diffusion tensor technology
Technical field
The invention belongs to medical image aided diagnosis technique field, more particularly to it is a kind of based on diffusion tensor technology Medical diagnosis on disease accessory system and method.
Background technology
Diffusion tensor (Diffusion Tensor Imaging, DTI) technology is the white matter of current unique hurtless measure Nerve fibre bundle living imaging method, it is Magnetic resonance imaging (MRI) special shape.It is tracking different from Magnetic resonance imaging Hydrogen atom in hydrone, diffusion tensor are according to the drawing of hydrone moving direction.Diffusion tensor figure can reveal that How brain tumor influences nerve cell connection, and guiding healthcare givers carries out operation on brain.It can also disclose same apoplexy, multiple hard Change the trickle unusual change about brain and spinal cord such as disease, schizophrenia, Dyslexia.Diffusion tensor data are substantially It is second-order tensor structure, its each voxel contains the three dimensions letter of hydrone disperse in white matter nerve fiber bundles Breath.
The effective information for judging human body diseases can be extracted from original dispersion tensor image by machine learning method, from And provide strong help for the forecast analysis brain disease related to spinal cord.But traditional machine learning method is all based on The algorithm of vector pattern, such as SVMs, linear discriminant analysis and neutral net etc., otherwise these algorithms are just only handled Some scalar indexs of diffusion tensor data, and the structure space information of dispersion tensor image can not be made full use of； Before analyzing and processing, tensor is first expanded into vector.However, this way can bring problems with：1st, initial data is destroyed Structure and tensor structured data inherent correlation.2nd, brain diffusion tensor data are after vectorization, its system generated It is very big to count the dimension of parameter (being typically covariance matrix), the structure of initial data can be destroyed, lose the inherent correlation of data, So as to cause high computation complexity and storage cost.
The content of the invention
The invention provides a kind of medical diagnosis on disease accessory system and method based on diffusion tensor technology, it is intended to solves Existing vector pattern learning algorithm can not make full use of the structure space information of dispersion tensor view data, and tensor number During according to vectorization, the structure of initial data can be destroyed, loses the inherent correlation of data, increases computation complexity and deposits Store up the technical problem of cost.
The present invention is achieved in that a kind of medical diagnosis on disease accessory system based on diffusion tensor technology, including figure As pretreatment module, expertise library module, tensor study module and diagnostic result output module；Described image pretreatment module Collection, dispersion tensor image weight registration and dispersion tensor image characteristics extraction for dispersion tensor image；The expert knows Know the expert knowledge library that library module is used to establish brain and spinal cord relevant disease；The tensor study module is used to know using expert Know storehouse and tensor learning algorithm to carry out dispersion tensor image the training of tensor disaggregated model and the optimization of tensor disaggregated model and survey Examination；The diagnostic result output module is used to export diagnostic result.
The technical scheme that the embodiment of the present invention is taken also includes：Described image pretreatment module include image acquisition units, Image registration unit and feature extraction unit；
Described image collecting unit is used to gather dispersion tensor image；
Described image registration unit is used to carry out the dispersion tensor image registration based on tensor similarity, then carries out based on mark Measure the dispersion tensor image registration of similarity；
The feature extraction unit is used for the polyteny core principle component analysis method based on tensor algebra and carries out dispersion tensor The feature extraction of image and dimensionality reduction.
The technical scheme that the embodiment of the present invention is taken also includes：The feature extraction of the dispersion tensor image and dimensionality reduction are specific For：Appropriate kernel function is chosen first, and initial data is mapped to respective feature space；Again by tensor product by all spies Sign is mapped to multilinear subspace so that each subspace can capture the amount of variability of most of orthogonal multidimensional；Handed over according to minimum For a square principle, calculate and solve new feature.
The technical scheme that the embodiment of the present invention is taken also includes：The tensor study module includes tensor disaggregated model training Unit and tensor disaggregated model optimization unit；
The tensor disaggregated model training unit is used to support tensor machine to disperse using expert knowledge library and optimal projection Tensor image carries out tensor disaggregated model training；
The tensor disaggregated model optimization unit is used to carry out the optimization of tensor disaggregated model and test.
The technical scheme that the embodiment of the present invention is taken also includes：The tensor disaggregated model training is specially：Based on optimal Projection supports tensor machine to carry out pattern-recognition to dispersion tensor image, finds optimal projection of the image on boundary direction, and lead to Cross and maximize after projection between the class of sample in matrix and class the ratio between distribution of matrix to calculate optimal projection vector, the calculating Optimal projection vector specifically includes：Matrix S between calculating classbWith matrix S in classwMaximum the max [(V of the ratio between distributionTSbV)/ (VTSwV)], the optimal projection vector v on first boundary direction is solved；Training sample is projected by v, solves quadratic programming Problem, similarly try to achieve the optimal projection vector on second boundary direction；Defined with scatter matrix in scatter matrix between class and class Project coefficient of dispersion R=Tr (Sb×Sb T)/Tr(Sw×Sw T), R represents training sample sample after the projection of optimal projection vector Between separating degree, R values it is bigger represent training sample distance is bigger between sample after projection.
The technical scheme that the embodiment of the present invention is taken also includes：The tensor disaggregated model training also includes：Based on optimal Projection supports tensor machine to analyze two kinds of projection schemes and be：The training sample of each sub-classifier to same direction projection, or The each sub-classifier of person determines different projection vectors, and then training sample is respectively to projecting on respective projection vector direction.
Another technical scheme that the embodiment of the present invention is taken is：A kind of medical diagnosis on disease based on diffusion tensor technology is auxiliary Aid method, including：
Step a：Dispersion tensor image is gathered, and carries out the heavy registering and dispersion tensor characteristics of image of dispersion tensor image and carries Take；
Step b：Establish the expert knowledge library of brain and spinal cord relevant disease；
Step c：Tensor disaggregated model training is carried out to dispersion tensor image using expert knowledge library and tensor learning algorithm And the optimization of tensor disaggregated model and test；
Step d：Export diagnostic result.
The technical scheme that the embodiment of the present invention is taken also includes：In the step a, the dispersion tensor image weight registration Including the dispersion tensor image registration based on tensor similarity and the dispersion tensor image registration based on scalar similarity.
The technical scheme that the embodiment of the present invention is taken also includes：In the step a, the dispersion tensor characteristics of image carries It is taken as：Polyteny core principle component analysis method based on tensor algebra carries out feature extraction and the dimensionality reduction of dispersion tensor image；Tool Body includes：Appropriate kernel function is chosen first, and initial data is mapped to respective feature space, then will be all by tensor product Feature Mapping to multilinear subspace so that each subspace can capture the amount of variability of most of orthogonal multidimensional, according to most Small alternating square principle, calculate and solve new feature.
The technical scheme that the embodiment of the present invention is taken also includes：In the step c, the tensor disaggregated model training tool Body is：Support tensor machine to carry out pattern-recognition to dispersion tensor image based on optimal projection, find image on boundary direction Optimal projection, and pass through and maximize after projection between the class of sample in matrix and class the ratio between distribution of matrix to calculate optimal projection Vector, the optimal projection vector of the calculating specifically include：Matrix S between calculating classbWith matrix S in classwThe maximum of the ratio between distribution max[(VTSbV)/(VTSwV)], the optimal projection vector v on first boundary direction is solved；Training sample is projected by v, Quadratic programming problem is solved, similarly tries to achieve the optimal projection vector on second boundary direction；With between class in scatter matrix and class Scatter matrix definition projection coefficient of dispersion R=Tr (Sb×Sb T)/Tr(Sw×Sw T), R represents training sample and sweared by optimal projection Separating degree after amount projection between sample, distance is bigger between sample after projection for the bigger expression training sample of R values.
The medical diagnosis on disease accessory system based on diffusion tensor technology and method of the embodiment of the present invention are for disperse The characteristics of directional information is more more sensitive than half-tone information in spirogram picture, propose double measurement method for registering；Secondly, propose a kind of multi-thread Property core principle component analysis method realizes feature extraction and the dimensionality reduction of dispersion tensor image；Finally, for vector pattern learning algorithm Limitation in terms of tensor structured data is handled, the structural risk minimization principle of combining classification device and existing tensor pattern Learning algorithm, and optimal projection vector criterion is based on, propose to support tensor machine based on optimal projection, realize and calculated based on tensor pattern The medical diagnosis on disease Real time identification of method and dispersion tensor image, can fully it be excavated on the premise of initial data structure is not destroyed The original information of image, pattern classification precision is improved, amount of calculation can be greatly reduced and calculate cost, and effectively protected whole The real-time of medical diagnosis on disease.The present invention can be used for modification, spiritual demyelinating lesion, central nervous system between astrocytoma The diseases analysis associated with brain and spinal cord such as dysplasia, schizophrenia, depression, cervical spondylotic myelopathy and prediction side Face, while it is contemplated that the present invention may be use with terms of the prevention and health care of the brain and spinal cord relevant disease of normal person.
Brief description of the drawings
Fig. 1 is the structural representation of the medical diagnosis on disease accessory system based on diffusion tensor technology of the embodiment of the present invention Figure；
Fig. 2 is the flow chart of the medical diagnosis on disease householder method based on diffusion tensor technology 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 knot of the medical diagnosis on disease accessory system based on diffusion tensor technology of the embodiment of the present invention Structure schematic diagram.The medical diagnosis on disease accessory system based on diffusion tensor technology of the embodiment of the present invention includes image preprocessing mould Block, expertise library module, tensor study module and diagnostic result output module；Wherein, image pre-processing module is used for disperse The collection of tensor image, the dispersion tensor image based on tensor pattern are again registering and based on polyteny core principle component analysis method Dispersion tensor image characteristics extraction；Expertise library module is used for the expert knowledge library for establishing brain and spinal cord relevant disease； Tensor study module is used to carry out tensor disaggregated model instruction to dispersion tensor image using expert knowledge library and tensor learning algorithm Practice and tensor disaggregated model optimizes and test；Diagnostic result output module is used to export diagnostic result.
Specifically, image pre-processing module includes image acquisition units, image registration unit and feature extraction unit,
Image acquisition units are used to gather dispersion tensor image；
Image registration unit is used to be directed to the characteristics of directional information is more more sensitive than half-tone information in dispersion tensor image, first The dispersion tensor image registration based on tensor similarity is carried out, then carries out the dispersion tensor image registration based on scalar similarity.
Feature extraction unit realizes dispersion tensor image for the polyteny core principle component analysis method based on tensor algebra Feature extraction and dimensionality reduction, be specially：Appropriate kernel function is chosen first, and initial data is mapped to respective feature space, then By tensor product by all Feature Mappings to multilinear subspace so that each subspace can capture most of orthogonal multidimensional Amount of variability, according to minimum alternately square principle, calculate solve new feature.Wherein, principal component analysis (Principal Component Analysis, PCA) method is to select the one of less number significant variable by multiple variables by linear transformation Kind Multielement statistical analysis method, also known as principal component analysis.
Tensor study module includes tensor disaggregated model training unit and tensor disaggregated model optimization unit,
Tensor disaggregated model training unit is used to support tensor machine (OPSTM) to more using expert knowledge library and optimal projection Dissipate tensor image and carry out tensor disaggregated model training；Specially：Using Fisher criterions, tensor machine pair is supported based on optimal projection Dispersion tensor image carries out pattern-recognition, finds optimal projection of the image on boundary direction, and pass through sample after maximization projection The ratio between distribution of matrix calculates optimal projection vector in matrix and class between this class；Algorithm is divided into two steps to realize：(1)、 Matrix S between calculating classbWith matrix S in classwMaximum the max [(V of the ratio between distributionTSbV)/(VTSwV)], first border side is solved Upward optimal projection vector v.(2), training sample is projected by v, quadratic programming problem is solved, can similarly try to achieve second Optimal projection vector on boundary direction.The present invention utilizes scatter matrix definition projection coefficient of dispersion in scatter matrix between class and class R=Tr (Sb×Sb T)/Tr(Sw×Sw T), R represents separating degree of the training sample after the projection of optimal projection vector between sample, R Distance is bigger between sample after projection for the bigger expression training sample of value.Optimal projection, which will be based on, supports tensor machine from two classification It is generalized in classify more, analyzes two kinds of projection schemes：1st, the training sample of each sub-classifier is to same direction projection.2、 Each sub-classifier determines different projection vectors, and then training sample is respectively to projecting on respective projection vector direction.By Projected in training sample in tensor subspace, inhomogeneous training sample can be separated from each other, and similar training sample can phase It is mutually close.The scheme discrimination that this to choose same projection vector is higher than the scheme for choosing non-same projection vector.
Tensor disaggregated model optimization unit is used to carry out the optimization of tensor disaggregated model and test；Specially：With reference to polyteny Composition operation rule, for flow data feature, tensor machine is supported to be generalized to online form optimal projection, and by optimization problem Inequality constraints be converted into equality constraint, for diffusion tensor data (second-order tensor), it is only necessary to solve two lines Property equation group, carries out simple matrix inversion operation, amount of calculation can be greatly reduced.In objective function optimization method, fully examine The characteristics of considering flow data, Lagrange multiplier iteration is completed using stochastic gradient descent method.In order to keep the original of flow data Structural information simultaneously reduces calculating cost, and auxiliary inner product of tensor computing is decomposed using the CP of tensor, and this can keep original tensor Natural structure information while, reduce amount of calculation and simultaneously save memory space.
Referring to Fig. 2, it is the stream of the medical diagnosis on disease householder method based on diffusion tensor technology of the embodiment of the present invention Cheng Tu.The medical diagnosis on disease householder method based on diffusion tensor technology of the embodiment of the present invention comprises the following steps：
Step 100：Gather dispersion tensor image；
Step 200：Registration is carried out to the dispersion tensor image of the collection.Specifically：
For directional information in dispersion tensor image it is more more sensitive than half-tone information the characteristics of, carry out first based on tensor it is similar The dispersion tensor image registration of degree, then carry out the dispersion tensor image registration based on scalar similarity
Step 300：The feature that polyteny core principle component analysis method based on tensor algebra carries out dispersion tensor image carries Take and dimensionality reduction；
In step 300, the feature extraction of dispersion tensor image and dimensionality reduction are specially：Appropriate kernel function handle is chosen first Initial data is mapped to respective feature space, then by tensor product by all Feature Mappings to multilinear subspace, make The amount of variability of most of orthogonal multidimensional can be captured by obtaining each subspace, according to minimum alternating square principle, calculate solution new feature.
Step 400：Establish the expert knowledge library of brain and spinal cord relevant disease；
Step 500：Find optimal projection of the above-mentioned dispersion tensor image on boundary direction, and by after maximizing and projecting The ratio between distribution of matrix calculates optimal projection vector in matrix and class between the class of sample.Specifically：
It is divided into two steps to realize：(1), matrix S between calculating classbWith matrix S in classwThe maximum max of the ratio between distribution [(VTSbV)/(VTSwV)], the optimal projection vector v on first boundary direction is solved.(2), training sample is projected by v, Quadratic programming problem is solved, can similarly try to achieve the optimal projection vector on second boundary direction.
Step 600：Tensor machine is supported to carry out tensor disaggregated model training using expert knowledge library and optimal projection；
In step 600, tensor disaggregated model training is specially：Using Fisher criterions, support to open based on optimal projection Amount machine carries out pattern-recognition to dispersion tensor image.The present invention utilize between class in scatter matrix and class scatter matrix definition projection from Dissipate coefficients R=Tr (Sb×Sb T)/Tr(Sw×Sw T), R represents point of the training sample after the projection of optimal projection vector between sample From degree, distance is bigger between sample after projection for the bigger expression training sample of R values.Will be based on it is optimal projection support tensor machine from Two classification are generalized in more classification, analyze two kinds of projection schemes：1st, the training sample of each sub-classifier is to same direction Projection.2nd, each sub-classifier determines different projection vectors, and then training sample is respectively on respective projection vector direction Projection.Due to training sample is projected in tensor subspace, inhomogeneous training sample can be separated from each other, similar training sample This meeting is close to each other.The scheme discrimination that this to choose same projection vector is higher than the scheme for choosing non-same projection vector.
Step 700：Carry out the optimization of tensor disaggregated model and test；
In step 700, the optimization of tensor disaggregated model and test are specially：With reference to multilinear algebra operation rule, for Flow data feature, support tensor machine to be generalized to online form optimal projection, and the inequality constraints in optimization problem is converted For equality constraint, for diffusion tensor data (second-order tensor), it is only necessary to solve two systems of linear equations, carry out simple Matrix inversion operation, amount of calculation can be greatly reduced.In objective function optimization method, the characteristics of taking into full account flow data, profit Lagrange multiplier iteration is completed with stochastic gradient descent method.In order to keep the prototype structure information of flow data and reduce calculating Cost, auxiliary inner product of tensor computing is decomposed using the CP of tensor, this can keep the same of the natural structure information of original tensor When, reduce amount of calculation and save memory space.
Step 800：Export diagnostic result.
The medical diagnosis on disease accessory system based on diffusion tensor technology and method of the embodiment of the present invention are for disperse The characteristics of directional information is more more sensitive than half-tone information in spirogram picture, propose double measurement method for registering；Secondly, propose a kind of multi-thread Property core principle component analysis method realizes feature extraction and the dimensionality reduction of dispersion tensor image；Finally, for vector pattern learning algorithm Limitation in terms of tensor structured data is handled, the structural risk minimization principle of combining classification device and existing tensor pattern Learning algorithm, and optimal projection vector criterion is based on, propose to support tensor machine based on optimal projection, realize and calculated based on tensor pattern The medical diagnosis on disease Real time identification of method and dispersion tensor image, can fully it be excavated on the premise of initial data structure is not destroyed The original information of image, pattern classification precision is improved, amount of calculation can be greatly reduced and calculate cost, and effectively protected whole The real-time of medical diagnosis on disease.The present invention can be used for modification, spiritual demyelinating lesion, central nervous system between astrocytoma The diseases analysis associated with brain and spinal cord such as dysplasia, schizophrenia, depression, cervical spondylotic myelopathy and prediction side Face, while it is contemplated that the present invention may be use with terms of the prevention and health care of the brain and spinal cord relevant disease of normal person.
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 (4)

  1. A kind of 1. medical diagnosis on disease accessory system based on diffusion tensor technology, it is characterised in that：Including image preprocessing mould Block, expertise library module, tensor study module and diagnostic result output module；Described image pretreatment module is for disperse Collection, dispersion tensor image weight registration and the dispersion tensor image characteristics extraction of spirogram picture；The expertise library module is used In the expert knowledge library for establishing brain and spinal cord relevant disease；The tensor study module is used to utilize expert knowledge library and tensor The training of tensor disaggregated model is carried out to dispersion tensor image for learning algorithm and tensor disaggregated model optimizes and test；The diagnosis As a result output module is used to export diagnostic result；
    Wherein：The tensor study module includes tensor disaggregated model training unit and tensor disaggregated model optimization unit；It is described Tensor disaggregated model training unit is used to support tensor machine to carry out dispersion tensor image using expert knowledge library and optimal projection Tensor disaggregated model training；The tensor disaggregated model optimization unit is used to carry out the optimization of tensor disaggregated model and test；
    The tensor disaggregated model training is specially：Tensor machine is supported to enter row mode to dispersion tensor image and know based on optimal projection Not, optimal projection of the image on boundary direction is found, and passes through and maximizes after projection between the class of sample matrix in matrix and class The ratio between distribution calculate optimal projection vector, it is described to calculate optimal projection vector and specifically include：Matrix S between calculating classb With matrix S in classwMaximum the max [(V of the ratio between distributionTSbV)/(VTSwV)], the optimal projection on first boundary direction is solved Vector v；Training sample is projected by v, quadratic programming problem is solved, similarly tries to achieve the optimal projection on second boundary direction Vector；With scatter matrix definition projection coefficient of dispersion R=Tr (S in scatter matrix between class and classb×Sb T)/Tr(Sw×Sw T), R generations Separating degree of the table training sample after the projection of optimal projection vector between sample, R values are bigger to represent training sample after projection Distance is bigger between sample；
    The tensor disaggregated model optimization includes：With reference to multilinear algebra operation rule, for flow data feature, optimal projection Support tensor machine to be generalized to online form, and the inequality constraints in optimization problem is converted into equality constraint.
  2. 2. the medical diagnosis on disease accessory system according to claim 1 based on diffusion tensor technology, it is characterised in that institute Stating image pre-processing module includes image acquisition units, image registration unit and feature extraction unit；
    Described image collecting unit is used to gather dispersion tensor image；
    Described image registration unit is used to carry out the dispersion tensor image registration based on tensor similarity, then carries out being based on scalar phase Like the dispersion tensor image registration of degree；
    The feature extraction unit is used for the polyteny core principle component analysis method based on tensor algebra and carries out dispersion tensor image Feature extraction and dimensionality reduction.
  3. 3. the medical diagnosis on disease accessory system according to claim 2 based on diffusion tensor technology, it is characterised in that institute The feature extraction and dimensionality reduction for stating dispersion tensor image be specially：Appropriate kernel function is chosen first to be mapped to each initial data Feature space；Again by tensor product by all Feature Mappings to multilinear subspace so that each subspace can capture The amount of variability of most of orthogonal multidimensional；According to minimum, alternately square principle, calculating solve new feature.
  4. 4. the medical diagnosis on disease accessory system according to claim 1 based on diffusion tensor technology, it is characterised in that institute Stating tensor disaggregated model training also includes：Tensor machine is supported to analyze two kinds of projection schemes and be based on optimal projection：Each subclassification The training sample of device is all to same direction projection, or each sub-classifier determines different projection vectors, then trains sample This is respectively to projecting on respective projection vector direction.
CN201410627555.9A 2014-11-10 2014-11-10 A kind of medical diagnosis on disease accessory system based on diffusion tensor technology Active CN104361318B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410627555.9A CN104361318B (en) 2014-11-10 2014-11-10 A kind of medical diagnosis on disease accessory system based on diffusion tensor technology

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410627555.9A CN104361318B (en) 2014-11-10 2014-11-10 A kind of medical diagnosis on disease accessory system based on diffusion tensor technology

Publications (2)

Publication Number Publication Date
CN104361318A CN104361318A (en) 2015-02-18
CN104361318B true CN104361318B (en) 2018-02-06

Family

ID=52528576

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410627555.9A Active CN104361318B (en) 2014-11-10 2014-11-10 A kind of medical diagnosis on disease accessory system based on diffusion tensor technology

Country Status (1)

Country Link
CN (1) CN104361318B (en)

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104715260B (en) * 2015-03-05 2016-06-01 中南大学 Based on the multi-modal fusion image classification method of RLS-ELM
CN104866713B (en) * 2015-05-12 2018-02-13 南京霁云信息科技有限公司 Locally differentiate the Kawasaki disease and fever diagnostic system of subspace insertion based on increment
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
CN105654110A (en) * 2015-12-04 2016-06-08 深圳先进技术研究院 Supervised learning optimization method under tensor mode and system thereof
SG11201609625WA (en) * 2015-12-04 2017-07-28 Shenzhen Inst Of Adv Tech Cas Optimization method and system for supervised learning under tensor mode
TWI639830B (en) * 2017-03-17 2018-11-01 長庚大學 Method for identifying neurological diseases using magnetic resonance imaging images
CN107330267A (en) * 2017-06-28 2017-11-07 首都医科大学宣武医院 Utilize the white matter fiber brain map construction method of diffusion tensor medical image
CN107590806B (en) * 2017-09-19 2021-06-01 陈烨 Detection method and system based on brain medical imaging
CN108198622A (en) * 2018-01-30 2018-06-22 上海蓬海涞讯数据技术有限公司 The system that disease complexity quantitatively evaluating is realized based on computer software
CN108320804A (en) * 2018-01-30 2018-07-24 上海蓬海涞讯数据技术有限公司 The method for realizing disease complexity quantitatively evaluating based on disease fault network complexity evaluation system
CN109978871B (en) * 2019-03-30 2021-06-25 西安电子科技大学 Fiber bundle screening method integrating probability type and determination type fiber bundle tracking
CN110993094B (en) * 2019-11-19 2023-05-23 中国科学院深圳先进技术研究院 Intelligent auxiliary diagnosis method and terminal based on medical image
CN111081372B (en) * 2019-12-14 2023-06-20 中国科学院深圳先进技术研究院 Disease diagnosis device, terminal device, and computer-readable storage medium

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103383716A (en) * 2013-06-26 2013-11-06 北京理工大学 Senile dementia computer-aided diagnosis method based on 18F-FDG PET image
CN103425999A (en) * 2013-08-27 2013-12-04 西安电子科技大学 Brain cognitive state judgment method based on non-negative tensor projection operator decomposition algorithm

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140293069A1 (en) * 2013-04-02 2014-10-02 Microsoft Corporation Real-time image classification and automated image content curation

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103383716A (en) * 2013-06-26 2013-11-06 北京理工大学 Senile dementia computer-aided diagnosis method based on 18F-FDG PET image
CN103425999A (en) * 2013-08-27 2013-12-04 西安电子科技大学 Brain cognitive state judgment method based on non-negative tensor projection operator decomposition algorithm

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于张量学习的目标识别技术研究;牛少波;《中国优秀硕士学位论文全文数据库 信息科技辑》;20101225;第2010年卷(第12期);I136-347 *
磁共振扩散张量图像配准的研究;李文;《中国优秀硕士学位论文全文数据库 信息科技辑》;20100115;第2010年卷(第1期);I138-288 *

Also Published As

Publication number Publication date
CN104361318A (en) 2015-02-18

Similar Documents

Publication Publication Date Title
CN104361318B (en) A kind of medical diagnosis on disease accessory system based on diffusion tensor technology
US9454641B2 (en) Systems and methods for modeling and processing functional magnetic resonance image data using full-brain vector auto-regressive model
US11062450B2 (en) Systems and methods for modeling neural architecture
Gupta et al. Natural image bases to represent neuroimaging data
CN102737250B (en) To the method and system of spinal injury automatic detection in 3D medical images
CN110139597A (en) The system and method for being iterated classification using neuro-physiological signals
Rezaei et al. Deep neural network with l2-norm unit for brain lesions detection
CN110534195B (en) Alzheimer disease detection method based on data space transformation
Mokhtari et al. Decoding brain states using backward edge elimination and graph kernels in fMRI connectivity networks
Kim et al. Riemannian nonlinear mixed effects models: Analyzing longitudinal deformations in neuroimaging
Zhao et al. Connectome-scale group-wise consistent resting-state network analysis in autism spectrum disorder
CN105184794B (en) A kind of CSM Computer Aided Analysis Systems and method based on tensor image
CN104281856A (en) Image preprocessing method and system for brain medical image classification
Wilms et al. Invertible modeling of bidirectional relationships in neuroimaging with normalizing flows: application to brain aging
Jie et al. Sub-network based kernels for brain network classification
Lord et al. Simultaneous registration and parcellation of bilateral hippocampal surface pairs for local asymmetry quantification
Kwaśniewska et al. Real-time facial features detection from low resolution thermal images with deep classification models
CN104408713B (en) A kind of method and system of dispersion tensor image characteristics extraction
CN112561935B (en) Intelligent classification method, device and equipment for brain images
Zhan et al. Multiple stages classification of Alzheimer’s disease based on structural brain networks using generalized low rank approximations (GLRAM)
Kanawade et al. A Deep Learning Approach for Pneumonia Detection from X− ray Images
Lakra et al. A comparative analysis of MRI brain tumor segmentation technique
Kaur et al. A Study of Particle Swarm Optimization based K-means Clustering for Detection of Dental Caries in Dental X-ray Images.
Seo Deep Learning and Medical Applications
Biradar Efficient Brain Tumor Segmentation in MRI Images: A ResNet-based Approach with SegNet Classification

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C53 Correction of patent of invention or patent application
CB03 Change of inventor or designer information

Inventor after: Wang Shuqiang

Inventor after: Tan Weiqi

Inventor after: Shen Yanyan

Inventor after: Hu Jinxing

Inventor after: Yin Ling

Inventor after: Zeng Chunxia

Inventor before: Wang Shuqiang

Inventor before: Hu Yong

Inventor before: Tan Weiqi

Inventor before: Shen Yanyan

Inventor before: Hu Jinxing

Inventor before: Yin Ling

Inventor before: Zeng Chunxia

COR Change of bibliographic data

Free format text: CORRECT: INVENTOR; FROM: WANG SHUQIANG HU YONG TAN WEIQI SHEN YANYAN HU JINXING YIN LING CENG CHUNXIA TO: WANG SHUQIANG TAN WEIQI SHEN YANYAN HU JINXING YIN LING CENG CHUNXIA

GR01 Patent grant
GR01 Patent grant