CN105787494B - Cerebral disease identification device based on multistage subregion bag of words - Google Patents
Cerebral disease identification device based on multistage subregion bag of words Download PDFInfo
- Publication number
- CN105787494B CN105787494B CN201610102378.1A CN201610102378A CN105787494B CN 105787494 B CN105787494 B CN 105787494B CN 201610102378 A CN201610102378 A CN 201610102378A CN 105787494 B CN105787494 B CN 105787494B
- Authority
- CN
- China
- Prior art keywords
- words
- bag
- subregion
- brain
- multistage
- 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
Links
- 208000018152 Cerebral disease Diseases 0.000 title claims abstract description 24
- 210000004556 brain Anatomy 0.000 claims abstract description 70
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 238000000034 method Methods 0.000 claims abstract description 22
- 239000012141 concentrate Substances 0.000 claims abstract description 10
- 239000000284 extract Substances 0.000 claims abstract description 9
- 208000014644 Brain disease Diseases 0.000 claims abstract description 6
- 210000003625 skull Anatomy 0.000 claims description 6
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 abstract description 5
- 238000003745 diagnosis Methods 0.000 abstract description 4
- 201000010099 disease Diseases 0.000 abstract description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 abstract description 2
- 239000013598 vector Substances 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000011282 treatment Methods 0.000 description 2
- 208000024827 Alzheimer disease Diseases 0.000 description 1
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000002595 magnetic resonance imaging Methods 0.000 description 1
- 230000002969 morbid Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000001755 vocal effect Effects 0.000 description 1
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
- G06V10/20—Image preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
- G06V10/464—Salient features, e.g. scale invariant feature transforms [SIFT] using a plurality of salient features, e.g. bag-of-words [BoW] representations
-
- 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
Abstract
The invention discloses a kind of cerebral disease identification devices based on multistage subregion bag of words.Wherein, which includes concentrating the brain magnetic resonance configurations image of various kinds sheet to pre-process training sample set and test sample；Then it concentrates various kinds sheet and test sample to concentrate the twodimensional magnetic resonance structural images of various kinds sheet based on standard brain template, training sample, extracts its structure feature respectively；Multistage brain subregion is carried out to standard brain template again；It is directed to the structure feature of standard form again, tectonic reverse bag of words are distinguished using unsupervised clustering in each subregion；Then, using each grade subregion bag of words, each grade bag of words histogram of each sample is constructed；Finally, establishing multistage classifier using each grade bag of words histogram and classifying to test sample, and then realize brain diseases identification.The embodiment of the present invention carries out disease identification and individual attribute determines by the brain magnetic resonance configurations image classification method based on multistage subregion bag of words, auxiliary cerebral disease clinical analysis diagnosis.
Description
Technical field
The present embodiments relate to technical field of medical image processing, more particularly, to one kind based on multistage subregion bag of words mould
The cerebral disease identification device of type.
Background technique
A kind of Brian Imaging method of the magnetic resonance imaging as non-intrusion type, has been widely used in the research of cerebral disease.
The Accurate Diagnosis of cerebral disease plays positive effect to disease treatment.Using magnetic resonance configurations image, objective evaluation morbid state,
Assisted medical diagnosis becomes the hot spot of current research.
Magnetic resonance configurations image is indicated using image local structural feature and is attracted attention.Bag of words are as a kind of
Image representing method based on local feature, basic thought are by the probability distribution of image local feature (vision word)
(histogram) indicates.Bag of words are widely used in target classification and field of target recognition, are gradually available for magnetic resonance in recent years
Structural images analysis.Daliri is in document (Daliri MR.Automated diagnosis of Alzheimer disease
using the scale-invariant feature transforms in magnetic resonance images[J]
.Journal of medical systems, 2012,36 (2): 995-1000.) in from magnetic resonance image extract SIFT feature,
Propose a kind of magnetic resonance configurations image classification algorithms based on bag of words and support vector machines.In addition, other methods will
Grey density values, Laguerre Circular Harmonic Functions etc. are used as local feature, establish bag of words.On
Predicate bag model is mainly based upon full brain or fixed area is implemented to complete, and location information is not fully utilized.
Summary of the invention
In view of the above problems, it proposes on the present invention overcomes the above problem or at least be partially solved in order to provide one kind
State a kind of cerebral disease identification device based on multistage subregion bag of words of problem.
To achieve the goals above, the present invention provides following technical schemes:
A kind of cerebral disease identification device based on multistage subregion bag of words, described device include at least:
Preprocessing module is configured as concentrating training sample set and test sample the brain magnetic resonance configurations of various kinds sheet
Image is pre-processed, and the pretreatment includes: removal skull and background and the brain magnetic resonance configurations image is affine
Transform to standard brain template；
Structure feature extraction module is configured as concentrating various kinds sheet based on the standard brain template, the training sample
Various kinds sheet two-dimentional brain magnetic resonance configurations each of on cross section, sagittal plane and coronal-plane direction are concentrated with the test sample
Image extracts its structure feature respectively；
Multistage brain subregion constructs module, is configured as establishing multistage brain subregion for the standard brain template；
Bag of words construct module, are configured as the structure feature for the standard brain template, use Unsupervised clustering side
Method respectively clusters structure feature in the subregions at different levels, constructs each grade subregion bag of words；
Bag of words histogram constructs module, is configured as concentrating various kinds for the training sample set and the test sample
This two-dimentional brain magnetic resonance configurations feature on the cross section, the sagittal plane and the coronal-plane, using described each etc.
Grade subregion bag of words, construct each grade bag of words histogram of each sample；
Cerebral disease identification module is configured as concentrating each grade bag of words histogram of various kinds sheet using the training sample
Figure establishes multistage classifier, and classifies to the test sample, and then realize the identification of brain diseases.
Compared with prior art, above-mentioned technical proposal at least has the advantages that
The embodiment of the present invention establishes multistage brain subregion using standard brain template；For the structure feature of standard brain template, make
Feature in partitioned organizations at different levels is clustered respectively with unsupervised clustering, to construct each grade subregion bag of words；Then needle
Two-dimentional brain magnetic resonance configurations of the various kinds sheet on cross section, sagittal plane and coronal-plane are concentrated to training sample set and test sample
Feature, using each grade subregion bag of words, to construct each grade bag of words histogram of each sample；It is finally concentrated using training sample each
Each grade bag of words histogram of sample establishes multistage classifier and classifies to test sample, and then realizes the identification of brain diseases.
Certainly, it implements any of the products of the present invention and is not necessarily required to realize all the above advantage simultaneously.
Above description is only the general introduction of technical solution of the present invention, in order to better understand technology hand of the invention
Section, can be implemented in accordance with the contents of the specification, and in order to allow above and other objects of the present invention, feature and advantage energy
Enough more obvious and easy to understand, the followings are specific embodiments of the present invention.
Detailed description of the invention
Attached drawing is as a part of the invention, and for providing further understanding of the invention, of the invention is schematic
Examples and descriptions thereof are used to explain the present invention, but does not constitute an undue limitation on the present invention.Obviously, the accompanying drawings in the following description
Only some embodiments to those skilled in the art without creative efforts, can be with
Other accompanying drawings can also be obtained according to these attached drawings.In the accompanying drawings:
Fig. 1 is the stream according to the cerebral disease recognition methods based on multistage subregion bag of words shown in an exemplary embodiment
Journey schematic diagram；
Fig. 2 is the multistage brain subregion schematic diagram shown according to an exemplary embodiment；
Fig. 3 is according to the schematic diagram for constructing grade bag of words histogram using subregion bag of words shown in an exemplary embodiment.
These attached drawings and verbal description are not intended to the conception range limiting the invention in any way, but by reference to
Specific embodiment is that those skilled in the art illustrate idea of the invention.
Specific embodiment
The technical issues of with reference to the accompanying drawing and specific embodiment is solved to the embodiment of the present invention, used technical side
Case and the technical effect of realization carry out clear, complete description.Obviously, described embodiment is only one of the application
Divide embodiment, is not whole embodiments.Based on the embodiment in the application, those of ordinary skill in the art are not paying creation
Property labour under the premise of, all other equivalent or obvious variant the embodiment obtained is fallen within the scope of protection of the present invention.
The embodiment of the present invention can be embodied according to the multitude of different ways being defined and covered by claim.
It should be noted that in the following description, understanding for convenience, giving many details.But it is very bright
Aobvious, realization of the invention can be without these details.
It should be noted that in the absence of clear limitations or conflicts, each embodiment in the present invention and its
In technical characteristic can be combined with each other and form technical solution.
The embodiment of the present invention provides a kind of cerebral disease recognition methods based on multistage subregion bag of words, carries out cerebral disease knowledge
Not.As shown in Figure 1, this method at least may include step S1 to step S6.
S1: the brain magnetic resonance configurations image of various kinds sheet is concentrated to pre-process training sample set and test sample, in advance
Processing includes: removal skull and background and by brain magnetic resonance configurations image affine transformation to standard brain template.
In this step, training sample set includes patient's sample collection and healthy sample set.Test sample concentrate sample be
Patient's sample or healthy sample.
Pretreatment specifically includes:
Step S1.1: removal skull and background.
Wherein it is possible to utilize the skull and back of the softwares such as FreeSurfer removal original two dimensional brain magnetic resonance configurations image
Scape.
Step S1.2: the brain magnetic resonance configurations image affine transformation of skull and background will be eliminated to standard brain template.
Wherein it is possible to using softwares such as FSL by brain magnetic resonance configurations image affine transformation to standard brain template (i.e. standard
Between Naokong).
The embodiment of the present invention obtains standard brain template, training sample from three cross section, sagittal plane and coronal-plane directions respectively
This collection and test sample concentrate the two-dimensional sequence brain image of various kinds sheet.Two-dimensional sequence brain image herein refers to include unidirectional more
Zhang Erwei brain image covers full brain.
S2: various kinds sheet and test sample is concentrated to concentrate various kinds sheet in cross section, sagittal based on standard brain template, training sample
Two-dimentional brain magnetic resonance configurations image, extracts its structure feature respectively each of on face and coronal-plane direction.
Illustrate the process for extracting structure feature for extracting SIFT feature below.
Although it will be apparent to one skilled in the art that the embodiment of the present invention be using SIFT feature as structure feature for come into
Row explanation, but structure feature referred to herein includes but is not limited to SIFT feature.
SIFT feature extraction is broadly divided into four steps: detection key point first, in difference of Gaussian Pyramid technology structure
The middle extreme point for extracting image is as candidate key point；Then key point is accurately positioned, and key point is screened,
Remove low contrast or in borderline key point；Then principal direction is specified using the gradient information of key point surrounding pixel；
Description of 128 dimensions is finally generated for each key point.Wherein, each SIFT feature corresponds to its location information.
S3: being directed to standard brain template, establishes multistage brain subregion.
Fig. 2 schematically illustrates multistage brain subregion schematic diagram.
In this step, multistage brain subregion can be used any one following mode and carry out subregion: fixed dimension subregion, base
In the brain subregion and sliding window subregion of function.
This step is described in detail by taking ICBM-152 standard brain template as an example below.
ICBM-152 standard brain template is carried out multistage subregion by the embodiment of the present invention in the following manner.
Level 1: full brain is analyzed as a whole.
2: two subregion of Level, i.e., according to the central plane in cross section, sagittal plane or coronal-plane direction to standard brain template
It is divided.According to the central plane of different directions, Level 2 can be subdivided into Level 2-1, Level 2-2 and Level again
2-3。
3: four subregion of Level, i.e., according to the two of them central plane in three cross section, sagittal plane or coronal-plane directions
Standard brain template is divided.It is combined according to different median planes, Level 3 can be subdivided into Level 3-1, Level 3-2 again
With Level 3-3.
Level 4: octant, i.e., according to cross section, three directions of sagittal plane and coronal-plane central plane to standard brain
Template is divided.
S4: special to structure in subregions at different levels respectively using unsupervised clustering for the structure feature of standard brain template
Sign is clustered, and each grade subregion bag of words are constructed.
In one embodiment of the invention, sorting procedure may include the location information according to structure feature, by structure
Feature incorporates into subregions at different levels, and is clustered respectively in subregions at different levels.
In a preferred embodiment of the invention, this step may include the structure feature for subregions at different levels, use
Unsupervised clustering extracts cluster centre, subregion bag of words is added using cluster centre as basic vocabulary, and to different grades of
Each subregion constructs subregion bag of words respectively.
As an example, constructing each grade subregion bag of words and may include:
For the SIFT feature that standard brain template extraction arrives, incorporate into according to its position in different brain subregions, by each brain point
Using AP (Affinity Propagation), algorithm is clustered the SIFT feature in area, and extracts cluster centre.It extracts
Subregion bag of words will be added as basic vocabulary in cluster centre.Then, subregion bag of words are constructed respectively to different grades of each subregion.
For example, eight subregions in Level 4 respectively cluster the feature for falling into each subregion, 8 subregion bag of words finally can be obtained.
S5: two dimension of the various kinds sheet on cross section, sagittal plane and coronal-plane is concentrated for training sample set and test sample
Brain magnetic resonance configurations feature constructs each grade bag of words histogram of each sample using each grade subregion bag of words.
In some optional implementations of the embodiment of the present invention, this step be can specifically include for training sample set
In two-dimentional brain magnetic resonance configurations feature of each sample on cross section, sagittal plane and coronal-plane, according to structure feature position believe
Breath is replaced using basic vocabulary most like therewith in corresponding subregion bag of words, constructs subregion bag of words histogram, and then constructed each etc.
Grade bag of words histogram.
Wherein, constructing each grade bag of words histogram can further include: by the bag of words histogram of ad eundem different subregions
Figure is combined in order, obtains the grade bag of words histogram of the sample.
Fig. 3 is eight subregion bag of words using Level 4, constructs grade bag of words histogram schematic diagram.
Illustratively, it can specifically include using subregion bag of words building grade bag of words histogram:
Step 5.1: finding and replace word
Two-dimentional brain of the various kinds sheet on cross section, sagittal plane and coronal-plane is concentrated for training sample set and test sample
Magnetic resonance configurations feature finds the correspondence brain subregion in subregions at different levels according to the location information of structure feature, uses corresponding point
Basic vocabulary most like therewith replaces in area's bag of words.
Preferably, some embodiments of the present invention measure the similitude using Euclidean distance, i.e., distance gets over modern age
Table feature is more similar.
Preferably, some embodiments of the present invention are corresponded in subregion bag of words in each grade using the method that arest neighbors is searched and are looked into
Look for the basic vocabulary nearest apart from this feature vector.
Step 5.2: building subregion bag of words histogram
Two-dimentional brain of the various kinds sheet on cross section, sagittal plane and coronal-plane is concentrated for training sample set and test sample
Magnetic resonance configurations feature corresponds in each grade according to its position and finds substitution word in subregion bag of words, and calculates basic vocabulary and go out
The existing frequency.That is: the structure feature (preferably SIFT feature) that each sample extraction arrives, can be with different etc. according to its location information
The different basicvocabularies of the corresponding subregion bag of words of grade replace.
Illustratively, a SIFT feature is located at the second subregion of Level 3-1, can be by one of two subregion of grade
Basicvocabulary replaces.Further, since the SIFT feature is located at the first subregion of Level 2-1, therefore the SIFT feature can also be with
It is replaced by a basicvocabulary of one subregion of grade.
Step 5.3: building grade bag of words histogram
Specifically, various kinds sheet is concentrated to training sample set and test sample, obtains its each grade subregion bag of words histogram respectively
Figure, the bag of words histogram of ad eundem different subregions is combined in order, obtains the grade bag of words histogram of the sample.
For example, 8 subregions that Fig. 3 gives Level 4 divide, subregion bag of words histogram, subregion bag of words are calculated in each subregion
Histogram combines to obtain grade bag of words histogram in order.Again using grade bag of words histogram as the word of image under the grade classification
Bag indicates vector.
Those skilled in the art will be understood that the mode of each grade bag of words histogram of each sample of above-mentioned building is only to lift
Example, if the mode of each grade bag of words histogram of each sample of building that is any other existing or being likely to occur from now on can be suitably used for
The present invention then should also be included within protection scope of the present invention and be herein incorporated by reference herein.
S6: each grade bag of words histogram of various kinds sheet is concentrated using training sample, establishes multistage classifier, and to test sample
Classification, and then realize the identification of brain diseases.
Established in this step multistage classifier can specifically include using training sample concentrate each grade bag of words of various kinds sheet it is straight
Side's figure constructs each grade classifier respectively.Each grade classifier building multistage classifier is comprehensively utilized, to carry out test specimens one's duty
Class.
Illustratively, test sample classification can specifically include:
Step 6.1: training classifier
In embodiments of the present invention, the grade bag of words histogram vectors based on training sample, set according to the affiliated group of sample
Calibration number (sample belongs to health group and corresponds to marked as 1, and it is corresponding marked as -1 to belong to patient's group), is respectively trained one in each grade
Hold vector machine disaggregated model.Such as supporting vector can be respectively trained for 8 kinds of partitioned modes shown in Fig. 2 in the embodiment of the present invention
Classifier.
Step 6.2: sample classification
Based on each supporting vector machine model that the training of each grade subregion obtains, according to each grade classifier as a result, dividing
Class new samples, and then realize the identification of brain diseases.The embodiment of the present invention is according to Level 2-1, Level 2-2, Level 2-
3, Level 3-1, Level 3-2,4 seven grades of classifiers of Level 3-3 and Level as a result, using most Voting principles classification
New samples.Using the classification results of Level 1 as control.
Cerebral disease recognition methods based on multistage subregion bag of words described in the embodiment of the present invention utilizes public data collection
OASIS and PPMI are assessed.The experiment results show that compared to the bag of words for not doing multidomain treat-ment, the embodiment of the present invention
The cerebral disease recognition methods based on multistage subregion bag of words achieves better classifying quality.
Although each step is described in the way of above-mentioned precedence in the present embodiment, this field skill
Art personnel are appreciated that the effect in order to realize the present embodiment, execute between different steps not necessarily in such order,
It (parallel) simultaneously can execute or be executed with reverse order, these simple variations are all within protection scope of the present invention.
Technical solution is provided for the embodiments of the invention above to be described in detail.Although applying herein specific
A example the principle of the present invention and embodiment are expounded, still, the explanation of above-described embodiment be only applicable to help manage
Solve the principle of the embodiment of the present invention；Meanwhile to those skilled in the art, according to an embodiment of the present invention, it is being embodied
It can be made a change within mode and application range.
It, can be with it should be noted that the flow chart being referred to herein is not limited solely to form shown in this article
It is divided and/or is combined.
It should be understood that the label and text in attached drawing are intended merely to be illustrated more clearly that the present invention, it is not intended as to this
The improper restriction of invention protection scope.
The terms "include", "comprise" or any other like term are intended to cover non-exclusive inclusion, so that
Process, method, article or equipment/device including a series of elements not only includes those elements, but also including not bright
The other elements really listed, or further include the intrinsic element of these process, method, article or equipment/devices.
Each step of the invention can be realized with general computing device, for example, they can concentrate on it is single
On computing device, such as: personal computer, server computer, handheld device or portable device, laptop device or more
Processor device can also be distributed over a network of multiple computing devices, they can be to be different from sequence herein
Shown or described step is executed, perhaps they are fabricated to each integrated circuit modules or will be more in them
A module or step are fabricated to single integrated circuit module to realize.Therefore, the present invention is not limited to any specific hardware and soft
Part or its combination.
" some embodiments ", " embodiment " described herein means that: the technical characteristic that describes in conjunction with the embodiments, structure
Or characteristic is included at least one embodiment of the present invention.
In the above embodiment of the invention, it all emphasizes particularly on different fields to the description of each embodiment, does not have in some embodiment
The part of detailed description, reference can be made to the related descriptions of other embodiments.
Although a large amount of detail is described herein.However, you should be able to understand, the embodiment of the present invention can not have
Implement in the case where there are these details.In some embodiments, well-known methods, structures and techniques have not been shown in detail,
So as not to obscure the understanding of this specification.
Present invention is not limited to the embodiments described above, and without departing substantially from substantive content of the present invention, this field is common
Any deformation, improvement or the replacement that technical staff is contemplated that each fall within protection scope of the present invention.
Claims (7)
1. a kind of cerebral disease identification device based on multistage subregion bag of words, which is characterized in that described device includes at least:
Preprocessing module is configured as concentrating training sample set and test sample the brain magnetic resonance configurations image of various kinds sheet
It is pre-processed, the pretreatment includes: removal skull and background and by the brain magnetic resonance configurations image affine transformation
To standard brain template；
Structure feature extraction module is configured as concentrating various kinds sheet and institute based on the standard brain template, the training sample
It states test sample and concentrates various kinds sheet two-dimentional brain magnetic resonance configurations figure each of on cross section, sagittal plane and coronal-plane direction
Picture extracts its structure feature respectively；
Multistage brain subregion constructs module, is configured as establishing multistage brain subregion for the standard brain template；
Bag of words construct module, are configured as the structure feature for the standard brain template, use unsupervised clustering point
It is other that structure feature in the subregions at different levels is clustered, construct each grade subregion bag of words；Wherein, described to the subregions at different levels
Interior structure feature is clustered, and the location information according to structure feature in the subregions at different levels is specifically included, and the structure is special
Sign incorporates into the subregions at different levels, and is clustered respectively in the subregions at different levels；
Bag of words histogram constructs module, is configured as concentrating various kinds sheet to exist for the training sample set and the test sample
Two-dimentional brain magnetic resonance configurations feature on the cross section, the sagittal plane and the coronal-plane, utilizes each ranking score
Area's bag of words construct each grade bag of words histogram of each sample；
Cerebral disease identification module is configured as being concentrated each grade bag of words histogram of various kinds sheet using the training sample, be built
Vertical multistage classifier, and classify to the test sample, and then realize the identification of brain diseases.
2. the cerebral disease identification device according to claim 1 based on multistage subregion bag of words, which is characterized in that described
Structure feature includes but is not limited to SIFT feature.
3. the cerebral disease identification device according to claim 1 based on multistage subregion bag of words, which is characterized in that described
Multistage brain subregion building module is further configured to using fixed dimension subregion, the brain subregion based on function or sliding window point
Area establishes the multistage brain subregion.
4. the cerebral disease identification device according to claim 1 based on multistage subregion bag of words, which is characterized in that
The bag of words building module is further configured to the structure feature for the subregions at different levels, using described unsupervised poly-
Class method extracts cluster centre, subregion bag of words is added using the cluster centre as basic vocabulary, and to different grades of each point
Area constructs subregion bag of words respectively.
5. the cerebral disease identification device according to claim 1 based on multistage subregion bag of words, which is characterized in that
The bag of words histogram building module is further configured to concentrate various kinds sheet described cross-section for the training sample
The two-dimentional brain magnetic resonance configurations feature on face, the sagittal plane and the coronal-plane, according to the structure feature position
Information is replaced using basic vocabulary most like therewith in corresponding subregion bag of words, constructs subregion bag of words histogram, and then construct instruction
Practice sample set and test sample concentrates each grade bag of words histogram of various kinds sheet.
6. the cerebral disease identification device according to claim 5 based on multistage subregion bag of words, which is characterized in that
The bag of words histogram building module includes grade bag of words histogram building submodule, the grade bag of words histogram building
Submodule is configured as in order combining the bag of words histogram of ad eundem different subregions, obtains the grade bag of words histogram of the sample
Figure.
7. the cerebral disease identification device according to claim 1 based on multistage subregion bag of words, which is characterized in that
The cerebral disease identification module includes classifier building module, and the classifier building module is constructed including the first classifier
Submodule and the second classifier construct submodule；
The first classifier building submodule is configured as concentrating each grade bag of words of various kinds sheet straight using the training sample
Side's figure constructs each grade classifier respectively；
The second classifier building submodule is configured as comprehensively utilizing each grade classifier building multistage classifier, with
Carry out test sample classification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610102378.1A CN105787494B (en) | 2016-02-24 | 2016-02-24 | Cerebral disease identification device based on multistage subregion bag of words |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610102378.1A CN105787494B (en) | 2016-02-24 | 2016-02-24 | Cerebral disease identification device based on multistage subregion bag of words |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105787494A CN105787494A (en) | 2016-07-20 |
| CN105787494B true CN105787494B (en) | 2019-04-23 |
Family
ID=56402734
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610102378.1A Active CN105787494B (en) | 2016-02-24 | 2016-02-24 | Cerebral disease identification device based on multistage subregion bag of words |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105787494B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108596228B (en) * | 2018-04-13 | 2022-08-23 | 江南大学 | Brain function magnetic resonance image classification method based on unsupervised fuzzy system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663446A (en) * | 2012-04-24 | 2012-09-12 | 南方医科大学 | Building method of bag-of-word model of medical focus image |
| EP2570970A1 (en) * | 2011-09-16 | 2013-03-20 | Technische Universität Berlin | Method and system for the automatic analysis of an image of a biological sample |
| CN104915561A (en) * | 2015-06-11 | 2015-09-16 | 万达信息股份有限公司 | Intelligent disease attribute matching method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9076196B2 (en) * | 2010-05-13 | 2015-07-07 | Loma Linda University Medical Center | Iron quantification of brain microbleeds |
-
2016
- 2016-02-24 CN CN201610102378.1A patent/CN105787494B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2570970A1 (en) * | 2011-09-16 | 2013-03-20 | Technische Universität Berlin | Method and system for the automatic analysis of an image of a biological sample |
| CN102663446A (en) * | 2012-04-24 | 2012-09-12 | 南方医科大学 | Building method of bag-of-word model of medical focus image |
| CN104915561A (en) * | 2015-06-11 | 2015-09-16 | 万达信息股份有限公司 | Intelligent disease attribute matching method |
Non-Patent Citations (1)
| Title |
|---|
| 基于词袋模型的MR图像脑肿瘤分割方法的研究;赵建奇;《中国优秀硕士学位论文全文数据库医药卫生科技辑》;20150115;正文第19-48页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105787494A (en) | 2016-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106570505B (en) | Method and system for analyzing histopathological images | |
| Laosai et al. | Classification of acute leukemia using medical-knowledge-based morphology and CD marker | |
| CN106340021B (en) | Blood vessel extraction method | |
| Al-Kharraz et al. | Automated system for chromosome karyotyping to recognize the most common numerical abnormalities using deep learning | |
| Nguyen et al. | 3D-brain segmentation using deep neural network and Gaussian mixture model | |
| CN109086809A (en) | Based on the improved early stage Alzheimer disease classification method supervised and be locally linear embedding into | |
| Wang et al. | Morphological analysis of dendrites and spines by hybridization of ridge detection with twin support vector machine | |
| Nazir et al. | Ecsu-net: an embedded clustering sliced u-net coupled with fusing strategy for efficient intervertebral disc segmentation and classification | |
| Mao et al. | A deep convolutional neural network trained on representative samples for circulating tumor cell detection | |
| CN106250701B (en) | The system and method for Lung neoplasm degree of danger classification are realized in computer software | |
| CN110264454A (en) | Cervical cancer tissues pathological image diagnostic method based on more hidden layer condition random fields | |
| Reddy et al. | Brain and pancreatic tumor classification based on GLCM—k-NN approaches | |
| Mohapatra et al. | Unsupervised leukocyte image segmentation using rough fuzzy clustering | |
| Sahli et al. | Supervised classification approach of biometric measures for automatic fetal defect screening in head ultrasound images | |
| CN105787494B (en) | Cerebral disease identification device based on multistage subregion bag of words | |
| CN113936172A (en) | Disease classification method and device based on ensemble learning and multi-mode feature fusion | |
| Yang et al. | Classification of human epithelial type 2 cell images using independent component analysis | |
| CN105528791B (en) | A kind of quality evaluation device and its evaluation method towards touch screen hand-drawing image | |
| CN107194319A (en) | The mitotic mapping sorted based on SVMs and knowledge method for distinguishing | |
| Liu et al. | Refined segmentation R-CNN: a two-stage convolutional neural network for punctate white matter lesion segmentation in preterm infants | |
| He et al. | Automated diagnosis of neural foraminal stenosis using synchronized superpixels representation | |
| Kalaiselvi et al. | Fast brain abnormality detection method for magnetic resonance images (mri) of human head scans using k-means clustering technique | |
| Suresha et al. | Relieff feature selection based Alzheimer disease classification using hybrid features and support vector machine in magnetic resonance imaging | |
| Krawczyk et al. | One-class classification decomposition for imbalanced classification of breast cancer malignancy data | |
| Pan et al. | Leukocyte image segmentation using novel saliency detection based on positive feedback of visual perception |
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 |