CN104933712A - Graph model displaying method based on cerebral CT image - Google Patents
Graph model displaying method based on cerebral CT image Download PDFInfo
- Publication number
- CN104933712A CN104933712A CN201510324245.4A CN201510324245A CN104933712A CN 104933712 A CN104933712 A CN 104933712A CN 201510324245 A CN201510324245 A CN 201510324245A CN 104933712 A CN104933712 A CN 104933712A
- Authority
- CN
- China
- Prior art keywords
- roi
- vertex
- region
- image
- brain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000002490 cerebral effect Effects 0.000 title abstract 4
- 210000004556 brain Anatomy 0.000 claims abstract description 39
- 230000003902 lesion Effects 0.000 claims abstract description 18
- 230000001747 exhibiting effect Effects 0.000 claims description 10
- 230000011218 segmentation Effects 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 4
- 210000003140 lateral ventricle Anatomy 0.000 abstract 2
- 238000007917 intracranial administration Methods 0.000 abstract 1
- 230000007170 pathology Effects 0.000 abstract 1
- 238000007781 pre-processing Methods 0.000 abstract 1
- 230000000007 visual effect Effects 0.000 abstract 1
- 238000013459 approach Methods 0.000 description 2
- 239000003814 drug Substances 0.000 description 2
- 238000010191 image analysis Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000036285 pathological change Effects 0.000 description 1
- 231100000915 pathological change Toxicity 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 230000000505 pernicious effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 210000000115 thoracic cavity Anatomy 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/168—Segmentation; Edge detection involving transform domain methods
-
- 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
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention belongs to the technical field of medical treatment information, and in particular relates to a graph model displaying method based on a cerebral CT image. The method comprises the following steps: a to-be-modeled image making a modeling request: using a to-be-modeled image as an original cerebral CT image; pre-processing an image; segmenting and marking ROIs (regions of Interest); determining a region to which the ROI belongs; determining a region priority of the ROI; establishing a vertex and defining a feature vector on the vertex; defining a establishing rule of a side and the feature vector on the side; and then displaying a result. The invention provides a TRVL graph model displaying method based on the cerebral CT image, the model is a topological relation graph established according to the spatial relation between lateral ventricles and the influence of the lesion region to the ventricles; not only the relation between intracranial lateral ventricles of a brain can be correctly represented, but also the pathology information such as the influence of the lesion region to the ventricles can be represented at the same time; image information is well converted into a computer visual information.
Description
Technical field
The invention belongs to medical information technical field, be specifically related to a kind of graph model methods of exhibiting based on brain CT image.
Background technology
The focus of medical science and the research of computing machine cross discipline is become in recent years towards the research of medical image.Along with the fast development of medical digital equipment, medical information database is widely used.The structured text information of patient, and a large amount of destructuring medical images, data mining for medical image provides abundant data resource, thus make medical image effectively assist physicians can detect, locate and judge the good pernicious of it to pathological change region, therefore medical image is widely used in clinical diagnostic process.But, even if different judgements may be there is to same medical image in the doctor with different knowledge background, so, modeling is carried out to medical image, the image information contained by medical image and this relational structure of spatial information figure show, then on figure, carry out the operation that a series of data volume excavates, assist physician makes diagnostic result better objectively.Therefore based on brain CT graph model, there is higher learning value and actual application prospect.
At present, the method that some images of domestic and international existence represent, mainly comprise 2D String (Two-Dimensional String), 9D-SPA (9-Direction SPannng Area), RAG (Region Adjacency Graph), Irregular pyramid etc., these image representing method are the models set up for general pattern, do not consider medical domain knowledge.And in medical image method for expressing, in the paper A graph-based approach for the retrievalof multi-modality medical images that Kumar A. etc. deliver at Medical image analysis, CAPP (Complete-Anatomy Proximal-Pathology) graph model is proposed, but this model is for multi-modal thoracic cavity image, does not consider the ins and outs of brain CT image.In the paper Retrieval of brain tumorswith region-specific bag-of-visual-words representations in contrast-enhanced MRI images that Huang M. etc. deliver at Computational and mathematical methods in medicine, a Region-Specific BoW model is proposed, but do not consider the spatial relationship between area-of-interest (ROIs, Regins Of Interest) in this model.Therefore propose a kind of based on brain CT image, can represent that image information can represent that again the model of picture structure is a problem demanding prompt solution.
Summary of the invention
The object of the invention is to propose a kind of graph model methods of exhibiting based on brain CT image.
The object of the present invention is achieved like this:
(1) treat that modeled images proposes modeling request: treat that modeled images is original brain CT image;
(2) Image semantic classification: first, extracts brain encephalic part; Secondly, correct brain angle and extract brain center line ML simultaneously; Then, according to the external matrix cutting image of the vertical direction of encephalic part; Finally, normalized image size is Row × Column;
(3) segmentation of ROIs and mark: first successively split telocoele, gray scale presents more black lesion region and gray scale presents whiter lesion region, then marks three class ROIs respectively with different gray-scale values;
(4) region residing for ROI is determined: encephalic part is divided into upper left, lower-left, upper right, region, four, bottom right by brain center line ML and image level center line LL, each ROI is in a region in these four coordinates regionals or crosses over multiple region, determines region residing for ROI according to the pixel coordinate of ROI and the relation of ML, LL;
(5) area priorities of ROI is determined: the spatial relationship of region and ML, LL defines an area priorities to each ROI residing for ROI;
(6) definition of proper vector on the foundation on summit and summit: set up vertex set V according to ROIs, calculate the set of eigenvectors F on summit simultaneously
v;
(7) definition of proper vector on the foundation rule on limit and limit: region and the relation between ROI and other regions ROI set up limit collection E residing for the ROI generic of vertex correspondence, ROI, calculate the set of eigenvectors F on limit simultaneously
e;
(8) result is shown: show that the TRVL of a width brain CT image schemes G=(V, E, F
v, F
e), wherein, V is vertex set, and E is limit collection, F
vfor the set of eigenvectors on summit, F
efor the set of eigenvectors on limit.
Described ROIs is divided into: first, utilizes bayesian theory in brain CT image, to extract telocoele by the classification results of grey scale pixel value with according to the method that the segmentation result of ventricles of the brain collection of illustrative plates combines; Then, global threshold method is performed to encephalic part, obtains threshold value T1, according to T1 to image binaryzation, obtain gray scale and present more black lesion region; Finally, pixel encephalic part being removed to the ROIs split performs global threshold method, obtains threshold value T2, according to T2 to image binaryzation, obtains gray scale and present whiter lesion region.
Residing for described ROI, region is: first judge ROI and ML position relationship ROI in the left side of ML, right side or cross over ML, brain center line ML is outwardness, can obtain according to the pixel coordinate of ROI and the pixel coordinate of ML; Then judge the position relationship ROI of ROI and LL on the top of LL, bottom or cross over LL, LL is relevant with the size of image, and the normalized of image size makes directly to judge to there is error according to the pixel coordinate of ROI and the pixel coordinate of LL, introduce Two Variables y for this reason
1and y
2, y
1=y
min+ | y
max-y
min|/3, y
2=y
max-| y
max-y
min|/3, wherein y
maxfor the maximum ordinate of ROI, y
minfor the minimum ordinate of ROI, if y
1and y
2all be less than Row/2, then this ROI is in LL top, if y
1and y
2all be greater than Row/2, then this ROI is in LL bottom, otherwise this ROI crosses over LL, can judge ROI affiliated area according to this two step.
The area priorities of described ROI is: if ROI is in a region, then 1.1) region residing for ROI is first priority regions of this ROI, is designated as Area1; 1.2) ROI region is second priority regions of this ROI about the region of LL symmetry, is designated as Area2; 1.3) ROI region is the 3rd priority regions of this ROI about the region of ML symmetry, is designated as Area3; 1.4) the 4th priority regions of last region ROI for this reason, is designated as Area4, if ROI crosses over multiple region, then 2.1) first priority regions of the region crossed over of ROI ROI all for this reason, is designated as Area1; 2.2) second priority regions of other regions ROI for this reason, is designated as Area2.
On the foundation on described summit and summit, proper vector is defined as: each ROI
ibe defined as a vertex v
i, i.e. ROI
iand vertex v
iit is one-to-one relationship; Vertex v
ion proper vector be designated as F
v(v
i), then F
v(v
i)=(lab_vertex, area_vertex, s_vertex, leng_vertex, ht_vertex, c_vertex, r_vertex); Wherein, lab_vertex is the ROI generic of vertex correspondence, area_vertex be vertex correspondence ROI residing for region, s_vertex is the number of pixels in the ROI of vertex correspondence, leng_vertex is the ultimate range in the ROI of vertex correspondence between two pixels, ht_vertex is the homogeneity coefficient of the ROI of vertex correspondence, and c_vertex is the barycenter of the ROI of vertex correspondence, and r_vertex is the round and smooth degree of the ROI of vertex correspondence.
Described limit foundation rule and limit on proper vector be defined as: any two vertex v
iand v
jbetween the existence on limit meet following rule: (3.1) are if vertex v
icorresponding ROI
ibe telocoele and ROI
ibe in a region, if at ROI
ifirst, second and third priority regions in there is ROIs and these ROIs are telocoele, then v
ilimit is built, if otherwise ROI between the summit corresponding with each ROI
ito there is ROIs be telocoele in the 4th priority regions, v
ilimit is built on the summit corresponding with each ROI; If vertex v
icorresponding ROI
ibe telocoele and ROI
icross over multiple region, if at ROI
ithe first priority regions in there is ROIs and these ROIs are telocoele, then v
ilimit is built, if otherwise ROI between the summit corresponding with each ROI
ito there is ROIs be telocoele in the second priority regions, v
ilimit is built on the summit corresponding with each ROI; (3.2) if v
icorresponding ROI
ilesion region, then v
iwith at least one v
jbetween there is limit.V
jmeet following two conditions: (I) v
jcorresponding ROI
jit is telocoele; (II) if ROI
ithe first priority regions there is ROI
jand ROI
jtelocoele, then v
iwith v
jbetween build limit, otherwise check ROI
inext priority regions, until there is ROI in this priority regions
jand ROI
jtelocoele, then v
iwith v
jbetween build limit.
Beneficial effect of the present invention is:
The present invention proposes a kind of TRVL graph model methods of exhibiting based on brain CT image, this model is the topological relation figure set up the impact of telocoele according to the spatial relationship between telocoele and lesion region, it not only represents the relation between brain encephalic telocoele exactly, also present lesion region to pathological informations such as the impacts of telocoele simultaneously, change image information into computer vision information well.
Accompanying drawing explanation
Fig. 1 is Image semantic classification example;
Fig. 2 is that ROIs splits example;
Fig. 3 is the example that image transfers figure to;
Fig. 4 is image modeling process flow diagram.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is further illustrated:
(1) modeled images proposes modeling request: treat that modeled images is original brain CT image;
(2) Image semantic classification process: first, extracts brain encephalic part; Secondly, correct brain angle and extract brain center line ML (Middle Line) simultaneously; Then, according to the external matrix cutting image of the vertical direction of encephalic part; Finally, normalized image size is Row Column.
(3) segmentation of ROIs and mark: successively split telocoele, gray scale presents more black lesion region and gray scale presents whiter lesion region, then mark three class ROIs respectively with different gray-scale value.
(4) region residing for ROI: determine region residing for ROI according to the pixel coordinate of ROI and the relation of ML, LL.
(5) area priorities of ROI: encephalic part is divided into upper left, lower-left, upper right, region, four, bottom right by brain center line ML and image level center line LL, the area priorities of each ROI of contextual definition in region and these four regions residing for each ROI.
(6) definition of proper vector on the foundation on summit and summit: set up vertex set V according to ROIs, calculate the set of eigenvectors F on summit simultaneously
v.
(7) definition of proper vector on the foundation rule on limit and limit: region and the relation between ROI and other regions ROI set up limit collection E residing for the ROI generic of vertex correspondence, ROI, calculate the set of eigenvectors F on limit simultaneously
e.
(8) result is shown: show that the TRVL of a width brain CT image schemes G=(V, E, F
v, F
e), wherein, V is vertex set, and E is limit collection, F
vfor the set of eigenvectors on summit, F
efor the set of eigenvectors on limit.
Embodiment
(1) modeled images proposes modeling request: treat that modeled images is original brain CT image, as Fig. 1 (a);
(2) pre-service is carried out to medical image: first, utilize canny operator extraction brain encephalic part, as Fig. 1 (b); Secondly, the method proposed in the paper Automatic recognition ofmidline shift on brain CT images utilizing Liao C.C. etc. to deliver at Computers in biology and medicine is corrected brain angle and is extracted brain center line ML (Middle Line) simultaneously; Then, according to the external matrix cutting image of the vertical direction of encephalic part, as Fig. 1 (d) and 1 (e); Last normalized image size is Row × Column, as Fig. 1 (f).The size of image in the image data base that statistics uses, get Row=285, Column=260 is more reasonable.
(3) segmentation of ROIs and mark: successively split telocoele respectively, gray scale presents more black lesion region and gray scale presents whiter lesion region and mark three class ROIs with different gray-scale value.First, the ventricles of the brain dividing method that the paper Combined pixel classification and atlas-based segmentation of theventricular system in brain CT Images utilizing Vos P.C. etc. to deliver on SPIE MedicalImaging proposes splits telocoele on Fig. 2 (a), and in Fig. 2 (b), white portion is the telocoele extracted; Secondly, the encephalic partial pixel of global threshold method to Fig. 2 (a) utilizing Gonzalez R C to propose in its works Digital image processing processes, obtain threshold value T1, according to T1 to image binaryzation, the ROIs obtained comprises gray scale and presents more black ROIs and telocoele, remove telocoele, remaining ROIs is that gray scale presents more black lesion region, as white portion in Fig. 2 (c); Then, at the ROIs that the upper removal of Fig. 2 (a) has been split, Fig. 2 (d) is obtained; Then, global threshold method is performed again to the encephalic pixel removing the ROIs split, obtains threshold value T2, according to T2 to image binaryzation, obtain gray scale and present whiter lesion region, as white portion in Fig. 2 (e); Finally, the image of an initialization Row Column, marks the different gray-scale value of three class ROIs obtained, wherein telocoele gray-scale value 255 marks, gray scale presents more black ROIs gray-scale value 75 and marks, and gray scale presents whiter ROIs gray-scale value 150 and marks, as Fig. 2 (f).
(4) region residing for ROI: encephalic part is divided into upper left, lower-left, upper right, region, four, bottom right by brain center line ML and image level center line (LL, Level Line), each ROI be in these four regions one or more.Judge region residing for ROI by following two steps: first, judge ROI and ML position relationship, according to the location of pixels of ROI and the location of pixels of ML judge ROI in the left side of ML, right side or cross over ML; Then, judge the position relationship of ROI and LL, introduce Two Variables y for this reason
1and y
2if, y
1and y
2all be less than Row/2 (Row is the ordinate of LL), then this ROI is in LL top, if y
1and y
2all be greater than Row/2, then this ROI is in LL bottom, otherwise this ROI crosses over LL.Y
1and y
2be calculated as follows:
Wherein, y
maxfor the maximum ordinate of ROI, y
minfor the minimum ordinate of ROI.ROI affiliated area can be judged according to this two step.As Fig. 3 (a), it obtains with red line markings ML and LL on Fig. 2 (f), and encephalic part is divided into four regions by ML and LL.Location of pixels according to ROI and ML, LL can judge region residing for each ROI.
(5) encephalic part is divided into upper left, lower-left, upper right, region, four, bottom right by the area priorities of ROI: ML and LL, the decision rule of the area priorities of a contextual definition ROI in region and these four regions residing for each ROI: 1) if ROI is in a region, region then residing for ROI is first priority regions of this ROI, is designated as Area1; ROI region is second priority regions of this ROI about the region of LL symmetry, is designated as Area2; ROI region is the 3rd priority regions of this ROI about the region of ML symmetry, is designated as Area3; 4th priority regions of last region ROI for this reason, is designated as Area4.2) if ROI crosses over multiple region, then first priority regions of region ROI all for this reason crossed over of ROI, is designated as Area1; Second priority regions of other regions ROI for this reason, is designated as Area2.
(6) definition of proper vector on the foundation on summit and summit: first, set up vertex set V according to ROIs, each ROI
ibe defined as a vertex v
i, thus obtain a vertex set, if Fig. 3 (d) is the vertex set set up according to the ROIs in Fig. 3 (a).Then, composition graphs 3 (a) and Fig. 3 (b) calculate each ROI
ifeature, it can be used as corresponding vertex v
ion proper vector F
v(v
i), and continuous type feature is normalized makes its codomain for [0,1], here F
v(v
i)=(lab_vertex, area_vertex, s_vertex, leng_vertex, ht_vertex, c_vertex, r_vertex).Wherein, lab_vertex and area_vertex is discrete type feature, and other feature is all continuous type.
1) lab_vertex is the mark of the ROI of vertex correspondence: { 255,150,75};
2) area_vertex be vertex correspondence ROI residing for region: { Area1, Area2, Area3, Area4};
3) s_vertex is the number of pixels in the ROI of vertex correspondence, and normalization s_vertex=s_vertex/ (Row Column) makes s_vertex (0,1);
4) leng_vertex is the ultimate range in the ROI of vertex correspondence between two pixels, and normalization leng_vertex=1/leng_vertex is leng_vertex (0,1);
5) ht_vertex [0,1] be the homogeneity coefficient of ROI of vertex correspondence, for weighing the consistance of ROI pixel grey scale, in the 3-D image proposed in the paper A graph-based approach for the retrieval ofmulti-modality medical images delivered at Medical image analysis by Kumar A. etc., the formula of the homogeneity coefficient of ROI expands in 2-D image, obtains following formula:
Wherein, nd (x
i, y
j) represent pixel (x
i, y
j) the set of adjacent pixels point, Row Column is the number of pixels of image.
6)
be the barycenter of the ROI of vertex correspondence, c_vertex is two tuples, is mainly used in the calculating of proper vector on hereafter limit;
Wherein, M
10and M
01for the first moment of image, M
00for the zeroth order square of image.
7) r_vertex ∈ [0,1] is the round and smooth degree of the ROI of vertex correspondence, when r_vertex 1, ROI is rounder and more smooth.
Wherein, μ
20, μ
02, μ
11it is the second-order moment around mean of image.
(7) definition of proper vector on the foundation rule on limit and limit: first, set up limit collection E:1 according to following rule) if F
v(v
i, 1)=255 and v
ibe in a region, if ROI
iarea1, Area2 and Area3 in there is the ROIs of lab_vertex=255, then v
iand build limit between each ROI, otherwise v
iwith ROI
iarea4 in each ROI of lab_vertex=255 build limit; If F
v(v
i, 1)=255 and ROI
icross over multiple region, then v
ilimit is built between the summit corresponding with the ROIs of all lab_vertex=255.2) if F
v(v
i, 1) and 255, v
iwith the vertex v of at least one lab_vertex=255
jbetween there is limit.V
jmeet following condition: if ROI
iarea1 in there is the ROIs of lab_vertex=255, then v
iand build limit between each ROI, otherwise check ROI
inext priority regions, until there is the ROIs of lab_vertex=255 in this priority regions, v
iand build limit between each ROI.Limit rule of building like this makes the figure set up be connected graph.Then, the proper vector F on every bar limit e is calculated
e(e)=(d_edge, md_edge, ro_edge}, wherein:
1) d_edge is the distance between two ROIs barycenter.Given two barycenter
with
2) md_edge is the bee-line between two ROIs.Given two ROIs:ROI and ROI ',
3) ro_edge=(sin θ, cos θ) is the angle tolerance of two ROIs barycenter lines and horizontal direction.As Fig. 3 (c).
(8) show that the TRVL that result: Fig. 3 (f) is an original brain CT image transforms schemes G=(V, E, F
v, F
e), wherein, V is vertex set, and E is limit collection, F
vfor the set of eigenvectors on summit, F
efor the set of eigenvectors on limit.
Claims (6)
1., based on the graph model methods of exhibiting of brain CT image, it is characterized in that, comprise the steps:
(1) treat that modeled images proposes modeling request: treat that modeled images is original brain CT image;
(2) Image semantic classification: first, extracts brain encephalic part; Secondly, correct brain angle and extract brain center line ML simultaneously; Then, according to the external matrix cutting image of the vertical direction of encephalic part; Finally, normalized image size is Row × Column;
(3) segmentation of ROIs and mark: first successively split telocoele, gray scale presents more black lesion region and gray scale presents whiter lesion region, then marks three class ROIs respectively with different gray-scale values;
(4) region residing for ROI is determined: encephalic part is divided into upper left, lower-left, upper right, region, four, bottom right by brain center line ML and image level center line LL, each ROI is in a region in these four coordinates regionals or crosses over multiple region, determines region residing for ROI according to the pixel coordinate of ROI and the relation of ML, LL;
(5) area priorities of ROI is determined: the spatial relationship of region and ML, LL defines an area priorities to each ROI residing for ROI;
(6) definition of proper vector on the foundation on summit and summit: set up vertex set V according to ROIs, calculate the set of eigenvectors F on summit simultaneously
v;
(7) definition of proper vector on the foundation rule on limit and limit: region and the relation between ROI and other regions ROI set up limit collection E residing for the ROI generic of vertex correspondence, ROI, calculate the set of eigenvectors F on limit simultaneously
e;
(8) result is shown: show that the TRVL of a width brain CT image schemes G=(V, E, F
v, F
e), wherein, V is vertex set, and E is limit collection, F
vfor the set of eigenvectors on summit, F
efor the set of eigenvectors on limit.
2. the graph model methods of exhibiting based on brain CT image according to claim 1, it is characterized in that, described ROIs is divided into: first, utilizes bayesian theory in brain CT image, to extract telocoele by the classification results of grey scale pixel value with according to the method that the segmentation result of ventricles of the brain collection of illustrative plates combines; Then, global threshold method is performed to encephalic part, obtains threshold value T1, according to T1 to image binaryzation, obtain gray scale and present more black lesion region; Finally, pixel encephalic part being removed to the ROIs split performs global threshold method, obtains threshold value T2, according to T2 to image binaryzation, obtains gray scale and present whiter lesion region.
3. the graph model methods of exhibiting based on brain CT image according to claim 1, it is characterized in that, residing for described ROI, region is: first judge ROI and ML position relationship ROI in the left side of ML, right side or cross over ML, brain center line ML is outwardness, can obtain according to the pixel coordinate of ROI and the pixel coordinate of ML; Then judge the position relationship ROI of ROI and LL on the top of LL, bottom or cross over LL, LL is relevant with the size of image, and the normalized of image size makes directly to judge to there is error according to the pixel coordinate of ROI and the pixel coordinate of LL, introduce Two Variables y for this reason
1and y
2, y
1=y
min+ | y
max-y
min|/3, y
2=y
max-| y
max-y
min|/3, wherein y
maxfor the maximum ordinate of ROI, y
minfor the minimum ordinate of ROI, if y
1and y
2all be less than Row/2, then this ROI is in LL top, if y
1and y
2all be greater than Row/2, then this ROI is in LL bottom, otherwise this ROI crosses over LL, can judge ROI affiliated area according to this two step.
4. the graph model methods of exhibiting based on brain CT image according to claim 1, it is characterized in that, the area priorities of described ROI is: if ROI is in a region, then 1.1) region residing for ROI is first priority regions of this ROI, is designated as Area1; 1.2) ROI region is second priority regions of this ROI about the region of LL symmetry, is designated as Area2; 1.3) ROI region is the 3rd priority regions of this ROI about the region of ML symmetry, is designated as Area3; 1.4) the 4th priority regions of last region ROI for this reason, is designated as Area4, if ROI crosses over multiple region, then 2.1) first priority regions of the region crossed over of ROI ROI all for this reason, is designated as Area1; 2.2) second priority regions of other regions ROI for this reason, is designated as Area2.
5. the graph model methods of exhibiting based on brain CT image according to claim 1, is characterized in that, on the foundation on described summit and summit, proper vector is defined as: each ROI
ibe defined as a vertex v
i, i.e. ROI
iand vertex v
iit is one-to-one relationship; Vertex v
ion proper vector be designated as F
v(v
i), then F
v(v
i)=(lab_vertex, area_vertex, s_vertex, leng_vertex, ht_vertex, c_vertex, r_vertex); Wherein, lab_vertex is the ROI generic of vertex correspondence, area_vertex be vertex correspondence ROI residing for region, s_vertex is the number of pixels in the ROI of vertex correspondence, leng_vertex is the ultimate range in the ROI of vertex correspondence between two pixels, ht_vertex is the homogeneity coefficient of the ROI of vertex correspondence, and c_vertex is the barycenter of the ROI of vertex correspondence, and r_vertex is the round and smooth degree of the ROI of vertex correspondence.
6. the graph model methods of exhibiting based on brain CT image according to claim 1, is characterized in that, on the foundation on described limit rule and limit, proper vector is defined as: any two vertex v
iand v
jbetween the existence on limit meet following rule: (3.1) are if vertex v
icorresponding ROI
ibe telocoele and ROI
ibe in a region, if at ROI
ifirst, second and third priority regions in there is ROIs and these ROIs are telocoele, then v
ilimit is built, if otherwise ROI between the summit corresponding with each ROI
ito there is ROIs be telocoele in the 4th priority regions, v
ilimit is built on the summit corresponding with each ROI; If vertex v
icorresponding ROI
ibe telocoele and ROI
icross over multiple region, if at ROI
ithe first priority regions in there is ROIs and these ROIs are telocoele, then v
ilimit is built, if otherwise ROI between the summit corresponding with each ROI
ito there is ROIs be telocoele in the second priority regions, v
ilimit is built on the summit corresponding with each ROI; (3.2) if v
icorresponding ROI
ilesion region, then v
iwith at least one v
jbetween there is limit; v
jmeet following two conditions: (I) v
jcorresponding ROI
jit is telocoele; (II) if ROI
ithe first priority regions there is ROI
jand ROI
jtelocoele, then v
iwith v
jbetween build limit, otherwise check ROI
inext priority regions, until there is ROI in this priority regions
jand ROI
jtelocoele, then v
iwith v
jbetween build limit.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510324245.4A CN104933712B (en) | 2015-06-12 | 2015-06-12 | Graph model methods of exhibiting based on brain CT images |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510324245.4A CN104933712B (en) | 2015-06-12 | 2015-06-12 | Graph model methods of exhibiting based on brain CT images |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104933712A true CN104933712A (en) | 2015-09-23 |
CN104933712B CN104933712B (en) | 2017-12-01 |
Family
ID=54120866
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510324245.4A Expired - Fee Related CN104933712B (en) | 2015-06-12 | 2015-06-12 | Graph model methods of exhibiting based on brain CT images |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104933712B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110415219A (en) * | 2019-07-04 | 2019-11-05 | 杭州深睿博联科技有限公司 | Medical image processing method and device, equipment, storage medium based on depth segmentation network |
CN110443808A (en) * | 2019-07-04 | 2019-11-12 | 杭州深睿博联科技有限公司 | Medical image processing method and device, equipment, storage medium for the detection of brain middle line |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100322496A1 (en) * | 2008-02-29 | 2010-12-23 | Agency For Science, Technology And Research | Method and system for anatomy structure segmentation and modeling in an image |
CN103247073A (en) * | 2013-04-18 | 2013-08-14 | 北京师范大学 | Three-dimensional brain blood vessel model construction method based on tree structure |
CN104217213A (en) * | 2014-08-20 | 2014-12-17 | 哈尔滨工程大学 | Medical image multi-stage classification method based on symmetry theory |
-
2015
- 2015-06-12 CN CN201510324245.4A patent/CN104933712B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100322496A1 (en) * | 2008-02-29 | 2010-12-23 | Agency For Science, Technology And Research | Method and system for anatomy structure segmentation and modeling in an image |
CN103247073A (en) * | 2013-04-18 | 2013-08-14 | 北京师范大学 | Three-dimensional brain blood vessel model construction method based on tree structure |
CN104217213A (en) * | 2014-08-20 | 2014-12-17 | 哈尔滨工程大学 | Medical image multi-stage classification method based on symmetry theory |
Non-Patent Citations (3)
Title |
---|
ASHNIL KUMAR 等: "A graph-based approach for the retrieval of multi-modality medical images", 《MEDICAL IMAGE ANALYSIS》 * |
PIETER C. VOS 等: "Combined Pixel Classification and Atlas-based segmentation of the Ventricular System in brain CT images", 《MEDICAL IMAGING 2013:IMAGE PROCESSING》 * |
潘海为 等: "一种新颖的医学图像建模及相似性搜索方法", 《计算机学报》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110415219A (en) * | 2019-07-04 | 2019-11-05 | 杭州深睿博联科技有限公司 | Medical image processing method and device, equipment, storage medium based on depth segmentation network |
CN110443808A (en) * | 2019-07-04 | 2019-11-12 | 杭州深睿博联科技有限公司 | Medical image processing method and device, equipment, storage medium for the detection of brain middle line |
CN110443808B (en) * | 2019-07-04 | 2022-04-01 | 杭州深睿博联科技有限公司 | Medical image processing method and device for brain midline detection, equipment and storage medium |
CN110415219B (en) * | 2019-07-04 | 2022-04-05 | 杭州深睿博联科技有限公司 | Medical image processing method and device based on depth segmentation network, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN104933712B (en) | 2017-12-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Gao et al. | Classification of CT brain images based on deep learning networks | |
Xue et al. | Chest x-ray image view classification | |
Banerjee et al. | Automated 3D segmentation of brain tumor using visual saliency | |
CN103353986B (en) | A kind of brain MR image segmentation based on super-pixel fuzzy clustering | |
CN103218605B (en) | A kind of fast human-eye positioning method based on integral projection and rim detection | |
CN103914697B (en) | A kind of extracting method of breast 3-D view area-of-interest | |
CN104091337A (en) | Deformation medical image registration method based on PCA and diffeomorphism Demons | |
CN105760874A (en) | CT image processing system and method for pneumoconiosis | |
US20230005140A1 (en) | Automated detection of tumors based on image processing | |
CN105488800A (en) | Feature extracting method and computer-aided diagnosis method and system | |
CN102663762B (en) | The dividing method of symmetrical organ in medical image | |
CN114359277B (en) | Brain image processing method and system for stroke patient | |
CN103150723A (en) | Lymph gland detection system and method based on stomach computed tomography (CT) image with shape fitting with oval | |
Abdel-Nasser et al. | Temporal mammogram image registration using optimized curvilinear coordinates | |
CN103903238A (en) | Method for fusing significant structure and relevant structure of characteristics of image | |
CN104318559A (en) | Quick feature point detecting method for video image matching | |
Schwier et al. | Automated spine and vertebrae detection in CT images using object‐based image analysis | |
CN101727568B (en) | Foreground action estimation device and foreground action estimation method | |
CN104933712A (en) | Graph model displaying method based on cerebral CT image | |
CN105426808A (en) | Intra-brain sagittal line measurement method and system | |
Kim et al. | Automatic localization of anatomical landmarks in cardiac MR perfusion using random forests | |
Taghizadeh et al. | The fast and accurate approach to detection and segmentation of melanoma skin cancer using fine-tuned YOLOV3 and SegNet based on deep transfer learning | |
CN106446923A (en) | Medical image classification method based on corner matching | |
US10825174B2 (en) | Breast image analysis method, breast image analysis system, and non-transitory computer-readable medium | |
CN113989269B (en) | Traditional Chinese medicine tongue image tooth trace automatic detection method based on convolutional neural network multi-scale feature fusion |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171201 |
|
CF01 | Termination of patent right due to non-payment of annual fee |