CN104809736B - Medical Slice Images closure bone contours computational methods based on priori - Google Patents
Medical Slice Images closure bone contours computational methods based on priori Download PDFInfo
- Publication number
- CN104809736B CN104809736B CN201510239512.8A CN201510239512A CN104809736B CN 104809736 B CN104809736 B CN 104809736B CN 201510239512 A CN201510239512 A CN 201510239512A CN 104809736 B CN104809736 B CN 104809736B
- Authority
- CN
- China
- Prior art keywords
- template
- bone
- reference template
- contours
- same type
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
- G06T7/0014—Biomedical image inspection using an image reference approach
-
- 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
-
- 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
- G06T2207/30008—Bone
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Image Analysis (AREA)
Abstract
Bone contours computational methods are closed the invention discloses a kind of Medical Slice Images based on priori；First, the bone contours extracted from Medical Slice Images are pre-processed, without unnecessary branch line and the single loop bone contours of intersecting situation will be occurred without as reference template, remaining is non-reference template；Then, introduce same type bone template, the profile matched the most with reference template is searched in same type bone template by method of characteristic and both affine transformations are set up, using non-reference template and the neighbouring relations of reference template, in the equidirectional same type bone template for finding and being matched the most with non-reference template；Finally, using reference template and same type bone template is as priori completion non-reference template cut edge and removes uncorrelated contour line.The invention is applied in Reconstructing Three-Dimensional Medical Section design, to often occur in contours extract intersecting, adjacent, cut edge the problems such as carry out Taxonomic discussion, the quality and efficiency of contours extract can be greatly improved.
Description
Technical field
The invention belongs to computer graphics techniques field, more particularly to during Medical Slice Images bone three-dimensional reconstruction
Need the bone contours line extracted.
Background technology
During the bone three-dimensional reconstruction of faultage image, bone contours identification is particularly important, and it is directly connected to finally
The levels of precision of reconstructed results, but due to the influence of the factors such as ageing equipment, picture noise, the fine cracks of bone surface, obtain
Bone contours line can not truly reduce its style.There are many documents to be directed to finding better method identification both at home and abroad
Image contour line, such as automatically obtains high-low threshold value, and a variety of detection technique phases in edge detection stage using adaptive mode
Fusion obtains the best edge wheel profile of effect；In the Boundary Extraction stage edge is used as by the use of marginal point two side portions gray average
Contrast characteristic's identification edge line trend, or constrain itself size and Orientation by the average vector in the field of marginal point 8.These sides
Method is all based on the identification of single image edge line, and because Medical Slice Images have sequentiality, neighbouring similitude, same type
There is global similitude, the bone contours of extraction have the features such as closed, for the bone wheel of Medical Slice Images between bone
Exterior feature identification can using adjacent layer either existing same type bone collectively as priori auxiliary completion do not close cut edge or
Assist in identifying and close wrong profile.
Using adjacent layer or existing same type bone collectively as priori, it is necessary to be found according to feature described function
The same type bone contours line matched the most with current layer.Profile starting point registration is a kind of highly effective method, by following
Ring displacement can obtain most like in differently contoured starting Point matching one.Affine transformation can realize characteristic point image space
Coordinate transform, the corresponding relation set up between current layer and same type bone contours line.
The content of the invention
For due to machinery equipment aging, picture noise, the fine gap of bone surface, many bones of Medical Slice Images
The phenomenons such as scanning cause contour line intersecting, adjacent, cut edge situation occur；Priori is based on it is an object of the invention to provide one kind
The bone Medical Slice Images closed contour computational methods of knowledge, have sequentiality, neighbouring similitude according to Medical Slice Images,
There is global similitude, the bone contours of extraction have the features such as continuity, closed between same type bone；Propose with existing
The preferable single loop profile of same type bone contours, effect is priori auxiliary, solve to occur in other contour lines it is intersecting,
Adjacent, cut edge problem, compared with existing outline extraction technique, is all improved in the precision and efficiency of extraction.
To achieve these goals, the technical solution used in the present invention is：
Medical Slice Images closure bone contours computational methods based on priori, it is characterised in that：Imitated using extracting
Fruit preferably single loop bone contours, existing same type bone template contours are that priori is guided jointly, and other are obtained automatically
Medical Slice Images close the process of bone contours.
First, the bone contours all Medical Slice Images extracted pre-process and are categorized into reference template and non-ginseng
Template is examined, the step includes as follows：
Step S01：Medical Slice Images are read in successively, and the bone contours of Medical Slice Images are obtained using Canny methods
Map, some continuous or discrete bone contours of every Medical Slice Images are obtained with reference to edge following algorithm；
Step S02：By obtained bone contours referred to as objective contour, if can searching out closure from objective contour or close
The best bone contours of conjunction property, then obtain closure bone contours as referring to mould in advance by the use of cubic spline interpolation connection breakpoint
Version；Otherwise it is labeled as non-reference masterplate；
Step S03：The unnecessary branch line produced in pre- reference template due to interpolation is removed, the pre- reference template of single loop is marked
For reference template；Otherwise it is labeled as non-reference masterplate；
Then, non-reference template and same type bone template corresponding relation are set up, the step includes as follows：
Step S04：Same type bone template contour interval is set, resampling obtains one group of closure bone contours；
Step S05：Make similitude judgement between correspondence reference template and same type bone template using method of characteristic, so that
Find the same type bone template most matched with non-reference template；
Finally, using neighboring reference template, same type bone template as priori, completion cut edge simultaneously removes non-reference masterplate
In incoherent contour line, the step include it is as follows：
Step S06：May situation classification to uncorrelated contour line；
Step S07：Utilize the reference vector in public angle calcu-lation same type bone template, neighboring reference template and non-
Constraint calculates the interpolated coordinates of breakpoint subsequent point with completion cut edge to reference template breakpoint average vector jointly；
Step S08：Removed out by the use of the reference vector in same type bone template, neighboring reference template as priori
Existing uncorrelated contour line；
Step S09：Repeat step S07 to step S08 directly causes non-reference template to turn into single loop closure bone contours and incite somebody to action
It is labeled as reference template.
Bone contours map described in S01 refers to set binaryzation threshold to Medical Slice Images using Canny edge detection methods
Value, filters out the pixels such as skin, muscle, blood vessel, retains the bone contours of all bone informations（Including bone internal layer profile）.
The continuous or discrete bone contours refer to from Canny rim detections obtain two-value bone contours map in carry
Outermost layer bone contours are taken out, due to the factor such as fine cracks on noise in original image, bone, the bone contours of extraction can go out
Existing non-continuous event.
The bone contours closed described in S02 refer to one or one group of bone contours is closure state；
The best bone contours of the closed refer to that contour line has Euclidean distance between breakpoint and adjacent breakpoint and is respectively less than
Distance threshold T, distortionless closure bone contours can be obtained by interpolation；
The best bone contours of the closed refer to that contour line has Euclidean distance between breakpoint and adjacent breakpoint and is respectively less than
Distance threshold T, distortionless closure bone contours are obtained by interpolation；
The pre- reference template refers to that interpolated rear formed carries unnecessary branch line and multiloop closure bone contours.
The non-reference masterplate meets following at 2 points：1. do not close or closed is bad, 2. meet closure or closed most
Get well but feeder number is more than 1 after cubic spline interpolation, the unnecessary branch line of removal.
If described can search out bone contours closure or that closed is best from obtained contour line, three are utilized
Secondary spline method connection breakpoint obtains closing bone contours, concretely comprises the following steps：
Step i01：Search for the breakpoint number of current layer bone contours and record coordinate position, closure is considered if without breakpoint
Bone contours；
Step i02：The each breakpoint neighborhood of this layer of bone contours is searched for one by one judges whether other contour line breakpoints, if
In the presence of then calculating Euclidean distance between two breakpoints；
Step i03：Given distance threshold T, if Euclidean distance is respectively less than distance threshold T between all adjacent breakpoints, then it is assumed that
It is the best bone contours of closed, breakpoint is connected using cubic spline interpolation；
Step i04：Repeat step i01 to step i03 directly causes to find all bone contours that can be closed, and is marked as
Pre- reference template, records pre- reference template numbering.
Branch line described in S03 can be divided into cut edge, intersecting two classes, due to machinery equipment aging, picture noise, bone surface it is thin
What the factors such as small gap were caused, cut edge refers to that branch line is path and certain end point is breakpoint, intersecting to refer to that branch line is loop.
The unnecessary branch line refers to there is intersection point in closure bone contours, and cut edge is branched into from a certain intersection point.
The reference template refers to filter out from pre- reference template without unnecessary branch line and occurred without single time of intersecting situation
Road bone contours.
The unnecessary branch line produced in pre- reference template due to interpolation is closed in the removal, and the pre- reference template of single loop is labeled as
Reference template, is concretely comprised the following steps：
Step q01：All intersection points that there is branch line and coordinate is recorded in the pre- reference template of search positioning, if failure is positioned
Think that the pre- reference template, without unnecessary branch line, continues step q03；Otherwise step q02 is continued；
Step q02：According to orientation, track every branch of each intersection point and record follow-up summit, if follow-up summit
Then think that the branch line belongs to cut edge class for breakpoint（It is labeled as）, all vertex values of the branch line are directly set to 0 i.e. removable this and broken
The unnecessary branch line of side class；Think that the branch line belongs to intersecting class and records coordinate if being another intersection point if follow-up summit；
Step q03：Calculate pre- reference template feeder number and update intersection point record sheet, be 0 by number of hits in pre- reference template
And feeder number is labeled as reference template for 1 pre- reference template, otherwise labeled as non-reference masterplate.
Step q04：Repeat step q01 to step q03 is straight to cause to filter out all ginsengs for meeting condition from pre- reference template
Examine masterplate.
Existing same type bone template refers to identical with Medical faultage image position and separated by joint described in S04
The known skeleton data obtained afterwards, single time with the interlamellar spacing such as Medical faultage image can be obtained by contour resampling
Road contour line collection.
Make similitude judgement between correspondence reference template and same type bone template using using method of characteristic described in S05,
So as to find the same type bone template most matched with non-reference template, concretely comprise the following steps：
Step t01：Profile point on reference template and same type bone template is utilized into the minimum method polymerization of mean error
Non-significant characteristic point, retains identical quantative attribute point using threshold value in two templates；
Step t02：Both, which are obtained, using method of characteristic matches starting point；
Step t03：It can be found if it can obtain best match equidirectional with adjacent non-reference template contours the most
The same type bone template contours of matching；Otherwise same type bone template contour first sampling point is adjusted, continues step S04；
Step t04：Using center of gravity as reference, reference template and same type bone template affine transformation equation are calculated；
Step t05：The template adjacent with the same type bone template, as the non-reference template adjacent with reference template are most
For matching；
Step t06：Repeat step t01 to step t05, finds the same type bone mould that all non-reference templates are matched the most
Plate.
The method of characteristic refers to calculate both characteristic functions, using amplitude maximal point as reference template matching starting point,
This time matching that all characteristic points of loopy moving obtain minimum extreme value difference in same type bone template is best match, this time
It is same type bone template matches starting point to match starting point.
The affine transformation refers to the matching starting point obtained using step t02, to same type on the basis of reference template
Bone template do rotate, translation, alignd so that both starting points be done to center of gravity line；It is secondary to utilize the affine transformation
With the alignment step u02 non-reference template contours referred to and same type bone template contours.
Uncorrelated contour line described in S06 is made up of following 2 class situation：1. it is equal to the intersecting class that step S03 is mentioned, due to
The factors such as machinery equipment aging, picture noise, the fine gap of bone surface cause bone contours intersecting situation occur；2. by
It is many bones and bone contours adjacencies occurs in what Medical Slice Images was scanned.
Public angle described in S07 is used to inquire about non-reference template and pair in same type template bone contours is most matched with it
Ying Dian, is concretely comprised the following steps：
Step u01：Make non-reference template and same type template bone contours are most matched with it and be denoted as A, B, barycentric coodinates note
Make C, C '；
Step u02：The affine equation obtained using step t04, profile A, B center of gravity are alignd；
Step u03：Certain point P to center of gravity C line PC and x-axis formation angle theta from A；
Step u04：Sent from B centers of gravity C ' with x-axis angle be all θ ray, ray and B intersect at point P ', then (P, P ')
As required point pair；
Step u05：The corresponding points pair in neighboring reference template and non-reference template can also similarly be calculated（P, P "）
The utilization non-reference template breakpoint average vector is constrained jointly, calculates the interpolated coordinates of breakpoint subsequent point to mend
Full cut edge, is concretely comprised the following steps：
Step e01：Its vector between preceding n summit is done distance weighted flat by positioning to a certain breakpoint P of non-reference template
, vectorial a is obtained as the most probable direction discreet value of P+1 points；
Step e02：The corresponding points P ' and P " obtained using step u01- steps u05, calculating reference vector b=P ' P+1 ', c=
P”P+1”；
Step e03：Using average vector a, reference vector b and c average weighted result inserting as P point subsequent points P+1
It is worth direction, repeats this step and directly cause subsequent point P+1 to be non-breakpoint；
Step e04：Repeat step e01 and step e03, directly causes all contour line cut edge connections to finish；
The reference vector refers to due to having global similitude between same type bone；Adjacent interlayer bone contours have office
It is follow-up with non-reference template intersection point/breakpoint P corresponding points in portion's similitude, therefore same type bone template, neighboring reference template
Point can be used to differentiating/instructing that subsequent point may moving direction as reference vector.
The average vector refers to that P is breakpoint situation, because bone contours have continuity and flatness, therefore non-reference
The vector of template breakpoint P n front and continued point has certain guidance meaning to point P subsequent point coordinate direction, it is possible to use average
Vector estimates point P+1 may moving direction.
Removed out described in S08 by the use of the reference vector in same type bone template, neighboring reference template as priori
Existing uncorrelated contour line, is concretely comprised the following steps：
Step v01：If not there is intersection point P in reference template, then situation is 1. planted for described in right 8；Calculate intersection point P
With each subsequent point P+1 vector value d, utilize corresponding points P's ' in public angle calcu-lation same type bone template to subsequent point P+1 '
Vector value c, removes the d maximum with the c deviations of directivity；
Step v02：If not reference template is multiloop and is then 2. to plant situation described in right 8 without breakpoint；Calculate non-
Each distance between center of gravity and reference template center of gravity, removes the contour line farther out from reference template center of gravity in reference template, you can protect
Bone contours line, realizes bone quick separating needed for staying；
The present invention is obtained without unnecessary branch line and occurred without single time of intersecting situation by being pre-processed to Medical Slice Images
Road bone contours are used as reference template；Same type bone template is re-introduced into, is searched by method of characteristic in same type bone template
Rope sets up both affine transformations with the profile that reference template is matched the most；Found using affine transformation adjacent with reference template non-
The same type bone template that reference template is matched the most, reference template and same type bone template is non-as priori completion
Reference template cut edge simultaneously removes uncorrelated contour line.The invention is applied in Reconstructing Three-Dimensional Medical Section design, for wheel
The problems such as intersecting, adjacent, cut edge that exterior feature occurs in extracting, makees classification processing, can greatly improve contours extract quality, it is ensured that three-dimensional
The accuracy of reconstruction.
Brief description of the drawings
Fig. 1 is the bone Medical Slice Images closed contour evaluation work flow figure based on priori in the present invention；
Fig. 2 is using Canny algorithms to extract bone contours process schematic in the present invention；
Fig. 3 is the best bone contours flow chart of objective contour pretreatment stage identification closure/closed in the present invention；
Fig. 4 is three sample interpolation schematic diagrames of objective contour pretreatment stage in the present invention；
Fig. 5 is the unnecessary branch line schematic diagram of objective contour pretreatment stage removing in the present invention；
Fig. 6 is that objective contour pretreatment stage filters out reference template flow chart from pre- reference template in the present invention；
Fig. 7 be mention in the present invention same type bone template, non-reference masterplate, reference template triadic relation figure；
Fig. 8 is to make affine transformation schematic diagram to same type bone template on the basis of reference template in the present invention；
Fig. 9 is to remove uncorrelated contour line in non-reference masterplate in the present invention by priori of same type bone template
（Intersecting situation）Schematic diagram；
Figure 10 is to remove uncorrelated contour line in non-reference masterplate in the present invention by priori of same type bone template
（There are the adjacencies of useless breakpoint）Schematic diagram；
Figure 11 is to do cut edge completion to non-reference masterplate with same type bone template, reference template priori in the present invention
Schematic diagram；
Figure 12 is to use the inventive method to some non-reference template for breakpoint and uncorrelated contour line occur in the present invention
Branch's schematic diagram afterwards.
Embodiment
Below according to accompanying drawing, with reference to specific implementation step, the invention will be further described：
As shown in figure 1, the Medical Slice Images closure bone contours computational methods based on priori, it is characterised in that：
It is that priori is guided jointly using the preferable single loop bone contours of extraction effect, existing same type bone template contours, from
It is dynamic to obtain the process that other Medical Slice Images close bone contours.
First, the bone contours all Medical Slice Images extracted are pre-processed（Walked as shown in Fig. 1 dotted line frames
Suddenly）And reference template and non-reference template are categorized into, comprise the following steps：
Step S01：Medical Slice Images are read in successively, and the bone contours of Medical Slice Images are obtained using Canny methods
Map, some continuous or discrete bone contours of every Medical Slice Images are obtained with reference to edge following algorithm；
Step S02：By obtained bone contours referred to as objective contour, if can searching out closure from objective contour or close
The best bone contours of conjunction property, then obtain closure bone contours as referring to mould in advance by the use of cubic spline interpolation connection breakpoint
Version；Otherwise it is labeled as non-reference masterplate；
Step S03：The unnecessary branch line produced in pre- reference template due to interpolation is removed, the pre- reference template of single loop is marked
For reference template；Otherwise it is labeled as non-reference masterplate；
Then, non-reference template and same type bone template corresponding relation are set up, the step includes as follows：
Step S04：Same type bone template contour interval is set, resampling obtains one group of closure bone contours；
Step S05：Make similitude judgement between correspondence reference template and same type bone template using method of characteristic, so that
Find the same type bone template most matched with non-reference template；
Finally, using neighboring reference template, same type bone template as priori, completion cut edge simultaneously removes non-reference masterplate
In incoherent contour line, the step include it is as follows：
Step S06：May situation classification to uncorrelated contour line；
Step S07：Utilize the reference vector in public angle calcu-lation same type bone template, neighboring reference template and non-
Constraint calculates the interpolated coordinates of breakpoint subsequent point with completion cut edge to reference template breakpoint average vector jointly；
Step S08：Removed out by the use of the reference vector in same type bone template, neighboring reference template as priori
Existing uncorrelated contour line；
Step S09：Repeat step S07 to step S08 directly causes non-reference template to turn into single loop closure bone contours and incite somebody to action
It is labeled as reference template.
As shown in Fig. 2 1. the 2- is Medical Slice Images；2. 2- is to set binary-state threshold to Medical Slice Images
Afterwards, the pixels such as skin in artwork, muscle, blood vessel have been filtered out, have retained all bone pixels；3. 2- is to utilize Canny rim detections
The bone contours map that method is obtained（Including bone internal layer profile）；4. 2- is to obtain outer layer bone wheel using edge following algorithm
Exterior feature, because the factor such as fine cracks influences to cause the bone contours extracted to occur discontinuously on noise in original image, bone
Phenomenon.
As shown in figure 3, recognizing the best bone contours flow chart of closure/closed, tool for objective contour pretreatment stage
Body step is：
Step i01：Search for the breakpoint number of current layer bone contours and record coordinate position, closure is considered if without breakpoint
Bone contours；
Step i02：The each breakpoint neighborhood of this layer of bone contours is searched for one by one judges whether other contour line breakpoints, if
In the presence of then calculating Euclidean distance between two breakpoints；
Step i03：Given distance threshold T, if Euclidean distance is respectively less than distance threshold T between all adjacent breakpoints, then it is assumed that
It is the best bone contours of closed, breakpoint is connected using cubic spline interpolation；
Step i04：Repeat step i01 to step i03 directly causes to find all bone contours that can be closed, and is marked as
Pre- reference template, records pre- reference template numbering.
The bone contours of the closure refer to one or one group of bone contours is closure state；The best bone of closed
Profile refers to that contour line has Euclidean distance between breakpoint and adjacent breakpoint and is respectively less than distance threshold T, can be obtained not by interpolation
The closure bone contours of distortion；
It is described do not close or bone contours that closed is bad refer to contour line exist between breakpoint and adjacent breakpoint it is European away from
From being all higher than distance threshold, it is impossible to obtain distortionless closure bone contours by interpolation；
The pre- reference template refers to that interpolated rear formed carries unnecessary branch line and multiloop closure bone contours；
The non-reference masterplate meets following at 2 points：1. do not close or closed is bad, 2. meet closure or closed most
Get well but feeder number is more than 1 after cubic spline interpolation, the unnecessary branch line of removal.
It is described do not close or bone contours that closed is bad refer to contour line exist between breakpoint and adjacent breakpoint it is European away from
From being all higher than distance threshold, it is impossible to obtain distortionless closure bone contours by interpolation.
As shown in figure 4, left figure with 3 dashed circles significantly represent breakpoint between distance meet distance threshold T, it is possible to use three
Secondary sample interpolation；Right figure is result schematic diagram after interpolation.
As shown in figure 5, there is unnecessary branch line after significantly representing three sample interpolations with 3 dashed circles in left figure；Right figure is
The result schematic diagram after removing unnecessary branch line.
As shown in fig. 6, filtering out reference template flow chart from pre- reference template for objective contour pretreatment stage, specifically
Step is：
Step q01：All intersection points that there is branch line and coordinate is recorded in the pre- reference template of search positioning, if failure is positioned
Think that the pre- reference template, without unnecessary branch line, continues step q03；Otherwise step q02 is continued；
Step q02：According to orientation, track every branch of each intersection point and record follow-up summit, if follow-up summit
Then think that the branch line belongs to cut edge class for breakpoint, it is unnecessary that all vertex values of the branch line directly are set into 0 i.e. removable cut edge class
Branch line；Think that the branch line belongs to intersecting class and records coordinate if being another intersection point if follow-up summit；
Step q03：Calculate pre- reference template feeder number and update intersection point record sheet, be 0 by number of hits in pre- reference template
And feeder number is labeled as reference template for 1 pre- reference template, otherwise labeled as non-reference masterplate.
Step q04：Repeat step q01 to step q03 is straight to cause to filter out all ginsengs for meeting condition from pre- reference template
Examine masterplate.
The branch line can be divided into cut edge, intersecting two classes, due to machinery equipment aging, picture noise, bone surface it is tiny
What the factors such as gap were caused.
The cut edge refers to that branch line is path and certain end point is breakpoint.
It is described intersecting to refer to that branch line is loop.
The unnecessary branch line refers to there is intersection point in closure bone contours, and cut edge is branched into from a certain intersection point.
The reference template refers to filter out from pre- reference template without unnecessary branch line and occurred without single time of intersecting situation
Road bone contours.
As shown in fig. 7, describing same type bone template, non-reference masterplate, reference template triadic relation.
The same type bone template refers to identical with Medical faultage image position and obtained after joint is separated
Known skeleton data, the single loop profile with the interlamellar spacing such as Medical faultage image can be obtained by contour resampling
Line collection.
The characteristic matching refers to make similitude between correspondence reference template and same type bone template using method of characteristic
Judge, so as to find the same type bone template most matched with non-reference template, concretely comprise the following steps：
Step t01：Profile point on reference template and same type bone template is utilized into the minimum method polymerization of mean error
Non-significant characteristic point, retains identical quantative attribute point using threshold value in two templates；
Step t02：Both, which are obtained, using method of characteristic matches starting point；
Step t03：It can be found if it can obtain best match equidirectional with adjacent non-reference template contours the most
The same type bone template contours of matching；Otherwise same type bone template contour first sampling point is adjusted, continues step S04；
Step t04：Using center of gravity as reference, reference template and same type bone template affine transformation equation are calculated；
Step t05：The template adjacent with the same type bone template, as the non-reference template adjacent with reference template are most
For matching；
Step t06：Repeat step t01 to step t05, finds the same type bone mould that all non-reference templates are matched the most
Plate.
The method of characteristic refers to calculate both characteristic functions, using amplitude maximal point as reference template matching starting point,
This time matching that all characteristic points of loopy moving obtain minimum extreme value difference in same type bone template is best match, this time
It is same type bone template matches starting point to match starting point.
As shown in figure 8, the affine transformation refers to the matching starting point obtained using step t02, using reference template as base
Standard, same type bone template is done rotate, translation, alignd so that both starting points be done to center of gravity line；It is secondary to utilize
The non-reference template contours that the affine transformation can be referred to alignment step u02 and same type bone template contours.
As shown in Fig. 9-Figure 10, upper left corner subgraph is non-reference template, and various situations not phase is significantly represented with dashed circle
Close the position that contour line occurs；Lower left corner subgraph is neighboring reference template；Lower right corner subgraph is having for being matched with reference template
Same type bone template；Upper right corner subgraph is the same type bone template on equidirectional, the template with the subgraph of the upper left corner
Non-reference template is matched the most.
The uncorrelated contour line is made up of following 2 class situation：1. the intersecting class that equivalent steps S03 is mentioned, due to machine
The factors such as ageing equipment, picture noise, the fine gap of bone surface cause bone contours intersecting situation occur；2. due to medical science
Being many bones and bone contours adjacencies occur for faultage image scanning, eliminates the uncorrelated contour line of above two and specifically walks
Suddenly it is：
Step v01：If not there is intersection point P in reference template, then situation is 1. planted for described in right 8（9）；Calculate and hand over
Point P and each subsequent point P+1 vector value d, subsequent point P+ is arrived using corresponding points P ' in public angle calcu-lation same type bone template
1 ' vector value c, removes the d maximum with the c deviations of directivity；
Step v02：If not reference template is multiloop and is then 2. to plant situation described in right 8 without breakpoint（As schemed
10）；Each distance between center of gravity and reference template center of gravity in non-reference template is calculated, the profile farther out from reference template center of gravity is removed
Line, you can bone contours line needed for retaining, realizes bone quick separating；
The public angle refers to for inquiring about two adjacent profiles or through the corresponding points pair in affine transformation rear profile, tool
Body step is：
Step u01：Make non-reference template and same type template bone contours are most matched with it and be denoted as A, B, barycentric coodinates note
Make C, C '；
Step u02：The affine equation obtained using step t04, profile A, B center of gravity are alignd；
Step u03：Certain point P to center of gravity C line PC and x-axis formation angle theta from A；
Step u04：Sent from B centers of gravity C ' with x-axis angle be all θ ray, ray and B intersect at point P ', then (P, P ')
As required point pair；
As shown in figure 11, the interpolation refers to using in adjacent reference template and most like same type bone template
Average vector at reference vector and non-reference template breakpoint itself is constrained jointly, obtain the interpolation direction of breakpoint subsequent point with
Determine interpolated coordinates.
The reference vector refers to due to having global similitude, adjacent interlayer bone contours to have sequence between same type bone
Subsequent point in row, therefore same type bone template, neighboring reference template with non-reference template intersection point/breakpoint P corresponding points can
It is used to differentiating/instructing that subsequent point may moving direction as reference vector.
The average vector refers to that P is breakpoint situation, because bone contours have continuity and flatness, therefore non-reference
The vector of template breakpoint P n front and continued point has certain guidance meaning to point P subsequent point coordinate direction, it is possible to use average
Vector estimates point P+1 may moving direction.
The priori interpolation, is concretely comprised the following steps：
Step e01：Its vector between preceding n summit is done distance weighted flat by positioning to a certain breakpoint P of non-reference template
, vectorial a is obtained as the most probable direction discreet value of P+1 points；
Step e02：On step v01 bases, do similar to step v02 operations, obtain non-reference template and same type
Bone template corresponding points, to (P, P "), and are calculated corresponding points between (P, P ') and the reference template adjacent with non-reference template
Reference vector b=P ' P+1 ', c=P " P+1 "；
Step e03：Using average vector a, reference vector b and c result of weighted average as P point subsequent points P+1 interpolation
Direction, repeats this step and directly causes subsequent point P+1 to be non-breakpoint；
Step e04：Repeat step e01 and step e03, directly causes all contour line cut edge connections to finish；
First example is pre-processed below by the inventive method, then repaired into example by priori closest
The process of the reference template of actual profile.
Embodiment one：
It is preprocessed to obtain a non-reference template as shown in Figure 12-1., because non-reference template is not handed in case study on implementation
Point, but have breakpoint, therefore be according to the most like same type bone template obtained after affine transformation and adjacent reference template
Priori, does cut edge completion operation（As a result such as Figure 12-2.）.Then the unnecessary branch line that cut edge completion operation is produced is removed（Knot
Fruit such as Figure 12-3.）.Although without breakpoint in current outline, occurring in that intersection point and feeder number being not 1, be still non-ginseng
Examine template.Therefore removing uncorrelated contour line operation is also continued executing with, according to there is intersection point situation（1. plant situation）, with most phase
As same type bone template be that priori removes unnecessary branch；According to without intersection point without breakpoint situation（2. plant situation）, with
Neighboring reference template is that priori removes redundant loop（As a result such as Figure 12-3.）.
The general principle and principal character and advantages of the present invention of the present invention has been shown and described above.The technology of the industry
Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the simply explanation described in above-described embodiment and specification is originally
The principle of invention, without departing from the spirit and scope of the present invention, various changes and modifications of the present invention are possible, these changes
Change and improvement all fall within the protetion scope of the claimed invention.The claimed scope of the invention by appended claims and its
Equivalent thereof.
Claims (10)
1. the Medical Slice Images closure bone contours computational methods based on priori, it is characterised in that：Utilize list
Loop bone contours, existing same type bone template contours are that priori is guided jointly, and other doctors are obtained automatically
Learn the process that faultage image closes bone contours；All Medical Slice Images closure bone contours are calculated, it is calculated
Method comprises the following steps：
（1）The bone contours that all Medical Slice Images are extracted pre-process and are categorized into reference template and non-ginseng
Template is examined, the step includes as follows：
Step S01：Medical Slice Images are read in successively, and the bone wheel of Medical Slice Images is obtained using Canny methods
Wide map, the continuous or discrete bone wheel of some of every Medical Slice Images is obtained with reference to edge following algorithm
It is wide；
Step S02：By obtained bone contours referred to as objective contour, if can be searched out from objective contour close to closure
Bone contours, then by the use of cubic spline interpolation connection breakpoint obtain closure bone contours be used as pre- reference template；
Otherwise it is labeled as non-reference template；Wherein, the bone contours close to closure refer to that contour line has breakpoint and adjacent
Euclidean distance is respectively less than distance threshold T between breakpoint, and distortionless closure bone contours are obtained by interpolation；
Step S03：The unnecessary branch line produced in pre- reference template due to interpolation is removed, by the pre- reference template mark of single loop
It is designated as reference template；Otherwise it is labeled as non-reference template；
（2）Non-reference template and same type bone template corresponding relation are set up, the step includes as follows：
Step S04：Same type bone template contour interval is set, resampling obtains one group of closure bone contours；
Step S05：Make similitude judgement between correspondence reference template and same type bone template using method of characteristic, from
And find the same type bone template with non-reference template matches；
（3）Using neighboring reference template, same type bone template as priori, completion cut edge simultaneously removes non-reference mould
Incoherent contour line in plate, the step includes as follows：
Step S06：Uncorrelated contour line situation is classified；
Step S07：Using the reference vector in public angle calcu-lation same type bone template, neighboring reference template and
Constraint calculates the interpolated coordinates of breakpoint subsequent point with completion cut edge to non-reference template breakpoint average vector jointly；
Step S08：Removed by the use of the reference vector in same type bone template, neighboring reference template as priori
The uncorrelated contour line occurred；
Step S09：Repeat step S07 to step S08 is until non-reference template turns into single loop closure bone contours simultaneously
It is marked as reference template.
2. the Medical Slice Images closure bone contours computational methods according to claim 1 based on priori,
It is characterized in that：In step S02, the closure bone contours refer to one or one group of bone contours is closure
State；
Pre- reference template refers to that interpolated rear formed carries unnecessary branch line and multiloop closure bone contours；
Inc bone contours refer to that contour line has Euclidean distance between breakpoint and adjacent breakpoint and is all higher than distance
Threshold value, it is impossible to which distortionless closure bone contours are obtained by interpolation.
3. the Medical Slice Images closure bone contours computational methods according to claim 1 based on priori,
It is characterized in that：In the step S02, if the bone contours close to closure can be searched out from objective contour,
Then obtain closing bone contours using cubic spline interpolation connection breakpoint, concretely comprise the following steps：
Step i01：Search for the breakpoint number of current layer bone contours and record coordinate position, be considered to close if without breakpoint
The bone contours of conjunction；
Step i02：The each breakpoint neighborhood of this layer of bone contours is searched for one by one judges whether other contour line breakpoints,
The Euclidean distance between two breakpoints are calculated in the presence of if；
Step i03：Given distance threshold T, if Euclidean distance is respectively less than distance threshold T between all adjacent breakpoints,
It is considered the bone contours close to closure, breakpoint is connected using cubic spline interpolation；
Step i04：Repeat step i01 to step i03 is until find all bone contours close to closure, and marked
Pre- reference template is designated as, pre- reference template numbering is recorded.
4. the closure bone contours computational methods of the Medical Slice Images based on priori according to claim 1,
It is characterized in that：In step S03, the branch line is divided into cut edge, intersecting two classes, and cut edge refers to that branch line is path
It is intersecting to refer to that branch line is loop and certain end point is breakpoint；Unnecessary branch line, which refers to exist in closure bone contours, to be handed over
Point, cut edge is branched into from a certain intersection point；Reference template refers to filter out without unnecessary from pre- reference template
Branch line and the single loop bone contours for occurring without intersecting situation.
5. the Medical Slice Images closure bone contours computational methods according to claim 1 based on priori,
It is characterized in that：It is described to remove the unnecessary branch line produced in pre- reference template due to interpolation in the step S03,
The pre- reference template of single loop is labeled as reference template, concretely comprised the following steps：
Step q01：All intersection points that there is branch line and coordinate is recorded in the pre- reference template of search positioning, if positioning failure
Then think that the pre- reference template, without unnecessary branch line, continues step q03；Otherwise step q02 is continued；
Step q02：According to orientation, track every branch of each intersection point and record follow-up summit, if follow-up top
Put and then think that the branch line belongs to cut edge class for breakpoint, all vertex values of the branch line directly are set into 0 can remove this
The unnecessary branch line of cut edge class；Think that the branch line belongs to intersecting class and records coordinate if being another intersection point if follow-up summit；
Step q03：Calculate pre- reference template feeder number and update intersection point record sheet, be by number of hits in pre- reference template
0 and feeder number for 1 pre- reference template be labeled as reference template, otherwise labeled as non-reference template；
Step q04：Repeat step q01 all meets condition to step q03 until being filtered out from pre- reference template
Reference template.
6. the closure bone contours computational methods of the Medical Slice Images based on priori according to claim 1,
It is characterized in that：Same type bone template described in step S04 refer to it is identical with Medical faultage image position and
The known skeleton data obtained after joint is separated, can obtain disconnected with Medical by contour resampling
The single loop contour line collection of the interlamellar spacings such as tomographic image.
7. the closure bone contours computational methods of the Medical Slice Images based on priori according to claim 1,
It is characterized in that：Made described in step S05 using method of characteristic between correspondence reference template and same type bone template
Similitude judges, so as to find the same type bone template with non-reference template matches, concretely comprises the following steps：
Step t01：Profile point on reference template and same type bone template is utilized into the minimum method polymerization of mean error
Non-significant characteristic point, retains identical quantative attribute point using threshold value in two templates；
Step t02：Both, which are obtained, using method of characteristic matches starting point；
Step t03：It can be matched if it can obtain best match in equidirectional find with adjacent non-reference template contours
Same type bone template contours；Otherwise same type bone template contour first sampling point is adjusted, continues step
t04；
Step t04：Using center of gravity as reference, reference template and same type bone template affine transformation equation are calculated；
Step t05：The template adjacent with the same type bone template, is the non-reference template adjacent with reference template
Matching；
Step t06：Repeat step t01 to step t05, finds the same type bone template of all non-reference template matches.
8. the closure bone contours computational methods of the Medical Slice Images based on priori according to claim 4,
It is characterized in that：In the step S06, the uncorrelated contour line is made up of following 2 class situation：1. it is equivalent
The intersecting class mentioned in step S03, due to machinery equipment aging, picture noise, bone surface fine gap
Factor cause bone contours intersecting situation occur；2. it is many bones and goes out due to what Medical Slice Images were scanned
Existing bone contours adjacencies.
9. the Medical Slice Images closure bone contours computational methods according to claim 7 based on priori,
It is characterized in that：Public angle described in step S07 is used to inquire about non-reference template and matched same type bone
Corresponding points in template contours, are concretely comprised the following steps：
Step u01：Non-reference template and matched same type bone template contours are made to be denoted as A, B, center of gravity respectively
Coordinate is denoted as C, C ' respectively；
Step u02：The affine transformation equation obtained using step t04, profile A, B center of gravity are alignd；
Step u03：A certain breakpoint P to center of gravity C line PC and x axles formation angle theta from A；
Step u04：Sent from B centers of gravity C ' with x axle clamps angle be all θ ray, ray and B intersect at point P ',
Then (P, P ') it is required point pair；
Step u05：The corresponding points in neighboring reference template and non-reference template are calculated using step u01 to step u04
It is right（P, P "）；
The utilization non-reference template breakpoint average vector is constrained jointly, calculates the interpolated coordinates of breakpoint subsequent point to mend
Full cut edge, is concretely comprised the following steps：
Step e01：Positioning is done distance to a certain breakpoint P of non-reference template, by its vector between preceding n summit and added
Weight average, obtains vectorial a as the most probable direction discreet value of P+1 points；
Step e02：The corresponding points P ' and P " obtained using step u01- steps u05, calculates reference vector b=P '
P+1 ', c=P " P+1 "；
Step e03：Using average vector a, reference vector b and c average weighted result are used as P point subsequent points P+1
Interpolation direction；
Step e04：Repeat step e01 and step e03, until the connection of all contour line cut edges is finished.
10. the Medical Slice Images closure bone contours computational methods according to claim 8 based on priori,
It is characterized in that：In the step S08, the ginseng in the utilization same type bone template, neighboring reference template
Examine vector and the uncorrelated contour line occurred is removed as priori, concretely comprise the following steps：
Step v01：If not there is intersection point in reference template, then situation is 1. planted in claim 8；
Step v02：If not reference template is multiloop and is then 2. to plant situation in claim 8 without breakpoint；Meter
Each distance between center of gravity and reference template center of gravity in non-reference template is calculated, the wheel farther out from reference template center of gravity is removed
Profile, you can bone contours line needed for retaining, realizes bone quick separating.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510239512.8A CN104809736B (en) | 2015-05-12 | 2015-05-12 | Medical Slice Images closure bone contours computational methods based on priori |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510239512.8A CN104809736B (en) | 2015-05-12 | 2015-05-12 | Medical Slice Images closure bone contours computational methods based on priori |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104809736A CN104809736A (en) | 2015-07-29 |
| CN104809736B true CN104809736B (en) | 2017-09-12 |
Family
ID=53694541
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510239512.8A Expired - Fee Related CN104809736B (en) | 2015-05-12 | 2015-05-12 | Medical Slice Images closure bone contours computational methods based on priori |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104809736B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107993277A (en) * | 2017-11-28 | 2018-05-04 | 河海大学常州校区 | Damage location artificial skelecton patch formation model method for reconstructing based on priori |
| CN109949341A (en) * | 2019-03-08 | 2019-06-28 | 广东省智能制造研究所 | A kind of pedestrian target tracking based on human skeleton structured features |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104881875A (en) * | 2015-06-09 | 2015-09-02 | 河海大学常州校区 | Priori-knowledge-based reconstruction method of closed skeleton contour in medical fault image |
| CN106408648A (en) * | 2015-08-03 | 2017-02-15 | 青岛海信医疗设备股份有限公司 | Medical-tissue slice-image three-dimensional reconstruction method and equipment thereof |
| CN105653887A (en) * | 2016-03-24 | 2016-06-08 | 河海大学常州校区 | Method for recovering damaged skeleton model based on parametric template |
| CN108596917B (en) * | 2018-04-19 | 2021-10-22 | 湖北工业大学 | Method for extracting target main skeleton line |
| CN113313690B (en) * | 2021-05-31 | 2022-08-09 | 齐鲁工业大学 | Same root region intelligent detection method for leather fiber MCT sequence slice image |
| CN114638828B (en) * | 2022-05-18 | 2022-09-23 | 数聚(山东)医疗科技有限公司 | Radiological image intelligent segmentation method based on computer vision |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102509338A (en) * | 2011-09-20 | 2012-06-20 | 北京航空航天大学 | Contour and skeleton diagram-based video scene behavior generation method |
| CN103632371A (en) * | 2013-12-06 | 2014-03-12 | 河海大学常州校区 | Compatibility mesh segmentation based skeleton parameter computation method |
| CN104200524A (en) * | 2014-09-11 | 2014-12-10 | 河海大学常州校区 | Three-dimensional mesh skeleton model averaging method oriented to bone plate design |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2886057B1 (en) * | 2013-10-24 | 2020-01-01 | Samsung Electronics Co., Ltd | Medical imaging apparatus and method of reconstructing medical image |
-
2015
- 2015-05-12 CN CN201510239512.8A patent/CN104809736B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102509338A (en) * | 2011-09-20 | 2012-06-20 | 北京航空航天大学 | Contour and skeleton diagram-based video scene behavior generation method |
| CN103632371A (en) * | 2013-12-06 | 2014-03-12 | 河海大学常州校区 | Compatibility mesh segmentation based skeleton parameter computation method |
| CN104200524A (en) * | 2014-09-11 | 2014-12-10 | 河海大学常州校区 | Three-dimensional mesh skeleton model averaging method oriented to bone plate design |
Non-Patent Citations (2)
| Title |
|---|
| 基于先验形状信息核图割模型的肝脏分割方法;罗清 等;《计算机工程与设计》;20140630;第35卷(第6期);全文 * |
| 面向RP骨骼CT图像的轮廓提取与精简;任国成 等;《山东大学学报》;20040630;第34卷(第3期);全文 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107993277A (en) * | 2017-11-28 | 2018-05-04 | 河海大学常州校区 | Damage location artificial skelecton patch formation model method for reconstructing based on priori |
| CN107993277B (en) * | 2017-11-28 | 2019-12-17 | 河海大学常州校区 | priori knowledge-based reconstruction method of artificial bone repair model of damaged part |
| CN109949341A (en) * | 2019-03-08 | 2019-06-28 | 广东省智能制造研究所 | A kind of pedestrian target tracking based on human skeleton structured features |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104809736A (en) | 2015-07-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104809736B (en) | Medical Slice Images closure bone contours computational methods based on priori | |
| CN110163853B (en) | Edge defect detection method | |
| Szeptycki et al. | A coarse-to-fine curvature analysis-based rotation invariant 3D face landmarking | |
| CN109544612B (en) | Point cloud registration method based on feature point geometric surface description | |
| CN109446895B (en) | Pedestrian identification method based on human head features | |
| CN104732536A (en) | Sub-pixel edge detection method based on improved morphology | |
| CN104299008A (en) | Vehicle type classification method based on multi-feature fusion | |
| CN113706537B (en) | Irregular workpiece detection method based on machine vision | |
| CN105740945A (en) | People counting method based on video analysis | |
| CN104091155A (en) | Rapid iris positioning method with illumination robustness | |
| CN105574527A (en) | Quick object detection method based on local feature learning | |
| CN112464847B (en) | Human body action segmentation method and device in video | |
| CN114299275A (en) | Hough transform-based license plate inclination correction method | |
| CN109781737B (en) | Detection method and detection system for surface defects of hose | |
| CN106504262A (en) | A kind of small tiles intelligent locating method of multiple features fusion | |
| CN101826209A (en) | Canny model-based method for segmenting three-dimensional medical image | |
| CN104299009A (en) | Plate number character recognition method based on multi-feature fusion | |
| CN104077775A (en) | Shape matching method and device combined with framework feature points and shape contexts | |
| CN106327451A (en) | Image restorative method of ancient animal fossils | |
| CN113393426A (en) | Method for detecting surface defects of rolled steel plate | |
| CN104298993B (en) | A kind of bar number positioning and recognition methods suitable under complex scene along track | |
| CN116358449A (en) | Aircraft rivet concave-convex amount measuring method based on binocular surface structured light | |
| CN111986157A (en) | Digital pathological image quality evaluation system | |
| GB2503331A (en) | Aligning garment image with image of a person, locating an object in an image and searching for an image containing an object | |
| CN104881875A (en) | Priori-knowledge-based reconstruction method of closed skeleton contour in medical fault image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate 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: 20170912 Termination date: 20210512 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |