CN106127849B - Three-dimensional fine vascular method for reconstructing and its system - Google Patents
Three-dimensional fine vascular method for reconstructing and its system Download PDFInfo
- Publication number
- CN106127849B CN106127849B CN201610305637.0A CN201610305637A CN106127849B CN 106127849 B CN106127849 B CN 106127849B CN 201610305637 A CN201610305637 A CN 201610305637A CN 106127849 B CN106127849 B CN 106127849B
- Authority
- CN
- China
- Prior art keywords
- repaired
- dimensional fine
- fine vascular
- image
- blocks
- 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
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2210/00—Indexing scheme for image generation or computer graphics
- G06T2210/41—Medical
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The present invention discloses a kind of three-dimensional fine vascular method for reconstructing.The three-dimensional fine vascular method for reconstructing is the following steps are included: Step 1: be loaded into three-dimensional fine vascular original image；Step 2: the image preprocessing based on vessel properties；Step 3: the characteristic matching based on vessel properties；Step 4: image completion and BORDER PROCESSING.The present invention also discloses a kind of three-dimensional fine vascular reconstructing systems.Using three-dimensional fine vascular method for reconstructing provided by the invention and its system, the filling region for repairing deleted areas is obtained by mapping position relationship and calculating whole matching degree, deleted areas is repaired by filling region and obtains complete high precision image, and the accuracy that three-dimensional fine vascular is rebuild is high.
Description
Technical field
The present invention relates to the crossing domains of Digital Image Processing and medical imaging, and in particular, to a kind of fine blood of three-dimensional
Pipe method for reconstructing and its system.
Background technique
In field of biomedicine, different imaging techniques is observed that different phenomenon and data, can comprehensively consider
The advantage of different scanning equipment obtains the subtle sight image data under a variety of scales of true biologic soft tissue.CT image is to people
The scan image of body a part, can be to imaging of tissue such as blood vessel, tumours, although imaging is complete, its shortcoming is that scanning
The details quality of data is obviously not so good as histotomy, and the organization type that can be differentiated is limited.Utilize optics and electron microscope
Histotomy is imaged, the subtle sight multi-scale image number that sharpness of definition is higher, institutional framework type is richer can be obtained
According to, but easily there is regional area missing in the image data obtained, therefore, it is necessary to the part in high-precision scan image
Absent region carries out reconstruction.
Three-dimensional fine vascular image in medicine has similar tree-like complex topology, and especially there are bifurcateds on vessel branch
Region, and how much different the numbers of branches of bifurcation region is, while the thickness dimensional variation of blood vessel is big, in visual research,
Volume drawing to blood vessel is always a stubborn problem, and needs in surgical operation to main vascular morphology and accurately retouch
It states, to assist doctor to select rational therapy method in time, for example from the three-dimensional structure of blood vessel and determines lesion locations, do
Quick diagnosis out；Intuitive reference frame is provided for surgical planning.Therefore once localized loss occurs in three-dimensional fine vascular image,
It just will affect the diagnosis of doctor.
It a kind of is rebuild for the high accuracy three-dimensional fine vascular image for having deleted areas thus, it is necessary to provide
Technical solution.
Summary of the invention
The present invention provides a kind of three-dimensional fine vascular method for reconstructing and its system, passes through mapping position relationship and calculates whole
Matching degree obtains the filling region for rebuilding deleted areas, repairs the complete high-precision of deleted areas acquisition by filling region and schemes
Picture, the accuracy that three-dimensional fine vascular is rebuild are high.
A kind of three-dimensional fine vascular method for reconstructing, comprising the following steps:
Step 1: being loaded into three-dimensional fine vascular original image:
It is described three-dimensional fine vascular original image include scan same target complex pattern and reference picture to be repaired, it is described to
Repairing image is the blood-vessel image for having deleted areas, and the reference picture is complete blood-vessel image, the complex pattern to be repaired
Precision is higher than the precision of the reference picture；
Area to be repaired is specified according to the complex pattern to be repaired, and defines the multiblock to be repaired comprising the area to be repaired,
Determine that position is corresponding in the reference picture, size is identical with reference to mapping block according to the multiblock to be repaired, it is described to be repaired
Area to be repaired in block corresponds to the reference mapping area with reference in mapping block；
Step 2: the image preprocessing based on vessel properties:
The three-dimensional fine vascular original image of loading is analyzed, the angiosomes for needing to enhance are determined, to institute
Stating three-dimensional fine vascular original image is enhanced, and three-dimensional fine vascular enhancing image, the three-dimensional fine vascular enhancing are obtained
Image includes enhanced complex pattern to be repaired and enhanced reference picture；
Step 3: the characteristic matching based on vessel properties:
Enhance image according to the three-dimensional fine vascular, calculate whole matching degree, obtains the reference mapping block described
Blocks and optimal matching blocks in complex pattern to be repaired, the reference mapping area with reference in mapping block correspond in the blocks and optimal matching blocks
Filling region；
Step 4: image completion and BORDER PROCESSING:
By the corresponding filling of the filling region of the blocks and optimal matching blocks to the area to be repaired of the multiblock to be repaired, and to boundary
It is smoothed, completes reconstruction.
In a kind of preferred embodiment of three-dimensional fine vascular method for reconstructing provided by the invention, the step 2 includes such as
Lower step:
The three-dimensional fine vascular original image is analyzed based on Hessian matrix: choosing Hessian matrix to institute
It states three-dimensional fine vascular original image and does convolution algorithm, acquire characteristic value and feature vector；
It establishes angiosomes characteristic function: establishing area vasculosa based on vessel properties and according to the characteristic value and feature vector
Characteristic of field function；
It determines and needs the angiosomes enhanced: determining the area vasculosa for needing to enhance according to the angiosomes characteristic function
Domain, and then the three-dimensional fine vascular original image is enhanced, obtain three-dimensional fine vascular enhancing image.
It is described to be based on Hessian in a kind of preferred embodiment of three-dimensional fine vascular method for reconstructing provided by the invention
The step of matrix analyzes the three-dimensional fine vascular original image include the following:
Calculate the second-order partial differential coefficient and mixed partial derivative of each pixel in the three-dimensional fine vascular original image；
Hessian matrix and the three-dimensional fine vascular original graph are formed with the second-order partial differential coefficient and mixed partial derivative of calculating
As doing convolution algorithm；
Acquire eigenvalue λ1、λ2And λ3With feature vector γ corresponding with the characteristic value1、γ2And γ3, and | λ1|<|λ2|<
|λ3|。
In a kind of preferred embodiment of three-dimensional fine vascular method for reconstructing provided by the invention, the angiosomes feature
Function are as follows:
Define RA=| λ2|/|λ3|,Wherein, a, c and d
Respectively planar structure parameter, background differentiation parameter and three-dimensional blood vessel structure parameter, k1、k2And k3Respectively along the feature
Vector γ1、γ2、γ3The size of the rate of gray level of the gradient direction of opposite direction.
In a kind of preferred embodiment of three-dimensional fine vascular method for reconstructing provided by the invention, the step 3 includes such as
Lower step:
The complex pattern to be repaired is divided into multiple candidate blocks identical with the reference mapping block size, calculates the time
Select block and the global registration degree S with reference to mapping blockM；
Calculate vessel segment similarity ST；
According to the global registration degree SMWith the vessel segment similarity STCalculate whole matching degree Sves, the whole matching
Spend Sves=μ SM+ηST, wherein μ and η is respectively SMAnd STWeight coefficient, and+η=1 μ；
According to the whole matching degree Sves, obtain the Optimum Matching with reference to mapping block in the complex pattern to be repaired
Block, the reference mapping area with reference in mapping block correspond to the filling region in the blocks and optimal matching blocks.
In a kind of preferred embodiment of three-dimensional fine vascular method for reconstructing provided by the invention, the calculating vessel segment phase
Like degree STThe step of include the following steps:
Based on the angiosomes for needing to enhance described in method refinement step two are successively stripped, single pixel center line is extracted, by
The three-dimensional fine vascular enhancing image obtains three-dimensional fine vascular binary map, and the three-dimensional fine vascular binary map includes described
The binary map of the binary map of complex pattern to be repaired and the reference picture；
The end point set that the candidate blocks are extracted according to the binary map of the complex pattern to be repaired, according to the two of the reference picture
Value figure extracts the end point set with reference to mapping block, seeks endpoint feature vector respectively；
Vessel segment similarity S is calculated according to the endpoint feature vectorT。
In a kind of preferred embodiment of three-dimensional fine vascular method for reconstructing provided by the invention, the step 4 includes such as
Lower step:
By the corresponding filling of the filling region of the blocks and optimal matching blocks to the area to be repaired of the multiblock to be repaired；
The boundary of the filling region is subjected to interpolation iteration, is constantly approached until convergence, by the filling region and
Reconstruction is completed in the edge smoothing transition of the area to be repaired.
The present invention also provides a kind of three-dimensional fine vascular reconstructing systems, including getImage module, for being loaded into three-dimensional essence
Thin blood vessel original image, the three-dimensional fine vascular original image include scanning the complex pattern to be repaired of same target and with reference to figure
Picture, the complex pattern to be repaired are the blood-vessel image of tool missing, and the reference picture is complete blood-vessel image, the figure to be repaired
The precision of picture is higher than the reference picture；And area to be repaired is specified according to the complex pattern to be repaired, and define comprising it is described to
The multiblock to be repaired of restoring area, corresponding, the identical ginseng of size according to the determining position in the reference picture of the multiblock to be repaired
Mapping block is examined, the area to be repaired in the multiblock to be repaired corresponds to the reference mapping area with reference in mapping block；
Image pre-processing module, for described image insmod loading the three-dimensional fine vascular original image into
Row analysis, determines the angiosomes for needing to enhance, and enhances the three-dimensional fine vascular original image, obtains three-dimensional fine
Blood vessel enhances image, and the three-dimensional fine vascular enhancing image includes enhanced complex pattern to be repaired and enhanced with reference to figure
Picture；
Characteristic matching module, the three-dimensional fine vascular for being obtained according to described image preprocessing module enhance image, meter
Whole matching degree is calculated, the blocks and optimal matching blocks with reference to mapping block in the complex pattern to be repaired are obtained, it is described to refer to mapping block
In reference mapping area correspond to the filling region in the blocks and optimal matching blocks；
Processing module is filled, the filling region of the blocks and optimal matching blocks for obtaining the characteristic matching module is corresponding
It fills to the area to be repaired of the multiblock to be repaired, and boundary is smoothed, complete reconstruction.
In a kind of preferred embodiment of three-dimensional fine vascular reconstructing system provided by the invention, described image pre-processes mould
Block includes:
First arithmetic element chooses Hessian matrix and does convolution algorithm to the three-dimensional fine vascular original image, acquires
Characteristic value and feature vector；
Function creation unit, the characteristic value and feature vector acquired according to first arithmetic element are simultaneously based on vessel properties
Establish angiosomes characteristic function；
Blood vessel enhancement unit, the angiosomes characteristic function established according to the function creation unit determine what needs enhanced
Angiosomes, and then the three-dimensional fine vascular original image is enhanced, obtain three-dimensional fine vascular enhancing image.
In a kind of preferred embodiment of three-dimensional fine vascular reconstructing system provided by the invention, the characteristic matching module
Including the first computing unit, the second computing unit and third computing unit, it is similar to be respectively used to calculating global registration degree, vessel segment
Degree and whole matching degree.
Compared to the prior art, three-dimensional fine vascular method for reconstructing provided by the invention and its system, have beneficial below
Effect:
One, three-dimensional fine vascular method for reconstructing provided by the invention has carried out characteristic value to the angiosomes that needs enhance and has mentioned
Take, three-dimensional fine vascular original image be subjected to image enhancement, convenient for derived from same blood vessel sample complex pattern to be repaired and
Reference picture carries out structure matching, improves the efficiency that three-dimensional fine vascular is rebuild, and can guarantee precision；
Two, the present invention has redefined the angiosomes characteristic function based on Hessian matrix, it is contemplated that three-dimensional fine blood
The grayscale information of pipe original image can effectively improve the effect of image enhancement；Simultaneously consider blood vessel bifurcation region and
The influence of the individual differences such as the biggish region of radian, can be good at solving the problems, such as individual difference, even if in special feelings
Condition will not judge incorrectly, and have wider applicability；
Three, three-dimensional fine vascular method for reconstructing provided by the invention is formed based on vessel properties, it is contemplated that blood vessel structure
Tubular character and stronger self-similarity, thus of the guidance derived from the complex pattern and reference picture to be repaired of same blood vessel sample
Match, improve the reliability and accuracy of blood vessel structure characteristic matching, further increases the precision that three-dimensional fine vascular is rebuild；
Four, the present invention is obtained with reference to mapping block in complex pattern to be repaired by the matching of complex pattern to be repaired and reference picture
Blocks and optimal matching blocks, boundary is smoothed again after corresponding filling, it is high to repair precision.
Detailed description of the invention
Fig. 1 is the Principles figure of three-dimensional fine vascular method for reconstructing provided by the invention；
Fig. 2 is the flow chart of three-dimensional fine vascular method for reconstructing provided by the invention；
Fig. 3 (a) is the reference picture in three-dimensional fine vascular original image provided by the invention；
Fig. 3 (b) is the complex pattern to be repaired in three-dimensional fine vascular original image provided by the invention；
Fig. 4 is the flow chart of step S2 in three-dimensional fine vascular method for reconstructing shown in Fig. 2；
Fig. 5 is the flow chart of step S21 in step S2 shown in Fig. 4；
Fig. 6 (a) is enhanced reference picture shown in Fig. 3 (a)；
Fig. 6 (b) is enhanced complex pattern to be repaired shown in Fig. 3 (b)；
Fig. 7 is the flow chart of step S3 in three-dimensional fine vascular method for reconstructing shown in Fig. 2；
Fig. 8 is the flow chart of step S32 in step S3 shown in Fig. 7；
Fig. 9 (a) is the reference picture after refinement shown in Fig. 6 (a)；
Fig. 9 (b) is the complex pattern to be repaired after refinement shown in Fig. 6 (b)；
Figure 10 is the flow chart of step S4 in three-dimensional fine vascular method for reconstructing shown in Fig. 2；
Figure 11 (a) is a kind of three-dimensional fine vascular structure figure to be repaired；
Figure 11 (b) is the three-dimensional fine vascular structure figure after repairing shown in Figure 11 (a)；
Figure 12 is the structural block diagram of three-dimensional fine vascular reconstructing system provided by the invention；
Figure 13 is the structural block diagram of image pre-processing module shown in Figure 12；
Figure 14 is the structural block diagram of characteristic matching module shown in Figure 12；
Figure 15 is the structural block diagram of filling processing module shown in Figure 12.
Specific embodiment
The utility model (invention) is described further below in conjunction with drawings and embodiments.Referring to Fig. 1, for this
The Principles figure of the three-dimensional fine vascular method for reconstructing provided is provided.Present invention technical problem to be solved is to repair tool missing
The three-dimensional fine vascular image at position, is denoted as complex pattern 1 to be repaired.The present invention also provides for repairing the complex pattern to be repaired 1
Reference picture 3.Specifically, under the premise of not destroying blood vessel sample, in order to obtain the microcosmic multi-scale image of blood vessel sample
Data are carried out Sequentially continuous slice to a blood vessel sample, are scanned and are obtained using the optics and electron microscope of high resolution
The complex pattern 1 to be repaired is obtained, since resolution ratio is higher, the range in energy scanned samples region is smaller, carries out to blood vessel sample subtle
The image data of multiple dimensioned scanning acquisition is seen there are deleted areas, i.e., the described complex pattern to be repaired 1 is the high-precision for having deleted areas
Blood-vessel image；Same blood vessel sample is directly scanned using Micro-CT scanning/MRI and Synchrotron Radiation Computed Tomography, obtains the reference picture 3, institute
Stating reference picture 3 is complete low precision blood-vessel image；This low precision and high-precision are relative concepts, i.e., the described complex pattern to be repaired
1 scanning accuracy is higher than the scanning accuracy of the reference picture 3.
T1, it specifies deleted areas described in the complex pattern to be repaired 1 to be area to be repaired 10, defines from the area to be repaired
Domain 10 extend outwardly acquisition cube extended region be multiblock 12 to be repaired；
T2, size corresponding to 12 position of multiblock to be repaired is obtained in the reference picture 3 by mapping position relationship
It is identical to refer to mapping block 32, the corresponding reference with reference in mapping block 32 in area to be repaired 10 in the multiblock 12 to be repaired
Mapping area 30；
T3, the search matching blocks and optimal matching blocks 52 similar with the reference mapping block 32 in the complex pattern 1 to be repaired, then
Filling region 50 corresponding with the area to be repaired 10 is obtained by mapping position relationship；
T4, the filling region 50 is corresponded into filling to the area to be repaired 10, the filling region 50 is repaired described
Deleted areas is smoothed the boundary of the filling region 50 and the area to be repaired 10, obtains nature transition
Texture image completes reconstruction, obtains complete high-precision three-dimensional fine vascular image.
Referring to Fig. 2, for the flow chart of three-dimensional fine vascular method for reconstructing provided by the invention.The present invention provides one kind three
Fine vascular method for reconstructing is tieed up, is included the following steps:
Step S1, it is loaded into three-dimensional fine vascular original image:
The three-dimensional fine vascular original image includes the complex pattern to be repaired 1 and reference picture 3 for scanning same target, described
Complex pattern 1 to be repaired is the blood-vessel image for having deleted areas, and the reference picture 3 is complete blood-vessel image, the figure to be repaired
As 1 precision is higher than the precision of the reference picture 3.
In the present embodiment, specifically, Fig. 3 (a) and Fig. 3 (b) is referred to, wherein Fig. 3 (a) is provided by the invention three
Tie up the reference picture in fine vascular original image；Fig. 3 (b) be in three-dimensional fine vascular original image provided by the invention to
Repair image.It should be noted that the present invention is to upper for the committed step in clearer description method provided by the invention
It states and has carried out appropriate interception derived from the reference picture of same blood vessel sample and complex pattern to be repaired, and be illustrated as example,
But it is not limited to image disclosed above.
Area to be repaired 10 is specified according to the complex pattern 1 to be repaired, and is defined to be repaired comprising the area to be repaired 10
Multiblock 12 determines that position is corresponding in the reference picture 3, size is identical according to the multiblock 12 to be repaired and refers to mapping block
32, the corresponding reference mapping area 30 with reference in mapping block 32 in area to be repaired 10 in the multiblock 12 to be repaired.
In the present embodiment, specifying deleted areas described in the complex pattern to be repaired 1 is area to be repaired 10, is defined from institute
State area to be repaired 10 extend outwardly acquisition cube extended region be multiblock 12 to be repaired.By mapping position relationship described
Acquisition is corresponding to 12 position of multiblock to be repaired in reference picture 3, size is identical refers to mapping block 32, the multiblock 12 to be repaired
In the corresponding reference mapping area 30 with reference in mapping block 32 in area to be repaired 10.
Step S2, based on the image preprocessing of vessel properties:
The three-dimensional fine vascular original image of loading is analyzed, the angiosomes for needing to enhance are determined, to institute
Stating three-dimensional fine vascular original image is enhanced, and three-dimensional fine vascular enhancing image, the three-dimensional fine vascular enhancing are obtained
Image includes enhanced complex pattern to be repaired 1 and enhanced reference picture 3.
Step S3, based on the characteristic matching of vessel properties:
Enhance image according to the three-dimensional fine vascular, calculate whole matching degree, obtains the reference mapping block 32 in institute
The blocks and optimal matching blocks 52 in complex pattern 1 to be repaired are stated, it is described described optimal with reference to the correspondence of reference mapping area 30 in mapping block 32
Filling region 50 in match block 52.
Step S4, image completion and BORDER PROCESSING:
The filling region 50 of the blocks and optimal matching blocks 52 is corresponded to and is filled to the area to be repaired 10 of the multiblock 12 to be repaired,
And boundary is smoothed, complete reconstruction.
Referring to Fig. 4, for the flow chart of step S2 in three-dimensional fine vascular method for reconstructing shown in Fig. 2.Due to acquiring equipment
The error of light intensity, in the medical images such as three-dimensional fine vascular the gray value of region of interest (i.e. angiosomes) may and it is unknown
It is aobvious, for the ease of being repaired to image, needs first to carry out enhancing processing to image, enhance the matching of key structure.
It can identify that edge, angle point and the flat site in image, structure tensor are in image according to structure tensor
Pixel utilize matrix organization data structure, form is Hessian matrix.The vessel graph special for shape feature
Picture, the characteristic of the Grey imaging Gaussian Profile of vascular cross-section can use structure tensor and effectively obtain feature needed for blood vessel, in turn
Feature enhancing is carried out to it.
The image preprocessing step S2 based on vessel properties includes:
Step S21, the three-dimensional fine vascular original image is analyzed based on Hessian matrix: chooses Hessian
Matrix does convolution algorithm to the three-dimensional fine vascular original image, acquires characteristic value and feature vector.
It is the flow chart of step S21 in step S2 shown in Fig. 4 please refer to Fig. 5.
The step S21 analyzed based on Hessian matrix the three-dimensional fine vascular original image includes:
Step S21-1, it is inclined that the second-order partial differential coefficient of each pixel and mixing in the three-dimensional fine vascular original image are calculated
Derivative；
Step S21-2, Hessian matrix is formed with the second-order partial differential coefficient of calculating and mixed partial derivative and the three-dimensional is fine
Blood vessel original image does convolution algorithm；
Step S21-3, eigenvalue λ is acquired1、λ2And λ3With feature vector γ corresponding with the characteristic value1、γ2And γ3,
And | λ1|<|λ2|<|λ3|。
Detailed process are as follows:
If Vx,Vy,VzThe partial derivative in the direction respectively x, y, z,Calculate described three
The second-order partial differential coefficient and mixed partial derivative of each pixel in fine vascular original image are tieed up, wherein Vσ=Gσ* V, * represent convolution calculation
Son obtains the linear enhancing in different scale σ by Gaussian function (1) and filters.When carrying out local characteristics analysis, with currently processed
Centered on pixel, in the image data for being presently in reason, taking half-breadth is the rectangular window of 3 σ, it is sufficient to include the straight of blood vessel
Diameter.
The Gaussian function (1) are as follows:
Structure tensor(being a three-dimensional Hessian matrix) is defined as follows formula (2):
WhereinIt is Descartes's inner product；There are a three-dimensional orthogonal matrix S, so thatWherein Λ=diag (λm) diagonal element beCharacteristic value, form matrix S each row to
Amount is structure tensorFeature vector, be denoted as λ respectively1、λ2And λ3, and meet | λ1|<|λ2|<|λ3|, with three characteristic values
One-to-one three feature vectors are denoted as γ respectively1、γ2And γ3, wherein γ1⊥λ2,γ1⊥λ3,γ2⊥λ3.Structure tensorEigenvalue λ1、λ2And λ3Pace of change of the image grayscale in individual features vector direction can be embodied.The edge of image
Intensity is big, can be quantified as structure tensorBiggish characteristic value.
According to three characteristic values | λ1|<|λ2|<|λ3|, characteristic value and corresponding shape and structure to matrix are divided as follows
Analysis:
a.|λ1|≈0,|λ2|≈0,|λ3| ≈ 0 can determine that as flat site；b.|λ1|≈0,|λ2|≈0,|λ3| > >
λ1, it is possible to determine that it is planar structure；|λ1|≈0,|λ2| > > | λ1|,|λ3| > > | λ1|, it is possible to determine that it is tubular structure；c.|
λ1| > > 0, | λ2| > > 0, | λ3| > > 0 can determine that as chondritic.
By above-mentioned analysis, for ideal three-dimensional fine vascular image, the relationship of characteristic value and feature vector should
Such as following formula (3):
Wherein λ2,λ3It is negative, illustrates that region of interest is highlight bar in three-dimensional fine vascular image and background is
Dark area；Two feature vector γ of matrix2,γ3It is the section of three-dimensional fine vascular representated by the face of composition, and it is minimum special
Feature vector γ corresponding to value indicative1What is represented is the trend of three-dimensional fine vascular, i.e., the extending direction of three-dimensional fine vascular.
Step S22, it establishes angiosomes characteristic function: being built based on vessel properties and according to the characteristic value and feature vector
Vertical angiosomes characteristic function.
Existing angiosomes enhance algorithm: according to the eigenvalue λ1、λ2、λ3With described eigenvector γ1、γ2、γ3,
First define variable RA、RBAnd S, respectively such as following formula (4), (5), (6):
RA=| λ2|/|λ3| (4)；
Variable RAValue can estimate a possibility that currently processed image belongs to tubular structure, value is bigger, more may belong to
In tubular structure；
Variable RBValue can estimate a possibility that currently processed image belongs to disk-like structure, value is bigger, more may belong to
In disk-like structure；
Variable S takes the quadratic sum of three characteristic values, can be belonged to the currently processed image of fuzzy Judgment structural relatively strong
Region.
According to the variable RA、RBAnd the existing angiosomes function that S is defined is following formula (7):
Wherein, a, b and c are respectively planar structure parameter, chondritic parameter and background differentiation parameter, rule of thumb with
And selected scale is chosen.
For true three-dimensional fine vascular image, blood vessel thickness and topology are all irregular, and blood vessel is of different thickness, for non-
Often thin blood vessel, crosspoint can almost be ignored, but for thicker blood vessel, bifurcation region just be can not be ignored.
Variable R is calculated according to formula (5)BWhen, if currently processed image is disk-like structure or bifurcation region, λ1It is and λ2、λ3
Sizable negative, under both of these case, variable RBValue it is larger；If currently processed image is the biggish area of blood vessel radian
Domain, RBValue also can be larger.
But when calculating existing angiosomes function according to formula (7), RBSmaller, currently processed image belongs to angiosomes
Weight is bigger, thus under special circumstances, judge that currently processed image will appear mistake according to formula (7), so should give for change
Ignored bifurcation region and the biggish region of blood vessel radian.
Present invention assumes that described eigenvector γ1、γ2、γ3Opposite direction respectively correspond as τ1,τ2,τ3, the feature to
Measure γ1、γ2、γ3It is directed toward vascular wall, under certain probability, τ1,τ2,τ3In have at least one direction be directed toward blood vessel extending direction,
If along τ1,τ2,τ3Wherein the variation of a direction grey scale pixel value slowly, then can be determined that as effective angiosomes.
Define the gradient direction along direction τFor scale σ, it is calculated along this gradient direction
Change rate size k, if three direction τ1,τ2,τ3Rate of gray level size be respectively k1,k2,k3, present invention definition is new to be become
Amount is blood vessel angle point region characteristic RD, such as following formula (8):
According to the variable RA、RBAnd new variables RD, defining new angiosomes function is following formula (9):
Wherein, a, c and d are respectively planar structure parameter, background differentiation parameter and three-dimensional blood vessel structure parameter, according to
Experience and selected scale are chosen.
If currently processed image is bifurcation region, k1、k2And k3As long as any one value is smaller in, variable RDValue compared with
Greatly；If currently processed image is the biggish region of blood vessel radian, k2And k3With λ2And λ3Sizableness, k1Value it is relatively small, then
Variable RDValue it is larger；If currently processed image is typical Vasa recta region, k1≈ 0, variable RDValue it is still larger；Therefore
New variables RDA variety of angiosomes are applicable in, interference will not be generated to other characteristic items.
Step S23, it determines and needs the angiosomes enhanced: determining what needs enhanced according to the angiosomes characteristic function
Angiosomes, and then the three-dimensional fine vascular original image is enhanced, obtain three-dimensional fine vascular enhancing image.It is described
Three-dimensional fine vascular enhancing image includes enhanced complex pattern to be repaired 1 and enhanced reference picture 3.
In the present embodiment, specifically, Fig. 6 (a) and Fig. 6 (b) is referred to, wherein Fig. 6 (a) is enhancing shown in Fig. 3 (a)
Reference picture afterwards；Fig. 6 (b) is enhanced complex pattern to be repaired shown in Fig. 3 (b).Comparison discovery utilizes blood provided by the invention
Area under control characteristic of field function, which carries out image enhancement, has preferable effect.
Referring to Fig. 7, for the flow chart of step S3 in three-dimensional fine vascular method for reconstructing shown in Fig. 2.It is described to be based on blood vessel
The character matching step S3 of characteristic includes the following steps:
Step S31, the complex pattern 1 to be repaired is divided into multiple candidates identical with reference 32 size of mapping block
Block calculates global registration degree SM。
Enhance image, i.e., the described complex pattern to be repaired 1 and the reference picture 3 according to the three-dimensional fine vascular, to two width
The overall distribution of image medium vessels carries out similarity mode, calculates the candidate blocks and the global registration with reference to mapping block 32
Spend SM。
Newly-built 0-1 matrix B1, the corresponding position that blood vessel is belonged in the candidate blocks of the complex pattern 1 to be repaired is set as 1,
The corresponding position for belonging to background is set as 0；Newly-built 0-1 matrix B2, will belong in the reference mapping block 32 of the reference picture 3
The corresponding position of blood vessel is set as 1, and the corresponding position for belonging to background is set as 0；To the matrix B1With the matrix B2On corresponding position
It carries out " XOR operation ", obtains matrix of consequence MR, the matrix MRIn " 0 " it is more, i.e., something in common is more, then explanation it is described to
The reference mapping block 32 of the candidate blocks and the reference picture 3 of repairing image 1 is more similar from profile.
Define global registration degree SMFor following formula (10):
Wherein, defining u (x, y, z) is following formula (11):
Step S32, vessel segment similarity S is calculatedT。
Referring to Fig. 8, for the flow chart of step S32 in step S3 shown in Fig. 7.The calculating vessel segment similarity STStep
Suddenly S32 includes:
Step S32-1, based on the angiosomes for needing to enhance described in method refinement step S2 are successively stripped, single pixel is extracted
Center line obtains three-dimensional fine vascular binary map, the three-dimensional fine vascular two-value by the three-dimensional fine vascular enhancing image
Figure includes the binary map of the complex pattern 1 to be repaired and the binary map of the reference picture 3；
Step S32-2, the end point set that the candidate blocks are extracted according to the binary map of the complex pattern 1 to be repaired, according to described
The binary map of reference picture 3 extracts the end point set with reference to mapping block 32, seeks endpoint feature vector respectively；
Step S32-3, vessel segment similarity S is calculated according to the endpoint feature vectorT。
Detailed process are as follows:
For blood-vessel image, segmentation, bifurcation and the endpoint of blood vessel are the features for especially needing to pay close attention to, therefore utilize bone
Frame thought, the center line of the single pixel wide for the angiosomes for needing to enhance described in extraction step S2, by described three-dimensional fine
Blood vessel enhancing image has obtained three-dimensional fine vascular binary map, and the three-dimensional fine vascular binary map includes the complex pattern to be repaired
The binary map of 1 binary map and the reference picture 3；Fig. 9 (a) and Fig. 9 (b) specifically is referred to, wherein Fig. 9 (a) is Fig. 6 (a)
Reference picture after shown refinement；Fig. 9 (b) is the complex pattern to be repaired after refinement shown in Fig. 6 (b).
Single pixel image after refinement is made of a section curved line segment, and wherein endpoint can be divided into terminal and crosspoint；
Some endpoints are from two or more line itself is separated, these endpoints are exactly crosspoint, and some endpoints are at itself
It terminates, these endpoints are terminals.
If the binary map of candidate blocks described in the complex pattern to be repaired 1 is volume data V1, extract the endpoint of the candidate blocks
Collection, the end point set includes terminal collection ES1={ E11,E12,...,E1mAnd bifurcation set FS1={ F11,F12,...,F1n, if institute
State volume data V1Side length be a, then be divided into 8 × 8 × 8 small small cubic blocks (side length a/8), statistics falls in each
Terminal quantity e in small cubic block1n(n=1,2 ... 512) and bifurcation quantity f1n(n=1,2 ... 512), form one
The volume data V with location information1Endpoint feature vector, the endpoint feature vector includes terminal feature vectorWith bifurcation feature vectorIf in the reference picture 3
The binary map with reference to mapping block 32 is volume data V2, the same procedure acquisition volume data V2Endpoint feature vector, it is described
Endpoint feature vector includes terminal feature vectorWith bifurcation feature vector
The terminal feature vector of two groups of volume datas is sought using the cosine lawWithDegree of correlation STe, such as following formula
(12)；And bifurcation feature vectorWithDegree of correlation STf, such as following formula (13)；According to the degree of correlation STeWith it is described
Degree of correlation STfCalculate the vessel segment similarity ST, such as following formula (14):
ST=STe·STf (14)。
Step S33, according to the global registration degree SMWith the vessel segment similarity STCalculate whole matching degree Sves, described
Whole matching degree Sves=μ SM+ηST, wherein μ and η is respectively SMAnd STWeight coefficient, and+η=1 μ.
Step S34, according to the whole matching degree Sves, the reference mapping block 32 is obtained in the complex pattern 1 to be repaired
Blocks and optimal matching blocks 52, the filling in the corresponding blocks and optimal matching blocks 52 of the reference mapping area 30 with reference in mapping block 32
Region 50.
The blocks and optimal matching blocks 52 are the highest candidate blocks of whole matching degree.The blocks and optimal matching blocks 52 pass through mapping
Positional relationship obtains filling region 50 corresponding with the area to be repaired 10.
Referring to Fig. 10, for the flow chart of step S4 in three-dimensional fine vascular method for reconstructing shown in Fig. 2.Described image filling
And BORDER PROCESSING step S4 includes:
Step S41, by the to be repaired of the corresponding filling of the filling region 50 of the blocks and optimal matching blocks 52 to the multiblock 12 to be repaired
Multiple region 10；
Step S42, the boundary of the filling region 50 is subjected to interpolation iteration, constantly approached until convergence, is filled out described
The edge smoothing transition of region 50 and the area to be repaired 10 is filled, reconstruction is completed.
Detailed process are as follows:
The filling region 50 is filled into the area to be repaired 10, the filling region 50 covers the area to be repaired
Complete the reparation of the deleted areas in domain 10.
The boundary of the area to be repaired 10 and the pixel grey scale of the filling region 50 are put down using the thought of " approaching "
It slips over and crosses, weaken the edge of piece between it.The multiblock 12 to be repaired and the restriction range of the blocks and optimal matching blocks 52 are distinguished
Vector field is established, gets off to carry out interpolation iteration in the guidance of vector field.
The p (x, y, z) that sets up an office is a bit in the multiblock to be repaired 12, and the collection of 6 neighborhood points is combined into N6, define its 6 neighbours
The differential vector collection in domain direction is combined intoUnit vector on 6 neighborhood directions is respectivelyIt is obtained according to the definition of difference To every in the multiblock 12 to be repaired
A pixel all carries out aforesaid operations, obtains the corresponding set of all pixels point in the multiblock to be repaired 12It is all to obtain
Vector constitutes the vector field D of the multiblock to be repaired 121, data iteration is carried out, pixel set P is finally obtained.
Same procedure constructs the vector field D of the blocks and optimal matching blocks 525, data iteration is carried out, pixel set is finally obtained
Q。
In order to make the pixel set P approach the most pixel set Q, minimum value of the problem equivalent in following formula (15) is asked
Topic:
Wherein, Δ is the difference operator on certain six neighborhood direction of pixel, D1It is the vector field of the multiblock to be repaired 12.Root
According to Euler-Lagrange equation, the above problem is equivalent to following formula (16):
Δ p=D1, p ∈ D1(16)；
It is reference to realize infinitely approaching for pixel that its meaning, which is with vector field,.Over-relaxation iterative method is used below, is passed through
Neighborhood territory pixel disappears member to realize.
If initialisation image primitive definition value is p0=p (x, y, z), (x, y, z) ∈ D1, the process of iteration is following formula (17):
Wherein, Δ pt(x, y, z) such as following formula (18):
Δpt(x, y, z)=pt(x+1, y, z)+pt(x-1, y, z)+pt(x, y+1, z)+pt(x,y-1,z)+pt(x,y,z+
1)+pt(x,y,z-1)-6×pt(x, y, z) (18)；
Iteration each time can all update the pixel value of the multiblock to be repaired 12, be allowed to gradually approach in the case where vector field guides
The pixel value of corresponding position in the blocks and optimal matching blocks 52, until restraining the effect that gray scale can be realized and coordinate transition.
Three-dimensional fine vascular is repaired using the three-dimensional fine vascular method for reconstructing, the fine blood of three-dimensional to be repaired
Three-dimensional fine vascular structure figure after pipe structure chart and reparation is detailed in Figure 11 (a) and Figure 11 (b), it is known that application of the present invention
This method can be good at repairing the three-dimensional fine vascular of tool deleted areas, and precision is higher.
Referring to Figure 12, for the structural block diagram of three-dimensional fine vascular reconstructing system provided by the invention.It is described three-dimensional fine
Reconstructing blood vessel system 100 includes getImage module 11, image pre-processing module 13, characteristic matching module 15 and filling processing mould
Block 17.
Described image insmods 11, and for being loaded into three-dimensional fine vascular original image, the three-dimensional fine vascular is original
Image includes the complex pattern to be repaired 1 and reference picture 3 for scanning same target, and the complex pattern 1 to be repaired is the vessel graph of tool missing
Picture, the reference picture 3 are complete blood-vessel image, and the precision of the complex pattern 1 to be repaired is higher than the reference picture 3；And root
Area to be repaired 10 is specified according to the complex pattern 1 to be repaired, and defines the multiblock to be repaired 12 comprising the area to be repaired 10, according to
The multiblock to be repaired 12 determines that position is corresponding in the reference picture 3, size is identical with reference to mapping block 32, described to be repaired
The corresponding reference mapping area 30 with reference in mapping block 32 in area to be repaired 10 in block 12.Described image insmods 11
Detailed execution process correspond to described in step S1 as above, repeat no more.
Described image preprocessing module 13, for described image insmod 11 loadings the three-dimensional fine vascular it is former
Beginning image is analyzed, and determines the angiosomes for needing to enhance, and is enhanced the three-dimensional fine vascular original image, is obtained
Three-dimensional fine vascular enhances image, and the three-dimensional fine vascular enhancing image includes the enhanced complex pattern 1 to be repaired and increases
The reference picture 3 after strong.
The characteristic matching module 15, the three-dimensional fine vascular enhancing for being obtained according to described image preprocessing module 13
Image calculates whole matching degree, obtains the blocks and optimal matching blocks 52 with reference to mapping block 32 in the complex pattern 1 to be repaired, institute
It states with reference to the filling region 50 in the corresponding blocks and optimal matching blocks 52 of reference mapping area 30 in mapping block 32.
The filling processing module 17, blocks and optimal matching blocks 52 for obtaining the characteristic matching module 15 are filled out
It fills the correspondence of region 50 to fill to the area to be repaired 10 of the multiblock 12 to be repaired, and boundary is smoothed, complete to repair
It rebuilds.
Figure 13 is please referred to, is the structural block diagram of image pre-processing module shown in Figure 12.Described image preprocessing module 13 is wrapped
Include the first arithmetic element 131, function creation unit 133 and blood vessel enhancement unit 135.
First arithmetic element 131 chooses Hessian matrix and does convolution fortune to the three-dimensional fine vascular original image
It calculates, acquires characteristic value and feature vector.
The function creation unit 133, the characteristic value and feature vector acquired according to first arithmetic element 131 and base
Angiosomes characteristic function is established in vessel properties.
The blood vessel enhancement unit 135 is determined according to the angiosomes characteristic function that the function creation unit 133 is established
The angiosomes for needing to enhance, and then the three-dimensional fine vascular original image is enhanced, it obtains three-dimensional fine vascular and increases
Strong image.Wherein, the detailed execution process of each unit corresponds to described in step S21 to S23 as above, repeats no more.
Figure 14 is please referred to, is the structural block diagram of characteristic matching module shown in Figure 12.The characteristic matching module 15 includes the
One computing unit 151, the second computing unit 153 and third computing unit 155 are respectively used to calculate global registration degree SM, blood vessel
Section similarity STAnd whole matching degree Sves.Wherein, the detailed execution process of each unit corresponds to step S31 to S34 institute as above
It states, repeats no more.
Figure 15 is please referred to, is the structural block diagram of filling processing module shown in Figure 12.The filling processing module 17 includes figure
As fills unit 171 and BORDER PROCESSING unit 173.
The filling region 50 of the blocks and optimal matching blocks 52 is corresponded to and is filled to described to be repaired by described image fills unit 171
The area to be repaired 10 of multiblock 12.
The boundary of the filling region 50 is carried out interpolation iteration by the BORDER PROCESSING unit 173, is constantly approached until receiving
It holds back, by the edge smoothing transition of the filling region 50 and the area to be repaired 10, completes reconstruction.Wherein, each
The detailed execution process of unit corresponds to described in step S41 to S42 as above, repeats no more.
Three-dimensional fine vascular method for reconstructing provided by the invention and its system have the advantages that
One, the angiosomes that the three-dimensional fine vascular method for reconstructing provided by the invention enhances needs have carried out feature
Value is extracted, and the three-dimensional fine vascular original image has been carried out image enhancement, convenient for derived from described in same blood vessel sample
Complex pattern 1 and the reference picture 3 to be repaired carry out structure matching, improve the efficiency that three-dimensional fine vascular is rebuild, and can protect
Demonstrate,prove precision；
Two, the present invention has redefined the angiosomes characteristic function based on Hessian gusts, it is contemplated that the three-dimensional
The grayscale information of fine vascular original image can effectively improve the effect of image enhancement；The crotch region of blood vessel is considered simultaneously
The influence of the individual differences such as domain and the biggish region of radian, can be good at solving the problems, such as individual difference, even if
Special circumstances will not judge incorrectly, and have wider applicability；
Three, the three-dimensional fine vascular method for reconstructing provided by the invention is formed based on vessel properties, it is contemplated that blood vessel knot
The tubular character of structure and stronger self-similarity, so that guidance is derived from the complex pattern 1 to be repaired of same blood vessel sample and described
The matching of reference picture 3 improves the reliability and accuracy of blood vessel structure characteristic matching, further increases three-dimensional fine vascular
The precision of reconstruction；
Four, the present invention is obtained described with reference to mapping block by the matching of the complex pattern 1 and the reference picture 3 to be repaired
32 blocks and optimal matching blocks 52 in the complex pattern 1 to be repaired are again smoothed boundary after corresponding filling, repair
Precision is high.
Person of ordinary skill in the field, which is understood that, realizes that all or part of the steps of above method embodiment can be with
It being done through the relevant hardware of the program instructions, program above-mentioned can store in computer-readable storage medium, and by
Processor executes, program above-mentioned when executed processor can execute it is including the steps of the foregoing method embodiments all or part of
Step.Wherein, the processor can be used as the implementation of one or more processors chip, or can be one or more dedicated
A part of integrated circuit (Application Specific Integrated Circuit, ASIC)；And storage above-mentioned is situated between
Matter can include but is not limited to following kind of storage medium: flash memory (Flash Memory), read-only memory (Read-Only
Memory, ROM), random access memory (Random Access Memory, RAM), mobile hard disk, magnetic or disk etc.
The various media that can store program code.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (6)
1. a kind of three-dimensional fine vascular method for reconstructing, which comprises the following steps:
Step 1: being loaded into three-dimensional fine vascular original image:
The three-dimensional fine vascular original image includes the complex pattern and reference picture to be repaired for scanning same target, described to be repaired
Image is the blood-vessel image for having deleted areas, and the reference picture is complete blood-vessel image, the precision of the complex pattern to be repaired
Higher than the precision of the reference picture；
Area to be repaired is specified according to the complex pattern to be repaired, and defines the multiblock to be repaired comprising the area to be repaired, according to
The multiblock to be repaired determines that position is corresponding in the reference picture, size is identical and refers to mapping block, in the multiblock to be repaired
Area to be repaired correspond to the reference mapping area with reference in mapping block；
Step 2: the image preprocessing based on vessel properties:
The three-dimensional fine vascular original image of loading is analyzed, the angiosomes for needing to enhance are determined, to described three
Dimension fine vascular original image is enhanced, and three-dimensional fine vascular enhancing image is obtained, and the three-dimensional fine vascular enhances image
Including enhanced complex pattern to be repaired and enhanced reference picture；Specifically, the step 2 includes the following steps:
The three-dimensional fine vascular original image is analyzed based on Hessian matrix: choosing Hessian matrix to described three
Dimension fine vascular original image does convolution algorithm, acquires characteristic value and feature vector；
It establishes angiosomes characteristic function: establishing angiosomes spy based on vessel properties and according to the characteristic value and feature vector
Levy function；
It determines and needs the angiosomes enhanced: determining the angiosomes for needing to enhance according to the angiosomes characteristic function, into
And the three-dimensional fine vascular original image is enhanced, obtain three-dimensional fine vascular enhancing image
Step 3: the characteristic matching based on vessel properties:
Enhance image according to the three-dimensional fine vascular, calculate whole matching degree, obtains the reference mapping block described to be repaired
Blocks and optimal matching blocks in complex pattern, the reference mapping area with reference in mapping block correspond to the filling in the blocks and optimal matching blocks
Region；Specifically, the step 3 includes the following steps:
The complex pattern to be repaired is divided into multiple candidate blocks identical with the reference mapping block size, calculates the candidate blocks
With the global registration degree S with reference to mapping blockM；
Calculate vessel segment similarity ST；
According to the global registration degree SMWith the vessel segment similarity STCalculate whole matching degree Sves, the whole matching degree
Sves=μ SM+ηST, wherein μ and η is respectively SMAnd STWeight coefficient, and+η=1 μ；
According to the whole matching degree Sves, obtain the blocks and optimal matching blocks with reference to mapping block in the complex pattern to be repaired, institute
State the filling region corresponded in the blocks and optimal matching blocks with reference to the reference mapping area in mapping block；
Step 4: image completion and BORDER PROCESSING:
By the corresponding filling of the filling region of the blocks and optimal matching blocks to the area to be repaired of the multiblock to be repaired, and boundary is carried out
Smoothing processing completes reconstruction.
2. three-dimensional fine vascular method for reconstructing according to claim 1, which is characterized in that described to be based on Hessian matrix
The step of analyzing the three-dimensional fine vascular original image include the following:
Calculate the second-order partial differential coefficient and mixed partial derivative of each pixel in the three-dimensional fine vascular original image；
Hessian matrix is formed with the second-order partial differential coefficient and mixed partial derivative of calculating to do with the three-dimensional fine vascular original image
Convolution algorithm；
Acquire eigenvalue λ1、λ2And λ3With feature vector γ corresponding with the characteristic value1、γ2And γ3, and | λ1|<|λ2|<|λ3
|。
3. three-dimensional fine vascular method for reconstructing according to claim 2, which is characterized in that the angiosomes characteristic function
Are as follows:
Define RA=| λ2|/|λ3|,Wherein, a, c and d distinguish
For planar structure parameter, background differentiation parameter and three-dimensional blood vessel structure parameter, k1、k2And k3Respectively along described eigenvector
γ1、γ2、γ3The size of the rate of gray level of the gradient direction of opposite direction.
4. three-dimensional fine vascular method for reconstructing according to claim 1, which is characterized in that the calculating vessel segment similarity
STThe step of include the following steps:
Based on the angiosomes for needing to enhance described in method refinement step two are successively stripped, single pixel center line is extracted, by described
Three-dimensional fine vascular enhancing image obtains three-dimensional fine vascular binary map, and the three-dimensional fine vascular binary map includes described to be repaired
The binary map of the binary map of complex pattern and the reference picture；
The end point set that the candidate blocks are extracted according to the binary map of the complex pattern to be repaired, according to the binary map of the reference picture
The end point set with reference to mapping block is extracted, seeks endpoint feature vector respectively；
Vessel segment similarity S is calculated according to the endpoint feature vectorT。
5. three-dimensional fine vascular method for reconstructing according to claim 1, which is characterized in that the step 4 includes following step
It is rapid:
By the corresponding filling of the filling region of the blocks and optimal matching blocks to the area to be repaired of the multiblock to be repaired；
The boundary of the filling region is subjected to interpolation iteration, is constantly approached until convergence, by the filling region and described
Reconstruction is completed in the edge smoothing transition of area to be repaired.
6. a kind of three-dimensional fine vascular reconstructing system characterized by comprising
GetImage module, for being loaded into three-dimensional fine vascular original image, the three-dimensional fine vascular original image includes sweeping
The complex pattern and reference picture to be repaired of same target are retouched, the complex pattern to be repaired is the blood-vessel image of tool missing, described with reference to figure
As being complete blood-vessel image, the precision of the complex pattern to be repaired is higher than the reference picture；And according to the complex pattern to be repaired
Specified area to be repaired, and the multiblock to be repaired comprising the area to be repaired is defined, it is determined according to the multiblock to be repaired described
Position is corresponding in reference picture, size is identical refers to mapping block, and the area to be repaired in the multiblock to be repaired corresponds to the ginseng
Examine the reference mapping area in mapping block；
Image pre-processing module, for dividing the insmod three-dimensional fine vascular original image of loading of described image
Analysis determines the angiosomes for needing to enhance, and enhances the three-dimensional fine vascular original image, obtains three-dimensional fine vascular
Enhance image, the three-dimensional fine vascular enhancing image includes enhanced complex pattern to be repaired and enhanced reference picture；Its
In, described image preprocessing module includes:
First arithmetic element chooses Hessian matrix and does convolution algorithm to the three-dimensional fine vascular original image, acquires feature
Value and feature vector；
Function creation unit, the characteristic value and feature vector acquired according to first arithmetic element are simultaneously established based on vessel properties
Angiosomes characteristic function；
Blood vessel enhancement unit, the angiosomes characteristic function established according to the function creation unit determine the blood vessel for needing to enhance
Region, and then the three-dimensional fine vascular original image is enhanced, obtain three-dimensional fine vascular enhancing image
Characteristic matching module, the three-dimensional fine vascular for being obtained according to described image preprocessing module enhance image, calculate whole
Body matching degree obtains the blocks and optimal matching blocks with reference to mapping block in the complex pattern to be repaired, described with reference in mapping block
The filling region in the blocks and optimal matching blocks is corresponded to reference to mapping area；Wherein, the characteristic matching module includes the first calculating
Unit, the second computing unit and third computing unit are respectively used to calculate global registration degree, vessel segment similarity and whole matching
Degree；
Fill processing module, the corresponding filling of the filling region of the blocks and optimal matching blocks for obtaining the characteristic matching module
The extremely area to be repaired of the multiblock to be repaired, and boundary is smoothed, complete reconstruction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610305637.0A CN106127849B (en) | 2016-05-10 | 2016-05-10 | Three-dimensional fine vascular method for reconstructing and its system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610305637.0A CN106127849B (en) | 2016-05-10 | 2016-05-10 | Three-dimensional fine vascular method for reconstructing and its system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106127849A CN106127849A (en) | 2016-11-16 |
| CN106127849B true CN106127849B (en) | 2019-01-11 |
Family
ID=57269939
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610305637.0A Expired - Fee Related CN106127849B (en) | 2016-05-10 | 2016-05-10 | Three-dimensional fine vascular method for reconstructing and its system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106127849B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106875375B (en) * | 2016-12-28 | 2019-07-30 | 浙江工业大学 | A kind of three-dimensional blood vessel axis detection method based on tubulose signature tracking |
| CN107067398B (en) * | 2017-04-28 | 2020-03-24 | 青岛海信医疗设备股份有限公司 | Completion method and device for missing blood vessels in three-dimensional medical model |
| CN109242964B (en) * | 2018-11-01 | 2023-04-11 | 青岛海信医疗设备股份有限公司 | Processing method and device of three-dimensional medical model |
| CN109829489A (en) * | 2019-01-18 | 2019-05-31 | 刘凯欣 | A kind of cultural relic fragments recombination method and device based on multilayer feature |
| CN110120027B (en) * | 2019-04-16 | 2021-05-14 | 华南理工大学 | CT slice image enhancement method and device for machine learning system data |
| CN111968051A (en) * | 2020-08-10 | 2020-11-20 | 珠海普生医疗科技有限公司 | Endoscope blood vessel enhancement method based on curvature analysis |
| CN112330708B (en) * | 2020-11-24 | 2024-04-23 | 沈阳东软智能医疗科技研究院有限公司 | Image processing method, device, storage medium and electronic equipment |
| CN112837288B (en) * | 2021-02-01 | 2021-11-23 | 数坤(北京)网络科技股份有限公司 | Blood vessel centerline extraction method and device and readable storage medium |
| KR102629331B1 (en) * | 2021-08-05 | 2024-01-29 | 숭실대학교 산학협력단 | Method for coronary arteries non-rigid registration using hierarchical deformation in computed tomography angiography images recording medium and device for performing the method |
| CN116596950B (en) * | 2023-05-31 | 2023-11-17 | 东北林业大学 | Retina fundus blood vessel tracking method based on feature weighted clustering |
| CN116721760B (en) * | 2023-06-12 | 2024-04-26 | 东北林业大学 | Biomarker-fused multitasking diabetic retinopathy detection algorithm |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130027389A1 (en) * | 2011-07-27 | 2013-01-31 | International Business Machines Corporation | Making a two-dimensional image into three dimensions |
| CN102999887A (en) * | 2012-11-12 | 2013-03-27 | 中国科学院研究生院 | Sample based image repairing method |
| CN104361626A (en) * | 2014-09-29 | 2015-02-18 | 北京理工大学 | Subcutaneous vein three-dimensional reconstruction method based on hybrid matching strategy |
| CN104933756A (en) * | 2014-03-21 | 2015-09-23 | 北京冠生云医疗技术有限公司 | Construction method of three-dimensional coronary artery analysis model and system thereof |
-
2016
- 2016-05-10 CN CN201610305637.0A patent/CN106127849B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130027389A1 (en) * | 2011-07-27 | 2013-01-31 | International Business Machines Corporation | Making a two-dimensional image into three dimensions |
| CN102999887A (en) * | 2012-11-12 | 2013-03-27 | 中国科学院研究生院 | Sample based image repairing method |
| CN104933756A (en) * | 2014-03-21 | 2015-09-23 | 北京冠生云医疗技术有限公司 | Construction method of three-dimensional coronary artery analysis model and system thereof |
| CN104361626A (en) * | 2014-09-29 | 2015-02-18 | 北京理工大学 | Subcutaneous vein three-dimensional reconstruction method based on hybrid matching strategy |
Non-Patent Citations (2)
| Title |
|---|
| "基于紧小波框架的多幅盲运动图像修复算法研究";王敏敏;《中国优秀硕士学位论文全文数据库 信息科技辑》;20160415;论文摘要 |
| "增强现实中的图像修复算法";顾晨;《中国优秀硕士学位论文全文数据库 信息科技辑》;20141115;论文第26-32页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106127849A (en) | 2016-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106127849B (en) | Three-dimensional fine vascular method for reconstructing and its system | |
| Saha et al. | Digital topology and geometry in medical imaging: a survey | |
| Tyrrell et al. | Robust 3-D modeling of vasculature imagery using superellipsoids | |
| CN106339998B (en) | Multi-focus image fusing method based on contrast pyramid transformation | |
| CN109215033A (en) | The method and system of image segmentation | |
| US7315639B2 (en) | Method of lung lobe segmentation and computer system | |
| Niessen et al. | Multiscale segmentation of three-dimensional MR brain images | |
| CN109478327B (en) | Method for automatic detection of systemic arteries in Computed Tomography Angiography (CTA) of arbitrary field of view | |
| CN113112609A (en) | Navigation method and system for lung biopsy bronchoscope | |
| Kanitsar et al. | Computed tomography angiography: a case study of peripheral vessel investigation | |
| Tian et al. | Medical image processing and analysis | |
| Sinko et al. | 3D registration of the point cloud data using ICP algorithm in medical image analysis | |
| Kutka et al. | Extraction of line properties based on direction fields | |
| Braude et al. | Contour-based surface reconstruction using mpu implicit models | |
| CN112862833A (en) | Blood vessel segmentation method, electronic device and storage medium | |
| Yasmin et al. | Brain image analysis: a survey | |
| Descoteaux et al. | Geometric flows for segmenting vasculature in MRI: Theory and validation | |
| CN106023094B (en) | Bone tissue microstructure repair system based on image and its restorative procedure | |
| Selle et al. | Mathematical methods in medical imaging: analysis of vascular structures for liver surgery planning | |
| Li et al. | Vessels as 4D curves: Global minimal 4D paths to extract 3D tubular surfaces | |
| CN116051553B (en) | Method and device for marking inside three-dimensional medical model | |
| Pizaine et al. | Vessel geometry modeling and segmentation using convolution surfaces and an implicit medial axis | |
| Morigi et al. | 3D long bone reconstruction based on level sets | |
| Alom et al. | Automatic slice growing method based 3D reconstruction of liver with its vessels | |
| CN109712124A (en) | The label minimizing technology and device of ultrasound image |
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: 20190111 Termination date: 20210510 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |