CN103871057A - Magnetic resonance image-based bone segmentation method and system thereof - Google Patents
Magnetic resonance image-based bone segmentation method and system thereof Download PDFInfo
- Publication number
- CN103871057A CN103871057A CN201410087926.9A CN201410087926A CN103871057A CN 103871057 A CN103871057 A CN 103871057A CN 201410087926 A CN201410087926 A CN 201410087926A CN 103871057 A CN103871057 A CN 103871057A
- Authority
- CN
- China
- Prior art keywords
- magnetic resonance
- mri
- resonance image
- bone
- pixel
- 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.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Image Processing (AREA)
Abstract
The invention is suitable for the field of the medical image technology, and provides a magnetic resonance image-based bone segmentation method and a system thereof. The method comprises the following steps: calculating step: calculating the Hessian matrix of each pixel and a characteristic value thereof in a first magnetic resonance image; extracting step: extracting the flake structure of the first magnetic resonance image and performing gray level enhancement on the flake structure of the first magnetic resonance image to obtain a second magnetic resonance image; optimization step: performing threshold connected calculation on the second magnetic resonance image, extracting a bone contour of the second magnetic resonance image, and carrying out smooth optimization to obtain a bone contour segmentation result of the second magnetic resonance image. Therefore by applying the magnetic resonance image-based bone segmentation method and the system thereof, a bone needing to be segmented is completely and independently segmented from the magnetic resonance image.
Description
Technical field
The present invention relates to medical image technical field, relate in particular to a kind of method and system thereof of the bone segmentation based on magnetic resonance image (MRI).
Background technology
Bone is the most important support of human body and motion structure.Bone is mainly made up of sclerotin, marrow and periosteum three parts, bone periphery has abundant blood vessel, nerve and soft tissue, often needs each piece bone to separate separately and blood vessel, nerve and the soft tissue of adhesion bone periphery in the time carrying out bone segmentation.
MRI(Magnetic Resonance Imaging, Magnetic resonance imaging) be a kind of tomography technology of rebuilding human body information by obtain electromagnetic signal from human body, in today of digitized video rapid technological improvement, MRI is used in medical procedure more and more, becomes one of requisite technology in diagnostic imaging field.
MRI and CT(Computed Tomography, computed tomography) the same, be all the most frequently used technology in medical imaging diagnosis field.In MRI image, the soft-tissue imagings such as muscle, cartilage, ligament, meniscus are clear, and effect is better than CT far away; And MRI can multisequencing imaging, export multiple image type, for diagnostic imaging provides the more information of horn of plenty.But MRI, when to bone imaging, is easily disturbed by perienchyma, make easily to occur its hetero-organization of adhesion bone periphery and occur the problems such as hole in bone surface in the time cutting apart bone.
There is no at present the related art scheme that relates to MRI bone segmentation.Chinese Patent Application No. be 201110260717.6, denomination of invention is that in the application for a patent for invention of " a kind of brain CT image bone method and device of extracting ", applicant Neusoft Group Co., Ltd proposes a kind of method of brain CT image being carried out to bone extraction based on gray level threshold segmentation.The method is carried out separating of bone and background by analysis image half-tone information repeatedly to CT image, thereby reaches the object of extracting brain bone.This have artifact or bone boundaries clearly in situation, to be difficult to be partitioned into complete bone not based on gradation of image threshold segmentation method at CT image, can make bone occur hole or adhere to other bone piece or blood vessels.And the complexity of MRI image, higher than CT image, more cannot be used for MRI image to the dividing method of CT bone.
Therefore, merely rely on image threshold to cut apart to the bone in MRI image many problems that exist.In the not so good data of picture quality, adjacent bone can be adhesion shape in region close to each other.So these two adjacent bones will be extracted in the time of Threshold segmentation simultaneously, cause these two bone adhesions and independent.And there will be bone hole while carrying out Threshold segmentation in the inhomogeneous region of some bone sclerotin gray scales.The region of leaning on closelyer in some bones and bone peripheral soft tissues (as ligament, meniscus), easily related to its hetero-organization in the time extracting bone.
In summary, in actual use, obviously there is inconvenience and defect in the existing bone segmentation technology based on magnetic resonance image (MRI), so be necessary to be improved.
Summary of the invention
For above-mentioned defect, the object of the present invention is to provide a kind of method and system thereof of the bone segmentation based on magnetic resonance image (MRI), with complete in magnetic resonance image (MRI) and be independently partitioned into the bone that need to cut apart.
To achieve these goals, the invention provides a kind of method of the bone segmentation based on magnetic resonance image (MRI), described method comprises the steps:
Calculation procedure: Hessian matrix and the eigenwert thereof of calculating each pixel in the first magnetic resonance image (MRI);
Extraction step: extract the schistose texture in described the first magnetic resonance image (MRI), and the schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtain the second magnetic resonance image (MRI);
Optimization Steps: described the second magnetic resonance image (MRI) is carried out to Threshold-connected calculating, extract the bone contours in described, and carry out smooth optimization, obtain the bone contours segmentation result in described the second magnetic resonance image (MRI).
According to described method, describedly before described calculation procedure, also comprise:
Image acquisition step: magnetic resonance imaging is carried out in the position that need to carry out bone segmentation, obtain first magnetic resonance image (MRI) at described position.
According to described method, described calculation procedure comprises:
For each pixel structure Gaussian filter G in described the first magnetic resonance image (MRI)
σ, the variance that wherein σ is Gaussian filter, to Gaussian filter G
σask its second derivative to obtain L
σ;
Adopt L
σdo convolution with described the first magnetic resonance image (MRI) and obtain I
xx(σ), I
xy(σ), I
xz(σ), I
yy(σ), I
yz(σ), I
zz(σ);
By L
σthe Hessian matrix H that the result obtaining with described the first magnetic resonance image (MRI) convolution forms is:
Calculate three eigenvalue λ of each Hessian matrix H
1, λ
2, λ
3.
According to described method, described extraction step comprises:
Construct the schistose texture feature discriminant function V in described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3), described function carries out structural determination to the Hessian proper value of matrix of the each pixel value of former magnetic resonance image (MRI), judges whether pixel is positioned at graph outline region;
Calculate schistose texture fundamental function L (the λ)=V of all pixels in described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3) * H (λ
1), wherein H (λ
1) be to weigh λ
1the function of action effect;
Differentiate result according to the strip-like features of each pixel schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtain described the second magnetic resonance image (MRI).
According to described method, described Optimization Steps comprises:
Set the gray threshold scope [I of the bone contours in described the second magnetic resonance image (MRI)
l, I
h];
The contour edge of the bone extracting at needs is chosen first pixel, the starting point using described the first pixel as Threshold-connected, and second pixel of search within the scope of the gray threshold of described bone contours;
Smooth optimization is carried out in the region of the first pixel within the scope of the gray threshold of described bone contours and the first pixel composition, obtain the bone contours segmentation result in described the first magnetic resonance image (MRI).
In order to realize another goal of the invention of the present invention, the present invention also provides a kind of system of the bone segmentation based on magnetic resonance image (MRI), and described system comprises:
Computing module: for calculating Hessian matrix and the eigenwert thereof of each pixel of the first magnetic resonance image (MRI);
Extraction module: for extracting the schistose texture of described the first magnetic resonance image (MRI), and the schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtain the second magnetic resonance image (MRI);
Optimize module: for described the second magnetic resonance image (MRI) is carried out to Threshold-connected calculating, extract the bone contours in described, and carry out smooth optimization, obtain the bone contours segmentation result in described the second magnetic resonance image (MRI).
According to described system, described system also comprises:
Image collection module, for magnetic resonance imaging is carried out in the position that need to carry out bone segmentation, obtains first magnetic resonance image (MRI) at described position.
According to described system, described computing module comprises:
The first constructor module, is used to each pixel structure Gaussian filter G in described the first magnetic resonance image (MRI)
σ, the variance that wherein σ is Gaussian filter, to Gaussian filter G
σask its second derivative to obtain L
σ;
Convolution submodule, for adopting L
σdo convolution with described the first magnetic resonance image (MRI) and obtain I
xx(σ), I
xy(σ), I
xz(σ), I
yy(σ), I
yz(σ), I
zz(σ);
Matrix submodule, for by L
σthe result obtaining with described the first magnetic resonance image (MRI) convolution forms
Hessian matrix H is:
The first calculating sub module, for calculating three eigenvalue λ of each Hessian matrix H
1, λ
2, λ
3.
According to described system, described extraction module comprises:
The second constructor module, for constructing the schistose texture feature discriminant function V of described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3), described function carries out structural determination to the Hessian proper value of matrix of the each pixel value of former magnetic resonance image (MRI), judges whether pixel is positioned at graph outline region;
The second calculating sub module, for calculating schistose texture fundamental function L (the λ)=V of all pixels of described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3) * H (λ
1), wherein H (λ
1) be to weigh λ
1the function of action effect;
Gray scale enhancer module, carries out gray scale enhancing for differentiating result according to the strip-like features of each pixel to described the first magnetic resonance image (MRI) schistose texture, obtains described the second magnetic resonance image (MRI).
According to described system, described optimization module comprises:
Set submodule, for setting the gray threshold scope [I of bone contours of described the second magnetic resonance image (MRI)
l, I
h];
Search submodule, chooses first pixel for the contour edge of the bone in needs extraction, the starting point using described the first pixel as Threshold-connected, and second pixel of search within the scope of the gray threshold of described bone contours;
Optimize submodule, for smooth optimization is carried out in the region of the first pixel within the scope of the gray threshold of described bone contours and the first pixel composition, obtain the bone contours segmentation result in described the first magnetic resonance image (MRI).
The present invention is by calculating Hessian matrix and the eigenwert thereof of each pixel in the first magnetic resonance image (MRI); Extract the schistose texture in described the first magnetic resonance image (MRI), and the schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtain the second magnetic resonance image (MRI); Described the second magnetic resonance image (MRI) is carried out to Threshold-connected calculating, extract the bone contours in described, and carry out smooth optimization, obtain the bone contours segmentation result in described the second magnetic resonance image (MRI).Thus, system and method provided by the invention is determined the bone contours in image by calculating MRI image second order derivative, bone contours is carried out to gray scale enhancing, then bone contours is carried out to threshold value extraction, finally obtains MRI bone segmentation result.This system and method can effectively be removed bone hole problem in MRI image, even if still can accurately intactly extract the outline of bone in the inadequate clearly situation of bone boundaries; And can effectively solve the problem of bone and other bone pieces or the adhesion of bone perienchyma.
Accompanying drawing explanation
Fig. 1 is the system architecture schematic diagram that the bone segmentation based on magnetic resonance image (MRI) is provided in first embodiment of the invention;
Fig. 2 be the present invention second and third, the system architecture schematic diagram of the bone segmentation based on magnetic resonance image (MRI) is provided in four, five embodiment;
Fig. 3 is the method flow diagram of the bone segmentation based on magnetic resonance image (MRI) that provides of sixth embodiment of the invention;
Fig. 4 A is the magnetic resonance image (MRI) of the knee joint femur cut apart of needs that one embodiment of the invention provides;
Fig. 4 B is the sheet hum pattern of the knee joint femur in Fig. 4 A of providing of one embodiment of the invention;
Fig. 4 C is the flake reinforcement figure of the knee joint femur in Fig. 4 A of providing of one embodiment of the invention;
Fig. 4 D is the Threshold-connected figure of the knee joint femur in Fig. 4 C of providing of one embodiment of the invention;
Fig. 4 E is the final segmentation result figure of the knee joint femur in Fig. 4 A of providing of one embodiment of the invention.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not intended to limit the present invention.
Referring to Fig. 1, a kind of system 100 of the bone segmentation based on magnetic resonance image (MRI) is provided in the first embodiment of the present invention, the system 100 of the bone segmentation based on magnetic resonance image (MRI) comprises:
In this embodiment, first calculated Hessian matrix and the eigenwert thereof of each pixel of MRI image by computing module 10; Then extraction module 20 extracts schistose texture and schistose texture in MRI image is carried out to gray scale enhancing; Finally, optimize module 30 and carry out Threshold-connected calculating, extract bone contours and carry out smooth optimization.Thus, can solve the problem such as bone adhesion and bone hole occurring while extracting bone at present in MRI image.The system 100 of the described bone segmentation based on magnetic resonance image (MRI) can intactly extract bone contours and solve and in the time cutting apart MRI bone, occur bone hole and and the problem such as other bone pieces of adhesion and tissue.
Referring to Fig. 2, in the second embodiment of the present invention, the system 100 of the bone segmentation based on magnetic resonance image (MRI) also comprises:
In this embodiment, by image collection module 40, magnetic resonance imaging is carried out in the position that need to carry out bone segmentation, for example, need the bone of knee portions of leg regions to cut apart, knee portions of leg regions is scanned, obtain described the first magnetic resonance image (MRI), i.e. the MRI image of original knee portions of leg regions.
Referring to Fig. 2, in the third embodiment of the present invention, computing module 10 comprises:
The first constructor module 11, is used to each pixel structure Gaussian filter G in described the first magnetic resonance image (MRI)
σ, the variance that wherein σ is Gaussian filter, to Gaussian filter G
σask its second derivative to obtain L
σ;
Convolution submodule 12, for adopting L
σdo convolution with described the first magnetic resonance image (MRI) and obtain I
xx(σ), I
xy(σ), I
xz(σ), I
yy(σ), I
yz(σ), I
zz(σ);
Hessian matrix H is:
The first calculating sub module 14, for calculating three eigenvalue λ of each Hessian matrix H
1, λ
2, λ
3.
In this embodiment, calculate Hessian matrix and the eigenwert thereof of each pixel in described the first magnetic resonance image (MRI).The concrete first constructor module 11 that first needs is constructed Gaussian filter G
σ, the variance that wherein σ is Gaussian filter, to Gaussian filter G
σask its second derivative to obtain L
σ; Then convolution submodule 12 adopts L σ and original image (being described the first magnetic resonance image (MRI)) to do convolution and obtains I
xx(σ), I
xy(σ), I
xz(σ), I
yy(σ), I
yz(σ), I
zz(σ); The result formation Hessian matrix that then matrix submodule 13 obtains L σ and original image (being described the first magnetic resonance image (MRI)) convolution
Last the first calculating sub module 14 is calculated three eigenvalue λ of each Hessian matrix H
1, λ
2, λ
3; Obtain thus Hessian matrix and the eigenwert thereof of each pixel in the first magnetic resonance image (MRI).
Referring to Fig. 2, in the fourth embodiment of the present invention, extraction module 20 comprises:
The second constructor module 21, for constructing the schistose texture feature discriminant function V of described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3), described function carries out structural determination to the Hessian proper value of matrix of the each pixel value of former magnetic resonance image (MRI), judges whether pixel is positioned at graph outline region;
The second calculating sub module 22, for calculating schistose texture fundamental function L (the λ)=V of all pixels of described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3) * H (λ
1), wherein H (λ
1) be to weigh λ
1the function of action effect;
Gray scale enhancer module 23, carries out gray scale enhancing for differentiating result according to the strip-like features of each pixel to described the first magnetic resonance image (MRI) schistose texture, obtains described the second magnetic resonance image (MRI).
In this embodiment, realize in described the first magnetic resonance image (MRI), extract schistose texture and to image in schistose texture carry out gray scale enhancing, obtain described the second magnetic resonance image (MRI) after treatment.The second concrete constructor module 21 is constructed sheet architectural feature discriminant function V
s(λ
1, λ
2, λ
3), this function carries out structural determination to the Hessian proper value of matrix of the each pixel value of original image (being described the first magnetic resonance image (MRI)), judges whether pixel is positioned at graph outline region; The second calculating sub module 22 is calculated schistose texture fundamental function L (the λ)=V of all pixels of original image (being described the first magnetic resonance image (MRI))
s(λ
1, λ
2, λ
3) * H (λ
1), wherein H (λ
1) be to weigh λ
1the function of action effect; The strip-like features of gray scale enhancer module 23 corresponding each pixels is differentiated result former figure is carried out to the enhancing of schistose texture gray scale, obtains described the second magnetic resonance image (MRI) after gray scale strengthens.
Referring to Fig. 2, in the fifth embodiment of the present invention, optimize module 30 and comprise:
Optimize submodule 33, for smooth optimization is carried out in the region of the first pixel within the scope of the gray threshold of described bone contours and the first pixel composition, obtain the bone contours segmentation result in described the first magnetic resonance image (MRI).
In this embodiment, optimizing module 30 carries out Threshold-connected calculating extraction bone contours and carries out smooth optimization operation.Concrete, setting submodule 31 is set the roughly gray threshold scope [I of the rear bone contours of gradation of image enhancing
l, I
h]; Search submodule 32 is chosen a pixel at the contour edge of the bone of required extraction, the starting point using this pixel as Threshold-connected the pixel of search within the scope of bone contours gray threshold; Optimize submodule 33 smooth optimization is carried out in the region of bone contours pixel composition, after processing, the region of gained is the final segmentation result of MRI bone.
In above-mentioned multiple enforcement, multiple module softwares unit, hardware cell or the software and hardware combining unit of the system 100 of the described bone segmentation based on magnetic resonance image (MRI).
Referring to Fig. 3, in the sixth embodiment of the present invention, provide a kind of method of the bone segmentation based on magnetic resonance image (MRI), described method comprises the steps:
In step S301, computing module 10 calculates Hessian matrix and the eigenwert thereof of each pixel in the first magnetic resonance image (MRI); This step is calculation procedure;
In step S302, extraction module 20 extracts the schistose texture in described the first magnetic resonance image (MRI), and the schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtains the second magnetic resonance image (MRI); This step is extraction step;
In step S303, optimize module 30 described the second magnetic resonance image (MRI) is carried out to Threshold-connected calculating, extract the bone contours in described, and carry out smooth optimization, obtain the bone contours segmentation result in described the second magnetic resonance image (MRI); This step is Optimization Steps.
In this embodiment, from the angle of the first magnetic resonance image (MRI) gray scale second derivative, the first magnetic resonance image (MRI) is analyzed, by Hessian matrix, the second derivative feature of the first magnetic resonance image (MRI) is classified, thereby determine the schistose texture in the first magnetic resonance image (MRI), therefore area-of-interest is locked as to the edge contour of bone, coordinates bone contours gray scale to strengthen algorithm the bone contours in the first magnetic resonance image (MRI) is become to clear and distinctive.In the bone adhesion of the second magnetic resonance image (MRI) after bone contours enhancing being carried out produce when simple Threshold-connected algorithm can be effectively removed in Threshold-connected extraction to bone segmentation and the problem of bone hole; Obtain the bone contours segmentation result in described the second magnetic resonance image (MRI).The method is can be in complicated MRI image complete and be partitioned into individually bone, for the three-dimensional reconstruction of MRI image provides technical foundation.
In the seventh embodiment of the present invention, before described step S301, also comprise:
Image acquisition step: magnetic resonance imaging is carried out in the position that need to carry out bone segmentation, obtain first magnetic resonance image (MRI) at described position.
In the eighth embodiment of the present invention, described step S301 comprises:
The first constructor module 11 is each pixel structure Gaussian filter G in described the first magnetic resonance image (MRI)
σ, the variance that wherein σ is Gaussian filter, to Gaussian filter G
σask its second derivative to obtain L
σ;
Convolution submodule 12 adopts L
σdo convolution with described the first magnetic resonance image (MRI) and obtain I
xx(σ), I
xy(σ), I
xz(σ), I
yy(σ), I
yz(σ), I
zz(σ);
Matrix H is:
The first calculating sub module 14 is calculated three eigenvalue λ of each Hessian matrix H
1, λ
2, λ
3.
In the eighth embodiment of the present invention, described step S302 comprises:
The second constructor module 21 is constructed the schistose texture feature discriminant function V in described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3), described function carries out structural determination to the Hessian proper value of matrix of the each pixel value of former magnetic resonance image (MRI), judges whether pixel is positioned at graph outline region;
The second calculating sub module 22 is calculated schistose texture fundamental function L (the λ)=V of all pixels in described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3) * H (λ
1), wherein H (λ
1) be to weigh λ
1the function of action effect;
Gray scale enhancer module 23 is differentiated result according to the strip-like features of each pixel schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtains described the second magnetic resonance image (MRI).
In nine embodiment of the present invention, described step S303 comprises:
The contour edge of the bone that search submodule 32 extracts at needs is chosen first pixel, the starting point using described the first pixel as Threshold-connected, and second pixel of search within the scope of the gray threshold of described bone contours;
Optimize submodule 33 smooth optimization is carried out in the region of the first pixel within the scope of the gray threshold of described bone contours and the first pixel composition, obtain the bone contours segmentation result in described the first magnetic resonance image (MRI).
Referring to Fig. 4 A~Fig. 4 E, in one embodiment of the invention, describe and adopted system and method provided by the invention to realize the technical process that knee joint femur is cut apart, in this embodiment, system receives the knee joint femur MRI image of input, then calculates Hessian matrix and the eigenwert thereof of the each pixel of MRI image; Then calculate strip-like features functional value; And schistose texture is strengthened; Then carry out Threshold-connected and extract bone contours; Carry out after smooth optimization is processed obtaining bone contours segmentation result.Detailed process is as follows:
Be the MRI image that knee joint femur carries out magnetic resonance imaging acquisition referring to Fig. 4 A, in this realization, need this image to process, with complete and be independently partitioned into knee joint femur.
First, computing module 10 calculates Hessian matrix and eigenwert thereof
The first constructor module 11 is constructed Gaussian filter G
σ, the variances sigma of Gaussian filter is taken as 1.6, to Gaussian filter G
σask its second derivative to obtain L
σ;
Convolution submodule 12 adopts L
σdo convolution with original image and obtain I
xx(σ), I
xy(σ), I
xz(σ), I
yy(σ), I
yz(σ), I
zz(σ);
The first calculating sub module 14 is calculated three eigenwerts of Hessian matrix H, and is arranged as λ by order from small to large
1< λ
2< λ
3.
Then, extraction module 20 extracts schistose texture and schistose texture in image is carried out to gray scale enhancing
The second constructor module 21 is constructed sheet architectural feature discriminant function V
s(λ
1, λ
2, λ
3), this function carries out structural determination to the Hessian proper value of matrix of the each pixel value of original image;
The second calculating sub module 22 is calculated schistose texture fundamental function L (the λ)=V of all pixels of original image
s(λ
1, λ
2, λ
3) * H (λ
1), wherein H (λ
1) be to weigh λ
1the function of action effect, the image sheet information obtaining thus as shown in Figure 4 B.
The strip-like features of gray scale enhancer module 23 corresponding each pixels is differentiated result former figure is carried out to gray scale enhancing, and the schistose texture gray scale obtaining thus strengthens result as shown in Figure 4 C.
Finally, optimize module 30 and carry out Threshold-connected calculating, extract bone contours
Optimize submodule 33 smooth optimization processing is carried out in the region of bone contours pixel composition, the bone contours expansion area of gained is the final segmentation result of bone contours.Final segmentation result as shown in Figure 4 E.
In sum, the present invention is by calculating Hessian matrix and the eigenwert thereof of each pixel in the first magnetic resonance image (MRI); Extract the schistose texture in described the first magnetic resonance image (MRI), and the schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtain the second magnetic resonance image (MRI); Described the second magnetic resonance image (MRI) is carried out to Threshold-connected calculating, extract the bone contours in described, and carry out smooth optimization, obtain the bone contours segmentation result in described the second magnetic resonance image (MRI).Thus, system and method provided by the invention is determined the bone contours in image by calculating MRI image second order derivative, bone contours is carried out to gray scale enhancing, then bone contours is carried out to threshold value extraction, finally obtains MRI bone segmentation result.This system and method can effectively be removed bone hole problem in MRI image, even if still can accurately intactly extract the outline of bone in the inadequate clearly situation of bone boundaries; And can effectively solve the problem of bone and other bone pieces or the adhesion of bone perienchyma.
Certainly; the present invention also can have other various embodiments; in the situation that not deviating from spirit of the present invention and essence thereof; those of ordinary skill in the art are when making according to the present invention various corresponding changes and distortion, but these corresponding changes and distortion all should belong to the protection domain of the appended claim of the present invention.
Claims (10)
1. a method for the bone segmentation based on magnetic resonance image (MRI), is characterized in that, described method comprises the steps:
Calculation procedure: Hessian matrix and the eigenwert thereof of calculating each pixel in the first magnetic resonance image (MRI);
Extraction step: extract the schistose texture in described the first magnetic resonance image (MRI), and the schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtain the second magnetic resonance image (MRI);
Optimization Steps: described the second magnetic resonance image (MRI) is carried out to Threshold-connected calculating, extract the bone contours in described, and carry out smooth optimization, obtain the bone contours segmentation result in described the second magnetic resonance image (MRI).
2. method according to claim 1, is characterized in that, describedly before described calculation procedure, also comprises:
Image acquisition step: magnetic resonance imaging is carried out in the position that need to carry out bone segmentation, obtain first magnetic resonance image (MRI) at described position.
3. method according to claim 1, is characterized in that, described calculation procedure comprises:
For each pixel structure Gaussian filter G in described the first magnetic resonance image (MRI)
σ, the variance that wherein σ is Gaussian filter, to Gaussian filter G
σask its second derivative to obtain L
σ;
Adopt L
σdo convolution with described the first magnetic resonance image (MRI) and obtain I
xx(σ), I
xy(σ), I
xz(σ), I
yy(σ), I
yz(σ), I
zz(σ);
By L
σthe Hessian matrix H that the result obtaining with described the first magnetic resonance image (MRI) convolution forms is:
Calculate three eigenvalue λ of each Hessian matrix H
1, λ
2, λ
3.
4. method according to claim 3, is characterized in that, described extraction step comprises:
Construct the schistose texture feature discriminant function V in described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3), described function carries out structural determination to the Hessian proper value of matrix of the each pixel value of former magnetic resonance image (MRI), judges whether pixel is positioned at graph outline region;
Calculate schistose texture fundamental function L (the λ)=V of all pixels in described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3) * H (λ
1), wherein H (λ
1) be to weigh λ
1the function of action effect;
Differentiate result according to the strip-like features of each pixel schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtain described the second magnetic resonance image (MRI).
5. method according to claim 4, is characterized in that, described Optimization Steps comprises:
Set the gray threshold scope [I of the bone contours in described the second magnetic resonance image (MRI)
l, I
h];
The contour edge of the bone extracting at needs is chosen first pixel, the starting point using described the first pixel as Threshold-connected, and second pixel of search within the scope of the gray threshold of described bone contours;
Smooth optimization is carried out in the region of the first pixel within the scope of the gray threshold of described bone contours and the first pixel composition, obtain the bone contours segmentation result in described the first magnetic resonance image (MRI).
6. a system for the bone segmentation based on magnetic resonance image (MRI), is characterized in that, described system comprises:
Computing module: for calculating Hessian matrix and the eigenwert thereof of each pixel of the first magnetic resonance image (MRI);
Extraction module: for extracting the schistose texture of described the first magnetic resonance image (MRI), and the schistose texture in described the first magnetic resonance image (MRI) is carried out to gray scale enhancing, obtain the second magnetic resonance image (MRI);
Optimize module: for described the second magnetic resonance image (MRI) is carried out to Threshold-connected calculating, extract the bone contours in described, and carry out smooth optimization, obtain the bone contours segmentation result in described the second magnetic resonance image (MRI).
7. system according to claim 6, is characterized in that, described system also comprises:
Image collection module, for magnetic resonance imaging is carried out in the position that need to carry out bone segmentation, obtains first magnetic resonance image (MRI) at described position.
8. system according to claim 6, is characterized in that, described computing module comprises:
The first constructor module, is used to each pixel structure Gaussian filter G in described the first magnetic resonance image (MRI)
σ, the variance that wherein σ is Gaussian filter, to Gaussian filter G
σask its second derivative to obtain L
σ;
Convolution submodule, for adopting L
σdo convolution with described the first magnetic resonance image (MRI) and obtain I
xx(σ), I
xy(σ), I
xz(σ), I
yy(σ), I
yz(σ), I
zz(σ);
Matrix submodule, for by L
σthe result obtaining with described the first magnetic resonance image (MRI) convolution forms
Hessian matrix H is:
The first calculating sub module, for calculating three eigenvalue λ of each Hessian matrix H
1, λ
2, λ
3.
9. method according to claim 6, is characterized in that, described extraction module comprises:
The second constructor module, for constructing the schistose texture feature discriminant function V of described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3), described function carries out structural determination to the Hessian proper value of matrix of the each pixel value of former magnetic resonance image (MRI), judges whether pixel is positioned at graph outline region;
The second calculating sub module, for calculating schistose texture fundamental function L (the λ)=V of all pixels of described the first magnetic resonance image (MRI)
s(λ
1, λ
2, λ
3) * H (λ
1), wherein H (λ
1) be to weigh λ
1the function of action effect;
Gray scale enhancer module, carries out gray scale enhancing for differentiating result according to the strip-like features of each pixel to described the first magnetic resonance image (MRI) schistose texture, obtains described the second magnetic resonance image (MRI).
10. system according to claim 9, is characterized in that, described optimization module comprises:
Set submodule, for setting the gray threshold scope [I of bone contours of described the second magnetic resonance image (MRI)
l, I
h];
Search submodule, chooses first pixel for the contour edge of the bone in needs extraction, the starting point using described the first pixel as Threshold-connected, and second pixel of search within the scope of the gray threshold of described bone contours;
Optimize submodule, for smooth optimization is carried out in the region of the first pixel within the scope of the gray threshold of described bone contours and the first pixel composition, obtain the bone contours segmentation result in described the first magnetic resonance image (MRI).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410087926.9A CN103871057A (en) | 2014-03-11 | 2014-03-11 | Magnetic resonance image-based bone segmentation method and system thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410087926.9A CN103871057A (en) | 2014-03-11 | 2014-03-11 | Magnetic resonance image-based bone segmentation method and system thereof |
Publications (1)
Publication Number | Publication Date |
---|---|
CN103871057A true CN103871057A (en) | 2014-06-18 |
Family
ID=50909562
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410087926.9A Pending CN103871057A (en) | 2014-03-11 | 2014-03-11 | Magnetic resonance image-based bone segmentation method and system thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103871057A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105096332A (en) * | 2015-08-25 | 2015-11-25 | 上海联影医疗科技有限公司 | Medical image segmentation method and medical image segmentation device |
CN106023144A (en) * | 2016-05-06 | 2016-10-12 | 福建工程学院 | Femur segmenting method in tomographic image |
CN106815853A (en) * | 2016-12-14 | 2017-06-09 | 海纳医信(北京)软件科技有限责任公司 | To the dividing method and device of retinal vessel in eye fundus image |
US10181191B2 (en) | 2014-12-02 | 2019-01-15 | Shanghai United Imaging Healthcare Co., Ltd. | Methods and systems for identifying spine or bone regions in computed tomography image sequence |
CN110458850A (en) * | 2019-08-01 | 2019-11-15 | 北京灵医灵科技有限公司 | A kind of dividing method and segmenting system of large joint tissue |
-
2014
- 2014-03-11 CN CN201410087926.9A patent/CN103871057A/en active Pending
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10181191B2 (en) | 2014-12-02 | 2019-01-15 | Shanghai United Imaging Healthcare Co., Ltd. | Methods and systems for identifying spine or bone regions in computed tomography image sequence |
US11094067B2 (en) | 2014-12-02 | 2021-08-17 | Shanghai United Imaging Healthcare Co., Ltd. | Method and system for image processing |
CN105096332A (en) * | 2015-08-25 | 2015-11-25 | 上海联影医疗科技有限公司 | Medical image segmentation method and medical image segmentation device |
CN105096332B (en) * | 2015-08-25 | 2019-06-28 | 上海联影医疗科技有限公司 | Medical image cutting method and device |
CN106023144A (en) * | 2016-05-06 | 2016-10-12 | 福建工程学院 | Femur segmenting method in tomographic image |
CN106023144B (en) * | 2016-05-06 | 2019-05-31 | 福建工程学院 | Divide the method for femur in fault image |
CN106815853A (en) * | 2016-12-14 | 2017-06-09 | 海纳医信(北京)软件科技有限责任公司 | To the dividing method and device of retinal vessel in eye fundus image |
CN110458850A (en) * | 2019-08-01 | 2019-11-15 | 北京灵医灵科技有限公司 | A kind of dividing method and segmenting system of large joint tissue |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104809740B (en) | Knee cartilage image automatic segmentation method based on SVM and Hookean region growth | |
US9990743B2 (en) | Suppression of vascular structures in images | |
CN103871057A (en) | Magnetic resonance image-based bone segmentation method and system thereof | |
Hu et al. | Segmentation of brain from computed tomography head images | |
CN104794708A (en) | Atherosclerosis plaque composition dividing method based on multi-feature learning | |
Jabbar et al. | Using convolutional neural network for edge detection in musculoskeletal ultrasound images | |
CN110288698B (en) | Meniscus three-dimensional reconstruction system based on MRI | |
Barbieri et al. | Vertebral body segmentation of spine MR images using superpixels | |
Hacihaliloglu et al. | Statistical shape model to 3D ultrasound registration for spine interventions using enhanced local phase features | |
Ganvir et al. | Filtering method for pre-processing mammogram images for breast cancer detection | |
WO2011050454A1 (en) | Bone imagery segmentation method and apparatus | |
Soni et al. | CT scan based brain tumor recognition and extraction using Prewitt and morphological dilation | |
CN106780492A (en) | A kind of extraction method of key frame of CT pelvises image | |
Jalab et al. | Fractional Renyi entropy image enhancement for deep segmentation of kidney MRI | |
Xu et al. | Texture analysis improves level set segmentation of the anterior abdominal wall | |
CN103456004B (en) | Articular cartilage partitioning method and articular cartilage partitioning system based on image sheet structure enhancement | |
CN115131319A (en) | Method for deeply mining hippocampal sclerosis radiologic characteristics by combining filtering algorithm | |
Fiallos et al. | Automatic detection of injuries in mammograms using image analysis techniques | |
Ahmad et al. | Enhancement of MRI human thigh muscle segmentation by template-based framework | |
Zhang et al. | Blood vessel segmentation based on digital subtraction angiography sequence | |
Shaharuddin et al. | Feature analysis of kidney ultrasound image in four different ultrasound using gray level co-occurrence matrix (GLCM) and intensity histogram (IH) | |
Wang et al. | Automatic knee cartilage segmentation using multi-feature support vector machine and elastic region growing for magnetic resonance images | |
Howes et al. | Texture enhanced Statistical Region Merging with application to automatic knee bones segmentation from CT | |
Vasilache et al. | A unified method based on wavelet filtering and active contour models for segmentation of pelvic CT images | |
Dixit et al. | Proposing a Framework to Analyze Breast Cancer in Mammogram Images Using Global Thresholding, Gray Level Co‐Occurrence Matrix, and Convolutional Neural Network (CNN) |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140618 |