CN104239874B - A kind of organ blood vessel recognition methods and device - Google Patents
A kind of organ blood vessel recognition methods and device Download PDFInfo
- Publication number
- CN104239874B CN104239874B CN201410514806.2A CN201410514806A CN104239874B CN 104239874 B CN104239874 B CN 104239874B CN 201410514806 A CN201410514806 A CN 201410514806A CN 104239874 B CN104239874 B CN 104239874B
- Authority
- CN
- China
- Prior art keywords
- pixel
- blood vessel
- region
- target organ
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The embodiment of the invention discloses a kind of organ blood vessel recognition methods and device.This method includes：Judge whether each pixel in each sectioning image is located at target organ region respectively, each sectioning image is divided into target organ region and non-targeted organic region；The pixel in target organ region is pointed to, the corresponding blood vessel metric of pixel is calculated；Organ blood vessel is recognized according to obtained blood vessel metric.Judge whether each pixel in each sectioning image is located at target organ region respectively, each sectioning image is divided into target organ region and non-targeted organic region, corresponding blood vessel metric only is calculated to the pixel in target organ region, rather than corresponding blood vessel metric is calculated to whole pixels, blood vessel metric hour operation quantity is calculated to reduce, so as to improve the treatment effeciency of blood vessel identification, and need the data of storage to reduce in calculating process, save memory space.The problem of solving low organ blood vessel recognition efficiency and a large amount of internal memories of occupancy and video memory space.
Description
Technical field
The present invention relates to field of medical image processing, more particularly to a kind of organ blood vessel recognition methods and device.
Background technology
The organ information that CT Scan (Computer Tomography, CT) Angiography is obtained, only
It is the sectioning image of bidimensional.It is generally necessary to by image processor (Graphic Processing Unit, GPU) to slice map
As processing, 3-D view is obtained, the blood vessel structure in organ is shown.When GPU is handled sectioning image, first by computer center
Processor (Central Processing Unit, CPU) reads in all sectioning images among internal memory, then will need processing
Sectioning image dump in video memory.Usual GPU can utilize the corresponding Hai Sen of pixel (Hessian) matrixes and Hessian
The characteristic value of matrix, to identify the blood vessel structure in organ.Therefore, GPU is firstly the need of calculating sectioning image all pixels
The corresponding Hessian matrixes of point., it is necessary to be the corresponding Hessian matrixes of all pixels point first when calculating Hessian matrixes
6 elements, are distributed and all pixels point identical memory space respectively.Need to distribute the 6 of all pixels point memory space altogether
Times.For example, all sectioning images have 10,000 pixels, it is necessary to same when calculating the corresponding Hessian matrixes of 10,000 pixels
The memory space of Shi Wei 1 Elemental partition, 10,000 pixels, is the memory space of 10,000 pixels of the 2nd Elemental partition,
For the memory space of 10,000 pixels of the 3rd Elemental partition, likewise, remaining 3 element also distributes 10,000 pixels respectively
Memory space.Therefore, it is necessary to occupy substantial amounts of memory space during calculating Hessian matrixes.Calculate all pixels point correspondence
Hessian matrixes after, the characteristic value of the Hessian matrixes is gone out according to the corresponding Hessian matrix computations of each pixel.
After the characteristic value for calculating the corresponding Hessian matrixes of all pixels point, further according to this feature value, blood vessel metric function is utilized
Calculate the corresponding blood vessel metric of sectioning image all pixels point.Calculating is obtained into the corresponding blood vessel metric point of each pixel
Threshold value not with setting is compared, if greater than threshold value, it is determined that corresponding pixel is the pixel of vessel position, if
Less than threshold value, it is determined that corresponding pixel is not the pixel of vessel position, so as to identify organ blood vessel.Due to device
The sectioning image quantity of official is relatively more, when calculating blood vessel functional value using this method, takes the space of a large amount of internal memories and video memory,
It is huge and calculating overlong time is influenceed by amount of calculation again.
The content of the invention
The purpose of the embodiment of the present invention is to provide a kind of organ blood vessel recognition methods and device, to solve device in the prior art
The following problem that the identification of official's blood vessel is present：When carrying out the identification of organ blood vessel, cause blood vessel recognition efficiency low because computationally intensive, account for
With the space of a large amount of internal memories and video memory.The purpose of the present invention is achieved through the following technical solutions：
A kind of organ blood vessel recognition methods, including：
Judge whether each pixel in each sectioning image is located at target organ region, each sectioning image point respectively
For target organ region and non-targeted organic region；
The pixel in target organ region is pointed to, the corresponding blood vessel metric of pixel is calculated；
Organ blood vessel is recognized according to obtained blood vessel metric.
It is preferred that judge whether each pixel in each sectioning image is located at target organ region respectively, including：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ
Region and non-targeted organic region, wherein, the pixel value in the target organ region is the first pixel value, first pixel value
Different from the pixel value of nontarget area；
Morphological dilations are carried out to target organ region in each template image and the border of non-targeted organic region respectively；
The picture in the template image after the corresponding morphological dilations of each pixel in each sectioning image is judged respectively
Whether the pixel value of vegetarian refreshments is the first pixel value；If it is, the pixel of corresponding sectioning image is located at target organ region,
Otherwise, the pixel of corresponding sectioning image is located at non-targeted organic region.
It is preferred that judge whether each pixel in each sectioning image is located at target organ region respectively, including：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ
Region and non-targeted organic region, wherein, the pixel value of the non-targeted organic region is the first pixel value, first pixel
Pixel value of the value different from target area；
Morphological dilations are carried out to target organ region in each template image and the border of non-targeted organic region respectively；
The picture in the template image after the corresponding morphological dilations of each pixel in each sectioning image is judged respectively
Whether the pixel value of vegetarian refreshments is the first pixel value；If it is, the pixel of corresponding sectioning image is located at non-targeted organ area
Domain, otherwise, the pixel of corresponding sectioning image are located at target organ region.
It is preferred that judge whether each pixel in each sectioning image is located at target organ region respectively, including：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ
Region and non-targeted organic region, wherein, the pixel value in the target organ region is the first pixel value, first pixel value
Different from the pixel value of nontarget area；
Judging the pixel value of the pixel in the corresponding template image of each pixel in each sectioning image respectively is
No is the first pixel value；If it is, the pixel of corresponding sectioning image is located at target organ region, it is otherwise, corresponding to cut
The pixel of picture is located at non-targeted organic region.
It is preferred that judge whether each pixel in each sectioning image is located at target organ region respectively, including：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ
Region and non-targeted organic region, wherein, the pixel value of the non-targeted organic region is the first pixel value, first pixel
Pixel value of the value different from target area；
Judging the pixel value of the pixel in the corresponding template image of each pixel in each sectioning image respectively is
No is the first pixel value；If it is, the pixel of corresponding sectioning image is located at non-targeted organic region, it is otherwise, corresponding
The pixel of sectioning image is located at target organ region.
It is preferred that the corresponding blood vessel metric of pixel is calculated, including：
The corresponding extra large gloomy Hessian matrixes of pixel of predetermined quantity are calculated every time；And according to this calculating obtain it is every
The corresponding Hessian matrix computations Hessian matrix exgenvalues of individual pixel；It is corresponding according to the pixel that this calculating is obtained
Hessian matrix exgenvalues calculate the corresponding blood vessel metric of pixel；
N times are divided to complete the calculating of the corresponding blood vessel metric of whole pixels in target organ region, the N is not small
In 2 integer.
A kind of organ blood vessel identifying device, including：
Judging unit, for judging whether each pixel in each sectioning image is located at target organ region respectively,
Each sectioning image is divided into target organ region and non-targeted organic region；
Computing unit, the pixel for being pointed to target organ region calculates the corresponding blood vessel metric of pixel；
Recognition unit, for recognizing organ blood vessel according to obtained blood vessel metric.
It is preferred that the judging unit specifically for：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ
Region and non-targeted organic region, wherein, the pixel value in the target organ region is the first pixel value, first pixel value
Different from the pixel value of nontarget area；
Morphological dilations are carried out to target organ region in each template image and the border of non-targeted organic region respectively；
The picture in the template image after the corresponding morphological dilations of each pixel in each sectioning image is judged respectively
Whether the pixel value of vegetarian refreshments is the first pixel value；If it is, the pixel of corresponding sectioning image is located at target organ region,
Otherwise, the pixel of corresponding sectioning image is located at non-targeted organic region.
It is preferred that the judging unit specifically for：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ
Region and non-targeted organic region, wherein, the pixel value of the non-targeted organic region is the first pixel value, first pixel
Pixel value of the value different from target area；
Morphological dilations are carried out to target organ region in each template image and the border of non-targeted organic region respectively；
The picture in the template image after the corresponding morphological dilations of each pixel in each sectioning image is judged respectively
Whether the pixel value of vegetarian refreshments is the first pixel value；If it is, the pixel of corresponding sectioning image is located at non-targeted organ area
Domain, otherwise, the pixel of corresponding sectioning image are located at target organ region.
It is preferred that the computing unit specifically for：
The corresponding Hessian matrixes of pixel of predetermined quantity are calculated every time；And each picture obtained according to this calculating
The corresponding Hessian matrix computations Hessian matrix exgenvalues of vegetarian refreshments；It is corresponding according to the pixel that this calculating is obtained
Hessian matrix exgenvalues calculate the corresponding blood vessel metric of pixel；
N times are divided to complete the calculating of the corresponding blood vessel metric of whole pixels in target organ region, the N is not small
In 2 integer.
The embodiment of the present invention has the beneficial effect that：
In the embodiment of the present invention, judge whether each pixel in each sectioning image is located at target organ area respectively
Domain, each sectioning image is divided into target organ region and non-targeted organic region, and only the pixel in target organ region is calculated
Corresponding blood vessel metric, rather than corresponding blood vessel metric is calculated to whole pixels, calculate blood vessel metric hour operation quantity
Reduce, so as to improve the treatment effeciency of blood vessel identification, and need the data of storage to reduce in calculating process, save storage empty
Between.
Target organ region and the gray value contrast of non-targeted organic region boundary are larger, therefore, its blood vessel measurement
The feature of value is approximate with the feature of the blood vessel metric at blood vessel, easily by target organ region and non-targeted organic region border
It is mistakenly identified as blood vessel.Technical scheme provided in an embodiment of the present invention, further, by the target organ region in template image with
The border of non-targeted organic region carries out morphological dilations, obtains the border that pixel value is different from target organ area pixel value
Band, when being participated in the judgement of pixel in corresponding sectioning image using the template image by morphological dilations, then need not
Calculate sectioning image target organ region blood vessel metric corresponding with the pixel on non-targeted organic region border, from without
Target organ border can be mistakenly identified as to the blood vessel of target organ, target organ border is eliminated and the blood vessel of target organ is recognized
Interference.
Further, obtain each, it is necessary to calculate respectively first when calculating Hessian matrixes in the prior art
6 elements of Hessian matrixes, 6 elements of each Hessian matrixes then obtained according to calculating are obtained accordingly
Hessian matrixes.Therefore, it is necessary to preserve certain corresponding unitary of all pixels point successively during Hessian matrixes are calculated
Element.Technical scheme provided in an embodiment of the present invention, calculates the corresponding Hessian matrixes of pixel of predetermined quantity every time, and counts
Corresponding Hessian matrix exgenvalues are calculated, so that the corresponding blood vessel metric of the pixel for calculating predetermined quantity；According to the process
Divide the calculating of the corresponding blood vessel metric of whole pixels completed at least twice in target organ region, then, it is only necessary to protect
Each element of the corresponding Hessian matrixes of pixel of predetermined quantity is deposited, it is corresponding without preserving whole pixels
Each element of Hessian matrixes, saves the memory space that each element of Hessian matrixes needs to take.
Brief description of the drawings
Fig. 1 is a kind of flow chart of organ blood vessel recognition methods provided in an embodiment of the present invention；
Fig. 2 for it is provided in an embodiment of the present invention the first judge respectively each pixel in each sectioning image whether position
Method flow diagram in target organ region；
Fig. 3 be each pixel for judging respectively in each sectioning image for second provided in an embodiment of the present invention whether position
Method flow diagram in target organ region；
Fig. 4 for provided in an embodiment of the present invention the third judge respectively each pixel in each sectioning image whether position
Method flow diagram in target organ region；
Fig. 5 be 4th kind provided in an embodiment of the present invention each pixel for judging respectively in each sectioning image whether position
Method flow diagram in target organ region；
Fig. 6 is a kind of flow chart of organ blood vessel recognition methods by taking liver vessel as an example provided in an embodiment of the present invention；
Fig. 7 is the liver primary template image that reads in the embodiment of the present invention；
Fig. 8 a are the liver vessel recognition methods of the prior art by taking liver vessel as an example provided in the embodiment of the present invention
Effect；
Fig. 8 b are the effect of the organ blood vessel recognition methods by taking liver vessel as an example provided in the embodiment of the present invention；
Fig. 9 is a kind of organ blood vessel identifying device figure provided in an embodiment of the present invention.
Embodiment
A kind of organ blood vessel recognition methods provided with reference to the accompanying drawings and examples the present invention and device carry out more detailed
Carefully illustrate.
The embodiments of the invention provide a kind of organ blood vessel recognition methods, as shown in figure 1, comprising the following steps that：
Step 110：Judge whether each pixel in each sectioning image is located at target organ region respectively, each cut
Picture is divided into target organ region and non-targeted organic region.
Wherein, sectioning image can be gray level image, and accordingly, pixel value is represented with gray value.
Wherein, target organ refers to the organ of pending blood vessel identification.
Step 120：The pixel in target organ region is pointed to, the corresponding blood vessel metric of pixel is calculated.
Step 130：Organ blood vessel is recognized according to obtained blood vessel metric.
In the embodiment of the present invention, judge whether each pixel in each sectioning image is located at target organ area respectively
Domain, each sectioning image is divided into target organ region and non-targeted organic region, and only the pixel in target organ region is calculated
Corresponding blood vessel metric, rather than corresponding blood vessel metric is calculated to whole pixels, calculate blood vessel metric hour operation quantity
Reduce, so as to improve the treatment effeciency of blood vessel identification, and need the data of storage to reduce in calculating process, save storage empty
Between.
In the embodiment of the present invention, judge whether each pixel in each sectioning image is located at target organ region respectively
Method have a variety of.It is preferred that judging whether is each pixel in sectioning image by the corresponding template image of sectioning image
Positioned at target organ region.In the embodiment of the present invention, template image is divided into two regions, one is target organ region, secondly
It is non-target organ region.Template image is to use existing image dividing processing scheme, and corresponding sectioning image is divided into
Target organ region and non-targeted organic region, and by setting so that what the pixel value difference in the two regions was obtained.
Enumerate below and several judge whether is each pixel in sectioning image by the corresponding template image of sectioning image
Implementation positioned at target organ region.
Whether the first each pixel judged respectively in each sectioning image provided in an embodiment of the present invention is located at mesh
The method of organic region is marked as shown in Fig. 2 comprising the following steps that：
Step 210：Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into
Target organ region and non-targeted organic region.
Wherein, the pixel value in the target organ region is the first pixel value, and first pixel value is different from non-targeted organ
The pixel value in region.
Wherein, the pixel value in the target organ region is that the first pixel value refers to, each pixel in target organ region
Pixel value be the first pixel value.
The corresponding template image of each sectioning image in the step is to use existing image dividing processing scheme, will be right
The sectioning image answered is divided into target organ region and non-targeted organic region, and by the pixel value in target organ region therein
It is set to what the first pixel was worth to.
Wherein, template image can be gray level image, and accordingly, pixel value is represented with gray value.
Step 220：Shape is carried out to target organ region in each template image and the border of non-targeted organic region respectively
State expands.
In the step, when carrying out morphological dilations, the border in the target organ region is simultaneously to target organ region and non-
Target organ region expands.By morphological dilations, the pixel value for expanding the pixel being related to is configured differently than first
The pixel value of pixel value, so as to obtain the boundary strip that pixel value is different from the first pixel value.
Wherein it is determined that the implementation on the border of target area and nontarget area has a variety of, such as：By each template
The pixel value all for the pixel of nonzero value of any one four neighborhood of image is set to zero, so as to obtain target area and non-mesh
Mark the border in region.
In another example, search in template image, there is the pixel of the first pixel value and non-first pixel value in four neighborhoods, search
The pixel arrived constitutes the border of target area and nontarget area.
Step 230：The Prototype drawing after the corresponding morphological dilations of each pixel in each sectioning image is judged respectively
Whether the pixel value of the pixel as in is the first pixel value；If it is, the pixel of corresponding sectioning image is located at target
Organic region, otherwise, the pixel of corresponding sectioning image are located at non-targeted organic region.
In the embodiment of the present invention, if the pixel a of sectioning image is located at the line n m row in sectioning image, with template
The pixel a ' of image is located at the line n m row in template image, then the pixel a of sectioning image and the picture in template image
Vegetarian refreshments a ' correspondences.
Whether each pixel judged respectively in each sectioning image for second provided in an embodiment of the present invention is located at mesh
The method of organic region is marked as shown in figure 3, comprising the following steps that：
Step 310：Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into
Target organ region and non-targeted organic region.
Wherein, the pixel value of the non-targeted organic region is the first pixel value, and first pixel value is different from target organ
The pixel value in region.
Wherein, the pixel value of the non-targeted organic region is that the first pixel value refers to, each picture in non-targeted organic region
The pixel value of vegetarian refreshments is the first pixel value.
The corresponding template image of each sectioning image in the step is to use existing image dividing processing scheme, will be right
The sectioning image answered is divided into target organ region and non-targeted organic region, and by the pixel of non-targeted organic region therein
Value is set to what the first pixel was worth to.
Step 320：Shape is carried out to target organ region in each template image and the border of non-targeted organic region respectively
State expands.
In the step, when carrying out morphological dilations, the border in the target organ region is simultaneously to target organ region and non-
Target organ region expands.By morphological dilations, the pixel value for expanding the pixel being related to is set to the first pixel value,
So as to obtain the boundary strip that pixel value is the first pixel value.
Step 330：The Prototype drawing after the corresponding morphological dilations of each pixel in each sectioning image is judged respectively
Whether the pixel value of the pixel as in is the first pixel value；If it is, the pixel of corresponding sectioning image is located at non-mesh
Organic region is marked, otherwise, the pixel of corresponding sectioning image is located at target organ region.
Target organ region and the gray value contrast of non-targeted organic region boundary are larger, therefore, its blood vessel measurement
The feature of value is approximate with the feature of the blood vessel metric at blood vessel, easily by target organ region and non-targeted organic region border
It is mistakenly identified as blood vessel.In two kinds of implementations that above-described embodiment is provided, by the target organ region in template image and non-mesh
The border for marking organic region carries out morphological dilations, obtains the boundary strip that pixel value is different from target organ area pixel value,
During using participating in the judgement of pixel in corresponding sectioning image by the template image of morphological dilations, then it need not calculate and cut
The target organ region of picture blood vessel metric corresponding with the pixel on non-targeted organic region border, so that will not be by mesh
Mark organ boundaries are mistakenly identified as the blood vessel of target organ, eliminate target organ border to doing that the blood vessel of target organ is recognized
Disturb.
Whether the third each pixel judged respectively in each sectioning image provided in an embodiment of the present invention is located at mesh
The method of organic region is marked as shown in figure 4, comprising the following steps that：
Step 410：Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into
Target organ region and non-targeted organic region, wherein, the pixel value in the target organ region is the first pixel value, first picture
Pixel value of the element value different from non-targeted organic region；
Step 420：Pixel in the corresponding template image of each pixel in each sectioning image is judged respectively
Whether pixel value is the first pixel value；If it is, the pixel of corresponding sectioning image is located at target organ region, otherwise,
The pixel of corresponding sectioning image is located at non-targeted organic region.
Whether 4th kind provided in an embodiment of the present invention each pixel judged respectively in each sectioning image is located at mesh
The method of organic region is marked as shown in figure 5, comprising the following steps that：
Step 510：Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into
Target organ region and non-targeted organic region, wherein, the pixel value of the non-targeted organic region is the first pixel value, and this first
Pixel value is different from the pixel value in target organ region；
Step 520：Pixel in the corresponding template image of each pixel in each sectioning image is judged respectively
Whether pixel value is the first pixel value；If it is, the pixel of corresponding sectioning image is located at non-targeted organic region, it is no
Then, the pixel of corresponding sectioning image is located at target organ region.
It should be pointed out that judging whether each pixel in each sectioning image is located at target organ region respectively
Method is not limited only to approach described above.
In above-mentioned steps 120, calculating the method for the corresponding blood vessel metric of pixel has a variety of, and the embodiment of the present invention is enumerated
One of which method, is comprised the following steps that：
The corresponding Hessian matrixes of pixel of predetermined quantity are calculated every time；And each picture obtained according to this calculating
The corresponding Hessian matrix computations Hessian matrix exgenvalues of vegetarian refreshments；It is corresponding according to the pixel that this calculating is obtained
Hessian matrix exgenvalues calculate the corresponding blood vessel metric of pixel.
N times are divided to complete the calculating of the corresponding blood vessel metric of whole pixels in target organ region, the times N is not
Integer less than 2.
The quantity that predetermined quantity carries out the pixel that corresponding blood vessel metric is calculated according to actual needs is calculated every time
It is determined that.
Wherein, the times N can carry out according to actual needs the pixel that corresponding blood vessel metric is calculated quantity it is true
Fixed, N is the integer not less than 2.
In the embodiment of the present invention, obtain each, it is necessary to calculate first when calculating Hessian matrixes in the prior art
6 elements of Hessian matrixes, 6 elements of each Hessian matrixes then obtained according to calculating are obtained accordingly
Hessian matrixes.Therefore, it is necessary to preserve certain corresponding unitary of all pixels point successively during Hessian matrixes are calculated
Element.Technical scheme provided in an embodiment of the present invention, calculates the corresponding Hessian matrixes of pixel of predetermined quantity every time, and counts
Corresponding Hessian matrix exgenvalues are calculated, so that the corresponding blood vessel metric of the pixel for calculating predetermined quantity；According to the process
Divide the calculating of the corresponding blood vessel metric of whole pixels completed at least twice in target organ region, then, it is only necessary to protect
Each element of the corresponding Hessian matrixes of pixel of predetermined quantity is deposited, it is corresponding without preserving whole pixels
Each element of Hessian matrixes, saves the memory space that each element of Hessian matrixes needs to take.
Because the blood vessel of target organ has different yardsticks, for each pixel, in the yardstick that standard deviation is sigma
Lower calculating Hessian matrixes, the expression formula of Hessian matrixes is as follows：
I_sigma represent after filtering after pixel, x, y represent the position in sectioning image of the pixel, z
The numbering of sectioning image where representing the pixel.The matrix is made up of the second-order partial differential coefficient of the pixel, there is 9 in matrix
Partial derivative.Because the Hessian matrixes have symmetry, therefore it can only calculate 6 partial derivatives.When calculating, only 6 need to be calculated
Individual element can be obtained by Hessian matrixes.
Further, sigma is used2It is multiplied by Hessian matrixes to be normalized, so as to select most under multiple dimensioned
Big blood vessel metric, now, Hessian matrixes are rewritten into：
In the step, three eigenvalue λs of the corresponding Hessian matrixes of the pixel are calculated1、λ2、λ3(|λ1|≤|λ2|≤
|λ3|).For 3-dimensional (three Dimensional, 3D) voxel, the voxel is expression of the pixel in 3-D view
Mode, because the change of gray value in the direction of the blood vessel is smaller, and it is changed greatly perpendicular to vessel directions.Minimum characteristic value
λ1Corresponding characteristic vector represents the direction in the minimum direction, i.e. blood vessel of curvature；And larger eigenvalue λ2And λ3(both almost
It is equal and be negative value) corresponding characteristic vector constitutes a plane perpendicular to vessel directions.
In above-described embodiment, calculating the blood vessel metric function of blood vessel metric has a variety of, it is preferred that the blood vessel metric is used
Forlan Ji (Frangi) blood vessel metric function is calculated.Because the blood vessel structure in target organ is of different sizes characteristic, because
This needs to obtain its blood vessel metric with multiple dimensioned method, i.e.,：
Wherein, Vesselness represents blood vessel metric, and sigma_low represents the smallest dimension of sectioning image medium vessels,
Sigma_high represents the out to out of sectioning image medium vessels, and Vesselness_Frangi represents the blood vessel degree of different scale
Value.
In above-described embodiment, when calculating the corresponding blood vessel metric per a collection of pixel, serial computing can be both used,
Parallel computation can be used, it is preferred that use parallel computation.Selectable parallel calculating method has a variety of, it is preferred that using logical
With parallel computation framework (Computer Unified Device Architecture, CUDA) parallel computation.CUDA is to be based on
GPU universal parallel computing architecture, can carry out C language programming, improve the flexibility of programming.If judging, pixel is located at
In part or all of target organ region, then CUDA platforms are based on using the corresponding blood vessel of parallel algorithm calculating pixel by GPU
Metric；If pixel is judged not in part or all of target organ region, by the corresponding blood vessel metric of pixel
Zero is calculated as, or any operation is not carried out to the pixel.
The corresponding blood vessel metric of pixel is calculated by using the mode of parallel computation, so as to further increase organ
The treatment effeciency of the identification of blood vessel.
By taking liver vessel as an example, embodiment such as Fig. 6 of organ blood vessel recognition methods provided in an embodiment of the present invention
Shown, idiographic flow is as follows：
Step 610：Read in liver section image and liver template image.
Step 620：The corresponding liver mould of liver section image procossing each liver section image read according to step 610
Plate image, including：It is by any one four neighborhood that each liver section image of acquisition distinguishes corresponding liver template image
The pixel of nonzero value is set to zero, determines the border of liver area and non-liver area；To obtained liver area and non-liver
The border in dirty district domain carries out morphological dilations to liver area and non-liver area, obtains the liver boundary that pixel value is nonzero value
Band.In the Texture memory that template image after processing is read into GPU video memorys.
Wherein, when carrying out morphological dilations, the number of times of expansion can be set according to different liver section images, preferably
, 10 expansions are carried out to the border of liver area.
Step 630：The liver section image of reading is filtered and contrast enhancing etc. pretreatment operation.Will pretreatment
Liver section image afterwards is read into the Texture memory of GPU video memorys.
The step is used to suppress noise and strengthen the contrast of blood vessel and background, beneficial to the identification of liver medium vessels.
Step 640：Set and calculate the sigma yardsticks of liver vessel metric for blood vessel smallest dimension sigma_low.
In the embodiment of the present invention, sigma yardsticks refer to the blood vessel yardstick that standard deviation is sigma.
Step 650：Judge whether sigma yardsticks are less than or equal to the maximum yardstick sigma_high of blood vessel, if so, then
Perform step 660；Otherwise, step 6170 is performed.
Step 660：The 3D gaussian filterings of sigma yardsticks are carried out to liver section image.
In the step, when carrying out 3D gaussian filterings to pretreated liver section image, using its linear separability,
Three independent one-dimensional spaces for being broken down into x, y, z direction are respectively calculated, so as to reduce the complexity of calculating.
Above-mentioned steps 620 are with step 630~step 660 without timing requirements.
Step 670：Order reads a filtered pixel.
In the step, the quantity of the pixel of predetermined computation is one.Specifically, for the picture of different liver section images
Vegetarian refreshments, according to putting in order for liver section image；For the pixel in same liver section image, pushed up according to from upper left
The pixel of point to bottom right vertex puts in order.
Step 680：Whether the pixel read in judgment step 670 is located in partial liver region, if so, then performing step
Rapid 6100；Otherwise, step 690 is performed.
Step 690：Frangi blood vessel metric Vesselness (x, y, z) are set to 0, and perform step 6150.
Step 6100：Calculate the corresponding Hessian matrixes of pixel.
Step 6110：Calculate the characteristic value of the corresponding Hessian matrixes of pixel.
Step 6120：Calculate the Frangi blood vessel metrics Vesselness_Frangi of pixel.
Step 6130：Judge whether the blood vessel metric Vesselness_Frangi of sigma yardsticks is more than blood vessel metric
The maximum blood vessel metric that Vesselness, Vesselness obtain for calculating.If so, then performing step 6140, otherwise hold
Row step 6150.
Wherein, the Vesselness values at Vesselness ∈ [0,1], liver vessel are non-zero, for increasing blood vessel ash
Angle value and the contrast of other tissue gray values in liver, to realize the enhancing of blood vessel.
Step 6140：It is Vesselness_Frangi to set blood vessel metric Vesselness.
Step 6150：Judge whether all pixels point has traveled through, if so, then performing step 6160；Otherwise, step is performed
670。
Step 6160：Change sigma yardsticks, the scale-value changed every time is sigma_step, and performs step 650.
Step 6170：Final result is obtained, liver vessel identification is carried out.
Wherein, the process of processing liver template image is carried out in CPU；Pixel in partial liver region,
CUDA platforms are based on using the corresponding blood vessel metric of parallel algorithm calculating pixel using by GPU.
During calculating, parameter sigma_low, sigma_high, sigma_step are according to the size of actual liver section image
It is configured.
In above-described embodiment, specific liver primary template image is as shown in fig. 7, the region shown in rectangle frame is non-liver
Region.It is the liver vessel recognition effect obtained according to prior art as shown in Fig. 8 a rectangle frames.The gray value pair of liver boundary
More larger than degree, therefore, the feature of the blood vessel metric of liver boundary is approximate with the feature of the blood vessel metric at liver vessel, holds
Liver boundary is easily mistakenly identified as liver vessel.It is the liver vessel identification that the embodiment of the present invention is obtained as shown in Fig. 8 b rectangles
Effect.Due to the liver boundary in template image is carried out into morphological dilations, pixel value is obtained different from liver area pixel value
Liver boundary band, using after corresponding morphological dilations template image participate in sectioning image in pixel judgement when, liver
The pixel on dirty border is judged as not being located at target organ region, it is not necessary to calculate blood vessel metric, therefore, will not be by liver
Border is mistakenly identified as liver vessel.
In above-described embodiment, constitute 3D organic images one group of liver section image size can be expressed as (x × y ×
Z), wherein, x represents the quantity of a line in each liver section image (or one row) pixel, and y represents each liver section image
In row (or a line) pixel quantity, z represents the quantity of this group of liver section image.Such as this group liver section image is big
Small is (512 × 512 × 76), represents a line in this group of liver section image, each liver section image (or a row) pixel
Quantity be 512, the quantity of row (or a line) pixel is 512, this group of liver section image in each liver section image
There are 76 liver section images.When recognizing liver vessel, for less liver section image, using prior art, identification
Liver vessel needs 1375.56 seconds.The time that identification liver vessel of the embodiment of the present invention needs is 33.49 seconds, with prior art
The speed-up ratio compared is 41.07 times.For larger liver section image, such as one group liver section image size for (512 ×
512 × 301), the time that identification liver vessel of the embodiment of the present invention needs is 133.76 seconds.And prior art can not then carry out liver
Dirty blood vessel identification, because when calculating the Hessian matrixes of all liver section images, it is necessary to which the internal memory of distribution is at least 2107
MB internal memory, this can not be met in the Memory System and video memory space run now, so also can not just calculate knot
Really.
Embodiment also provides a kind of organ blood vessel identifying device, as shown in figure 9, specifically including judging unit 901, calculating single
Member 902 and recognition unit 903.
Judging unit 901, for judging whether each pixel in each sectioning image is located at target organ area respectively
Domain, each sectioning image is divided into target organ region and non-targeted organic region.
Computing unit 902, the pixel for being pointed to target organ region calculates the corresponding blood vessel measurement of pixel
Value.
Recognition unit 903, for recognizing organ blood vessel according to obtained blood vessel metric.
It is preferred that the judging unit 901 specifically for：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ
Region and non-targeted organic region, wherein, the pixel value in the target organ region is the first pixel value, and first pixel value is different
Pixel value in nontarget area.
Morphological dilations are carried out to target organ region in each template image and the border of non-targeted organic region respectively.
The picture in the template image after the corresponding morphological dilations of each pixel in each sectioning image is judged respectively
Whether the pixel value of vegetarian refreshments is the first pixel value；If it is, the pixel of corresponding sectioning image is located at target organ region,
Otherwise, the pixel of corresponding sectioning image is located at non-targeted organic region.
It is preferred that the judging unit 901 specifically for：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ
Region and non-targeted organic region, wherein, the pixel value of the non-targeted organic region is the first pixel value, and first pixel value is not
It is same as the pixel value of target area.
Morphological dilations are carried out to target organ region in each template image and the border of non-targeted organic region respectively.
The picture in the template image after the corresponding morphological dilations of each pixel in each sectioning image is judged respectively
Whether the pixel value of vegetarian refreshments is the first pixel value；If it is, the pixel of corresponding sectioning image is located at non-targeted organ area
Domain, otherwise, the pixel of corresponding sectioning image are located at target organ region.
It is preferred that the computing unit 902 specifically for：
The corresponding Hessian matrixes of pixel of predetermined quantity are calculated every time；And each picture obtained according to this calculating
The corresponding Hessian matrix computations Hessian matrix exgenvalues of vegetarian refreshments；It is corresponding according to the pixel that this calculating is obtained
Hessian matrix exgenvalues calculate the corresponding blood vessel metric of pixel.
Point n times complete the calculating of the corresponding blood vessel metric of whole pixels in target organ region, the N be not less than
2 integer.
It should be understood by those skilled in the art that, embodiments of the invention can be provided as method, system or computer program
Product.Therefore, the present invention can be using the reality in terms of complete hardware embodiment, complete software embodiment or combination software and hardware
Apply the form of example.Moreover, the present invention can be used in one or more computers for wherein including computer usable program code
The computer program production that usable storage medium is implemented on (including but is not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of product.
The present invention is the flow with reference to method according to embodiments of the present invention, equipment (system) and computer program product
Figure and/or block diagram are described.It should be understood that can be by every first-class in computer program instructions implementation process figure and/or block diagram
Journey and/or the flow in square frame and flow chart and/or block diagram and/or the combination of square frame.These computer programs can be provided
The processor of all-purpose computer, special-purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce
A raw machine so that produced by the instruction of computer or the computing device of other programmable data processing devices for real
The device for the function of being specified in present one flow of flow chart or one square frame of multiple flows and/or block diagram or multiple square frames.
These computer program instructions, which may be alternatively stored in, can guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works so that the instruction being stored in the computer-readable memory, which is produced, to be included referring to
Make the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one square frame of block diagram or
The function of being specified in multiple square frames.
These computer program instructions can be also loaded into computer or other programmable data processing devices so that in meter
Series of operation steps is performed on calculation machine or other programmable devices to produce computer implemented processing, thus in computer or
The instruction performed on other programmable devices is provided for realizing in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in individual square frame or multiple square frames.
, but those skilled in the art once know basic creation although preferred embodiments of the present invention have been described
Property concept, then can make other change and modification to these embodiments.So, appended claims are intended to be construed to include excellent
Select embodiment and fall into having altered and changing for the scope of the invention.
Obviously, those skilled in the art can carry out the essence of various changes and modification without departing from the present invention to the present invention
God and scope.So, if these modifications and variations of the present invention belong to the scope of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to comprising including these changes and modification.
Claims (7)
1. a kind of organ blood vessel recognition methods, it is characterised in that including：
If the pixel value of the pixel in the corresponding template image of each pixel in each sectioning image is the first pixel value,
Then the pixel of corresponding sectioning image be located at target organ region, wherein, each sectioning image be divided into target organ region and
Non-targeted organic region；
The pixel in target organ region is pointed to, the corresponding blood vessel metric of pixel is calculated；
Organ blood vessel is recognized according to obtained blood vessel metric.
2. according to the method described in claim 1, it is characterised in that if each pixel correspondence in each described sectioning image
Template image in pixel pixel value be the first pixel value, then the pixel of corresponding sectioning image be located at target organ
Region, is specifically included：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ region
With non-targeted organic region, wherein, the pixel value in the target organ region is the first pixel value, and first pixel value is different
Pixel value in nontarget area；
Morphological dilations are carried out to target organ region in each template image and the border of non-targeted organic region respectively；
If the pixel value of the pixel in the corresponding morphological dilations rear pattern plate image of each pixel in each sectioning image
For the first pixel value, then the pixel of corresponding sectioning image is positioned at target organ region, otherwise, the picture of corresponding sectioning image
Vegetarian refreshments is located at non-targeted organic region.
3. according to the method described in claim 1, it is characterised in that if each pixel correspondence in each described sectioning image
Template image in pixel pixel value be the first pixel value, then the pixel of corresponding sectioning image be located at target organ
Region, is specifically included：Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into mesh
Organic region and non-targeted organic region are marked, wherein, the pixel value in the target organ region is the first pixel value, described first
Pixel value is different from the pixel value of nontarget area；
If the pixel value of the pixel in the corresponding template image of each pixel in each sectioning image is the first pixel value,
Then the pixel of corresponding sectioning image is located at target organ region, otherwise, and the pixel of corresponding sectioning image is located at non-mesh
Mark organic region.
4. according to any one of claims 1 to 3 methods described, it is characterised in that calculate the corresponding blood vessel metric of pixel, bag
Include：
The corresponding extra large gloomy Hessian matrixes of pixel of predetermined quantity are calculated every time；And each picture obtained according to this calculating
The corresponding Hessian matrix computations Hessian matrix exgenvalues of vegetarian refreshments；It is corresponding according to the pixel that this calculating is obtained
Hessian matrix exgenvalues calculate the corresponding blood vessel metric of pixel；
N times are divided to complete the calculating of the corresponding blood vessel metric of whole pixels in target organ region, the N is not less than 2
Integer.
5. a kind of organ blood vessel identifying device, it is characterised in that including：
Judging unit, if the pixel value for the pixel in the corresponding template image of each pixel in each sectioning image
For the first pixel value, then the pixel of corresponding sectioning image is located at target organ region, wherein, each sectioning image is divided into mesh
Mark organic region and non-targeted organic region；
Computing unit, the pixel for being pointed to target organ region calculates the corresponding blood vessel metric of pixel；
Recognition unit, for recognizing organ blood vessel according to obtained blood vessel metric.
6. device according to claim 5, it is characterised in that the judging unit specifically for：
Obtain the corresponding template image of each sectioning image, the corresponding template image of each sectioning image is divided into target organ region
With non-targeted organic region, wherein, the pixel value in the target organ region is the first pixel value, and first pixel value is different
Pixel value in nontarget area；
Morphological dilations are carried out to target organ region in each template image and the border of non-targeted organic region respectively；
If the pixel value of the pixel in the corresponding morphological dilations rear pattern plate image of each pixel in each sectioning image
For the first pixel value, then the pixel of corresponding sectioning image is positioned at target organ region, otherwise, the picture of corresponding sectioning image
Vegetarian refreshments is located at non-targeted organic region.
7. the device according to claim 5 or 6, it is characterised in that the computing unit specifically for：
The corresponding Hessian matrixes of pixel of predetermined quantity are calculated every time；And each pixel obtained according to this calculating
Corresponding Hessian matrix computations Hessian matrix exgenvalues；The corresponding Hessian of pixel obtained according to this calculating
Matrix exgenvalue calculates the corresponding blood vessel metric of pixel；
N times are divided to complete the calculating of the corresponding blood vessel metric of whole pixels in target organ region, the N is not less than 2
Integer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410514806.2A CN104239874B (en) | 2014-09-29 | 2014-09-29 | A kind of organ blood vessel recognition methods and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410514806.2A CN104239874B (en) | 2014-09-29 | 2014-09-29 | A kind of organ blood vessel recognition methods and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104239874A CN104239874A (en) | 2014-12-24 |
| CN104239874B true CN104239874B (en) | 2017-11-03 |
Family
ID=52227900
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410514806.2A Active CN104239874B (en) | 2014-09-29 | 2014-09-29 | A kind of organ blood vessel recognition methods and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104239874B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6378715B2 (en) * | 2016-04-21 | 2018-08-22 | ゼネラル・エレクトリック・カンパニイ | Blood vessel detection device, magnetic resonance imaging device, and program |
| CN107610095A (en) * | 2017-08-04 | 2018-01-19 | 南京邮电大学 | Heart CT coronary artery full-automatic partition methods based on image co-registration |
| CN109758119B (en) * | 2019-01-28 | 2022-03-08 | 广东唯仁医疗科技有限公司 | Tumor microvascular image processing device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1943513A (en) * | 2006-11-08 | 2007-04-11 | 沈阳东软医疗系统有限公司 | Processing device method and system for CT image |
| CN101425186A (en) * | 2008-11-17 | 2009-05-06 | 华中科技大学 | Liver subsection method based on CT image and system thereof |
| CN101807257A (en) * | 2010-05-12 | 2010-08-18 | 上海交通大学 | Method for identifying information of image tag |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8488852B2 (en) * | 2011-02-15 | 2013-07-16 | Siemens Aktiengesellschaft | Medical imaging system for segementing blood vessel |
-
2014
- 2014-09-29 CN CN201410514806.2A patent/CN104239874B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1943513A (en) * | 2006-11-08 | 2007-04-11 | 沈阳东软医疗系统有限公司 | Processing device method and system for CT image |
| CN101425186A (en) * | 2008-11-17 | 2009-05-06 | 华中科技大学 | Liver subsection method based on CT image and system thereof |
| CN101807257A (en) * | 2010-05-12 | 2010-08-18 | 上海交通大学 | Method for identifying information of image tag |
Non-Patent Citations (1)
| Title |
|---|
| 基于血管增强分割的三维肺结节自动检测;潘烁等;《计算机应用与软件》;20140531;第31卷(第5期);期刊全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104239874A (en) | 2014-12-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104809723B (en) | The three-dimensional CT image for liver automatic division method of algorithm is cut based on super voxel and figure | |
| US20200265632A1 (en) | System and method for real-time rendering of complex data | |
| CN105389811B (en) | A kind of multi-modality medical image processing method split based on multilevel threshold | |
| CN107832807A (en) | A kind of image processing method and system | |
| TWI398158B (en) | Method for generating the depth of a stereo image | |
| CN106062824B (en) | edge detecting device and edge detection method | |
| US20140375631A1 (en) | Image generation apparatus, method, and medium with image generation program recorded thereon | |
| CN107369131A (en) | Conspicuousness detection method, device, storage medium and the processor of image | |
| CN112950554B (en) | Lung lobe segmentation optimization method and system based on lung segmentation | |
| CN102096939B (en) | Medical mass data-oriented multi-resolution volume rendering method | |
| CN110400299A (en) | A kind of method and device of lung's pleural effusion detection | |
| CN104239874B (en) | A kind of organ blood vessel recognition methods and device | |
| CN111462071B (en) | Image processing method and system | |
| CN103927730B (en) | Image noise reduction method based on Primal Sketch correction and matrix filling | |
| CN102419864B (en) | Method and device for extracting skeletons of brain CT (computerized tomography) image | |
| CN110992310A (en) | Method and device for determining partition where mediastinal lymph node is located | |
| WO2024002110A1 (en) | Methods and systems for determining image control point | |
| CN112435211B (en) | Method for describing and matching dense contour feature points in endoscope image sequence | |
| CN105956587B (en) | A kind of knee joint magnetic resonance image sequence meniscus extraction method based on shape constraining | |
| US20120313942A1 (en) | System and method for digital volume processing with gpu accelerations | |
| CN109785283B (en) | Texture feature matching method and device for fabric segmentation | |
| CN116012393A (en) | Carton point cloud segmentation method, device and processing equipment | |
| CN111062473A (en) | Data calculation method, image processing method and device in neural network model | |
| CN116092035A (en) | Lane line detection method, lane line detection device, computer equipment and storage medium | |
| CN113744849A (en) | Semi-automatic labeling method and system for CT 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 |