CN107230204A - A kind of method and device that the lobe of the lung is extracted from chest CT image - Google Patents
A kind of method and device that the lobe of the lung is extracted from chest CT image Download PDFInfo
- Publication number
- CN107230204A CN107230204A CN201710374181.8A CN201710374181A CN107230204A CN 107230204 A CN107230204 A CN 107230204A CN 201710374181 A CN201710374181 A CN 201710374181A CN 107230204 A CN107230204 A CN 107230204A
- Authority
- CN
- China
- Prior art keywords
- pixel
- lung
- point
- lobe
- subelement
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20101—Interactive definition of point of interest, landmark or seed
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
Abstract
The present invention discloses a kind of method and device that the lobe of the lung is extracted from chest CT image, is related to field of computer technology.This method includes the pulmonary parenchyma extraction process based on 3D region growings, the left and right adhesion of lung tracheae based on provincial characteristics and rejects process, Pulmonary Vascular root rejecting process, the extracted process in Pulmonary Vascular Center Road based on topological thinning and the lobe of the lung partitioning algorithm based on support vector cassification, to obtain lobe of the lung tissue.The present invention accurately can extract the lobe of the lung from chest CT image, the accurate qualitative assessment completed to the severity extent of each lobe of the lung, and the diagnosis and treatment to PUD D are more accurate and effective.
Description
Technical field
The invention belongs to field of computer technology, it is related to a kind of method and device that the lobe of the lung is extracted from chest CT image.
Background technology
CT (Computed Tomography) be using Accurate collimation X-ray beam and the high detector of sensitivity together
Profile scanning one by one is done around a certain position of human body, the suction of organ and tissue to x-ray is represented with different gray scales
Receipts degree, for example, on chest CT image, the region representation tracheae of low-density, pulmonary parenchyma, highdensity region representation blood vessel,
Thoracic cavity, bone etc., it is fast with sweep time, it is doctor's inspections and examinations available for the inspection of a variety of diseases the features such as image clearly
Disease provides convenient and reliable foundation.
Because CT equipment can obtain clearly chest CT image, therefore by means of CT images into the diagnosis chronic obstructive pulmonary disease state of an illness
One Main Means, but current medical level can only accomplish the quantitative Diagnosis to single lung, for the single lobe of the lung state of an illness
Qualitative assessment can not be accomplished.The treatment of chronic obstructive pulmonary disease usually requires lobe of the lung volume reduction surgery, but operation consent can not but carry out the lobe of the lung subtracts
The qualitative assessment of appearance, this will certainly influence the therapeutic effect of chronic obstructive pulmonary disease.If pulmo and left and right can be accurately partitioned into
The various pieces of lung, the i.e. lobe of the lung, accurately complete the qualitative assessment to the severity extent of each lobe of the lung, chronic obstructive pulmonary disease will be examined
Disconnected and treatment is of great importance.
The content of the invention
(1) technical problem to be solved
In order to solve the above mentioned problem of prior art, the present invention provides a kind of side that the lobe of the lung is extracted from chest CT image
Method, this method accurately can extract the lobe of the lung from chest CT image, accurately complete to the severity extent of each lobe of the lung
Qualitative assessment, the diagnosis and treatment to PUD D is more accurate and effective.
The present invention also provides a kind of device that the lobe of the lung is extracted from chest CT image, and the device can be accurately from chest CT
The lobe of the lung is extracted in image, the qualitative assessment to the severity extent of each lobe of the lung is accurately completed, diagnosis to PUD D and
Treatment is more accurate and effective.
(2) technical scheme
In order to achieve the above object, the main technical schemes that the present invention is used include:
A kind of method that the lobe of the lung is extracted from chest CT image, comprises the following steps:
S1, the n-layer chest CT image for receiving input, obtain intermediate image layer, and the specified pixel point for choosing lung areas is
Seed point, carries out 3D region growths according to setting segmentation threshold and initial seed point, obtains without pulmonary vascular pulmonary parenchyma area
Domain, wherein, n is natural number;
S2, in the pulmonary parenchyma region obtained by step S1 by corroding the left and right adhesion of lung elimination method on approximate adhesion border,
Eliminating conglutination obtains two independent pulmonary parenchymas;
S3, using first expansion post-etching operation handlebar blood vessel be filled into the pulmonary parenchyma obtained by step S2, pass through maximum kind
Between variance method automatically obtain optimum segmentation threshold value T0;
S4, with the optimum segmentation threshold value T obtained by step S30As the threshold value of segmentation blood vessel, the point calculated more than threshold value is made
For vessel seed point, 3D region growths are carried out, Pulmonary Vascular is obtained;
S5, the blood vessel from the different lobes of the lung of vascular root disconnection, make the blood vessel in each lobe of the lung turn into an independent company
Logical domain;
S6, the Pulmonary Vascular according to obtained by step S5, Pulmonary Vascular center path is extracted using elimination approach;
S7, two category support vector machines models of training, obtain two category support vector machines graders, are looked for by the grader
Interface between different lobe of the lung blood vessels, using this interface as the interface of the lobe of the lung and calculates and obtains pulmonary parenchyma pixel
Each pixel in discriminant classification function pair pulmonary parenchyma is differentiated, obtains lobe of the lung tissue.
It is preferred that, the step S1 comprises the following steps:
S11, n (n is natural number) the layer chest CT image for receiving input, obtain intermediate image layer, on intermediate image layer
The specified pixel point for choosing lung areas is used as initial seed point;
S12, according to the characteristics of pulmonary parenchyma segmentation threshold is set;
26 neighborhoods of the labeled initial seed point of S13, search, choose one of pixel;
Whether the selected pixel of S14, judgement has been labeled, if then return to step S13, otherwise performs step
S15;
Whether the gray value of the selected pixel of S15, judgement is less than segmentation threshold, if so, then the pixel is marked
Remember and add mark queue, otherwise the stop flag pixel, perform step S16;
S16, judge whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish, if then performing step S17,
Otherwise return to step S13;
Whether S17, judge mark queue are empty, if being not sky, then a mark point are taken out from the queue as initial
Labeled seed point, return to step S13, the pixel point set being otherwise labeled is exactly to be partitioned into without pulmonary vascular lung
Parenchyma section.
It is preferred that, the step S2 comprises the following steps:
S21, to the pulmonary parenchyma region decision of acquisition, whether it is two approximately equalised connected domains of size, if then tying
Beam, otherwise performs step S22;
S22, statistics intermediate image layer connected domain number, reject the less connected domain region of pixel number, retain pixel
The larger connected domain region of points, and judge the number of connected domain and the size of each connected domain, if two connected domains simultaneously
And size approximately equal then terminates, step S23 is otherwise performed;
S23, successively statistics connected domain number, if only one of which connected domain, pulmonary parenchyma one is scheduled on this layer of adhesion, performs
Step S24, otherwise records the number of each connected domain pixel, two maximum connected domains of pixel number is found, if face
Quite, then this layer of CT image or so pulmonary parenchyma adhesion, performs step to two maximum connected region pixel number numbers of product
S26, otherwise adhesion and larger connected domain are pulmonary parenchyma, perform step S24;
S24, with line segment approximately replace adhesion region, be converted into determine straight line position, peak of the lower boundary in y directions
The as lower extreme point of near linear, coboundary a little in point from lower extreme point recently be then near linear upper extreme point, seek
Find and straight line is determined after this 2 points, perform step S25;
S25, the near linear border eroded on every layer of pulmonary parenchyma region adhesion image, and 26 around straight border
The mark value of neighborhood point, obtains the pulmo region of two adhesions, performs step S26;
S26, judge whether all layers are scanned, if then terminating, otherwise perform step S23.
It is preferred that, the step S3 comprises the following steps:
S31, first expansion post-etching is performed on all chest CT images layer by choosing suitable structural element close fortune
Operation is calculated, Pulmonary Vascular is filled into segmented good pulmonary parenchyma;
S32, one layer close to centre is selected in CT images, optimum segmentation is automatically obtained using maximum variance between clusters
Threshold value T0。
It is preferred that, the step S4 comprises the following steps:
Grey scale pixel value is more than segmentation threshold T in S41, the obtaining step S3 pulmonary parenchyma with blood vessel0All pixels point,
And labeled as blood vessel.
S42, in the blood vessel obtained by step S41 choose a pixel be used as initial seed point;
26 neighborhoods of the labeled initial seed point of S43, search, choose one of pixel;
Whether the selected pixel of S44, judgement has been labeled, if then return to step S43, otherwise performs step
S45;
Whether the gray value of the selected pixel of S45, judgement is more than segmentation threshold, if so, then the pixel is marked
Remember and add mark queue, otherwise the stop flag pixel, perform step S46;
S46, judge whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish, if then performing step S47,
Otherwise return to step S43;
Whether S47, judge mark queue are empty, if being not sky, then a mark point are taken out from the queue as initial
Labeled seed point, return to step S43, the pixel point set being otherwise labeled is exactly the Pulmonary Vascular being partitioned into.
It is preferred that, the step S5 comprises the following steps:
S51, the Pulmonary Vascular pixel for finding lung edge and close tracheae, step S52 is performed if finding, otherwise continues to sweep
Retouch lung tissue;
S52, Pulmonary Vascular disconnected using the blood vessel at the edge of corrosion pulmonary parenchyma, and count of the connected domain of vascularization
Number, if the number of connected domain is five, performs step S53, otherwise performs step S51;
S53, the color different to each connected component labeling, all mark finish beam.
It is preferred that, the step S6 comprises the following steps:
S61, scanning Pulmonary Vascular pixel, whether be boundary point, if the neighborhood of pixel 6 or 26 neighborhoods have the back of the body if judging it
Scenic spot pixel is then boundary point, performs step S62, otherwise marks scanned, skips and continue executing with step S61;
S62, the Euler's characteristic for judging boundary point, step S63 is performed if Euler's characteristic is constant, otherwise skips and continues executing with
Step S61;
S63, judge whether boundary point is simple point, if the point on former three dimensions topological structure without influence if be simple
Point, while by the point deletion, performing step S64;Otherwise step S61 is performed;
S64, judge blood vessel pixel whether also have other simple points, if nothing, centerline extraction terminates, otherwise perform step
S61。
It is preferred that, the step S7 comprises the following steps:
S71, according to the interface between the lobe of the lung closest to the track of RBF, selection Radial basis kernel function is as to height
The function of dimension space mapping, it is determined that after kernel function, performs step S72;
S72, for left lung, it is assumed that the pixel of left lung be N1, blood vessel pixel number be n1, the institute of the upper leaf of left lung
There is blood vessel pixel labeled as just, while all blood vessel pixels on inferior lobe are obtained a discriminant function labeled as negative, perform
Step S73;
S73, gone with discriminant function to judge the symbol of each pixel in left pulmonary parenchyma, if symbol is just, the point
Belong to leaf, otherwise the point belongs to inferior lobe, and the number of statistical pixel, if then left lung segmentation is completed N1, perform step S74,
Otherwise step S73 is continued executing with;
S74, for right lung, it is assumed that the pixel of right lung be N2, blood vessel pixel number be n2, for using two classification support
Vector machine is classified, and the upper leaf of right lung and middle period, which are combined into an entirety, to be regarded as " upper leaf ", execution step S75;
S75, obtain a discriminant classification function using splitting the same method with left lung, inferior lobe is extracted first, walking
All blood vessel pixels of " upper leaf " that rapid S74 is regarded as are labeled as just, while all blood vessel pixels on inferior lobe are labeled as
It is negative, a discriminant function is obtained, step S76 is performed;
S76, gone with discriminant function to judge the symbol of each substantial pixel of right lung, if symbol is just, the point
Belong to " upper leaf " that step S74 is regarded as, otherwise the point belongs to inferior lobe, and the number of statistical pixel, if N2 then split by inferior lobe
Complete, perform step S77, otherwise continue executing with step S76;
S77, segmentation are completed after inferior lobe, then obtain one with leaf and the blood vessel Training Support Vector Machines model in middle period again
Individual differentiation surface function, the partial segmentation of inferior lobe is removed to right lung, extracts leaf and middle period, performs step S78;
S78, for the upper leaf of right lung and middle period, it is assumed that the pixel of right lung is N3, and blood vessel pixel number is n3, right
All blood vessel pixels of the upper leaf of lung, while middle period upper all blood vessel pixels of right lung are labeled as bearing, are obtained labeled as just
To a discriminant function, step S79 is performed;
S79, gone with discriminant function to judge the symbol of each substantial pixel of right lung, if symbol is just, the point
Belong to leaf, otherwise the point belongs to middle period, and the number of statistical pixel, if then superior lobe of right lung and middle period segmentation are completed N3, extremely
This pulmo segmentation lobe of the lung terminates.
A kind of device that the lobe of the lung is extracted from chest CT image, including:
CT image input units, the n-layer chest CT image for receiving input, wherein n is natural number;
Without pulmonary vascular pulmonary parenchyma acquiring unit, its specified pixel point for choosing lung areas is seed point, according to setting
Determine segmentation threshold and initial seed point carries out 3D region growths, obtain without pulmonary vascular pulmonary parenchyma region;
Two independent pulmonary parenchyma acquiring units, its obtain left and right pulmonary parenchyma after, judge left and right pulmonary parenchyma whether adhesion, if
Adhesion, then obtain two independent pulmonary parenchymas by rejecting the adhesion border of tracheae and near linear;
Pulmonary Vascular segmentation threshold acquiring unit, it is filled into pulmonary parenchyma using the operation handlebar blood vessel of first expansion post-etching,
Optimum segmentation threshold value T is automatically obtained by maximum variance between clusters0;
Pulmonary Vascular extraction unit, it is with the optimum segmentation threshold value T of gained0As the threshold value of segmentation blood vessel, calculate and be more than threshold
The point of value carries out 3D region growths as vessel seed point, obtains Pulmonary Vascular;
Pulmonary Vascular is divided into independent communication domain unit, and it disconnects the blood vessel of the different lobes of the lung from vascular root, makes each lung
The blood vessel of leaf turns into an independent connected domain;
Pulmonary Vascular centerline extraction unit, its lung blood according to obtained by the Pulmonary Vascular is divided into independent communication domain unit
Pipe, Pulmonary Vascular center path is extracted using elimination approach;
Lobe of the lung cutting unit, the sorter model of blood vessel Training Support Vector Machines two of the adjacent lobe of the lung, left lung are had two by it
The individual lobe of the lung, trains one, right lung has three lobes of the lung, and twice, three graders of SVMs two are obtained in pulmo for training, lead to
Cross SVMs and find interface between different lobe of the lung blood vessels, be partitioned into the lobe of the lung.
(3) beneficial effect
The beneficial effects of the invention are as follows:
The invention provides a kind of method and device that the lobe of the lung is extracted from chest CT image, wherein, method includes being based on
The pulmonary parenchyma extraction process of 3D region growings, the left and right adhesion of lung tracheae based on provincial characteristics are rejected process, Pulmonary Vascular root and picked
The extracted process in Pulmonary Vascular Center Road except process, based on topological thinning and the lobe of the lung segmentation based on support vector cassification are calculated
Method, lobe of the lung tissue is obtained by journey processed above.The present invention can accurately extract the lobe of the lung from chest CT image, accurate
True completion diagnosis to PUD D and treats more accurate and effectively to the qualitative assessment of the severity extent of each lobe of the lung.
Brief description of the drawings
Fig. 1 is the committed step flow chart that the lobe of the lung is extracted from chest CT image of preferred embodiment.
Fig. 2 is the method flow diagram that the lobe of the lung is extracted from chest CT image of preferred embodiment.
Fig. 3 is the grey level histogram of preferred embodiment chest CT image.
Fig. 4 is pulmonary parenchyma area results schematic diagram of the preferred embodiment without blood vessel.
Fig. 5 is pulmonary parenchyma area results schematic diagram of the preferred embodiment with tracheae.
Fig. 6 is the pulmonary parenchyma area results schematic diagram that preferred embodiment rejects tracheae.
Fig. 7 is the pulmonary parenchyma area results schematic diagram that preferred embodiment border is adhered.
Fig. 8 is the pulmonary parenchyma area results schematic diagram of preferred embodiment blood vessel filling.
Fig. 9 is that preferred embodiment Pulmonary Vascular extracts result schematic diagram.
Figure 10 is preferred embodiment Pulmonary Vascular centerline extraction result schematic diagram.
Figure 11 is that preferred embodiment Pulmonary Vascular disconnects root extraction result schematic diagram.
Figure 12 is that the preferred embodiment lobe of the lung extracts result schematic diagram.
Embodiment
In order to preferably explain the present invention, in order to understand, below in conjunction with the accompanying drawings, by embodiment, to this hair
It is bright to be described in detail.
Preferred embodiment
As depicted in figs. 1 and 2, the present embodiment proposes a kind of preferred method that the lobe of the lung is extracted from chest CT image,
The method that the lobe of the lung is extracted from chest CT image comprises the following steps:
S1, n (n is natural number) the layer chest CT image for receiving input, obtain intermediate image layer, choose the finger of lung areas
Fixation vegetarian refreshments is seed point, carries out 3D region growths according to setting segmentation threshold and initial seed point, obtains without Pulmonary Vascular
Pulmonary parenchyma region.
Specifically, step S1 comprises the following steps:
S11, n (n is natural number) the layer chest CT image for receiving input, obtain intermediate image layer, on intermediate image layer
The specified pixel point for choosing lung areas is used as initial seed point;
S12, according to the characteristics of pulmonary parenchyma segmentation threshold is set;
26 neighborhoods of the labeled initial seed point of S13, search, choose one of pixel;
Whether the selected pixel of S14, judgement has been labeled, if then return to step S13, otherwise performs step
S15;
Whether the gray value of the selected pixel of S15, judgement is less than segmentation threshold, if so, then the pixel is marked
Remember and add mark queue, otherwise the stop flag pixel, perform step S16;
S16, judge whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish, if then performing step S17,
Otherwise return to step S13;
Whether S17, judge mark queue are empty, if being not sky, then a mark point are taken out from the queue as initial
Labeled seed point, return to step S13, the pixel point set being otherwise labeled is exactly to be partitioned into without pulmonary vascular lung
Parenchyma section.
S2, in the pulmonary parenchyma region obtained by step S1 by corroding the left and right adhesion of lung elimination method on approximate adhesion border,
Eliminating conglutination obtains two independent pulmonary parenchymas.
Some researchers are first corroded with morphologic method to the pulmonary parenchyma of adhesion, until pulmo adhesion, then
Mark respectively, reflation is returned original size.Although this method has successfully separated left lung and right lung essence, lose
The details on border, makes the border of some lungs deviate original position, adds the error of segmentation.Also some calculate complicated side
Although method can accurately separate left lung and right lung essence, but pay substantial amounts of calculating time cost, so that answering
Use it is difficult to meet the requirement of time.
By being found to substantial amounts of adhesion regional observation, the adhesion of left lung and right lung essence is simply present in some layers of CT figures
As upper, the situation of adhesion, even adhesion, also simply one piece of region adhesion of very little is not present in most of layers of CT images.Therefore
The specific layer CT images of adhesion can be found first, then successively eliminating conglutination.
Finding the method for adhering layer is, connected domain number is successively counted, if only one of which connected domain, pulmonary parenchyma is certain
In this layer of adhesion.If multiple connected domains, then the number of each connected domain pixel is recorded, find pixel number maximum
Two connected domains, pixel number uses N respectively1And N2To represent, if
Threshold is usually arranged as 10 in above formula, and the specific of formula means:If two maximum connected regions of area
Domain pixel number number quite, then this layer of CT image or so pulmonary parenchyma adhesion, because left lung is real on the CT images of same layer
The area of matter region and right lung parenchyma section is suitable;, whereas if two maximum two connected region pixel numbers
Ratio great disparity, then this layer of CT image or so pulmonary parenchyma adhesion, the pulmonary parenchyma region letter of guarantee of a now larger connection domain representation left side
Pulmonary parenchyma region and right lung parenchyma section, and it is less connection domain representation tissue regions some be noise interference, some are
Tiny bronchus, and these bronchuses will not make left and right adhesion of lung.
It is determined that being then to find specific adhesion boundary position after specific adhering layer CT images.Because adhesion only has very
Small region, and the border of adhesion is a very short curve, therefore can approximately be replaced with line segment.The benefit so done
It is, you can to reduce amount of calculation so as to reduce the run time of algorithm, and to have little influence on the precision of pulmonary parenchyma segmentation.
Specifically, step S2 comprises the following steps:
S21, to the pulmonary parenchyma region decision of acquisition, whether it is two approximately equalised connected domains of size, if then tying
Beam, otherwise performs step S22;
S22, statistics intermediate image layer connected domain number, reject the less connected domain region of pixel number, retain pixel
The larger connected domain region of points, and judge the number of connected domain and the size of each connected domain, if two connected domains simultaneously
And size approximately equal then terminates, step S23 is otherwise performed;
S23, successively statistics connected domain number, if only one of which connected domain, pulmonary parenchyma one is scheduled on this layer of adhesion, performs
Step S24, otherwise records the number of each connected domain pixel, two maximum connected domains of pixel number is found, if face
Quite, then this layer of CT image or so pulmonary parenchyma adhesion, performs step to two maximum connected region pixel number numbers of product
S26, otherwise adhesion and larger connected domain are pulmonary parenchyma, perform step S24;
S24, with line segment approximately replace adhesion region, be converted into determine straight line position, peak of the lower boundary in y directions
The as lower extreme point of near linear, coboundary a little in point from lower extreme point recently be then near linear upper extreme point, seek
Find and straight line is determined after this 2 points, perform step S25;
S25, the near linear border eroded on every layer of pulmonary parenchyma region adhesion image, and 26 around straight border
The mark value of neighborhood point, obtains the pulmo region of two adhesions, performs step S26;
S26, judge whether all layers are scanned, if then terminating, otherwise perform step S23.
S3, using first expansion post-etching operation handlebar blood vessel be filled into the pulmonary parenchyma obtained by step S2, pass through maximum kind
Between variance method automatically obtain optimum segmentation threshold value T0。
Morphologic first expansion post-etching computing referred to as closed operation is carried out on image, it is therefore an objective to which packing ratio structural element is small
Hole and smoothed image edge.
Expansion is " lengthening " or the operation of " thicker " in bianry image.This special mode and thicker degree are by one
The individual set control for being referred to as structural element, structural element is represented with the matrix of 0 and 1.For corrosion, its as in expansion,
Contraction mode and degree are controlled by a structural element.
Maximum variance between clusters are that original image is divided into foreground and background two parts using threshold value.It is prospect and background to remember T
Segmentation threshold, prospect points account for image scaled for w0, average gray is u0;Background points account for image scaled for w1, average gray
For u1, then the overall average gray scale of image be:
U=w0u0+w1u1 (1.1)
The variance of foreground and background image:
σ2=w0(u0-u)2+w1(u1-u)2 (1.2)
Formula (1.1) bring into formula (1.2) obtain inter-class variance calculation formula it is as follows:
σ2=w0w1(u0-u1)2 (1.3)
Another σ is obtained using formula (1.3)2Maximum threshold value Tmax, it is used as threshold value of the segmentation figure as foreground and background.
Specifically, step S3 comprises the following steps:
S31, first expansion post-etching is performed on all chest CT images layer by choosing suitable structural element close fortune
Operation is calculated, Pulmonary Vascular is filled into segmented good pulmonary parenchyma;
S32, one layer close to centre is selected in CT images, optimum segmentation is automatically obtained using maximum variance between clusters
Threshold value T0。
S4, with the optimum segmentation threshold value T obtained by step S30As the threshold value of segmentation blood vessel, the point calculated more than threshold value is made
For vessel seed point, 3D region growths are carried out, Pulmonary Vascular is obtained.Wherein, Pulmonary Vascular extraction effect is as shown in Figure 9.
Specifically, step S4 comprises the following steps:
Grey scale pixel value is more than segmentation threshold T in S41, the obtaining step S3 pulmonary parenchyma with blood vessel0All pixels point,
And labeled as blood vessel.
S42, in the blood vessel obtained by step S41 choose a pixel be used as initial seed point;
26 neighborhoods of the labeled initial seed point of S43, search, choose one of pixel;
Whether the selected pixel of S44, judgement has been labeled, if then return to step S43, otherwise performs step
S45;
Whether the gray value of the selected pixel of S45, judgement is more than segmentation threshold, if so, then the pixel is marked
Remember and add mark queue, otherwise the stop flag pixel, perform step S46;
S46, judge whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish, if then performing step S47,
Otherwise return to step S43;
Whether S47, judge mark queue are empty, if being not sky, then a mark point are taken out from the queue as initial
Labeled seed point, return to step S43, the pixel point set being otherwise labeled is exactly the Pulmonary Vascular being partitioned into.
S5, the blood vessel from the different lobes of the lung of vascular root disconnection, make the blood vessel in each lobe of the lung turn into an independent company
Logical domain.
Specifically, step S5 comprises the following steps:
S51, the Pulmonary Vascular pixel for finding lung edge and close tracheae, step S52 is performed if finding, otherwise continues to sweep
Retouch lung tissue;
S52, Pulmonary Vascular disconnected using the blood vessel at the edge of corrosion pulmonary parenchyma, and count of the connected domain of vascularization
Number, if the number of connected domain is five, performs step S53, otherwise performs step S51;
S53, the color different to each connected component labeling, all mark finish beam.
S6, the Pulmonary Vascular according to obtained by step S5, Pulmonary Vascular center path is extracted using elimination approach.
Specifically, step S6 comprises the following steps:
S61, scanning Pulmonary Vascular pixel, whether be boundary point, if the neighborhood of pixel 6 or 26 neighborhoods have the back of the body if judging it
Scenic spot pixel is then boundary point, performs step S62, otherwise marks scanned, skips and continue executing with step S61;
S62, the Euler's characteristic for judging boundary point, step S63 is performed if Euler's characteristic is constant, otherwise skips and continues executing with
Step S61;
S63, judge whether boundary point is simple point, if the point on former three dimensions topological structure without influence if be simple
Point, while by the point deletion, performing step S64;Otherwise step S61 is performed;
S64, judge blood vessel pixel whether also have other simple points, if nothing, centerline extraction terminates, otherwise perform step
S61。
S7, two category support vector machines models of training, obtain two category support vector machines graders, are looked for by the grader
Interface between different lobe of the lung blood vessels, using this interface as the interface of the lobe of the lung and calculates and obtains pulmonary parenchyma pixel
Each pixel in discriminant classification function pair pulmonary parenchyma is differentiated, obtains lobe of the lung tissue.Wherein, the lobe of the lung extracts result such as
Shown in Figure 12.
SVMs (SVM, Support Vector Machines) proposes first by Vapnik et al., be from
The optimal classification surface of linear separability is gradually developed.
In three dimensions, between the Pulmonary Vascular of two adjacent lobes of the lung and in the absence of a plane, make on a lobe of the lung
Each pulmonary vascular pixel makes each pulmonary vascular pixel on another lobe of the lung flat in the side of plane
The opposite side in face, therefore this is a linearly inseparable problem.
For solving the problems, such as linearly inseparable with svm classifier, first have to select an appropriate kernel function, make these three-dimensional
Data in space are mapped in the feature space of a more higher-dimension, so that these sample datas become linear separability.It is logical
Cross and substantial amounts of lung tissue CT data observations are found, the interface between the lobe of the lung is selected closest to the track of RBF
Radial basis kernel function is used as the function mapped to higher dimensional space.
The form of Radial basis kernel function is:
The discriminant function that SVMs now is constructed is:
Wherein, s is the number of supporting vector, and supporting vector can determine the center of RBF.Radial direction base core
The dimension of the corresponding feature space of function is that infinitely great, limited data sample must be linear separability in this feature space,
Therefore Radial basis kernel function is most commonly used.
Algorithm of support vector machine is finally attributed to the problem of seeking quadratic programming (QP), can be rewritten as following rectangular
Formula.Given sample (xi,yi), i=1,2,3...n, kernel function K (xi,xj) and regulation parameter C, minimizing.
Constraints is:
αTY=0,0≤α≤C (1.7)
Kernel matrix H definition is:
H=[hij]=yiyjK(xi,xj) (i, j=1,2,3...n) (1.8)
Wherein, α=(α1,α2,...αn)T, α=(y1,y2,...yn)T
If training data is on a grand scale, because the scale of matrix H is square of training data, matrix H is possible to big
To computer normal process can not be used.
For problem above, it has been proposed that below for the method for large-scale data sample training.
1.Chunking algorithms
Vapnik et al. first proposed a kind of method that solution SVM trains memory space problem, and referred to as Chunking is calculated
Method.In formula (1.6), if removing row and column corresponding with zero Lagrange multiplier, its value is constant.Therefore it will can solve
The QP PROBLEM DECOMPOSITIONs of SVMs are into a series of less QP problems.The final goal for solving these less QP problems is true
Fixed all non-zero Lagrange multipliers, and remove zero all Lagrange multipliers.
2. decomposition algorithm
Osuan et al. first proposed decomposition algorithm, decomposition algorithm be the QP PROBLEM DECOMPOSITIONs of solution SVMs into
A series of less QP problems, but its working set size keeps constant, and line is changed into quadratic relationship from s to the demand of internal memory
Sexual intercourse.Can up to 110,000 with processing data sample point, the problem of supporting vector is more than 100,000.
3.SMO algorithms
Serial minimum optimized algorithm (Sequential Minimal Optimization, SMO) is carried first by Platt
Go out, SMO algorithms fall within a kind of decomposition algorithm, its working space is only right in every single-step iteration only comprising two data samples
Two Lagrange multipliers are optimized.Although QP problems are increased in SMO, total calculating speed is substantially increased,
And this algorithm is completely without the big matrix of processing, thus there is no extra requirement to memory space, very big SVM training is asked
Topic can also be run on a personal computer.Advantage due to more than, SMO algorithms become widest a kind of in actual applications
Method.
In the present embodiment, for left lung, it is assumed that the blood vessel pixel number of left lung is n, because only that two lobes of the lung, because
All blood vessel pixels of this upper leaf are labeled as (xi,yi),i≤n,xi∈R3,yi=+1, while all blood vessel pictures on inferior lobe
Vegetarian refreshments is labeled as (xi,yi),i≤n,xi∈R3,yi=-1.Obtain a discriminant function f1(x).Use discriminant function f1(x) go to sentence
In left lung tissue of breaking often with the symbol of individual pixel, if symbol is just, the point belongs to leaf, and otherwise the point belongs to inferior lobe.
So upper lobe of left lung and inferior lobe just can be divided out.
And for right lung, because there is three lobes of the lung, respectively upper leaf, middle period and inferior lobe.In order to classify using two
The upper leaf of right lung and middle period, are combined into an entirety by support vector cassification first, can thus be used and left lung segmentation one
The method of sample obtains a discriminant classification function f2(x) inferior lobe, is extracted first, is then trained again with leaf and the blood vessel in middle period
Supporting vector machine model, obtains one and differentiates surface function f3(x) the part subseries again of inferior lobe, is removed to right lung, is extracted
Leaf and middle period.
Specifically, step S7 comprises the following steps:
S71, according to the interface between the lobe of the lung closest to the track of RBF, selection Radial basis kernel function is as to height
The function of dimension space mapping, it is determined that after kernel function, performs step S72;
S72, for left lung, it is assumed that the pixel of left lung be N1, blood vessel pixel number be n1, the institute of the upper leaf of left lung
There is blood vessel pixel labeled as just, while all blood vessel pixels on inferior lobe are obtained a discriminant function labeled as negative, perform
Step S73;
S73, gone with discriminant function to judge the symbol of each pixel in left pulmonary parenchyma, if symbol is just, the point
Belong to leaf, otherwise the point belongs to inferior lobe, and the number of statistical pixel, if then left lung segmentation is completed N1, perform step S74,
Otherwise step S73 is continued executing with;
S74, for right lung, it is assumed that the pixel of right lung be N2, blood vessel pixel number be n2, for using two classification support
Vector machine is classified, and the upper leaf of right lung and middle period, which are combined into an entirety, to be regarded as " upper leaf ", execution step S75;
S75, obtain a discriminant classification function using splitting the same method with left lung, inferior lobe is extracted first, walking
All blood vessel pixels of " upper leaf " that rapid S74 is regarded as are labeled as just, while all blood vessel pixels on inferior lobe are labeled as
It is negative, a discriminant function is obtained, step S76 is performed;
S76, gone with discriminant function to judge the symbol of each substantial pixel of right lung, if symbol is just, the point
Belong to " upper leaf " that step S74 is regarded as, otherwise the point belongs to inferior lobe, and the number of statistical pixel, if N2 then split by inferior lobe
Complete;Perform step S77;
S77, segmentation are completed after inferior lobe, then obtain one with leaf and the blood vessel Training Support Vector Machines model in middle period again
Individual differentiation surface function, the partial segmentation of inferior lobe is removed to right lung, extracts leaf and middle period, performs step S78;
S78, for the upper leaf of right lung and middle period, it is assumed that the pixel of right lung is N3, and blood vessel pixel number is n3, right
All blood vessel pixels of the upper leaf of lung, while middle period upper all blood vessel pixels of right lung are labeled as bearing, are obtained labeled as just
To a discriminant function, step S79 is performed;
S79, gone with discriminant function to judge the symbol of each substantial pixel of right lung, if symbol is just, the point
Belong to leaf, otherwise the point belongs to middle period, and the number of statistical pixel, if then superior lobe of right lung and middle period segmentation are completed N3, extremely
This pulmo segmentation lobe of the lung terminates.
Meanwhile, the present embodiment additionally provides a kind of preferred device that the lobe of the lung is extracted from chest CT image, including CT figures
Obtained as input block, without pulmonary vascular pulmonary parenchyma acquiring unit, two independent pulmonary parenchyma acquiring units, Pulmonary Vascular segmentation thresholds
Unit, Pulmonary Vascular extraction unit, Pulmonary Vascular is taken to be divided into independent communication domain unit, Pulmonary Vascular centerline extraction unit and the lobe of the lung
Cutting unit.
Wherein, CT image input units, the n-layer chest CT image for receiving input, wherein n is natural number.
Without pulmonary vascular pulmonary parenchyma acquiring unit, its specified pixel point for choosing lung areas is seed point, according to setting
Determine segmentation threshold and initial seed point carries out 3D region growths, obtain without pulmonary vascular pulmonary parenchyma region.
Specifically, including following each unit without pulmonary vascular pulmonary parenchyma acquiring unit:
Initial seed point chooses unit, and it chooses the specified pixel point of lung areas as initial kind on middle tomographic image
Sub- point.Segmentation threshold obtains subelement, and it sets segmentation threshold according to the characteristics of pulmonary parenchyma.Pixel chooses subelement, and it is searched
26 neighborhoods of the labeled initial seed point of rope, choose one of pixel.Pixel marker for judgment subelement, it judges
Whether selected pixel has been labeled, if so, notifying pixel to choose subelement, otherwise notifies pixel value to judge son
Unit.Pixel value judgment sub-unit, it judges whether the gray value of selected pixel is less than segmentation threshold, if so, then
The pixel is marked and adds mark queue, otherwise the stop flag pixel, notifies the neighborhood territory pixel point search of seed point 26
Subelement.The neighborhood search subelement of seed point 26, it judges whether 26 neighborhood territory pixel points of seed point are all searched for and judged
Finish, if so, then notification indicia queue judgment sub-unit, otherwise notifies the pixel to choose subelement.Mark queue judges son
Whether unit, its judge mark queue is empty, if being not sky, then a mark point is taken out from the queue as initial labeled
Seed point, notifies pixel to choose subelement, the pixel point set being otherwise labeled is exactly to be partitioned into without Pulmonary Vascular
Pulmonary parenchyma.
Two independent pulmonary parenchyma acquiring units, its obtain left and right pulmonary parenchyma after, judge left and right pulmonary parenchyma whether adhesion, if
Adhesion, then obtain two independent pulmonary parenchymas by rejecting the adhesion border of tracheae and near linear.
Specifically, two independent pulmonary parenchyma acquiring units include following each unit:
Pulmonary parenchyma connected domain number judgment sub-unit, its pulmonary parenchyma to acquisition judges whether it is the approximate phase of two sizes
Deng connected domain, if then terminating, otherwise notify reject tracheae subelement.Tracheae subelement is rejected, it counts intermediate image layer
Connected domain number, reject pixel number less connected domain region, retain the larger connected domain region of pixel number, and judge
The size of the number of connected domain and each connected domain, if two connected domains and size approximately equal then terminates, otherwise leads to
Know a layer connected domain number judgment sub-unit.Layer connected domain number judgment sub-unit, it successively counts connected domain number, if only
One connected domain, then pulmonary parenchyma one be scheduled on this layer of adhesion, notify straight line determination subelement, otherwise record each connected domain pixel
The number of point, finds two maximum connected domains of pixel number, if two maximum connected region pixel number numbers of area
Quite, then this layer of CT image or so pulmonary parenchyma adhesion, notifies pixel layer scanning judgment sub-unit, otherwise adhesion and larger company
Logical domain is pulmonary parenchyma, notifies straight line determination subelement.Straight line determination subelement, it approximately replaces adhesion region, conversion with line segment
To determine the position of straight line, peak of the lower boundary in y directions is the lower extreme point of near linear, the institute of coboundary a little in from
The point of lower extreme point recently is then the upper extreme point of near linear, searches out and straight line is determined after this 2 points, notifies corrosion near linear side
Boundary's subelement.Corrode near linear border subelement, it erodes the near linear side on every layer of pulmonary parenchyma region adhesion image
The mark value of 26 neighborhood points around boundary, and straight border, obtains the pulmo region of two adhesions, notifies pixel layer to sweep
Retouch judgment sub-unit.Pixel layer scans judgment sub-unit, and it judges whether all layers are scanned, if then terminating, otherwise led to
Know a layer connected domain number judgment sub-unit.
Pulmonary Vascular segmentation threshold acquiring unit, it is filled into pulmonary parenchyma using the operation handlebar blood vessel of first expansion post-etching,
Optimum segmentation threshold value T is automatically obtained by maximum variance between clusters0。
Specifically, Pulmonary Vascular segmentation threshold acquiring unit includes following each unit:
Closed operation operating unit, it is first expanded by choosing suitable structural element and being performed on all chest CT images layer
The closed operation operation of post-etching, Pulmonary Vascular is filled into segmented good pulmonary parenchyma.Optimum segmentation threshold value acquiring unit, its
One layer close to centre is selected in CT images, optimum segmentation threshold value T is automatically obtained using maximum variance between clusters0。
Pulmonary Vascular extraction unit, it is with the optimum segmentation threshold value T of gained0As the threshold value of segmentation blood vessel, calculate and be more than threshold
The point of value carries out 3D region growths as vessel seed point, obtains Pulmonary Vascular.
Specifically, Pulmonary Vascular extraction unit includes following each unit:
Intermediate layer Pulmonary Vascular extracts subelement, and it obtains in intermediate layer grey scale pixel value in the pulmonary parenchyma with blood vessel and is more than point
Cut threshold value T0All pixels point, and labeled as blood vessel.Initial seed point obtains subelement, and it is carried in the intermediate layer Pulmonary Vascular
Take and a pixel is chosen in the blood vessel obtained by subelement as initial seed point.Pixel obtains subelement, and its search is marked
26 neighborhoods of the initial seed point of note, choose one of pixel.Pixel marker for judgment subelement, it judges selected
Pixel whether be labeled, if notify pixel obtain subelement, otherwise notify pixel judgment sub-unit.Picture
Element value judgment sub-unit, it judges whether the gray value of selected pixel is more than segmentation threshold, if so, then the pixel
Mark and add mark queue, notify the neighborhood search subelement of seed point 26, otherwise the stop flag pixel.Seed point 26
Neighborhood search subelement, it judges whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish, if then notifying mark
Remember queue judgment sub-unit, otherwise notify pixel to obtain subelement.Mark queue judgment sub-unit, its judge mark queue is
It is no for sky, be not such as sky, then a mark point taken out from the queue and is used as initial labeled seed point, notice pixel acquisition
Subelement, the pixel point set being otherwise labeled is exactly the Pulmonary Vascular being partitioned into.
Pulmonary Vascular is divided into independent communication domain unit, and it disconnects the blood vessel of the different lobes of the lung from vascular root, makes each lung
The blood vessel of leaf turns into an independent connected domain.
Specifically, Pulmonary Vascular, which is divided into independent communication domain unit, includes following each unit:
Preliminary sweep pulmonary parenchyma unit, it is found lung edge and close to the Pulmonary Vascular pixel of tracheae, if finding, notified
Pulmonary Vascular root subelement is disconnected, preliminary sweep pulmonary parenchyma unit is otherwise notified.Pulmonary Vascular root subelement is disconnected, it uses corruption
The blood vessel for losing the edge of pulmonary parenchyma disconnects Pulmonary Vascular, and counts the number of the connected domain of vascularization, if the number of connected domain
For five, then Pulmonary Vascular mark subelement is notified, preliminary sweep pulmonary parenchyma unit is otherwise notified.Pulmonary Vascular marks subelement, its
The color different to the vascular marker of each connected domain, all mark finishes beam.
Pulmonary Vascular centerline extraction unit, its lung blood according to obtained by the Pulmonary Vascular is divided into independent communication domain unit
Pipe, Pulmonary Vascular center path is extracted using elimination approach.
Specifically, Pulmonary Vascular centerline extraction unit includes following each unit:
Boundary point subelement is judged, it scans Pulmonary Vascular pixel, and whether judge it is boundary point, if the pixel 6 is adjacent
Domain or 26 neighborhoods exist background area pixel then be boundary point, notify the first judgment sub-unit, otherwise mark it is scanned, skip after
It is continuous to notify to judge boundary point subelement.First judgment sub-unit, it judges Euler's characteristic of boundary point, if Euler's characteristic is constant
The second judgment sub-unit is notified, continuation is otherwise skipped and notifies to judge boundary point subelement.Second judgment sub-unit, it judges border
Whether point is simple point, if the point on former three dimensions topological structure without influence if be simple point, while by the point deletion, notifying
3rd judgment sub-unit, otherwise notifies to judge boundary point subelement.3rd judgment sub-unit, it judges whether blood vessel pixel also has
Other simple points, if nothing, centerline extraction terminates, otherwise notify to judge boundary point subelement.
Lobe of the lung cutting unit, the sorter model of blood vessel Training Support Vector Machines two of the adjacent lobe of the lung, left lung are had two by it
The individual lobe of the lung, trains one, right lung has three lobes of the lung, and twice, three graders of SVMs two are obtained in pulmo for training, lead to
Cross SVMs and find interface between different lobe of the lung blood vessels, be partitioned into the lobe of the lung.
Specifically, lobe of the lung cutting unit includes following each unit:
SVMs Selection of kernel function unit, its according to the interface between the lobe of the lung closest to the track of RBF,
Selection Radial basis kernel function is used as the function mapped to higher dimensional space, it is determined that after kernel function, notify left lung discriminant function to ask for
Subelement.Left lung discriminant function asks for subelement, and it is for left lung, it is assumed that the pixel of left lung is N1, blood vessel pixel number
For n1, all blood vessel pixels of the upper leaf of left lung are labeled as just, while all blood vessel pixels on inferior lobe are labeled as to bear,
A discriminant function is obtained, left lung upper and lower lobes segmentation subelement is notified.Left lung upper and lower lobes split subelement, and it is gone with discriminant function
The symbol of each pixel in left pulmonary parenchyma is judged, if symbol is just, the point belongs to leaf, under otherwise the point belongs to
Leaf, and the number of statistical pixel, if then left lung segmentation is completed N1, notify the superior lobe of right lung middle period to merge subelement.Superior lobe of right lung
Middle period merges subelement, and it is for right lung, it is assumed that the pixel of right lung is N2, and blood vessel pixel number is n2, to use two points
Class support vector machines are classified, and the upper leaf of right lung and middle period, which are combined into an entirety, to be regarded as " upper leaf ", notice right lung discriminant function
Ask for subelement.Right lung discriminant function asks for subelement, and it obtains a discriminant classification using the method as the segmentation of left lung
Function, extracts inferior lobe first, all blood vessel pixels of " the upper leaf " regarded as is labeled as just, while all on inferior lobe
Blood vessel pixel obtains a discriminant function labeled as negative, notifies inferior lobe of right lung segmentation subelement.Inferior lobe of right lung segmentation is single
Member, it removes the symbol for judging each substantial pixel of right lung with discriminant function, if symbol is just, the point, which belongs to, to be seen
" the upper leaf " done, otherwise the point belongs to inferior lobe, and the number of statistical pixel, if then inferior lobe segmentation is completed N2, notifies on right lung
Leaf middle period discriminant function asks for subelement.Superior lobe of right lung middle period discriminant function asks for subelement, and it, which is split, completes after inferior lobe, then
Again with leaf and the blood vessel Training Support Vector Machines model in middle period, a differentiation surface function is obtained, the portion of inferior lobe is removed to right lung
Segmentation, extracts leaf and middle period, notifies superior lobe of right lung middle period segmentation subelement.The superior lobe of right lung middle period splits subelement, its
For the upper leaf of right lung and middle period, it is assumed that the pixel of right lung is N3, and blood vessel pixel number is n3, the institute of the upper leaf of right lung
There is blood vessel pixel labeled as just, while middle period upper all blood vessel pixels of right lung are labeled as bearing, obtain a differentiation letter
Number, and gone with discriminant function to judge the symbol of each substantial pixel of right lung, if symbol is just, the point belongs to
Leaf, otherwise the point belongs to middle period, and the number of statistical pixel, if then superior lobe of right lung and middle period segmentation are completed N3.
In the present embodiment, because the Density Distribution of pulmonary parenchyma is than more uniform, therefore, the pulmonary parenchyma gray value on CT images
Than more uniform, whole pulmonary parenchyma can be approximately also replaced when obtaining initial segmentation threshold value with the threshold value of intermediate layer CT images for distribution
Initial segmentation threshold value.Advantage of this is that, in the case where influenceing very little to initial segmentation threshold accuracy, it is greatly reduced
The amount of calculation of the initial segmentation threshold value in whole pulmonary parenchyma region is calculated, so as to meet the requirement of practical application.Because pulmonary parenchyma area
The gray value in domain and the gray value of background area (pulmonary parenchyma surrounding tissue) have obvious difference, can be showed in image graph
One obvious trough.It therefore, it can the method using statistic histogram (see Fig. 3), find the trough between two crests,
It is exactly from pulmonary parenchyma gray value to the excessive region of surrounding tissue gray value, so that it is determined that going out initial segmentation threshold value T0。
Fig. 4 is the pulmonary parenchyma area results schematic diagram without blood vessel, and as can be seen from Figure 4, pulmonary parenchyma region has hole and asked
Topic, can carry out lung areas holes filling using morphology operations filling blood vessel method.Wherein, carried out on image morphologic
First expansion post-etching computing is referred to as closed operation, it is therefore an objective to the edge of the small hole of packing ratio structural element and smoothed image.Cause
This closing operation of mathematical morphology can fill Pulmonary Vascular.
Because tracheae makes pulmo be sticked together, want to separate pulmo and must reject first to make the gas of left and right adhesion of lung
Pipe, although the gray value of tracheae is smaller than pulmonary parenchyma, but between pulmonary parenchyma gray value and tracheae gray value do not have one it is bright
Aobvious boundary.But tracheae has oneself unique distribution characteristics, the centre that only one, the top of tracheae is distributed in two lungs,
Two are then divided into, each into left lung and right lung, from the top of tracheae to the portion being divided into before two tracheaes enter pulmo
Point be not with pulmonary parenchyma adhesion, as shown in figure 5, middle independent fritter connected region is exactly tracheae.Carried in Fig. 5 pulmonary parenchyma
On the basis of taking result, closing operation of mathematical morphology is performed by choosing suitable structural element on all chest CT image bearing layers, finally
Acquisition includes pulmonary vascular lung tissue region, as a result as shown in Figure 8.
By counting the number of each connected domain pixel, the less connected domain region of pixel number is rejected, retains pixel
Two larger connected domain regions of points.Reject after tracheae, for some data, pulmo has been two independent UNICOMs
Region, as shown in Figure 6.
Fig. 7 is given with numerical differentiation (DDA) near linear that pulmonary parenchyma adhesion region is drawn on individual layer CT images
Border.The near linear border on every layer of pulmonary parenchyma region adhesion image, and 26 neighbours around straight border are fallen in final etching
The mark value of domain point, it is possible to obtain the pulmo region of two adhesions, finally directly counts mark point in three-dimensional data
Connected domain, two disjunct connected domains can be obtained, and mark respectively.
The root of blood vessel and the root of tracheae are tightly adjacent, and all at the edge of lung, the diameter of pulmonary vascular root
Also the diameter than other parts blood vessel is big, therefore can corrode the blood vessel at the edge of lung tissue and disconnect pulmonary vascular root, real
The pulmonary vascular separation of the existing different lobes of the lung, obtains five mutual disjunct connected domains, and marked respectively with different colors, such as Figure 11
It is shown.
Because blood vessel has too many pixel, if directly training can influence the speed of training, therefore first have to reduce blood
The pixel number of pipe, but the distribution of blood vessel can not be influenceed.The extraction of center line can just meet the two conditions, because
Center path can remove the marginal portion of column blood vessel and not interfere with the tendency of blood vessel, can using eliminate thinning method come
Extract the center path of blood vessel.For smooth Pulmonary Vascular region, filling Pulmonary Vascular inside leak that may be present, first to lung blood
Pipe does a closing operation of mathematical morphology, i.e., first expand post-etching, then extracts center path, end product with cancellation thinning method again
As shown in Figure 10.
Pulmonary parenchyma extraction process of the invention based on 3D region growings, the left and right adhesion of lung tracheae based on provincial characteristics are rejected
Process, Pulmonary Vascular root reject process, the extracted process in Pulmonary Vascular Center Road based on topological thinning and based on SVMs
The processing procedures such as the lobe of the lung partitioning algorithm of classification accurately extract the lobe of the lung from chest CT image, accurately complete to each lung
The qualitative assessment of the severity extent of leaf, the diagnosis and treatment to PUD D is more accurate and effective, whether each single lung
Leaf, or entirely the average segmentation accuracy rate of five lobes of the lung of lung tissue is all more than 85%, meets expected segmentation and requires.
It is to be appreciated that the description carried out above to the specific embodiment of the present invention is simply to illustrate that the skill of the present invention
Art route and feature, its object is to allow those skilled in the art to understand present disclosure and implement according to this, but
The present invention is not limited to above-mentioned particular implementation.Every various change made within the scope of the claims is repaiied
Decorations, should all cover within the scope of the present invention.
Claims (10)
1. a kind of method that the lobe of the lung is extracted from chest CT image, it is characterised in that comprise the following steps:
S1, the n-layer chest CT image for receiving input, obtain intermediate image layer, and the specified pixel point for choosing lung areas is seed
Point, 3D region growths are carried out according to setting segmentation threshold and initial seed point, are obtained without pulmonary vascular pulmonary parenchyma region, its
In, n is natural number;
S2, in the pulmonary parenchyma region obtained by step S1 by corroding the left and right adhesion of lung elimination method on approximate adhesion border, reject
Adhesion obtains two independent pulmonary parenchymas;
S3, using first expansion post-etching operation handlebar blood vessel be filled into the pulmonary parenchyma obtained by step S2, pass through between maximum kind side
Poor method automatically obtains optimum segmentation threshold value T0;
S4, with the optimum segmentation threshold value T obtained by step S30As the threshold value of segmentation blood vessel, the point calculated more than threshold value is used as blood vessel
Seed point, carries out 3D region growths, obtains Pulmonary Vascular;
S5, the blood vessel from the different lobes of the lung of vascular root disconnection, make the blood vessel in each lobe of the lung turn into an independent connected domain;
S6, the Pulmonary Vascular according to obtained by step S5, Pulmonary Vascular center path is extracted using elimination approach;
S7, two category support vector machines models of training, obtain two category support vector machines graders, are found not by the grader
With the interface between lobe of the lung blood vessel, the classification of pulmonary parenchyma pixel is obtained as the interface of the lobe of the lung and calculating using this interface
Discriminant function differentiates to each pixel in pulmonary parenchyma, obtains lobe of the lung tissue.
2. the method according to claim 1 that the lobe of the lung is extracted from chest CT image, it is characterised in that:The step S1 bags
Include following steps:
S11, n (n is natural number) the layer chest CT image for receiving input, obtain intermediate image layer, are chosen on intermediate image layer
The specified pixel point of lung areas is used as initial seed point;
S12, according to the characteristics of pulmonary parenchyma segmentation threshold is set;
26 neighborhoods of the labeled initial seed point of S13, search, choose one of pixel;
Whether the selected pixel of S14, judgement has been labeled, if then return to step S13, otherwise performs step S15;
Whether the gray value of the selected pixel of S15, judgement is less than segmentation threshold, if so, then the pixel is marked simultaneously
Mark queue is added, otherwise the stop flag pixel, perform step S16;
S16, judge whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish, if then performing step S17, otherwise
Return to step S13;
Whether S17, judge mark queue are empty, are not such as sky, then from the queue one mark point of taking-up as initially being marked
Remember seed point, return to step S13, the pixel point set being otherwise labeled is exactly to be partitioned into without pulmonary vascular pulmonary parenchyma
Region.
3. the method according to claim 1 that the lobe of the lung is extracted from chest CT image, it is characterised in that:The step S2 bags
Include following steps:
S21, to the pulmonary parenchyma region decision of acquisition, whether it is two approximately equalised connected domains of size, no if then terminating
Then perform step S22;
S22, statistics intermediate image layer connected domain number, reject the less connected domain region of pixel number, retain pixel number
Larger connected domain region, and judge the number of connected domain and the size of each connected domain, if two connected domains and greatly
Small approximately equal then terminates, and otherwise performs step S23;
S23, successively statistics connected domain number, if only one of which connected domain, pulmonary parenchyma one is scheduled on this layer of adhesion, performs step
S24, otherwise records the number of each connected domain pixel, two maximum connected domains of pixel number is found, if area is most
Quite, then this layer of CT image or so pulmonary parenchyma adhesion performs step S26 to two big connected region pixel number numbers, no
Then adhesion and larger connected domain are pulmonary parenchyma, perform step S24;
S24, with line segment adhesion region is approximately replaced, be converted into the position for determining straight line, peak of the lower boundary in y directions be
The lower extreme point of near linear, coboundary a little in from the nearest point of lower extreme point be then near linear upper extreme point, search out
Straight line is determined after this 2 points, step S25 is performed;
S25, the near linear border eroded on every layer of pulmonary parenchyma region adhesion image, and 26 neighborhoods around straight border
The mark value of point, obtains the pulmo region of two adhesions, performs step S26;
S26, judge whether all layers are scanned, if then terminating, otherwise perform step S23.
4. the method according to claim 1 that the lobe of the lung is extracted from chest CT image, it is characterised in that:The step S3 bags
Include following steps:
S31, by choose suitable structural element all chest CT images layer on perform first expansion post-etching closed operation grasp
Make, Pulmonary Vascular is filled into segmented good pulmonary parenchyma;
S32, one layer close to centre is selected in CT images, optimum segmentation threshold value is automatically obtained using maximum variance between clusters
T0。
5. the method according to claim 1 that the lobe of the lung is extracted from chest CT image, it is characterised in that:The step S4 bags
Include following steps:
Grey scale pixel value is more than segmentation threshold T in S41, the obtaining step S3 pulmonary parenchyma with blood vessel0All pixels point, and mark
It is designated as blood vessel.
S42, in the blood vessel obtained by step S41 choose a pixel be used as initial seed point;
26 neighborhoods of the labeled initial seed point of S43, search, choose one of pixel;
Whether the selected pixel of S44, judgement has been labeled, if then return to step S43, otherwise performs step S45;
Whether the gray value of the selected pixel of S45, judgement is more than segmentation threshold, if so, then the pixel is marked simultaneously
Mark queue is added, otherwise the stop flag pixel, perform step S46;
S46, judge whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish, if then performing step S47, otherwise
Return to step S43;
Whether S47, judge mark queue are empty, are not such as sky, then from the queue one mark point of taking-up as initially being marked
Remember seed point, return to step S43, the pixel point set being otherwise labeled is exactly the Pulmonary Vascular being partitioned into.
6. the method according to claim 1 that the lobe of the lung is extracted from chest CT image, it is characterised in that:The step S5 bags
Include following steps:
S51, the Pulmonary Vascular pixel for finding lung edge and close tracheae, step S52 is performed if finding, lung is otherwise continued to scan on
Organize in portion;
S52, Pulmonary Vascular disconnected using the blood vessel at the edge of corrosion pulmonary parenchyma, and counts the number of the connected domain of vascularization,
If the number of connected domain is five, step S53 is performed, step S51 is otherwise performed;
S53, the color different to each connected component labeling, all mark finish beam.
7. the method according to claim 6 that the lobe of the lung is extracted from chest CT image, it is characterised in that:The step S6 bags
Include following steps:
S61, scanning Pulmonary Vascular pixel, whether be boundary point, if the neighborhood of pixel 6 or 26 neighborhoods have background area if judging it
Pixel is then boundary point, performs step S62, otherwise marks scanned, skips and continue executing with step S61;
S62, the Euler's characteristic for judging boundary point, step S63 is performed if Euler's characteristic is constant, otherwise skips and continues executing with step
S61;
S63, judge whether boundary point is simple point, if the point on former three dimensions topological structure without influence if be simple point, together
When by the point deletion, perform step S64;Otherwise step S61 is performed;
S64, judge whether blood vessel pixel also has other simple points, if nothing, centerline extraction terminates, otherwise perform step S61.
8. the method according to claim 1 that the lobe of the lung is extracted from chest CT image, it is characterised in that:The step S7 bags
Include following steps:
S71, according to the interface between the lobe of the lung closest to the track of RBF, selection Radial basis kernel function is as empty to higher-dimension
Between the function that maps, it is determined that after kernel function, perform step S72;
S72, for left lung, it is assumed that the pixel of left lung be N1, blood vessel pixel number be n1, all blood of the upper leaf of left lung
Pipe pixel, while all blood vessel pixels on inferior lobe are obtained a discriminant function labeled as negative, performs step labeled as just
S73;
S73, gone with discriminant function to judge the symbol of each pixel in left pulmonary parenchyma, if symbol is just, the point belongs to
Upper leaf, otherwise the point belongs to inferior lobe, and the number of statistical pixel, if then left lung segmentation is completed N1, execution step S74, otherwise
Continue executing with step S73;
S74, for right lung, it is assumed that the pixel of right lung be N2, blood vessel pixel number be n2, for use two class Support Vectors
Machine is classified, and the upper leaf of right lung and middle period, which are combined into an entirety, to be regarded as " upper leaf ", execution step S75;
S75, using with left lung segmentation as method obtain a discriminant classification function, inferior lobe is extracted first, step S74
All blood vessel pixels of " the upper leaf " regarded as are labeled as just, while all blood vessel pixels on inferior lobe are obtained labeled as negative
To a discriminant function, step S76 is performed;
S76, gone with discriminant function to judge the symbol of each substantial pixel of right lung, if symbol is just, the point belongs to
" upper leaf " that step S74 is regarded as, otherwise the point belongs to inferior lobe, and the number of statistical pixel, if then inferior lobe segmentation is completed N2,
Step S77 is performed, step S76 is otherwise continued executing with;
S77, segmentation are completed after inferior lobe, are then obtained one with leaf and the blood vessel Training Support Vector Machines model in middle period again and are sentenced
Other surface function, the partial segmentation of inferior lobe is removed to right lung, extracts leaf and middle period, performs step S78;
S78, for the upper leaf of right lung and middle period, it is assumed that the pixel of right lung be N3, blood vessel pixel number be n3, right lung
All blood vessel pixels of upper leaf, while middle period upper all blood vessel pixels of right lung are labeled as bearing, obtain one labeled as just
Individual discriminant function, performs step S79;
S79, gone with discriminant function to judge the symbol of each substantial pixel of right lung, if symbol is just, the point belongs to
Upper leaf, otherwise the point belongs to middle period, and the number of statistical pixel, so far left if then superior lobe of right lung and middle period segmentation are completed N3
The right lung segmentation lobe of the lung terminates.
9. a kind of device that the lobe of the lung is extracted from chest CT image, it is characterised in that:Including:
CT image input units, the n-layer chest CT image for receiving input, wherein n is natural number;
Without pulmonary vascular pulmonary parenchyma acquiring unit, its specified pixel point for choosing lung areas is seed point, according to setting point
Cut threshold value and initial seed point carries out 3D region growths, obtain without pulmonary vascular pulmonary parenchyma region;
Two independent pulmonary parenchyma acquiring units, its obtain left and right pulmonary parenchyma after, judge left and right pulmonary parenchyma whether adhesion, if glue
Even, then two independent pulmonary parenchymas are obtained by rejecting the adhesion border of tracheae and near linear;
Pulmonary Vascular segmentation threshold acquiring unit, it is filled into pulmonary parenchyma using the operation handlebar blood vessel of first expansion post-etching, passed through
Maximum variance between clusters automatically obtain optimum segmentation threshold value T0;
Pulmonary Vascular extraction unit, it is with the optimum segmentation threshold value T of gained0As the threshold value of segmentation blood vessel, the point more than threshold value is calculated
As vessel seed point, 3D region growths are carried out, Pulmonary Vascular is obtained;
Pulmonary Vascular is divided into independent communication domain unit, and it disconnects the blood vessel of the different lobes of the lung from vascular root, makes each lobe of the lung
Blood vessel turns into an independent connected domain;
Pulmonary Vascular centerline extraction unit, its Pulmonary Vascular according to obtained by the Pulmonary Vascular is divided into independent communication domain unit,
Pulmonary Vascular center path is extracted using elimination approach;
Lobe of the lung cutting unit, the sorter model of blood vessel Training Support Vector Machines two of the adjacent lobe of the lung, left lung are had two lungs by it
Leaf, trains one, right lung has three lobes of the lung, and twice, three graders of SVMs two are obtained in pulmo for training, pass through branch
Hold vector machine and find interface between different lobe of the lung blood vessels, be partitioned into the lobe of the lung.
10. the device according to claim 9 that the lobe of the lung is extracted from chest CT image, it is characterised in that:
It is described to include without pulmonary vascular pulmonary parenchyma acquiring unit:
Initial seed point chooses unit, and its specified pixel point that lung areas is chosen on middle tomographic image is used as initial seed
Point;
Segmentation threshold obtains subelement, and it sets segmentation threshold according to the characteristics of pulmonary parenchyma;
Pixel chooses subelement, and it searches for 26 neighborhoods of labeled initial seed point, chooses one of pixel;
Pixel marker for judgment subelement, it judges whether selected pixel has been labeled, if so, notifying pixel
Subelement is chosen, pixel value judgment sub-unit is otherwise notified;
Pixel value judgment sub-unit, it judges whether the gray value of selected pixel is less than segmentation threshold, if so, then this
Pixel is marked and adds mark queue, otherwise the stop flag pixel, notifies the neighborhood territory pixel point search of seed point 26
Unit;
The neighborhood search subelement of seed point 26, it judges whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish,
If so, then notification indicia queue judgment sub-unit, otherwise notifies the pixel to choose subelement;
Whether mark queue judgment sub-unit, its judge mark queue is empty, if being not sky, then a mark is taken out from the queue
Note point notifies pixel to choose subelement, the pixel point set being otherwise labeled is exactly as initially labeled seed point
Be partitioned into without pulmonary vascular pulmonary parenchyma;
It is preferred that, described two independent pulmonary parenchyma acquiring units include:
Pulmonary parenchyma connected domain number judgment sub-unit, its pulmonary parenchyma to acquisition judges whether it is that two sizes are approximately equalised
Connected domain, if then terminating, otherwise notifies to reject tracheae subelement;
Tracheae subelement is rejected, it counts the connected domain number of intermediate image layer, rejects the less connected domain region of pixel number,
Retain the larger connected domain region of pixel number, and judge the number of connected domain and the size of each connected domain, if two
Connected domain and size approximately equal then terminates, otherwise notifies layer connected domain number judgment sub-unit;
Layer connected domain number judgment sub-unit, it successively counts connected domain number, if only one of which connected domain, pulmonary parenchyma one
This layer of adhesion is scheduled on, straight line determination subelement is notified, otherwise records the number of each connected domain pixel, find pixel number
Two maximum connected domains, if two maximum connected region pixel number numbers of area are quite, this layer of CT image or so
Pulmonary parenchyma adhesion, notifies pixel layer scanning judgment sub-unit, and otherwise adhesion and larger connected domain are pulmonary parenchyma, notify straight line
Determination subelement;
Straight line determination subelement, it approximately replaces adhesion region with line segment, is converted into the position for determining straight line, lower boundary is in y side
To peak be near linear lower extreme point, coboundary a little in from the nearest point of lower extreme point be then near linear
Upper extreme point, searches out and straight line is determined after this 2 points, notifies corrosion near linear border subelement;
Corrode near linear border subelement, it erodes the near linear border on every layer of pulmonary parenchyma region adhesion image, with
And the mark value of 26 neighborhood points around straight border, the pulmo region of two adhesions is obtained, notifies pixel layer scanning to sentence
Disconnected subelement;
Pixel layer scans judgment sub-unit, and it judges whether all layers are scanned, if then terminating, and otherwise notifies layer connected domain
Number judgment sub-unit;
It is preferred that, the Pulmonary Vascular segmentation threshold acquiring unit includes:
Closed operation operating unit, it performs corruption after first expansion by choosing suitable structural element on all chest CT images layer
The closed operation operation of erosion, Pulmonary Vascular is filled into segmented good pulmonary parenchyma;
Optimum segmentation threshold value acquiring unit, it selects one layer close to centre in CT images, utilizes maximum variance between clusters
Automatically obtain optimum segmentation threshold value T0;
It is preferred that, the Pulmonary Vascular extraction unit includes:
Intermediate layer Pulmonary Vascular extracts subelement, and it is obtained in intermediate layer, and grey scale pixel value is more than segmentation threshold in the pulmonary parenchyma with blood vessel
Value T0All pixels point, and labeled as blood vessel;
Initial seed point obtains subelement, and it chooses a picture in the blood vessel that the intermediate layer Pulmonary Vascular extracts obtained by subelement
Vegetarian refreshments is used as initial seed point;
Pixel obtains subelement, and it searches for 26 neighborhoods of labeled initial seed point, chooses one of pixel;
Pixel marker for judgment subelement, it judges whether selected pixel has been labeled, if notifying pixel
Subelement is obtained, pixel judgment sub-unit is otherwise notified;
Pixel value judgment sub-unit, it judges whether the gray value of selected pixel is more than segmentation threshold, if so, then this
Pixel is marked and adds mark queue, notifies the neighborhood search subelement of seed point 26, otherwise the stop flag pixel;
The neighborhood search subelement of seed point 26, it judges whether 26 neighborhood territory pixel points of seed point are all searched for and judge to finish,
If then notification indicia queue judgment sub-unit, otherwise notifies pixel to obtain subelement;
Whether mark queue judgment sub-unit, its judge mark queue is empty, if being not sky, then a mark is taken out from the queue
Note point notifies pixel to obtain subelement, the pixel point set being otherwise labeled is exactly as initially labeled seed point
The Pulmonary Vascular being partitioned into;
It is preferred that, the Pulmonary Vascular, which is divided into independent communication domain unit, to be included:
Preliminary sweep pulmonary parenchyma unit, it finds lung edge and close to the Pulmonary Vascular pixel of tracheae, if finding, and notifies to disconnect
Pulmonary Vascular root subelement, otherwise notifies preliminary sweep pulmonary parenchyma unit;
Pulmonary Vascular root subelement is disconnected, it uses the blood vessel at the edge of corrosion pulmonary parenchyma to disconnect Pulmonary Vascular, and counts blood vessel
The number of the connected domain of formation, if the number of connected domain is five, notifies Pulmonary Vascular mark subelement, otherwise notifies initially to sweep
Retouch pulmonary parenchyma unit;
Pulmonary Vascular marks subelement, the different color of its vascular marker to each connected domain, and all mark finishes beam;
It is preferred that, the Pulmonary Vascular centerline extraction unit includes:
Judge boundary point subelement, its scan Pulmonary Vascular pixel, whether judge it is boundary point, if the neighborhood of pixel 6 or
It is then boundary point that 26 neighborhoods, which have background area pixel, notifies the first judgment sub-unit, otherwise marks scanned, skips continuation logical
Know and judge boundary point subelement;
First judgment sub-unit, it judges Euler's characteristic of boundary point, and the second judgment sub-unit is notified if Euler's characteristic is constant,
Otherwise continuation is skipped to notify to judge boundary point subelement;
Second judgment sub-unit, it judges whether boundary point is simple point, if the point on former three dimensions topological structure without influence
It is then simple point, while by the point deletion, notifying the 3rd judgment sub-unit, otherwise notifies to judge boundary point subelement;
3rd judgment sub-unit, it judges whether blood vessel pixel also has other simple points, no if nothing, centerline extraction terminates
Then notify to judge boundary point subelement;
It is preferred that, the lobe of the lung cutting unit includes:
SVMs Selection of kernel function unit, it, closest to the track of RBF, is selected according to the interface between the lobe of the lung
Radial basis kernel function is used as the function mapped to higher dimensional space, it is determined that after kernel function, notifies left lung discriminant function is asked for sub single
Member;
Left lung discriminant function asks for subelement, and it is for left lung, it is assumed that the pixel of left lung is N1, and blood vessel pixel number is
N1, all blood vessel pixels of the upper leaf of left lung is labeled as just, while all blood vessel pixels on inferior lobe are obtained labeled as negative
To a discriminant function, left lung upper and lower lobes segmentation subelement is notified;
Left lung upper and lower lobes split subelement, and it is gone to judge the symbol of each pixel in left pulmonary parenchyma with discriminant function, if
Symbol is just, then the point belongs to leaf, and otherwise the point belongs to inferior lobe, and the number of statistical pixel, if then left lung is segmented N1
Into the notice superior lobe of right lung middle period merges subelement;
The superior lobe of right lung middle period merges subelement, and it is for right lung, it is assumed that the pixel of right lung is N2, and blood vessel pixel number is
The upper leaf of right lung and middle period, to be classified using two category support vector machines, are combined into an entirety and regarded as " upper leaf ", notice by n2
Right lung discriminant function asks for subelement;
Right lung discriminant function asks for subelement, and it obtains a discriminant classification function using the method as the segmentation of left lung, first
Inferior lobe is first extracted, all blood vessel pixels of " the upper leaf " regarded as are labeled as just, while all blood vessel pictures on inferior lobe
Vegetarian refreshments obtains a discriminant function labeled as negative, notifies inferior lobe of right lung segmentation subelement;
Inferior lobe of right lung splits subelement, and it removes the symbol for judging each substantial pixel of right lung with discriminant function, if symbol
Number for just, then the point belongs to " the upper leaf " regarded as, and otherwise the point belongs to inferior lobe, and the number of statistical pixel, if N2 then under
Leaf segmentation is completed, and notifies superior lobe of right lung middle period discriminant function to ask for subelement;
Superior lobe of right lung middle period discriminant function asks for subelement, and it, which is split, completes after inferior lobe, then again with leaf and the blood vessel in middle period
Training Support Vector Machines model, obtains a differentiation surface function, the partial segmentation of inferior lobe is removed to right lung, Ye Hezhong is extracted
Leaf, notifies superior lobe of right lung middle period segmentation subelement;
The superior lobe of right lung middle period splits subelement, and it is for the upper leaf of right lung and middle period, it is assumed that the pixel of right lung is N3, blood vessel picture
Vegetarian refreshments number is n3, and all blood vessel pixels of the upper leaf of right lung are labeled as just, while middle period upper all blood vessels of right lung
Pixel obtains a discriminant function labeled as negative, and is gone with discriminant function to judge each pixel of right lung substantially
Symbol, it is if symbol is just, the point belongs to leaf, and otherwise the point belongs to middle period, and the number of statistical pixel, if N3, then right
Leaf and middle period segmentation are completed on lung.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710374181.8A CN107230204B (en) | 2017-05-24 | 2017-05-24 | A kind of method and device for extracting the lobe of the lung from chest CT image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710374181.8A CN107230204B (en) | 2017-05-24 | 2017-05-24 | A kind of method and device for extracting the lobe of the lung from chest CT image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107230204A true CN107230204A (en) | 2017-10-03 |
CN107230204B CN107230204B (en) | 2019-11-22 |
Family
ID=59934015
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710374181.8A Active CN107230204B (en) | 2017-05-24 | 2017-05-24 | A kind of method and device for extracting the lobe of the lung from chest CT image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107230204B (en) |
Cited By (25)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107909581A (en) * | 2017-11-03 | 2018-04-13 | 杭州依图医疗技术有限公司 | Lobe of the lung section dividing method, device, system, storage medium and the equipment of CT images |
CN108171703A (en) * | 2018-01-18 | 2018-06-15 | 东北大学 | A kind of method that tracheae tree is automatically extracted from chest CT image |
CN108961273A (en) * | 2018-07-03 | 2018-12-07 | 东北大学 | A kind of method and system for dividing pulmonary artery and pulmonary vein from CT images |
CN109191436A (en) * | 2018-08-15 | 2019-01-11 | 复旦大学 | The low-dose CT Lung neoplasm detection algorithm of view-based access control model conspicuousness spectrum residual error method |
CN109360219A (en) * | 2018-10-23 | 2019-02-19 | 东北大学 | A kind of augmented reality auxiliary operation method and system |
CN109410166A (en) * | 2018-08-30 | 2019-03-01 | 中国科学院苏州生物医学工程技术研究所 | Full-automatic partition method for pulmonary parenchyma CT image |
CN110246145A (en) * | 2019-06-21 | 2019-09-17 | 福州大学 | A kind of dividing method of abdominal CT images |
CN110246144A (en) * | 2019-06-17 | 2019-09-17 | 吉林大学第一医院 | A kind of the blood vessel Enhancement Method and system of lung CT image |
CN110473207A (en) * | 2019-07-30 | 2019-11-19 | 赛诺威盛科技(北京)有限公司 | A kind of method of the Interactive Segmentation lobe of the lung |
CN110942453A (en) * | 2019-11-21 | 2020-03-31 | 山东众阳健康科技集团有限公司 | CT image lung lobe identification method based on neural network |
CN110956604A (en) * | 2018-09-27 | 2020-04-03 | 上海蔚来汽车有限公司 | Method and system for counting effective bonding area of bonding glue |
CN111179298A (en) * | 2019-12-12 | 2020-05-19 | 深圳市旭东数字医学影像技术有限公司 | CT image-based three-dimensional lung automatic segmentation and left-right lung separation method and system |
CN111275722A (en) * | 2020-02-18 | 2020-06-12 | 广州柏视医疗科技有限公司 | Lung segment and liver segment segmentation method and system |
CN111563902A (en) * | 2020-04-23 | 2020-08-21 | 华南理工大学 | Lung lobe segmentation method and system based on three-dimensional convolutional neural network |
TWI713054B (en) * | 2019-04-18 | 2020-12-11 | 大陸商北京市商湯科技開發有限公司 | Image segmentation method and device, electronic equipment and storage medium |
CN112150477A (en) * | 2019-11-15 | 2020-12-29 | 复旦大学 | Full-automatic segmentation method and device for cerebral image artery |
CN112884792A (en) * | 2021-02-02 | 2021-06-01 | 青岛海信医疗设备股份有限公司 | Lung image segmentation method and device, electronic equipment and storage medium |
CN112950554A (en) * | 2021-02-05 | 2021-06-11 | 慧影医疗科技(北京)有限公司 | Lung lobe segmentation optimization method and system based on lung segmentation |
CN113012114A (en) * | 2021-03-02 | 2021-06-22 | 推想医疗科技股份有限公司 | Blood vessel identification method and device, storage medium and electronic equipment |
CN113129317A (en) * | 2021-04-23 | 2021-07-16 | 广东省人民医院 | Lung lobe automatic segmentation method based on watershed analysis technology |
CN113160186A (en) * | 2021-04-27 | 2021-07-23 | 青岛海信医疗设备股份有限公司 | Lung lobe segmentation method and related device |
CN113223013A (en) * | 2021-05-08 | 2021-08-06 | 推想医疗科技股份有限公司 | Method, device, equipment and storage medium for pulmonary vessel segmentation positioning |
CN115049807A (en) * | 2022-06-28 | 2022-09-13 | 上海市胸科医院 | Method and device for establishing pulmonary blood vessel model and server |
CN116228767A (en) * | 2023-05-09 | 2023-06-06 | 北京易优联科技有限公司 | X-ray lung tumor image processing method based on computer vision |
CN116402816A (en) * | 2023-06-08 | 2023-07-07 | 中国人民解放军海军青岛特勤疗养中心 | Physical examination CT image data management method and system |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1447772A1 (en) * | 2003-02-11 | 2004-08-18 | MeVis GmbH | A method of lung lobe segmentation and computer system |
US20070133894A1 (en) * | 2005-12-07 | 2007-06-14 | Siemens Corporate Research, Inc. | Fissure Detection Methods For Lung Lobe Segmentation |
JP2008142481A (en) * | 2006-12-13 | 2008-06-26 | Med Solution Kk | Apparatus and program for carrying out segmentation of lungs to units of pulmonary segment automatically |
CN102208105A (en) * | 2010-03-31 | 2011-10-05 | 富士胶片株式会社 | Medical image processing technology |
CN103310457A (en) * | 2013-06-18 | 2013-09-18 | 哈尔滨工程大学 | Lung parenchyma segmentation method based on convex hulls correcting by parabolic |
CN103310449A (en) * | 2013-06-13 | 2013-09-18 | 沈阳航空航天大学 | Lung segmentation method based on improved shape model |
-
2017
- 2017-05-24 CN CN201710374181.8A patent/CN107230204B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1447772A1 (en) * | 2003-02-11 | 2004-08-18 | MeVis GmbH | A method of lung lobe segmentation and computer system |
US20070133894A1 (en) * | 2005-12-07 | 2007-06-14 | Siemens Corporate Research, Inc. | Fissure Detection Methods For Lung Lobe Segmentation |
JP2008142481A (en) * | 2006-12-13 | 2008-06-26 | Med Solution Kk | Apparatus and program for carrying out segmentation of lungs to units of pulmonary segment automatically |
CN102208105A (en) * | 2010-03-31 | 2011-10-05 | 富士胶片株式会社 | Medical image processing technology |
CN103310449A (en) * | 2013-06-13 | 2013-09-18 | 沈阳航空航天大学 | Lung segmentation method based on improved shape model |
CN103310457A (en) * | 2013-06-18 | 2013-09-18 | 哈尔滨工程大学 | Lung parenchyma segmentation method based on convex hulls correcting by parabolic |
Non-Patent Citations (5)
Title |
---|
ADRIEN DEPEURSINGE 等: "3D lung image retrieval usinglocalized features", 《MEDICAL IMAGING 2011》 * |
GANESH SABLE 等: "Lung Segmentation and Tumor Identification from CT Scan Images Using SVM", 《INTERNATIONAL JOURNAL OF SCIENCE AND RESEARCH》 * |
NISAR AHMED MEMON 等: "Segmentation of Lungs from CT Scan Images for Early Diagnosis of Lung Cancer", 《PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY》 * |
冯宗雪,董晓: "基于漫水填充算法的肺实质分割方法", 《电脑知识与技术》 * |
谭启路 等: "基于区域生长法的三维心血管模型的分割", 《北京生物医学工程》 * |
Cited By (38)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107909581A (en) * | 2017-11-03 | 2018-04-13 | 杭州依图医疗技术有限公司 | Lobe of the lung section dividing method, device, system, storage medium and the equipment of CT images |
CN107909581B (en) * | 2017-11-03 | 2019-01-29 | 杭州依图医疗技术有限公司 | Lobe of the lung section dividing method, device, system, storage medium and the equipment of CT images |
CN109636811B (en) * | 2017-11-03 | 2020-06-12 | 杭州依图医疗技术有限公司 | Integration method and device for lung lobe segment segmentation of CT (computed tomography) image |
CN109636811A (en) * | 2017-11-03 | 2019-04-16 | 杭州依图医疗技术有限公司 | Integration method, the device of the lobe of the lung section segmentation of CT images |
CN108171703A (en) * | 2018-01-18 | 2018-06-15 | 东北大学 | A kind of method that tracheae tree is automatically extracted from chest CT image |
CN108171703B (en) * | 2018-01-18 | 2020-09-15 | 东北大学 | Method for automatically extracting trachea tree from chest CT image |
CN108961273A (en) * | 2018-07-03 | 2018-12-07 | 东北大学 | A kind of method and system for dividing pulmonary artery and pulmonary vein from CT images |
CN108961273B (en) * | 2018-07-03 | 2021-10-29 | 东北大学 | Method and system for segmenting pulmonary artery and pulmonary vein from CT image |
CN109191436A (en) * | 2018-08-15 | 2019-01-11 | 复旦大学 | The low-dose CT Lung neoplasm detection algorithm of view-based access control model conspicuousness spectrum residual error method |
CN109410166B (en) * | 2018-08-30 | 2022-09-09 | 中国科学院苏州生物医学工程技术研究所 | Full-automatic segmentation method for lung parenchyma CT image |
CN109410166A (en) * | 2018-08-30 | 2019-03-01 | 中国科学院苏州生物医学工程技术研究所 | Full-automatic partition method for pulmonary parenchyma CT image |
CN110956604B (en) * | 2018-09-27 | 2022-12-16 | 上海蔚来汽车有限公司 | Method and system for counting effective bonding area of bonding glue |
CN110956604A (en) * | 2018-09-27 | 2020-04-03 | 上海蔚来汽车有限公司 | Method and system for counting effective bonding area of bonding glue |
CN109360219A (en) * | 2018-10-23 | 2019-02-19 | 东北大学 | A kind of augmented reality auxiliary operation method and system |
TWI713054B (en) * | 2019-04-18 | 2020-12-11 | 大陸商北京市商湯科技開發有限公司 | Image segmentation method and device, electronic equipment and storage medium |
CN110246144A (en) * | 2019-06-17 | 2019-09-17 | 吉林大学第一医院 | A kind of the blood vessel Enhancement Method and system of lung CT image |
CN110246145B (en) * | 2019-06-21 | 2023-02-21 | 福州大学 | Segmentation method of abdominal CT image |
CN110246145A (en) * | 2019-06-21 | 2019-09-17 | 福州大学 | A kind of dividing method of abdominal CT images |
CN110473207A (en) * | 2019-07-30 | 2019-11-19 | 赛诺威盛科技(北京)有限公司 | A kind of method of the Interactive Segmentation lobe of the lung |
CN112150477A (en) * | 2019-11-15 | 2020-12-29 | 复旦大学 | Full-automatic segmentation method and device for cerebral image artery |
CN110942453A (en) * | 2019-11-21 | 2020-03-31 | 山东众阳健康科技集团有限公司 | CT image lung lobe identification method based on neural network |
CN111179298B (en) * | 2019-12-12 | 2023-05-02 | 深圳市旭东数字医学影像技术有限公司 | Three-dimensional lung automatic segmentation and left and right lung separation method and system based on CT image |
CN111179298A (en) * | 2019-12-12 | 2020-05-19 | 深圳市旭东数字医学影像技术有限公司 | CT image-based three-dimensional lung automatic segmentation and left-right lung separation method and system |
CN111275722A (en) * | 2020-02-18 | 2020-06-12 | 广州柏视医疗科技有限公司 | Lung segment and liver segment segmentation method and system |
CN111563902A (en) * | 2020-04-23 | 2020-08-21 | 华南理工大学 | Lung lobe segmentation method and system based on three-dimensional convolutional neural network |
CN111563902B (en) * | 2020-04-23 | 2022-05-24 | 华南理工大学 | Lung lobe segmentation method and system based on three-dimensional convolutional neural network |
CN112884792A (en) * | 2021-02-02 | 2021-06-01 | 青岛海信医疗设备股份有限公司 | Lung image segmentation method and device, electronic equipment and storage medium |
CN112950554A (en) * | 2021-02-05 | 2021-06-11 | 慧影医疗科技(北京)有限公司 | Lung lobe segmentation optimization method and system based on lung segmentation |
CN113012114B (en) * | 2021-03-02 | 2021-12-03 | 推想医疗科技股份有限公司 | Blood vessel identification method and device, storage medium and electronic equipment |
CN113012114A (en) * | 2021-03-02 | 2021-06-22 | 推想医疗科技股份有限公司 | Blood vessel identification method and device, storage medium and electronic equipment |
CN113129317A (en) * | 2021-04-23 | 2021-07-16 | 广东省人民医院 | Lung lobe automatic segmentation method based on watershed analysis technology |
CN113160186A (en) * | 2021-04-27 | 2021-07-23 | 青岛海信医疗设备股份有限公司 | Lung lobe segmentation method and related device |
CN113223013A (en) * | 2021-05-08 | 2021-08-06 | 推想医疗科技股份有限公司 | Method, device, equipment and storage medium for pulmonary vessel segmentation positioning |
CN115049807A (en) * | 2022-06-28 | 2022-09-13 | 上海市胸科医院 | Method and device for establishing pulmonary blood vessel model and server |
CN116228767A (en) * | 2023-05-09 | 2023-06-06 | 北京易优联科技有限公司 | X-ray lung tumor image processing method based on computer vision |
CN116228767B (en) * | 2023-05-09 | 2023-06-27 | 北京易优联科技有限公司 | X-ray lung tumor image processing method based on computer vision |
CN116402816A (en) * | 2023-06-08 | 2023-07-07 | 中国人民解放军海军青岛特勤疗养中心 | Physical examination CT image data management method and system |
CN116402816B (en) * | 2023-06-08 | 2023-08-15 | 中国人民解放军海军青岛特勤疗养中心 | Physical examination CT image data management method and system |
Also Published As
Publication number | Publication date |
---|---|
CN107230204B (en) | 2019-11-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107230204B (en) | A kind of method and device for extracting the lobe of the lung from chest CT image | |
CN109102502B (en) | Pulmonary nodule detection method based on three-dimensional convolutional neural network | |
CN110276745B (en) | Pathological image detection algorithm based on generation countermeasure network | |
CN110930416B (en) | MRI image prostate segmentation method based on U-shaped network | |
Chang et al. | Thyroid segmentation and volume estimation in ultrasound images | |
Suzuki et al. | Extraction of left ventricular contours from left ventriculograms by means of a neural edge detector | |
CN108257135A (en) | The assistant diagnosis system of medical image features is understood based on deep learning method | |
CN109727235B (en) | Organ automatic delineation algorithm based on deep learning | |
Maitra et al. | Automated digital mammogram segmentation for detection of abnormal masses using binary homogeneity enhancement algorithm | |
Maitra et al. | Detection of abnormal masses using divide and conquer algorithmin digital mammogram | |
CN106295639A (en) | A kind of virtual reality terminal and the extracting method of target image and device | |
CN109636810B (en) | Pulmonary nodule segmentation method and system of CT image | |
Maitra et al. | Accurate breast contour detection algorithms in digital mammogram | |
CN108765409A (en) | A kind of screening technique of the candidate nodule based on CT images | |
CN114066886A (en) | Bone segmentation boundary determining method and device, electronic equipment and storage medium | |
CN110232685A (en) | Space pelvis parameter auto-testing method based on deep learning | |
CN113096184A (en) | Diatom positioning and identifying method under complex background | |
CN115471701A (en) | Lung adenocarcinoma histology subtype classification method based on deep learning and transfer learning | |
CN111383759A (en) | Automatic pneumonia diagnosis system | |
CN110517766A (en) | Identify the method and device of encephalatrophy | |
Mahbub et al. | A modified CNN and fuzzy AHP based breast cancer stage detection system | |
Yang et al. | Efficient catheter segmentation in 3D cardiac ultrasound using slice-based FCN with deep supervision and f-score loss | |
JP2023517058A (en) | Automatic detection of tumors based on image processing | |
CN108765384B (en) | Significance detection method for joint manifold sequencing and improved convex hull | |
CN105956587A (en) | Method for automatically extracting meniscus from knee-joint magnetic resonance image sequence based on shape constraint |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |